Kluyvera ascorbata Strain from Hospital Sewage Carrying the mcr-1 Colistin Resistance Gene

Abstract

The newly identified plasmid-borne colistin resistance gene mcr-1 was found in a Kluyvera ascorbata isolate from hospital sewage in China. mcr-1 was carried by a 57-kb self-transmissible IncI2 plasmid. Unlike in a previous report, mcr-1 was not associated with ISApl1 and was inserted into a gene encoding a putative membrane protein by an unknown mechanism. This study highlights that mcr-1 has spread to multiple bacterial species.

TEXT

Colistin is the last-resort antimicrobial agent for treating infections caused by many Gram-negative bacteria. However, colistin resistance has emerged worldwide (1). The mechanisms responsible for colistin resistance have been associated mainly with mutations of or insertions in chromosomal genes, such as the phoP-Q two-component system and its regulator gene mgrB (1). However, a plasmid-borne colistin resistance gene, mcr-1, was recently found in Escherichia coli and Klebsiella pneumoniae isolated from humans and animals in eastern and southern China (2). As mcr-1 can be transferred by a plasmid (3), it may have disseminated to species other than E. coli and K. pneumoniae. In a study that screened for the presence of colistin-resistant Enterobacteriaceae in hospital sewage, we found mcr-1-carrying Kluyvera ascorbata, which is reported here.

K. ascorbata strain WCH1410 was recovered from hospital sewage, which was collected from the mainstream influx of a wastewater treatment plant in West China Hospital, Chengdu, western China, in November 2015. Strain WCH1410 grew on a CHROMagar orientation agar plate (bioMérieux, Lyon, France) containing 2 μg/ml colistin and 32 μg/ml linezolid. Linezolid was added to the plate to inhibit the growth of Gram-positive bacteria. Initial species identification was performed using the API 20E test kit (bioMérieux, Hazelwood, MO), which identified strain WCH1410 as a Kluyvera sp. Partial sequencing of the 16S rRNA and gyrB genes as described previously (4) revealed that strain WCH1410 belonged to K. ascorbata, as the 1,330-bp fragment of its 16S rRNA gene and 1,100-bp fragment of gyrB had 99.0% (1,317/1,330) and 93.5% (1,031/1,100) identities, respectively, with those of K. ascorbata type strain ATCC 33433 (draft genome, GenBank accession number JMPL00000000.1). K. ascorbata is a member of the Enterobacteriaceae family and an infrequent opportunistic pathogen.

Strain WCH1410 was found by PCR and sequencing of the amplicon as described previously (2) to be carrying mcr-1. Since the first report of mcr-1 in E. coli and K. pneumoniae (2), it has been shown that mcr-1 encodes a phosphoethanolamine transferase enzyme that can add phosphoethanolamine to the lipid A moiety of lipopolysaccharides (LPSs), which is the initial target of colistin, and therefore results in resistance to colistin (2). Several follow-up studies found mcr-1-carrying E. coli in Africa (Algeria) (5), Europe (including Belgium [6], Denmark [7], France [8], Germany [9], the Netherlands [10], and Switzerland [11]), Asia (including Bangladesh [12], Cambodia [12], Japan [13], Laos [5, 12], Malaysia [2], Thailand [5], and Vietnam [14]), and North America (including Canada [15] and the United States [12]). In addition, mcr-1 was also found in several species of Salmonella in France (16) and Portugal (17, 18). This suggests that mcr-1 is widely distributed and poses an emerging threat for clinical management and public and animal health. Nonetheless, mcr-1 has not yet been found in Kluyvera spp.

Antimicrobial susceptibility tests were performed using the Vitek II automated system (bioMérieux, Durham, NC). Strain WCH1410 was resistant to ampicillin, cefazolin, and nitrofurantoin but was susceptible to amikacin, ampicillin-sulbactam, aztreonam, cefepime, cefoxitin, ceftazidime, ceftriaxone, ciprofloxacin, gentamicin, imipenem, piperacillin-tazobactam, tobramycin, and trimethoprim-sulfamethoxazole. MICs of amikacin, ceftazidime, ciprofloxacin, colistin, imipenem, and polymyxin B were further determined using the microdilution broth method following recommendations of the Clinical and Laboratory Standards Institute (CLSI) (19). Breakpoints defined by the CLSI were applied except for those of colistin and polymyxin B, for which breakpoints defined by EUCAST were used. Strain WCH1410 was resistant to colistin (MIC, 8 μg/ml) and polymyxin B (MIC, 8 μg/ml) but was susceptible to amikacin (MIC, 0.5 μg/ml), ceftazidime (MIC, ≤0.5 μg/ml), ciprofloxacin (MIC, ≤0.008 μg/ml), and imipenem (MIC, 0.125 μg/ml). This suggests that mcr-1 is not necessarily carried by extensively drug-resistant strains.

Genomic DNA of strain WCH1410 that was prepared using a QIAamp DNA minikit (Qiagen, Hilden, Germany) was subjected to whole-genome sequencing with 100× coverage using a HiSeq 2500 sequencer (Illumina, San Diego, CA) following manufacturer protocol. A total of 4,534,884 reads and 680,232,600 clean bases were generated. The SPAdes program (20) was used for de novo assembly and generated 75 contigs ≥1,000 bp in length (N50; 158,869 bp) with a 54.82% GC content. Annotation of the genomic sequence was carried out using the Prokka program (21). Antimicrobial resistance genes were predicted using ResFinder from the Center for Genomic Epidemiology (http://genomicepidemiology.org/). In addition to mcr-1, strain WCH1410 has two known antimicrobial resistance genes, i.e., the extended-spectrum β-lactamase gene bla_CTX-M and the fosfomycin resistance gene fosA. The fosA gene was identical to fosA4 in E. coli (GenBank accession number AB908992). K. ascorbata has been thought to be a source of bla_CTX-M, as its chromosome contains this gene (22). A new variant of bla_CTX-M, which encodes a CTX-M enzyme with two amino acid substitutions, compared to CTX-M-78 (GenBank accession number AM982522) of the CTX-M-25 group, the closest match, was identified in a contig belonging to the chromosome of strain WCH1410. This variant of bla_CTX-M was assigned bla_CTX-M-185 by the β-lactamase database curated by the National Center for Biotechnology Information.

The 876-bp complete coding sequence of bla_CTX-M was amplified with primers 1410CTXM-upEcoRI (AGAATTCAAAAAGGGGGAAACAAGAGG, the EcoRI restriction site is underlined) and 1410CTXM-dwEcoRI (AGAATTCTCAGGCCAGGTACAAAAACA). The amplicon was restricted using EcoRI (NEB, Ipswich, MA) and then was ligated to the EcoRI-restricted pBC SK vector (Agilent, Santa Clara, CA) using T4 ligase (NEB). The ligated fragments were electroporated into the E. coli DH5α strain, and the bla_CTX-M-containing transformants were selected on LB agar plates containing 100 μg/ml ampicillin. The presence of bla_CTX-M in transformants was confirmed by PCR and sequencing. MICs of ampicillin, aztreonam, ceftazidime, cefotaxime, cefoxitin, cephalothin, and imipenem were determined using the broth microdilution method following the recommendations of CLSI (19). Transformants containing bla_CTX-M were resistant to ampicillin (MIC, >512 μg/ml), cefotaxime (MIC, 8 μg/ml), and cephalothin (MIC, >512 μg/ml) but were susceptible to aztreonam (MIC, 1 μg/ml), ceftazidime (MIC, 0.5 μg/ml), cefoxitin (MIC, 1 μg/ml), and imipenem (MIC, ≤0.25 μg/ml). The resistance spectrum conferred by CTX-M was consistent with those of most CTX-M enzymes.

mcr-1 was transferred to azide-resistant E. coli strain J53 at a frequency of 10⁻⁵ cells per recipient cell by mating, which was carried out in brain heart infusion broth (Oxoid, Basingstoke, United Kingdom). Transconjugants were selected on LB agar plates containing 2 μg/ml colistin and 150 μg/ml sodium azide, and the presence of mcr-1 in transconjugants was confirmed using PCR. The plasmid carrying mcr-1 in strain WCH1410 was designated pMCR_1410. pHNSHP45, the first plasmid found to carry mcr-1 (GenBank accession number KP347127), is an IncI2 plasmid (2). pMCR_1410 also had the IncI2 replicon as determined using PCR with primers I2-F1 (AGTCACAGTATCGCGCCTTT) and I2-R1 (TTACGAGCCGAGTGAACAGA), which bind to the repA gene of the IncI2 replicon on pHNSHP45. In a previous study, the mcr-1-carrying pHNSHP45 was not transferred to K. pneumoniae from E. coli by conjugation (2). Therefore, the ability of pMCR_1410 to be transferred between different species of the Enterobacteriaceae family is worrisome, as it might have the potential to transfer mcr-1 to more species.

The sequence of pMCR_1410 was completely circularized with gaps between contigs closed by Sanger sequencing of amplicons from PCR using primers designed on the basis of available contig sequences. pMCR_1410 was a 57-kb IncI2 plasmid. The 27.8-kb backbone (containing IncI1 replicon and conjugal module) of pMCR_1410 was closest (99% identity) to that of the unnamed plasmid 3 of E. coli FAP1 (GenBank accession number CP009581) and was also similar (98% identity) to that of pHNSHP45. IncI2 plasmids contain a tra module encoding the thick pilus, which is the primary pilus for conjugation, and a pil module encoding the thin pilus, which can enhance conjugation in liquid medium (23). When compared with the tra and pil loci of pMCR_1410 with those of pHNSHP45, it became evident that the traC gene, which encodes a DNA primase, was absent from pHNSHP45 (Fig. 1). The absence of traC may explain the failure in the conjugation transfer of pHNSHP45.

FIG 1.

Comparison of pMCR_1410 and pHNSHP45. The tra and pil loci, rci (encoding the plasmid-encoded site-specific recombinase), mcr-1, and repA (encoding the plasmid replication initiation protein) are indicated. The traC gene was absent from pHNSHP45 (GenBank accession number KP347127).

On pHNSHP45, mcr-1 and a gene encoding a putative phosphoesterase (designated pho here) were located downstream of the insertion sequence ISApl1, which was originally found in Actinobacillus pleuropneumoniae (formerly Haemophilus pleuropneumoniae), a pathogen of pigs. However, ISApl1 was absent from pMCR_1410. A gene encoding a putative membrane protein, which is also present in R721 and the unnamed plasmid 3 of E. coli FAP1, was interrupted by mcr-1 and pho in pMCR_1410 (Fig. 2). No direct repeats and no mobile genetic elements abutting the mcr-1-pho region were identified, and the mechanism mediating the mobilization of mcr-1 and pho remains unknown.

FIG 2.

Genetic context of mcr-1 in pMCR_1410. Genetic context of mcr-1 in pHNSHP45 (GenBank accession number KP347127) and the corresponding region on the unnamed plasmid 3 of E. coli FAP1 (GenBank accession number CP009581) are shown for comparison. Unnamed open reading frames (ORFs) are indicated by numbers 1 to 3, among which ORF2 encodes a putative membrane protein and is disrupted into two parts, labeled 2Δ1 and 2Δ2, in pMCR_1410. Other genes shown are nikB (encoding relaxase of the plasmid), ydgA (DNA topoisomerase III), and ydfA (transcriptional regulator). In pHNSHP45, the 2-bp direct repeat (GA) abutting the ISApl1-mcr-1-pho region is shown.

In conclusion, in this study we identified, for the first time, mcr-1 in K. ascorbata from hospital sewage. mcr-1 was carried by an IncI2 plasmid, which was transferred to E. coli by conjugation. mcr-1 and a gene encoding a putative phosphoesterase were inserted into a gene encoding a putative membrane protein by an unknown mechanism.

Accession number(s).

Reads and the whole-genome shotgun sequencing project of K. ascorbata strain WCH1410 were deposited into DDBJ/EMBL/GenBank under accession numbers SRR3168915 and LSME00000000. The sequence of pMCR_1410 was deposited into DDBJ/EMBL/GenBank under accession number KU922754.