A total of 515 non-duplicated clinical E. coli isolates were collected from two tertiary hospitals in Zhejiang Province, China. Among these isolates, 105, 173, 40, and 197 strains were isolated in 2015, 2016, 2017, and 2019, respectively. Based on the analytical profile index, all strains were identified as E. coli using the VITEK 2.0 system (bioMérieux, Durham NC, USA). The six mcr-1-positive E. coli strains were subsequently subjected to species identification verifying by 16S rDNA sequencing with polymerase chain reaction (PCR) primers (Forward: 5′-AGAGTTTGATCCTGGCTCAG-3′; Reverse: 5′-GGTTACCTTGTTACGACTT-3′), followed by genome sequencing.

The minimum inhibitory concentrations (MICs) of clinical routine antimicrobial agents, including piperacillin (PRL), piperacillin-tazobactam (TZP), ceftazidime (CAZ), ceftriaxone (CTRX), cefoxitin (CFX), cefepime (FEP), imipenem (IMP), meropenem (MEM), aztreonam (AZT), levofloxacin (LEV), ciprofloxacin (CIP), amikacin (AMK), gentamicin (GEN), chloramphenicol (CAP) and rifampicin (RIF) were determined by the agar dilution method. The MICs of colistin were determined by the broth microdilution in cation-adjusted Mueller–Hinton broth (CAMHB) (Gajdács et al., 2020). We interpreted the results in accordance with the Clinical and Laboratory Standards Institute (CLSI) guidelines (CLSI, 2020). E. coli ATCC 25922 was used as the standard reference strain.

The transferability of mcr-1 was tested by conjugation experiment with mcr-1-positive E. coli as donors and rifampicin-resistant E. coli C600 as a recipient. The MacConkey agar plates containing rifampicin (512 µg/mL) and colistin (1 µg/mL) were used to select mcr-1-positive transconjugants. PCR analysis of mcr-1 and antimicrobial susceptibility testing were carried out to confirm the transconjugants.

All E. coli strains were screened for the presence of the mcr-1 gene by PCR using primers (Forward: 5′-CGGTCAGTCCGTTTGTTC-3′, Reverse: 5′-CTTGGTCGGTCTGTA GGG-3′) (Liu et al., 2016). The PCR products were purified and sequenced by Sanger sequencing (Generay, Shanghai, China).

Genomic DNA of six mcr-1-harboring E. coli strains was extracted using Bacterial Genomic DNA Miniprep kit (Generay, Shanghai, China), and subsequently sequenced by Illumina HiSeq 2500 (Illumina, Inc., San Diego, CA, United States). The short reads were assembled using SPAdes software (version3.14.0). EC195 showed the widest resistance spectrum and highest MIC levels against all the antimicrobials tested in this study, while the other five strains exhibited similar MIC levels, so we chose EC195 and one of the other five EC7518 as the representative isolates for whole genome sequencing. These two isolates were further sequenced by Pacific Bioscience (PacBio) RSII systems at the Shanghai Personal Biotechnology Co., Ltd. (Shanghai, China). Then, hybrid assembly was performed using Unicycler v0.4.8, with both short and long reads. The assembled sequences were annotated using Prokka (Seemann, 2014) and then corrected by BLAST (Boratyn et al., 2013) searches against the UniProtKB/Swiss-Prot, RefSeq, ISfinder (Siguier et al., 2006), and CARD (Jia et al., 2017) databases. The multi-locus sequence typing (MLST) and plasmid replicon type (Inc groups) identification were conducted using MLST (https://pubmlst.org/) (Jolley and Maiden, 2010) and PlasmidFinder (https://cge.cbs.dtu.dk//services/PlasmidFinder/), respectively. The core genome phylogenetic tree was generated using kSNP3 (Gardner et al., 2015), and plasmid maps were generated using CGView (Grant and Stothard, 2008). Gene organization diagrams were generated using Python script and modified with Inkscape 1.0 (https://inkscape.org/).

The differences in the colistin MIC distribution of 509 mcr-1-negative E. coli isolates among the four years were tested using the χ² test. A P value <0.05 was considered to be statistically significant. All analyses were conducted using the SPSS software (version 22.0; SPSS Inc., Chicago, IL, USA).

The genome data of mcr-1-positive strains have been submitted to NCBI under the BioProject accession number PRJNA770868. The mcr-1.34 gene in this work have been submitted to GenBank under accession numbers MZ450868.

Approximately 17.14% (18/105), 19.65% (34/173), 17.5% (7/40), and 74.11% (146/197) of the isolates showed MIC values ≥1 mg/L against colistin in 2015, 2016, 2017 and 2019, respectively. However, only six isolates showed MIC values of 4 mg/L, including three strains isolated in 2019 and one each in 2015, 2016, and 2017 ( Figure 1A ).

Further analysis demonstrated that isolates from 2015, 2016, and 2017 exhibited similar percentage of MIC to colistin, ranging from 0.25 mg/L to 2 mg/L. The percentages of different MIC levels were 53.65% (169/315), 28.57% (90/315), 12.38% (39/315), and 5.40% (17/315) at 0.25, 0.5, 1, and 2 mg/L, respectively. In 2019, nearly half (47.42%, 92/194) of the strains had a MIC value of 1 mg/L, and approximately 26.29% (51/194) of the strain had a MIC of up to 2 mg/L. Moreover, the MIC value of 0.25 mg/L was found in only 6.7% (13/194) of the bacteria isolated in 2019, which was just one-eighth (53.65%) of that from the previous years ( Figure 1B ). This might be associated with the increase in use of polymyxin to treat bacterial infections in humans (http://app1.nmpa.gov.cn/data_nmpa/face3/dir.html), which contributes to the emergence and increase of polymyxin-resistant bacteria.

In this study, six (1.17%) mcr-1-positive E. coli were identified. Previous publications reported mcr-1 positive rates of 0.35% (12/3434), 0.69% (2/291), and 0.6% (4/700) in clinical E. coli isolates from Chinese hospitals between 2002 and 2016 (Lu et al., 2018), in 2018 (Liao et al., 2020), and from 2014 to 2015 (He et al., 2017), respectively. One study reported an mcr-1 positive rate of 0.87% (6/689) in clinical Salmonella spp. between 2009 and 2018 (Fan et al., 2020). A higher mcr-1-positive rate of 3.5% (24/688) in E. coli was reported from 2016 to 2018 in Guangzhou, China (Shen et al., 2020). These results revealed the increasing prevalence of the mcr-1 gene in clinical settings.

Of the six mcr-1-positive strains, EC3769, EC4968, and EC7518 were isolated in 2015, 2016, and 2017, respectively, whereas EC16, EC117, and EC195 were isolated in 2019. The results showed that the six mcr-1-harboring strains exhibited higher MICs against colistin than the negative strains. All six mcr-1 positive strains found in this study were resistant to colistin with a MIC value of 4 mg/L, which was in accordance with previous studies (He et al., 2017; Liao et al., 2020). Furthermore, all mcr-1-positive strains showed resistance to multiple antimicrobial agents, exhibiting similar MIC levels against 14 antimicrobials tested, except MEM and IMP ( Table 1 ). One of the mcr-1-positive isolates, namely, EC195, showed the broadest resistance spectrum and highest MIC levels against all 14 antimicrobials tested in this study, especially for MEM and IMP. EC195 showed MIC levels >8 mg/L to MEM and IMP, approximately ≥16-fold higher than the other five strains, which showed MIC ≤ 1 mg/L to either MEM or IMP ( Table 1 ).

Sequencing the mcr-1 gene revealed that five of the six isolates exhibited 100% identity to the primary reported mcr-1 gene (KP347127). By contrast, the mcr-1 gene of EC3769 was a novel variant (designated mcr-1.34 in this study) with a one-point mutation at nucleotide position 6 (G to A), which led to the substitution of methionine residue to isoleucine residue (Met to Ile) in comparison with the MCR-1.1 protein (AKF16168). The MCR-1 protein contains an N-terminal inner membrane-bound domain (residues 1–241) and a C-terminal soluble catalytic domain (residues 215–541) (Gao et al., 2016). The substitution of the novel variant in this study was very close to the start of the transmembrane domain.

The general features of the genomes of the six mcr-positive isolates (four draft genomes of EC3769, EC117, EC16, and EC4968 and two complete genomes of EC195 and EC7518) are summarized in Table 2 . In silico analysis revealed that these strains belonged to different MLST, including ST1011 (EC3769), ST93 (EC117), ST101 (EC7518), ST602 (EC16), ST410 (EC195), and ST2505 (EC4968) ( Figure 2 ). ST1011 was proposed to be closely related to the poultry sector, identified in Lebanon and duck farms in southeast coastal China (Wang et al., 2021). However, an mcr-1-carrying E. coli ST1011 has also been isolated from a fecal sample of a 21-year-old male patient (Liang et al., 2021). Even though ST101 has been reported as one of the most prevalent sequence types (STs) among bla _NDM-positive E. coli strains in poultry production, it has been found in hospital sewage water and is regarded as an infection-causing isolate (Wang Y. et al., 2017; Jin et al., 2018). In addition, ST410 has been frequently reported in animal husbandry-related epidemiology studies (Wang et al., 2020; Cheng et al., 2021).

ST93 was previously regarded as a similar ST to ST410; however, an isolate from the blood sample of an 84-year-old man in a Uruguay hospital with resistance to third-generation cephalosporins changed that perception (Papa-Ezdra et al., 2020). ST93 may pose a more significant threat to human health than ST410. ST602 is identified in animal lesion organs (Liu et al., 2021), whereas E. coli ST2505 is a novel ST carrying mcr-1 gene first reported in this study. However, some prevalent STs of mcr-1 carrying E. coli in previous publications such as ST10, ST131, and ST156 (He et al., 2017) were not found in this study, especially E. coli ST10, the most prevalent ST of hospital-associated E. coli (Shen et al., 2018).

Up to 41 different antibiotic resistance genes (ARGs) were identified in these mcr-1-carrying E. coli genomes, including β-lactam resistance genes, aminoglycoside, tetracycline, and sulfonamide genes ( Figure 2 ). Moreover, aadA1, aadA2, aadA5, and the efflux genes oqxA and oqxB were also identified. This enabled them to develop multidrug resistance phenotypes. According to previous reports, bla _CTX-M-65, bla _TEM-206, bla _CMY-2, bla _NDM-5, bla _CTX-M-14, bla _TEM-1B, and bla _CTX-M-55 belong to extended-spectrum β-lactamases, the main problem of E. coli drug resistance, which can decrease the activity of β-lactam ring and cause bacterial resistance to antimicrobials (Cullik et al., 2010; Smet et al., 2012). Genes sul1, sul2, and sul3 contribute to alleviating the damage of E. coli caused by sulfonamide, especially sul2, which is usually located in multiple resistance-determining regions of a large transferable plasmid and can thus spread along with its carrier plasmid (Enne et al., 2001). The acetyltransferase genes, including aac(3)-IIa, aac(3)-Iv, aac(3)-IId, and the phosphotransferase genes, including aph(6)-Id, aph(3′)-Ib, aph(4)-Ia, aph(3′)-Ia, and aph(3′)-IIa, are thought to increase E. coli resistance by changing the target position and decreasing medicine efficacy (Recht and Puglisi, 2001). Furthermore, since its first report in pig manure, in 2003, many studies have demonstrated that oqxA and oqxB are related to diversified drug resistance by extracellularly extruding antimicrobial poison (Sørensen et al., 2003; Hansen et al., 2007; Sato et al., 2011).

Among the six mcr-1-carrying E. coli genomes, EC195 carried the highest number (28) of resistance genes, nearly a half more than those (19) of the following EC117. Resistance genes such as aadA2, aac(3)-IIa, aadA5, and bla _NDM-5 only appeared in EC195, which might contribute to its broader resistance spectrum and higher MIC levels against the antimicrobials tested than those of the other five isolates. The resistance genotype of EC195 was consistent with the phenotype. It had 12 classes of resistance genes and showed resistance to the corresponding antimicrobials ( Table 1 ). Moreover, the specific resistance against MEM and IMP of EC195 might be related to the presence of bla _NDM-5, the carbapenem resistance-related gene only identified in EC195.

In the six mcr-1-positive genomes, the mcr-1 genes were all located on three types of incompatibility (Inc) group plasmids, IncHI2 (pEC195-MCR-1), IncI2 (pEC16-MCR-1, pEC117-MCR-1, pEC4968-MCR-1, and pEC7518-MCR-1), and IncX4 (pEC3769-MCR-1). Five (exclusive of pEC195-MCR-1) of the six mcr-1 carrying plasmids were transferable and the transconjugants were obtained. The transconjugants showed colistin MIC levels slightly lower than those of their corresponding original isolates ( Table 1 ). In addition, complete plasmid sequences carrying mcr-1 were retrieved from GenBank for further analysis. IncHI2 was the most common Inc type among them, followed by IncI2 and IncX4 ( Supplementary Figure S1 ). This result was consistent with a previous report, which showed that IncI2, IncHI2, and IncX4 accounted for more than 90% of the plasmids reported to carry mcr-1 (Berglund et al., 2018).

The IncHI2 plasmid pEC195-MCR-1 was 253,380 bp in length with a G+C content of 46.80%. It shared 99% nucleotide identity and over 95% coverage with plasmid pHNSHP45-2 (KU341381), which was the first reported IncHI2 plasmid carrying mcr-1 (Arcilla et al., 2016) ( Figure 3A ). The variable region of pEC195-MCR-1 consisted of a group of resistance genes including bla _CTX-M-14, fosA3, aac(3)-Iv, aph(4’)-Ia, sul2, floR, aadA, cmlA6, ant3Ia, sul3, aph(4’)-Ia, sul1, oqxAB, a tellurium resistance gene cluster and mcr-1. However, additional ARGs (mrx/mphA) and mobile genetic elements (MGEs; IS1203, ISEc25, IS1A, and TnEc1) were identified in pEC195-MCR-1. The IncHI2-type plasmid has been reported to be the most diverse plasmid that contains a large MDR region composed of various ARGs and MGEs (Li et al., 2017). Co-occurrence of mcr-1 and other ARGs, especially EBSL, in plasmids can cause difficulties in clinical antibacterial treatment.

IncX4 plasmid pEC3769-MCR-1 was 33,310 bp in length and had an average G+C content of 41.58% ( Figure 3B ). The plasmid was almost identical (100% coverage, 99% identity) to other mcr-1-harboring IncX4 plasmids such as pHNSHP49 (MF774188) (Wu et al., 2018). These IncX4 plasmids have no known ARGs other than mcr-1. However, self-transmissible IncX4-type plasmids are important vehicles responsible for disseminating the mcr-1 gene among Enterobacteriaceae worldwide (Fernandes et al., 2016; Wang Q. et al., 2017).

The length of the four IncI2 plasmids ranged from 60,960 to 63,359 bp ( Figure 3C ). These plasmids consisted of replication genes, horizontal transfer, maintenance, and stability region and only carried one ARG mcr-1. BLASTN results revealed they were similar to the first reported mcr-1-harboring IncI2 plasmid pHNSHP45 (NZ_KP347127). These IncI2 plasmids share a highly conserved backbone and contain minor differences in the repA neighboring region. The IncI2-type plasmid was the earliest reported vector of the plasmid-mediated mcr-1 gene, which plays a significant role in rapidly mobilizing and acquiring mcr-1 (Liu et al., 2016; Petrillo et al., 2016; Zelendova et al., 2021).

Comparative genomic analysis was performed on the ~20-kb sequences around the mcr-1 genes. The results revealed that an ~2.6-kb region encoding mcr-1-pap2 was conserved among these plasmids ( Figure 4 ), which was in accordance with previous reports about the mcr-1 gene (Poirel et al., 2016). This 2.6-kb region is always associated with the ISApl1 element and can be mobilized in the form of Tn6330 (ISApl1-mcr-1-pap2-ISApl1) (Snesrud et al., 2018). However, among these sequenced mcr-1-bearing plasmids, only pEC195-253K carried a single copy of ISApl1 beside mcr-1, which indicated that mcr-1 was stable on these plasmids. Among these four IncI2 plasmids, the mcr-1 gene was located between tral-nikA-nikB and ymoA-topB sequences. In addition, these regions of pEC4968 and pEC117 were almost identical, but two copies of IS1203 truncated the ymoA-topB genes of pEC7518-63K. IS26 flanked by an 8-bp direct repeat was identified upstream of mcr-1 on pEC3769-MCR-1, but was not related to the mobilization of the mcr-1 context (Manageiro et al., 2019).

Furthermore, the 5-kb region around the mcr-1 gene of IncHI2, IncI2, and IncX4 plasmids was retrieved from GenBank of NCBI and subsequently clustered with a threshold identity >95% ( Supplementary Figure S1 ). A complete Tn6330 transposon was identified on IncHI2 and IncI2 plasmids, demonstrating its role in the initial acquisition of mcr-1. It has been reported that the nikA-nikB-mcr-1-pap2 structure is conserved among IncI2 plasmids (Feng et al., 2019). Herein, we also found that nikB upstream of mcr-1 in all IncI2 plasmids was often truncated to different lengths. Furthermore, in Cluster 6 of IncI2 plasmids, we noticed that the pap2 gene, which encodes the PAP2-family protein to facilitate the transfer of mcr-1, was lost.