Complete Sequences of mcr-1-Harboring Plasmids from Extended-Spectrum-β-Lactamase- and Carbapenemase-Producing Enterobacteriaceae

Abstract

Here we completely sequenced four mcr-1-haboring plasmids, isolated from two extended-spectrum-β-lactamase (ESBL)-producing Escherichia coli and two carbapenemase-producing Klebsiella pneumoniae clinical isolates. The mcr-1-harboring plasmids from an E. coli sequence type 2448 (ST2448) isolate and two K. pneumoniae ST25 isolates were identical (all pMCR1-IncX4), belonging to the IncX4 incompatibility group, while the plasmid from an E. coli ST2085 isolate (pMCR1-IncI2) belongs to the IncI2 group. A nearly identical 2.6-kb mcr-1-pap2 element was found to be shared by all mcr-1-carrying plasmids.

TEXT

The plasmid-mediated colistin resistance gene, mcr-1, has recently been reported from animals and hospitalized patients in China (1). Since then, mcr-1 has been found in ∼20 countries on four different continents (2). Alarmingly, mcr-1 has also been identified in several multidrug-resistant bacteria, including extended-spectrum-β-lactamase (ESBL)-producing and carbapenemase-producing Enterobacteriaceae (CPE) (3–9). However, knowledge regarding the structure of mcr-1-harboring plasmids is limited. Here we completely sequenced four mcr-1-harboring plasmids (three of which are identical), isolated from two ESBL-producing Escherichia coli and two carbapenemase-producing Klebsiella pneumoniae clinical isolates (4).

In a recent study, we identified mcr-1 in two ESBL-producing E. coli (SZ01 and SZ02) and two carbapenemase-producing K. pneumoniae (SZ03 and SZ04) clinical isolates from a tertiary hospital in eastern China (4). SZ01, SZ02, and SZ04 carry ESBL gene bla_CTX-M-55, while SZ03 and SZ04 harbor carbapenemase gene bla_NDM-5. Multilocus sequence typing (MLST) (10, 11) showed that the two E. coli isolates, SZ01 and SZ02, belong to two unrelated sequence types (STs) (ST2448 and ST2085), while the two K. pneumoniae strains (isolated from the same patient) both belong to ST25. The mcr-1-harboring plasmids from all four isolates were subsequently transferred to recipient strain E. coli J53 AZ^r via conjugation, along with the bla_NDM-5-harboring plasmids from SZ03 and SZ04. Susceptibility testing revealed that the four mcr-1-harboring E. coli transconjugants were resistant to colistin but not to any of the other antimicrobial agents tested. The two bla_NDM-5 transconjugants were resistant to all β-lactams, except for aztreonam, but remained susceptible to other classes of antimicrobial agents (data not shown). The mcr-1- and bla_NDM-5-harboring plasmids from these transconjugants were extracted and subjected to sequencing using the Illumina MiSeq platform (12). The sequencing reads were assembled de novo using SPAdes (13), and gaps were closed by standard PCR and Sanger sequencing as described previously (12).

The mcr-1-harboring plasmids from SZ01, SZ03, and SZ04 (subsequently named pMCR1-IncX4) were all identical, belonging to the IncX4 incompatibility group, and were 33,287 bp in length with a G+C content of 41.8%. The backbone of pMCR1-IncX4 is similar to that of other IncX4 plasmids, including pJIE143 (GenBank accession no. JN194214) (14), pBS512_33 (CP001059), pCROD2 (FN543504) (15), and pSH146_32 (JX258655) (16). BLASTn analysis showed that pMCR1-IncX4 has a query coverage of 87% and maximal 97% identity to pSH146_32, isolated from a Salmonella enterica Heidelberg strain from a porcine diagnostic specimen from Minnesota in 2002 (16), and a query coverage of 77% and maximal 99% identity to pJIE143, isolated from an E. coli ST131 strain from Australia in 2006 (14) (Fig. 1). Plasmid pMCR1-IncX4 possesses a replication region highly similar to the one on pJIE143 (Fig. 1), including the identical replication initiation protein gene pir, vegetative origins oriV-α and oriV-β, and highly similar oriV-γ (one less iteron). The region from traM to oriV-γ, encompassing the majority of the transfer region and including the taxABC and pilX operons, shares >99.9% nucleotide identity with pMCR1-IncX4 and pJIE143 (Fig. 1). However, the region in pMCR1-IncX4 between taxD and the histone-like nucleotide-structuring protein gene hns, where colistin resistance gene mcr-1 is located, is absent in pJIE143. In contrast, this region is highly similar to that of pSH146_32, except for the insertion of an mcr-1-pap2 element and an IS26 element in pMCR1-IncX4 (Fig. 1). A 2,610-bp mcr-1-pap2 fragment (nucleotides [nt] 2339 to 4948 in pMCR1-IncX4) was inserted into a hypothetical gene (locus tag pSH146_32_13). Interestingly, insertion element ISApl1, initially found to be associated with mcr-1 in pHNSHP45 (1), was not present in pMCR1-IncX4.

FIG 1.

Structures of plasmids pMCR1-IncX4 and pMCR1-IncI2. (A) Comparison of IncX4 plasmids pSH146_32 (GenBank accession no. JX258655), pMCR1_IncX4 (KU761327, this study), and pJIE143 (JN194214); (B) comparison of IncI2 plasmids pHNSHP45 (KP347127) and pMCR1_IncI2 (KU761326, this study). Colored arrows represent open reading frames, with dark blue, yellow, green, red, purple, and orange arrows representing replication genes, mobile elements, plasmid transfer genes, the mcr-1 gene, other antimicrobial resistance genes, and plasmid backbone genes, respectively. Blue shading denotes regions of shared homology among different plasmids.

The mcr-1-harboring plasmid from E. coli strain SZ02, pMCR1-IncI2, belongs to the IncI2 group, the same plasmid incompatibility group as pHNSHP45, the first plasmid reported to harbor mcr-1 (1). Plasmid pMCR1-IncI2 is 64,964 bp in length and harbors 83 predicted open reading frames (ORFs), with a G+C content of 42.7%. The plasmid backbone of pMCR1-IncI2 is similar to that of other IncI2 plasmids, such as pSH146_65 (GenBank accession no. JN983044)(16), pBK15692 (KC845573) (17), and pHNSHP45 (KP347127) (1). BLASTn search results showed that pMCR1-IncI2 exhibits only 86% query coverage and 97% overall identity to pHNSHP45. In addition, sequence alignment of both plasmids identified >1,400 nucleotide differences (single nucleotide polymorphisms [SNPs]), suggesting that pMCR1-IncI2 and pHNSHP45 are distinct, a finding that suggests that the mcr-1 gene may be repeatedly acquired. Further analysis showed that the mcr-1 gene in pMCR1-IncI2 integrated downstream of the nikB gene, in the same location as in pHNSHP45. Similar to the analysis of pMCR1-IncX4, the mcr-1-associated ISApl1 element was not found in pMCR1-IncI2 (Fig. 1).

Thus far, mcr-1 has been found in different plasmid incompatibility groups, including IncI2 (1, 9), X4 (6, 18), HI2 (19), and P (18). However, little is known regarding the mechanism whereby this gene can be mobilized between different plasmids. We therefore compared the mcr-1 neighboring regions of pHNSHP45, pMCR1-IncI2, and pMCR1-IncX4, as well as additional mcr-1-harboring contig sequences from the NCBI WGS database (20, 21) (Fig. 2). The comparison identified a nearly identical 2,600-bp region (nt 2349 to 4948 on pMCR1-IncX4) shared by all sequences examined, encompassing the mcr-1 and pap2 (encoding a putative PAP family transmembrane protein) genes. Our analysis suggests that the 2.6-kb mcr-1-pap2 element has been horizontally transferred into different plasmid backbones (Fig. 2). Further inspection of the upstream and downstream junctions of this mcr-1-pap2 element failed to identify any direct or inverted repeat sequences. In pHNSHP45, ISApl1 is inserted directly upstream of the mcr-1-pap2 element, and a 25-bp inverted reverse repeat is located adjacent to the 2.6-kb element. The 25-bp invert repeat was also identified at the same position in strain 2013LSAL02374, which contains a contig with a sequence from an IncP plasmid (Fig. 2). Consistent with the analysis of pMCR1-IncI2 and pMCR1-IncX4, ISApl1 was not always associated with mcr-1, and it was absent in several contigs belonging to the IncI2, IncX4, and IncP plasmids (Fig. 2). One possible explanation regarding the transfer mechanism of mcr-1 is that the mcr-1-pap2 element was initially translocated by the integration of ISApl1 (20), and the latter was subsequently lost following integration. Nevertheless, the exact mechanism underpinning mcr-1 transfer requires additional study.

FIG 2.

Genetic structure of mcr-1-harboring elements in selected plasmids. Plasmid incompatibility groups are noted on the left, along with the plasmids and/or strains harboring the genetic structure depicted. Arrows denote open reading frames (ORF), with red, gray, yellow, and white arrows denoting mcr-1, pap2, ISApl1, and neighboring genes, respectively. Following BLAST analysis of NCBI draft genomes, only contigs long enough to unambiguously identify plasmid replicon groups were included in this figure. The dashed yellow outline of the IncP ISApl1 ORF therefore indicates that only a partial sequence is available. The small black arrows adjacent to ISApl1 denote the inverted reverse repeats (IRR). The GenBank accession numbers of the mcr-1 contig sequences for EC5, EC7, EC13, 2012CEB2196SAL, 2012CEB4337SAL, 2013LSAL02374, and 2013LSAL04524 are JWKF01000084, JWKG01000081, JUJZ01000081, LKJJ01000031, LKJK01000087, LNCZ01000024, and LKJD01000010, respectively.

In addition to the two aforementioned mcr-1 plasmids, we also sequenced the two bla_NDM-5-harboring plasmids from strains SZ03 and SZ04, which were found to be identical and subsequently named pNDM5-IncX3. Plasmid pNDM5-IncX3 is 46,161 bp in length, with a G+C content of 46.7%, and harbors 58 putative ORFs. The sequence of pNDM5-IncX3 showed 100% query coverage and overall >99.9% nucleotide identity to bla_NDM-5-harboring plasmids pEc1929 (GenBank accession no. KT824791) (22) and pNDM_MGR194 (KF220657) (23), as well as bla_NDM-4-harboring pJEG027 plasmid (KM400601) (24) and bla_NDM-7-harboring plasmid pKpN01-NDM7 (CP012990) (25). Notably, IncX3 plasmids harboring different bla_NDM variants were frequently found in different hospitals among isolates of different multilocus sequence types and species in China (22, 26–28), suggesting that IncX3 plasmids are the primary type of vector responsible for the wide dissemination of NDM metallo-β-lactamases in China. Alarmingly, pNDM5-IncX3 was found to coexist with pMCR1-IncX4 within the same clinical isolate (strains SZ03 and SZ04), resulting in resistance to both colistin and carbapenems (4).

In summary, this study characterizes two mcr-1-harboring plasmids from ESBL-producing E. coli and carbapenemase-producing K. pneumoniae. The identification of the same plasmid (pMCR1-IncX4) in isolates of different species (SZ01, SZ03, and SZ04) suggests that plasmid transfer is contributing to the dissemination of mcr-1 in hospital settings in China. The potential for further spread of mcr-1-harboring plasmids within multidrug-resistant bacterial strains poses significant challenges for successful clinical treatment and infection control strategies.

Nucleotide sequence accession numbers.

The complete nucleotide sequences of plasmids pMCR1-IncI2, pMCR1-IncX4, and pNDM5-IncX3 have been deposited in GenBank under accession no. KU761326, KU761327, and KU761328, respectively.