Locally Acquired mcr-1 in Escherichia coli, Australia, 2011 and 2013

We identified discrete importation events of the mcr-1 gene on incompatibility group IncI2 plasmids in Escherichia coli isolated from patients in New South Wales, Australia, in 2011 and 2013. mcr-1 is present in a small minority of colistin-resistant Enterobacteriaceae and appears not to be established locally.


The Study
We reviewed 4,555 isolates of the family Enterobacteriaceae from 2007-2016 that were available for further testing for colistin (polymyxin E) resistance, excluding species that are intrinsically resistant. These isolates were from specimens tested at or referred to the Centre for Infectious Diseases Microbiology Laboratory Services at Westmead Hospital, Sydney, New South Wales, Australia, and were all resistant to third-generation cephalosporins, carbapenems, or both. For antimicrobial drug susceptibility testing, we used the Phoenix Automated Microbiology System (panels NMIC/ID-80, NMIC/ID-101, and NMIC-203; Becton Dickinson, Franklin Lakes, NJ, USA). Of the 4,555 isolates, 96 (2.1%) had a colistin (polymyxin E) minimum inhibitory concentration (MIC) of >2 µg/mL, which corresponds to EUCAST (1), we identified mcr-1 in 2 of the colistin-resistant E. coli isolates; these isolates were from patients in different cities in New South Wales.
In September 2011, E. coli JIE2288 (colistin MIC of 4 µg/mL) was isolated from the urine of a 70-year-old woman after she had been in the intensive care unit (ICU) at Westmead Hospital, a large metropolitan hospital, for 2 months for management of a subarachnoid hemorrhage. The woman was not administered colistin/polymyxin antimicrobial drugs during hospitalization. The E. coli JIE2288 isolate was also resistant to amikacin, gentamicin, tobramycin, amoxicillin/clavulanic acid, cefotaxime, ceftazidime, cefepime, cefoxitin, trimethoprim/sulfamethoxazole, and ciprofloxacin, according to EUCAST breakpoints. The isolate was susceptible to piperacillin/tazobactam (MIC of <4/4 µg/mL) and meropenem (MIC of <1 µg/mL). The patient had no history of overseas travel in the previous 5 years, and while she was in the ICU, no other ICU patients had a recognized colistin-resistant infection and none were receiving colistin treatment. The patient was housed in a single room, and standard infection-control precautions (disposable gown and gloves) were used. The patient was treated with a course of meropenem, and subsequent cultures of her urine did not contain colistin-resistant E. coli.
In July 2013, a second colistin-resistant E. coli strain (herein designated JIE3685; colistin MIC of >4 µg/mL) was referred to our laboratory after being isolated from the urine of a 71-year-old woman with diabetes mellitus. The woman had sought care from a community physician in Newcastle, New South Wales, for a urinary tract infection. She had had no healthcare contact in the previous 2 years and no history of travel outside Australia in the previous 5 years. JIE3685 was also resistant to cefotaxime, cefepime, ciprofloxacin, gentamicin, tobramycin, and trimethoprim/ sulfamethoxazole, but it was susceptible to amoxicillin/ clavulanic acid, ceftazidime, piperacillin/tazobactam, meropenem, and amikacin.
Most of the IncI2 plasmid from E. coli JIE2288 that carries mcr-1 (designated pJIE2288-1) was assembled as a single contig, missing only the shufflon region that rearranges to change the end of the pilV gene (8), causing assembly issues that may result in apparent differences between plasmids carrying mcr-1 (9). We obtained the shufflon segments of pJIE2288-1 by mapping raw reads to individual shufflon segments from the archetypal IncI2 plasmid R721 (GenBank accession no. AP002527) and then assembled them in the same order as in R721 to generate a complete plasmid sequence (8). An additional plasmid, carrying the bla CTX-M-14a gene (IncI1, designated pJIE2288-2), was identified in JIE2288. BLAST (https://blast.ncbi.nlm.nih.gov/ Blast.cgi) searches with 2 contigs from JIE3685 identified several closely matching plasmids (Table). We used pEC5-1, which had been sequenced by long-read methods, as a reference to resolve a repeated region and as a template for the shufflon, thus enabling complete assembly of a plasmid designated pJIE3685-1. We submitted the sequences of pJIE2288-1 and pJIE3685-1 to GenBank under accession numbers KY795977 and KY795978, respectively.
pJIE2288-1 and pJIE3685-1 each carry mcr-1 as the only resistance gene (located in the same position), and both lack ISApl1, but they do have differences in the remainder of their IncI2 backbones. pJIE2288-1 (60.733 kb) is closely related to 9 other sequenced mcr-1 plasmids (differing by only a few single nucleotide changes in backbone regions), some of which carry ISApl1 upstream of mcr-1, another IS, or both (Table). pJIE3685-1 (60.960 kb) is closely related to 6 other IncI2 mcr-1 plasmids (again, differing by only a few single nucleotide changes), all of which also lack ISApl1 (Table).
We noted differences between pJIE2288-1 and pJIE3685-1 in the mcr-1 promoter (11) and putative ribosome binding site and these or other differences in this region in available mcr-1 sequences (Figure). We examined the chromosomal arnBCADTEF, pmrAB, and phoPQ genes of each isolate but did not identify any of the changes previously reported in association with colistin resistance in E. coli (12). This finding suggests that mcr-1 may be the sole phenotype determinant in these 2 strains.  (10). ‡This information is assumed from a GenBank entry rather than from a published report. §This plasmid has many missing or extra nucleotides in homopolymeric regions compared with all other related plasmids.

Conclusions
We describe 2 E. coli isolates collected from patients in different cities in Australia ≈2 years apart that belong to STs previously associated with mcr-1 and which carry this gene on plasmids that are virtually identical to previously described mcr-1-bearing plasmids from Asia and the Middle East. Multiple importations of resistant isolates into the community microflora are to be expected, and these 2 case-patients in Australia had no history of travel outside Australia or direct links to each other. Australia's regional neighbors use colistin extensively in agriculture (13), but in Australia, its usage in agriculture and healthcare is minimal, and neither colistin nor polymyxin drugs appear in the top 20 antimicrobial drugs prescribed in Australia (14). Notwithstanding the limitations of this opportunistic study, our findings indicate that colistin resistance is unusual among Enterobacteriaceae in Australia and that mcr-1 is neither a key mechanism nor yet widely disseminated in this country despite multiple importation events.

Figure.
Differences in promoter and ribosome binding site regions of mcr-1 in plasmids from Escherichia coli in Australia (indicated by arrows) and in other sequences available from GenBank. The sequences end with the ATG start codon of mcr-1 and a second ATG codon that follows it. The −35 and −10 regions of the proposed promoter (11) are indicated by arrows. The numbers to the right indicate how many times each variant has been seen among available sequences.