Introduction

The emergence and rapid growth of antibiotic-resistant bacteria pose major challenges to human health, veterinary health, and food safety on a global scale because of the improper use of antibiotics in clinical application and livestock farming [1]. A lack of novel antibacterial agents to combat multidrug-resistant (MDR) Gram-negative bacteria has led to the reuse of polymyxins, particularly colistin, in treatment programs [1]. However, resistance to colistin has been on the rise following the emergence and widespread distribution of the plasmid-borne colistin resistance gene mcr-1 [1,2,3,4]. To date, 10 mcr variants and many subvariants have been globally identified in diverse bacterial species, particularly Escherichia coli, Klebsiella pneumoniae, and Salmonella species, from various sources [1, 5]. Among the 10 mcr variants, mcr-1 stands out as the predominant type [1].

E. coli are Gram-negative bacteria. Commensal variants of E. coli are harmless, whereas pathogenic variants can trigger intestinal or parenteral infections in humans and many animal hosts [6]. Statistically, 48 per 100,000 individuals in high-income countries are diagnosed with E. coli bacteremia infections every year [6]. A major concern is that MDR E. coli may cause treatment failures because they are a potential reservoir of drug resistance genes, particularly mcr-1, which confers resistance to colistin [7]. E. coli isolates carrying mcr-1 prevail significantly among food-producing animals and exhibit widespread distribution across various sources, including humans, wildlife, companion animals, food products, and the environment [1,2,3,4, 8]. The rapid dissemination can be attributed to the use of colistin in veterinary medicine or as a growth promoter, along with the capability of mcr-1 for horizontal transfer [1, 3, 8]. The insertion sequence ISApl1, transposon Tn6330 (ISApl1–mcr-1–ISApl1), and many plasmids (such as IncX4, IncI2, and IncHI2) have been reported to be involved in mcr-1 transmission among various sources [1, 3, 4].

Following the prohibition of adding colistin to animal feed for growth promotion on April 30, 2017, significant effects have been observed in China, notably in reducing both colistin resistance and the prevalence of mcr-1-positive E. coli [9]. To further assess the prevalence of mcr-1, we examined the dissemination of mcr-1 in E. coli from food-producing animals and retail raw meats and analyzed the genetic environment of mcr-1 in this study.

Materials and methods

Detection of mcr and antimicrobial susceptibility testing

From June 2019 to November 2020, a total of 1,353 fecal samples were collected from food-producing animals including pigs (n = 212), chickens (n = 358), cattle (n = 752), and pigeons (n = 31) from farms located in Anhui, Henan, Liaoning, Jiangsu, Guangdong, Shandong, and Xinjiang in China and 836 food samples including pork (n = 377), chicken meat (n = 341), and beef (n = 118), were collected from a slaughterhouse, farmers’ markets, and supermarkets in the aforementioned provinces and Shanghai (Supplementary Table S1). The isolation of E. coli using previously described methods with minor modifications [10, 11]. The samples were cultured for 18–24 h in buffered peptone water (BPW) broth at 37 °C. The positive growth was further streaked on a MacConkey agar plate and incubated at 37 °C for 24 h. One pink colony from each plate was inoculated onto an eosin methylene blue (EMB) agar plate for 24 h at 37 °C. A colony with metallic sheen color (presumptive E. coli) was inoculated onto another EMB agar plate for purification. One E. coli isolate was randomly chosen from each plate and identified using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (Bruker Daltonik GmbH, Bremen, Germany).

The susceptibility of all E. coli isolates to colistin was assessed using the broth microdilution method according to the International Standards ISO 20776-1 (https://www.iso.org/standard/70464.html). The colistin-resistant E. coli isolates were screened for the presence of mcr by PCR analysis and sequencing using the primers listed in Supplementary Table S2. The susceptibility of the mcr-positive isolates to 15 antimicrobial agents was further assessed using the agar dilution method following the guidelines of the Clinical and Laboratory Standards Institute (CLSI) M07 [12]. The results were interpreted according to the 30th edition of the CLSI M100 [13]. E. coli ATCC 25,922 served as the quality control in antimicrobial susceptibility testing.

Whole genome sequencing and analysis

Genomic DNA was extracted from all mcr-1-positive E. coli isolates using the TIANamp Bacteria DNA Kit (Tiangen, Beijing, China), according to the manufacturer’s instructions. The Illumina HiSeq platform was used to sequence all mcr-1-positive isolates, and SPAdes v.3.8.2 was used to assemble the sequence reads into contigs. Multilocus sequence typing (MLST), acquired resistance genes, chromosomal mutations, and plasmid incompatibility groups were detected using the CGE database (http://www.genomicepidemiology.org/). PCR analysis and Sanger sequencing were performed to assemble plasmid contigs into a complete plasmid sequence (Supplementary Table S3). Initial analysis and annotation of contigs or plasmids containing mcr-1 were using the RAST (https://rast.nmpdr.org/rast.cgi), ISfinder (https://www-is.biotoul.fr/), BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi) and Gene Construction kit 4.5 (Textco BioSoftware, Inc., Raleigh, NC). The genetic structures of mcr-1 in plasmids were drawn using Vector NTI 11 (Thermo Fisher Scientific, Inc., Waltham, MA) and manually adjusted. All whole genome sequences and mcr-1-carrying plasmids have been deposited in GenBank under the accession numbers PRJNA967092 and PRJNA974499, respectively.

Conjugation experiments

As mentioned previously, conjugation experiments were performed using all mcr-1-positive E. coli isolates [14]. The mcr-1-positive E. coli isolates were used as the donors and high-level streptomycin-resistant E. coli C600 was used as the recipient; they were mixed in a ratio of 1:4. Transconjugants were selected on EMB agar containing colistin (2 µg/mL) and streptomycin (3000 µg/mL) and were further confirmed by detecting mcr-1 using PCR. The experiments were performed in triplicate, and the conjugal frequency of mcr-1 was estimated as the number of transconjugants per recipient.

Results

Characterization of mcr-1 -positive E. coli isolates

In total, 1,403 E. coli strains were isolated from food-producing animals and retail meat products (Supplementary Table S1). Of these, 13 isolates (0.93%) from chicken meat (n = 7), chickens (n = 3), and pigs (n = 3) exhibited resistance to colistin with minimum inhibitory concentration (MIC) values of 4 to 16 mg/L. No other mcr variants were detected, except for mcr-1 (Table 1). A low detection rate of mcr-1 among E. coli isolates originating from food-producing animals (0.62%, 6/974) and animal-derived food (1.63%, 7/429) was observed in this study.

Table 1 Characteristics of mcr-1-carrying E. coli isolates in this study
Full size table

As shown in Table 1, all colistin-resistant E. coli isolates, except for isolate AH20PE105, were also resistant to multiple antibiotics. Eleven isolates were resistant to ampicillin, tetracycline, and chloramphenicol. Additionally, among the 13 mcr-1-positive isolates, resistance to streptomycin, sulfamethoxazole/trimethoprim, and gentamicin was observed in 10, 10, and nine cases, respectively. Nine isolates contained blaCTX−M, including blaCTX−M−55 (n = 6), blaCTX−M−14 (n = 3), blaCTX−M−65 (n = 1), and blaCTX−M−123 (n = 1), and were resistant to cefazolin and cefotaxime. Two isolates, SD20MCE26 and LN19MCE7, carried both blaCTX−M−55 and blaCTX−M−14. Furthermore, nine MCR-1-producing E. coli strains exhibited resistance to florfenicol, a common veterinary antimicrobial agent, and carried the florfenicol resistance gene floR. The fosfomycin resistance gene fosA3 and 16S rRNA methylase gene rmtB were identified in six fosfomycin-resistant strains and one amikacin-resistant strain, respectively. In addition, gyrA (S83L and D87N/Y) and parC (S80I) mutations were observed in 10 E. coli strains, thereby explaining their nalidixic acid- and ciprofloxacin-resistant phenotypes. The presence of the multidrug efflux pump genes oqxAB and quinolone resistance gene qnrS2 and the existence of a single mutation in parC (A56T) in isolate SD20MCE26 were responsible for its resistance to nalidixic acid.

MLST analysis based on whole genome sequencing revealed that 13 mcr-1-positive E. coli isolates belonged to 10 sequence types (STs), including ST10 (n = 2), ST101 (n = 2), ST156 (n = 2), ST155, ST48, ST6388, ST1011, ST3871, ST1589, and a new ST (ST14521) (Table 1). Three to seven plasmid replicons were identified in mcr-1-positive E. coli strains (Table 1). By analyzing mcr-1-carrying contigs or plasmids, we found that mcr-1 was located on plasmids in all isolates and that IncI2 was predominant (n = 9), followed by IncX4 (n = 2) and IncHI2 (n = 2) (Table 1).

Characterization of mcr-1 -carrying IncI2 plasmids in nine E. coli isolates

The mcr-1-carrying IncI2 plasmids in this study were similar to the first reported mcr-1-carrying plasmid pHNSHP45 (E. coli, pig, China, KP347127) (Fig. 1). The transposable element ISApl1 observed upstream of mcr-1 in pHNSHP45 was not present in eight mcr-1-carrying plasmids. Moreover, only one plasmid (pYUYZMC13-MCR) from the isolate YZ19MCE13 contained an incomplete ISApl1 upstream of mcr-1 (Fig. 1). In addition to mcr-1, the plasmids pYUAHC37-MCR, pYUAHC39-MCR, and pYUSDMC15-MCR from the isolates AH20CE37, AH20CE39, and SD20MCE15, respectively, carried blaCTX−M−55. It was located in a typical transposition unit (ISEcp1–blaCTX−M−55–orf477) with 5-bp direct repeats (DRs, 5′-GAAAA-3′), while ISEcp1 was interrupted by IS1294 in pYUAHC37-MCR and pYUAHC39-MCR (Fig. 1).

Fig. 1
figure 1

Genetic structure of mcr-1 in nine plasmids and structural comparison with related mcr-1-carrying IncI2 plasmids. Regions with > 99% identity are shaded in gray. Thick arrow indicates ORF. Red thin red arrow indicates resistance genes. Box indicates insertion sequences. The delta symbol (Δ) indicates a truncated gene or mobile element. Direct repeats are indicated by arrows and sequences

Full size image

A multiple inversion system called shufflon was first identified in the IncI1 plasmid R64 and later discovered in other Inc types, including IncIγ, IncI2, IncK, and IncZ plasmids [15]. The rearrangement of shufflon regions results in the generation of different C-terminal ends of the PilV protein, which is involved in bacterial conjugation [15]. The shufflon region of nine IncI2 plasmids was distinct and included eight arrangements, and it was interrupted by ISEc8 in pYUYZMC28-MCR (Fig. 1). A similar ISEc8 insertion was observed in the recombinase gene rci in pYUYZMC34-MCR (Fig. 1). In pYUYZMC13-MCR, the mok–parA fragment (2007 bp) was absent because of the insertion of IS1294 and the type IV pilus transmembrane gene pilR was interrupted by IS2, generating 5-bp DRs (5′-CCGCG-3′) (Fig. 1). The insertion of mobile elements into the conjugal region may explain the failure of conjugation of the three abovementioned mcr-1-carrying plasmids. Three plasmids in this study (pYUYZMC6-MCR, pYUSDMC15-MCR, and pYUAHP105-MCR) could be successfully transferred to E. coli C600 at a frequency of 5.79 × 10− 6, 8.99 × 10− 3, and 1.9 × 10− 4 transconjugants per recipient, respectively.

Characterization of mcr -1-carrying IncX4 plasmids in two E. coli isolates

Two mcr-1-carrying IncX4 plasmids, pYUAHP7-MCR and pYUYZP15-MCR, were obtained from isolates AH20PE7 and YZ19PE15, respectively, with a size of 34,541–36,021 bp. Both plasmids had typical IncX4 plasmid backbones, including genes encoding replication proteins and conjugal transfer proteins and those responsible for maintenance and stability (Fig. 2). Moreover, their organization was similar to that of other mcr-1-carrying IncX4 plasmids from animals or food products in China, such as pHNSHP10 (pig, MF774182) and pPY1 (pork, KX711708) from E. coli (Fig. 2). However, ISApl1 was inserted downstream of mcr-1–pap2 in pY1, which was not present in our plasmids. Moreover, the mcr-1–pap2 segment was inserted in pPY1 in reverse orientation. Instead, one copy of IS26 was inserted into the backbone of the two plasmids at the same site, generating 8-bp DRs (5′-CTGTGTGA-3′) (Fig. 2). In addition to IS26, one copy of IS679-like was inserted into hns with 8-bp DRs in pYUAHP7-MCR. Moreover, topB was interrupted by the insertion of ISKpn40 in pYUYZP15-MCR (Fig. 2). As described previously [4, 16, 17], the IncX4 plasmids pYUAHP7-MCR and pYUYZP15-MCR did not carry any drug resistance genes, except for mcr-1. They were transferable at a frequency of 3.75 × 10− 4 and 2.03 × 10− 3 transconjugants per recipient, respectively.

Fig. 2
figure 2

Genetic structure of mcr-1 in two plasmids and structural comparison with related mcr-1-carrying IncX4 plasmids. Regions with > 99% identity are shaded in gray. Red arrow indicates resistance genes. Purple arrow represents conjugal transfer genes. Green arrow represents plasmid replication genes. Blue arrow represents genes for maintenance and stability. Black arrow indicates other genes. The delta symbol (Δ) indicates a truncated gene or mobile element. Direct repeats are indicated by arrows and sequences

Full size image

Characterization of mcr-1 -carrying contigs associated with IncHI2 plasmids in two E. coli isolates

In strains SD20MCE26 and AH20CE15, mcr-1-carrying contigs (101,306 bp and 106,757 bp, respectively) were highly similar to the IncHI2 plasmids pHN6DS2 (E. coli, MH459020), pSI-16E242 (Salmonella, ON960347), and pEC15-MCR-50 (E. coli, MG656414) (Fig. 3). The mcr-1-positive contigs in SD20MCE26 and AH20CE15 harbored tellurium resistance genes (terYXWZABCDEF) and genes responsible for conjugal transfer (trhINUWFOZCVBKELA and htdKFATVO). The insert sequence ISApl1 observed upstream of mcr-1 in pHN6DS2 was present in SD20MCE26 but absent in AH20CE15 (Fig. 3). As IncHI2 plasmid replicons were identified in SD20MCE26 and AH20CE15 (Table 1) and mcr-1 from SD20MCE26 and AH20CE15 could be transferred to E. coli C600 at a frequency of 3.24 × 10− 5 and 1.94 × 10− 5 transconjugants per recipient, respectively, mcr-1 in SD20MCE26 and AH20CE15 may be associated with IncHI2 plasmids.

Fig. 3
figure 3

Genetic structure of mcr-1 in SD20MCE26 and AH20CE15 in this study and comparison with other IncHI2 plasmids. Regions with > 99% identity are shaded in gray. Thick arrow indicates ORF. Red thin red arrow indicates resistance genes. Box indicates insertion sequences

Full size image

Discussion

Since its identification in 2015 in a porcine E. coli strain in China, the plasmid-mediated colistin resistance gene mcr-1 has rapidly disseminated worldwide, being identified in diverse bacterial species across various sources, predominantly hosted by E. coli [1,2,3,4, 8]. In our study, a low prevalence of mcr-1 was observed among E. coli isolates from food-producing animals and animal-derived food. It might be attributed to the random isolation of E. coli strains without employing a medium supplemented with colistin for selection. Furthermore, we identified colistin-resistant E. coli isolates from chickens (1.29%, 3/275) and pigs (1.09%, 3/232), while none were found in cattle (0/507) or pigeons (0/27). The absence in pigeons might be due to the limited numbers of E. coli isolates, and notably, colistin has never been approved for use in cattle in China. On the other hand, our findings align with the substantial decrease in colistin resistance and mcr prevalence following the ban of colistin-positive additives in China [9]. This underscores the critical significance and consequential effects of prohibiting colistin as an animal growth promoter in China. Although a low detection rate of E. coli isolates carrying mcr-1 gene was noted in this study, its presence in animals and food remains a considerable threat to public health, given the potential risk of zoonotic transmission to humans through the food chain and contact with backyard animals [1, 7]. In addition to mcr-1, 12 isolates investigated in this study contained multiple antimicrobial resistance genes, such as blaCTX−M, floR, oqxAB, qnrS, and mutations in gyrA and parC, consistent with their antibiotic-resistant phenotype. The emergence and spread of these MDR E. coli strains present a heightened risk, potentially resulting in difficult-to-treat infections and limiting therapeutic options against infections they cause [7]. More importantly, they serve as a significant reservoir of resistance determinants to most families of antimicrobial agents for animals and humans [7].

A high diversity of E. coli isolates with different STs carrying mcr-1 has been identified in animals, food products, and humans [4, 9, 16]. The STs identified in this study, such as ST10, ST48, ST101, ST155, ST156, ST1011 and ST1589, were also previously reported as common mcr-1 carriers in food-producing animals and humans in China [9, 16, 18]. These STs have also been recognized as mcr-1 carriers beyond China. For example, mcr-1-positive E. coli ST10, ST1011, and ST156 were identified in poultry samples from Poland [19]. In Egypt, two E. coli ST155 strains encoding Tet(X7) and MCR-1 were isolated from chicken meat, while three mcr-1-positive E. coli ST101 strains were recovered from pigs in Europe [20, 21]. The diversity observed in E. coli STs in our study and prior research suggests that horizontal transmission serves as the primary route for mcr-1 dissemination in animals and their food products. Nonetheless, it is noteworthy that the clonal spread of specific ST-type E. coli strains, such as E. coli ST93 among companion animals, and E. coli ST10 in swine farms, may also contribute to the spread of mcr-1 [22, 23].

Plasmids play an essential role in the global dissemination of resistance genes including mcr-1 in Enterobacteriaceae [1, 3, 4, 24]. While various plasmids, such as IncFII, IncY, IncP, and IncK2, have been described as vectors of mcr-1, the majority of identified plasmids were affiliated with three incompatibility groups: IncI2, IncX4, and IncHI2 [3, 4, 8, 17, 25]. These three prevalent plasmid types have served as the principal vehicles for disseminating mcr-1 globally, frequently detected in Enterobacteriaceae, particularly E. coli, across diverse origins [3, 4, 8]. Consistent with this trend, our study also detected the presence of mcr-1 on IncI2, IncX4, or IncHI2 plasmids, with high similarity to each other and previously reported mcr-1-carrying plasmids within the same incompatibility group. Furthermore, a diverse array of IS elements was found integrated into the plasmid backbone, leading to the loss or acquisition of genetic fragments and driving plasmid evolution among different lineages of E. coli strains. Seven mcr-1-carrying plasmids identified in our study possess conjugative capabilities, representing an increased risk of spreading mcr-1 between bacteria, even across different species. This conjugal transferability significantly contributes to the dissemination of both mcr-1 and colistin resistance among bacterial populations [8].

In this study, IncI2 plasmids emerged as the primary vehicle for mcr-1 transmission, sharing a similar backbone yet distinct shufflon regions. Intriguingly, a striking rise in the occurrence of IncI2-type plasmids was noticed among mcr-1-positive E. coli strains from animals and humans following the cessation of colistin as an animal growth promoter in China [9]. The precise reason behind this observation remains unclear. However, IncI2 plasmids often carry additional resistance genes such as blaCTX−M−55 in this study, besides mcr-1, potentially contributing to the preferential selection of these plasmids due to the extensive use of β-lactam antibiotics such as amoxicillin in animals and cephalosprins in human clinical settings [9]. Furthermore, the enhanced fitness conferred by mcr-1-carrying IncI2 and IncX4 plasmids supports their dissemination and persistence in bacterial populations even without antibiotic selection pressures [17]. Conversely, the acquisition of mcr-1-carrying IncHI2 plasmids imposes a competitive disadvantage [17, 26]. Nonetheless, these plasmids often carry diverse resistance genes (e.g., floR, blaCTX−M, and fosA3), co-selection by other antimicrobials might augment the further dissemination of IncHI2 plasmids carrying mcr-1 [8, 17, 22, 27].

The insertion sequence ISApl1 is involved in mobilizing mcr-1 among DNA molecules, such as plasmid or chromosomes [3, 4, 25, 28]. However, in this study, the presence of ISApl1, intact or incomplete, was observed upstream of mcr-1 in only two plasmids, while the mcr-1-pap2 structure (n = 11) was more prevalent. This observation aligns with a prior study indicating that ISApl1 upstream of the mcr-1 gene was present in 77.8% of IncHI2 plasmids, 37.9% of IncI2 plasmids, and absent in IncX4 plasmids [4]. The proposed hypothesis suggests that mcr-1 initially mobilizes through the mobile transposon unit Tn6330 (ISApl1-mcr-1-pap2-ISApl1), subsequently undergoing a gradual loss of ISApl1 at both ends, potentially ensuring the stability of mcr-1 before plasmid-mediated transmission [3, 29]. Consequently, this results in the formation of diverse genetic structures harboring mcr-1, with the mcr-1-pap2 structure being predominant, followed by the ISApl1-mcr-1-pap2 structure [3, 29].

However, our study has several limitations. We collected samples from only seven regions in China, however, certain sources (e.g., pigeons and beef) were limited in sample quantity and location. The uneven distribution of samples across different regions and types led to considerable variation in the number of strains isolated from different areas. For instance, there were notably fewer pig and pork samples and isolated strains from Guangdong province. Additionally, we exclusively detected mcr genes in colistin-resistant E. coli isolates. Notably, mcr-4.3 demonstrated a silent phenotype due to mutations (V179G and V236F), and silent transmission of inactivated mcr-1 and mcr-9 with inducible colistin resistance have been previously reported [30,31,32,33]. Hence, there remains a possibility that the colistin-susceptible E. coli strains yet to be identified in our study may carry mcr.

Conclusion

This study unveils a low detection rate (0.93%) of mcr-1 among E. coli isolates originating from food-producing animals and animal-derived food products, associated with the previously identified IncI2, IncX4 and IncHI2 epidemic plasmids. While the low prevalence of mcr-1 might not immediately appear threatening, its emergence merits attention given its implication for colistin resistance and public health. Considering the potential dissemination of mcr-1 facilitated by plasmids among bacteria, and the risk of co-selection with other commonly used antibiotics in animal husbandry, continuous surveillance of mcr-1 is imperative. This surveillance needs to monitor not only its prevalence and dissemination in E. coli but also in other Enterobacteriaceae species.