First Report of blaIMP-14 on a Plasmid Harboring Multiple Drug Resistance Genes in Escherichia coli Sequence Type 131

The blaIMP-14 carbapenem resistance gene has largely previously been observed in Pseudomonas aeruginosa and Acinetobacter spp. As part of global surveillance and sequencing of carbapenem-resistant Escherichia coli, we identified a sequence type 131 strain harboring blaIMP-14 within a class 1 integron, itself nested within an ∼54-kb multidrug resistance region on an epidemic IncA/C2 plasmid. The emergence of blaIMP-14 in this context in the ST131 lineage is of potential clinical concern.

T he emergence of carbapenemases in clinically prevalent Escherichia coli lineages such as sequence type 131 (ST131) is a major problem for the management of patients infected with these strains (1,2). Globally, five major transmissible carbapenemase enzymes predominate, represented by the KPC, OXA-48-like, NDM, VIM, and IMP families (2,3).
An IMP metallo-beta-lactamase enzyme (IMP-1) was first detected in Japan in Pseudomonas aeruginosa in the late 1980s (4); since then, 52 genetically diverse bla IMP gene variants 738 to 747 bp in length have been identified (5). Most bla IMP variants have been isolated from either Pseudomonas or Acinetobacter spp. and demonstrate a degree of geographic structuring (6); however, some, such as bla IMP-4 and bla IMP-8 , have emerged successfully in members of the family Enterobacteriaceae and are distributed over wider geographic regions (6,7). Associations of bla IMP with E. coli ST131 have, to date, been restricted to bla IMP-4 and bla IMP-8 in Taiwan, China, and Australia (8)(9)(10)(11). As part of the Merck Study for Monitoring Antimicrobial Resistance Trends (SMART) (1), we identified an IMP-14-producing ST131 E. coli isolate, Ecol_732, that was isolated in Bangkok, Thailand, in 2012 and was sequenced in order to ascertain the genetic structures associated with this IMP variant in ST131.
Ecoli_732 was obtained from the urine of a hospitalized elderly male with a lower urinary tract infection. The MICs of ampicillinsulbactam, piperacillin-tazobactam, cefoxitin, ceftriaxone, ceftazidime, cefepime, ertapenem, imipenem, amikacin, and ciprofloxacin were determined with microdilution panels prepared at International Health Management Associates, Inc. (Schaumburg, IL, USA), in accordance with 2015 CLSI guidelines. It tested nonsusceptible (i.e., either intermediate or resistant) to the abovementioned agents. The MICs of colistin and tigecycline (determined by E tests) were 0.12 and 1 mg/liter, respectively. DNA (chromosomal plus plasmid) was extracted from pure overnight subcultures of the isolate for both PacBio (long-read) sequencing and Illumina MiSeq (short-read) sequencing with the Qiagen Genomic-tip 100/G kit and the QIAamp DNA minikit (catalogue numbers 10243 and 51304; Qiagen, Valencia, CA), respectively. Preliminary de novo assembly of PacBio reads with HGAP3 was performed; resulting contigs were annotated with Prokka (12) and then trimmed on the basis of sequence/anno-tation overlaps in Geneious (version 9.04) (13). One-hundredfifty-base paired-end MiSeq reads for each of the isolates were trimmed with Trimmomatic (version 0.35) (14) and then mapped to the corresponding PacBio assemblies with BWA mem (version 0.7.9a-r786) (15). Read pileups were inspected to confirm the structural integrity of the contigs and correct any small errors in the assembled contigs. Unmapped MiSeq reads were assembled with A5MiSeq (16) in order to identify any small plasmids (Ͻ7 kb) that may have been filtered out during the size selection process implemented as part of PacBio library preparation. Additional annotation focused on resistance genes and insertion sequences was performed with reference to the ResFinder (17), PlasmidFinder (18), and ISFinder (19) databases.
An alignment of pEC732-IMP14, prototype IncA/C 2 type 1 plasmid pRMH760, and the only publicly available type 1 IncA/C 2 sequence from Thailand, pR148 (RefSeq database accession no. NC_019380, from Aeromonas hydrophila [21]), demonstrates the genetic similarity of these plasmids (Fig. 2). All three sequences were Ͼ99% similar in the 1-to 86,573-bp region and in the ϳ27.5-kb region downstream of ARI-A (Fig. 2). Differences in pEC732-IMP14 include a region of clustered single-nucleotide variants suggestive of a recombination event (region, 3,100 to 8,000 kb) and the acquisition of two integrase subunits (regions, 86,573 to 89,203 and 90,138 to 200,167 bp; Fig. 2). Interestingly, the pR148-containing A. hydrophila strain was identified on a Thai tilapia fish farm that had successively used several antimicrobial classes (21).
To date, bla IMP-14 has not been described in E. coli, to our knowledge, and has largely previously been reported in P. aeruginosa and Acinetobacter baumannii strains by several hospital centers in Thailand, in some cases as part of clonal outbreaks (22)(23)(24)(25). Although bla IMP-14 is similarly associated with class 1 integrons in these cases, as in pEC732-IMP14, the wider plasmid contexts and sequences of these integrons in P. aeruginosa and A. baumannii strains have not been investigated. It is, however, conceivable that the bla IMP-14 -harboring integron observed in pEC732-IMP14 and A. xylosoxidans strain R4 has been exchanged more widely with Pseudomonas and Acinetobacter spp. in Thailand. Class 1 integrons have been linked with the recent successful spread and expansion of another metallo-beta-lactamase, bla VIM , in IncA/C 2 plasmids in members of the family Enterobacteriaceae in Greece (26) and bla VIM and bla IMP in Spain (27).
The presence of the extensively drug-resistant region observed here on an epidemic IncA/C 2 plasmid in an E. coli ST131 strain from Thailand is therefore of concern and may represent wider, regional, horizontal dissemination of bla IMP-14 mediated by mobile genetic elements across bacterial families. The homology of pEC732-IMP14 with an A. hydrophila plasmid found on a fish farm and the presence of bla IMP -harboring plasmids in E. coli in other environmental (28) and animal sampling frames (29) suggest that the transmission network for IMP-positive E. coli may extend beyond the health care setting. Broad surveillance and control measures that are targeted at both community and health care contexts may be required to monitor and limit bla IMP dissemination.
Nucleotide sequence accession numbers. Complete sequence data for Ecol_732 have been deposited in GenBank under Bio-Project number PRJNA316786. The accession numbers of the sequences are CP015138 (chromosome), CP015139 (pEC732-IMP14), CP015140 (pEC732_2), CP015141 (pEC732_3), CP015142 (pEC732_4), CP015143 (pEC732_5), and CP015144 (pEC732_6 [partial sequence only]). The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. The views expressed are those of the authors and not necessarily those of the National Health Service, the NIHR, or the Department of Health.

FUNDING INFORMATION
This work, including the efforts of Nicole Stoesser, Anna E. Sheppard, Luke W. Anson, and Derrick W. Crook, was funded by The Wellcome Trust (099423/Z/12/Z and WT098615/Z/12/Z). This work, including the efforts of Nicole Stoesser, Anna E. Sheppard, Luke W. Anson, and Derrick W. Crook, was funded by The United Kingdom Department of Health (DH) (HICF-T5-358). This work, including the efforts of Gisele Peirano and Johann D. Pitout, was funded by Calgary Laboratory Services (CLS) (10006465).