Novel Plasmid-Mediated Colistin Resistance Gene mcr-3 in Escherichia coli

OBSERVATION

Since we first reported the mobile colistin resistance gene mcr-1 in China in 2016 (1), there have been reports of mcr-1 in Enterobacteriaceae isolated from animals, animal products, humans, and the environment in over 30 countries across five continents (2). More worrisome is the presence of mcr-1 in Enterobacteriaceae carrying carbapenem resistance genes, such as bla_NDM and bla_KPC, which would seriously compromise the treatment of infections caused by these extensively drug-resistant pathogens (2). Meanwhile, six variants of the mcr-1 gene have been described, including mcr-1.2 (GenBank accession no. KX236309) in KPC-3-producing Klebsiella pneumoniae isolated from a rectal swab of a leukemic child (3), mcr-1.3 (NG_052861) in Escherichia coli from chickens in China, mcr-1.4 (KY041856) in E. coli from sewage in China, mcr-1.5 (KY283125) in E. coli isolated from a human urinary tract in Argentina, mcr-1.6 (NG_052893) in Salmonella enterica serovar Typhimurium from a healthy human in China, and mcr-1.7 (KY488488) in E. coli from sewage in China. These gene variants encode phosphoethanolamine transferase enzymes but differ from MCR-1 at a single amino acid: Gln₃ to Leu in MCR-1.2, Ile₃₇ to Val in MCR-1.3, Asp₄₃₉ to Asn in MCR-1.4, His₄₅₁ to Tyr in MCR-1.5, Arg₅₃₅ to His in MCR-1.6, and Ala₂₁₄ to Thr in MCR-1.7. Moreover, a novel plasmid-borne colistin resistance gene showing 77.3% nucleotide identity and 81.0% amino acid identity to mcr-1 was also identified in porcine and bovine colistin-resistant E. coli isolates and was named mcr-2 (4). Here, we report the discovery of another novel mcr variant, mcr-3, which coexisted with 18 additional resistance genes on a conjugative plasmid from E. coli of pig origin.

In a routine surveillance study of antimicrobial resistance of bacteria from farm animals in 2015, a colistin-resistant (MIC, ≥4 µg/ml) isolate, WJ1, was obtained from a fecal sample of an apparently healthy pig at a conventional farm in Shandong Province, China. Matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) analysis (Bruker Daltonik GmbH, Bremen, Germany) and 16S rRNA sequencing identified WJ1 as E. coli, and multilocus sequence typing (2) confirmed the sequence type as 1642. Although PCR screening assays for the presence of mcr-1 and mcr-2 were negative, the colistin resistance determinant could be transferred to E. coli strain EC600 by conjugation, with transconjugant EC600-WJ1 also exhibiting a multidrug resistance profile (Table 1). A ca. 250-kb plasmid was observed in both WJ1 and EC600-WJ1 by S1 nuclease pulsed-field gel electrophoresis analysis (data not shown), suggesting that this multiresistance plasmid, designated pWJ1, carried an unknown colistin resistance gene.

pWJ1 was extracted from transconjugant EC600-WJ1 using a Plasmid Midi kit (Omega, Norcross, GA) and then sent for single-molecule real-time (SMRT) sequencing using a PacBio RSII system (Sinobiocore, Beijing, China). Plasmid assembly was performed using the hierarchical genome assembly process (HGAP) and Quiver as part of the SMRT analysis (version 2.3) using the HGAP3 protocol. Whole-cell DNA from original strain WJ1 was extracted using a Wizard genomic DNA purification kit (Promega, Beijing, China) and used as the template for whole-genome sequencing using the Illumina HiSeq 2500 system (Annoroad, Beijing, China). A draft assembly of the sequences was obtained using CLC Genomics Workbench 9 (CLC Bio-Qiagen, Aarhus, Denmark). Comparative analysis of the SMRT and HiSeq sequencing results revealed that pWJ1 was an IncHI2-type plasmid, with a size of 261,119 bp. It had a similar backbone as several mcr-1-carrying plasmids (Fig. 1A). For instance, a 222.3-kb section of pWJ1 (85.4% of the total sequence) exhibited high similarity to the corresponding region of the 251-kb plasmid pHNSHP45-2 from E. coli strain SHP45, from which the first mcr-1 gene was identified (5). pWJ1 and pHNSHP45-2 had 12 resistance genes in common (Fig. 1A), while pWJ1 appeared to have acquired six additional resistance genes (Fig. 1A; Table 1), two of which conferred resistance to antimicrobials important for human medicine, including quinolones [aac(6′)-Ib-cr] and rifampin (arr3).

• Open in new tab
• Download powerpoint

• Open in new tab
• Download powerpoint

FIG 1 

(A) BRIG analysis of the mcr-3-carrying plasmid pWJ1. Comparative analysis of pWJ1 with four closely related mcr-1-harboring plasmids from E. coli isolates using the BLAST Ring Image Generator (10). The concentric rings display similarity between the reference sequence in the inner ring and the other sequences in the outer rings. The various color levels indicate a BLAST result with a matched degree of shared regions, as shown to the right of the ring. (B) Comparison of the genetic environments of mcr-3 genes in different plasmids and shotgun sequences extracted from the GenBank database. Arrows indicate the positions and directions of the genes; Δ indicates the truncated gene. Regions with >99% homology are indicated in gray shadow, with homology of >85% shown by a lighter gray shadow. (C) Structure prediction for the mcr-3 gene product, MCR-3. Domain 1 was predicted to be a transmembrane domain, while domain 2 was predicted to be phosphoethanolamine transferase. (D) The five transmembrane α-helices predicted by the Philius transmembrane prediction server (type confidence, 0.99; topology confidence, 0.88).

Further analysis of pWJ1 revealed a 1,626-bp putative phosphoethanolamine transferase gene, designated mcr-3, which exhibited 45.0% and 47.0% nucleotide sequence identity to mcr-1 and mcr-2, respectively. To confirm the role of this putative polymyxin resistance gene, an 1,896-bp DNA fragment including mcr-3 and its upstream sequence was ligated into cloning vector pUC19, yielding pUC19-mcr-3. This recombinant vector was further transferred to E. coli W3110 by electroporation. Compared with E. coli W3110 containing pUC19 alone, an 8-fold increase in the MIC of colistin (from 0.5 µg/ml to 4 µg/ml) was observed for E. coli W3110 containing pUC19-mcr-3, suggesting that the mcr-3 product is responsible for the polymyxin resistance. We further screened colistin-resistant E. coli isolates collected from pig feces (n = 380) and chicken cloacae (n = 200) in our routine surveillance study in 2015 using the primers MCR3-F (5′-TTGGCACTGTATTTTGCATTT-3′) and MCR3-R (5′-TTAACGAAATTGGCTGGAACA-3′) and the following cycling conditions: 30 cycles of 95°C for 30 s, 50°C for 30 s, and 72°C for 45 s, followed by 1 cycle of 72°C for 7 min. The presumptive 542-bp PCR product of mcr-3 was sent for sequencing. We detected mcr-3 in 7 (1.8%) E. coli isolates from pigs only.

The deduced amino acid sequence of the mcr-3 gene product, MCR-3, showed 32.5% and 31.7% amino acid identity to MCR-1 and MCR-2, respectively. Similarly to MCR-1 and MCR-2, the MCR-3 protein was predicted to have two domains using RaptorX (Xu group, Chicago, IL) (Fig. 1C). Domain 1 (residues 1 to 172) was predicted to contain five transmembrane α-helices (Fig. 1D), while domain 2 (residues 173 to 541) was a predicted periplasmic domain containing the putative catalytic center. When using the Swiss Model server for homology modeling, the best-fit structure in Protein Data Bank for domain 2 of MCR-3 was 4KAV, which is demonstrated by phosphoethanolamine transferase LptA from Neisseria meningitidis and has been previously exhibited to be the best fit for both MCR-1 and MCR-2 (1, 4).

Compared with 28 other phosphoethanolamine transferases, including MCR-1 and MCR-2, using CLC Genomics Workbench 9, MCR-3 exhibited 99.8% to 100% amino acid identity to those found in an E. coli isolate from a pig in Malaysia, in K. pneumoniae isolates from human pus and urine samples in Thailand, and in S. enterica serovar Typhimurium from human stool in the United States (Fig. 2; see Table S1 in the supplemental material). This finding indicated that the mcr-3 gene was already present in at least three Enterobacteriaceae species in both agricultural and clinical settings in Southeast Asia and North America. MCR-3 also showed 94.1% to 94.8% amino acid identity to proteins found in three Aeromonas species, including one A. hydrophila isolate from human peritoneal fluid and one A. caviae isolate from lake water in Malaysia and one A. media isolate of unknown origin. In addition, MCR-3 aligned closely (75.6% to 84.5% amino acid identity) with MCR-3-like sequences from eight Aeromonas species derived from humans, fish, water, and wetland samples from 10 countries in Asia, Europe, and North and South America (Fig. 2; Table S1). The species A. hydrophila, A. caviae, and Aeromonas sobria were recognized as the most common Aeromonas pathogens for humans (6). Furthermore, the 801-bp fragment located immediately upstream of nimC/nimA-mcr-3 in pWJ1 exhibited 87% nucleotide sequence identity to the partial sequence of transposon TnAs2, which was identified only in Aeromonas salmonicida (7). This truncated form, ΔTnAs2, was also located upstream of mcr-3 in an E. coli isolate from pig feces from Malaysia, a K. pneumoniae isolate from a human pus sample in Thailand, and an S. enterica serovar Typhimurium isolate from human stool in the United States (Fig. 1B). These results suggested that the mcr-3 gene in Enterobacteriaceae might have originated from Aeromonas species, and the Gram-negative isolates carrying mcr-3 or mcr-3-like genes might already be widely disseminated in humans, animals, and the environment.

• Open in new tab
• Download powerpoint

FIG 2 

Phylogenetic tree of the deduced amino acid sequences of 28 putative phosphoethanolamine transferases from different bacterial species with MCR-3 using CLC Genomics Workbench 9 (CLC Bio-Qiagen, Aarhus, Denmark).

It should be noted that when using a breakpoint of ≥4 µg/ml, Aeromonas species are almost universally susceptible to colistin, except for A. jandaei and A. hydrophila. The former species seems to be intrinsically resistant to polymyxins (8), while A. hydrophila showed low-level resistance and exhibited induced resistance to colistin following preinduction with low concentrations (2.5 µg/ml) of colistin (9). However, the function of several mcr-3-like genes in different species of Aeromonas, including those isolated from human infections (Table S1), remains unknown and needs to be further investigated. Meanwhile, the possibility that these mcr-3-like genes in Aeromonas species have functions other than colistin resistance cannot be excluded.

In summary, we report the discovery of a new mobile colistin resistance gene, mcr-3, in E. coli of pig origin. Because of its resemblance to various other phosphoethanolamine transferases in Enterobacteriaceae and aeromonads identified mainly in Southeast Asia and North America in the GenBank database, this novel mobile colistin resistance gene may already be widely disseminated. Therefore, screening for the mcr-3 gene should be urgently included in the surveillance of colistin-resistant Gram-negative pathogens from animals, humans, and the environment.