Conjugal Transfer, Whole-Genome Sequencing, and Plasmid Analysis of Four mcr-1-Bearing Isolates from U.S. Patients

Four Enterobacteriaceae clinical isolates bearing mcr-1 gene-harboring plasmids were characterized. All isolates demonstrated the ability to transfer colistin resistance to Escherichia coli; plasmids were stable in conjugants after multiple passages on nonselective media.

ABSTRACT

Four Enterobacteriaceae clinical isolates bearing mcr-1 gene-harboring plasmids were characterized. All isolates demonstrated the ability to transfer colistin resistance to Escherichia coli; plasmids were stable in conjugants after multiple passages on nonselective media. mcr-1 was located on an IncX4 (n = 3) or IncN (n = 1) plasmid. The IncN plasmid harbored 13 additional antimicrobial resistance genes. Results indicate that the mcr-1-bearing plasmids in this study were highly transferable in vitro and stable in the recipients.

TEXT

Antibiotic resistance is an urgent threat to public health. Resistance to last-resort antibiotics such as colistin is concerning, particularly when found on mobile genetic elements, such as plasmids, that can easily spread horizontally. Mobile colistin resistance gene (mcr)-harboring plasmids were first reported in China in Enterobacteriaceae isolates from animals and humans in 2015 (1). The mcr-1 gene was subsequently identified in Enterobacteriaceae globally, including in an isolate from a U.S. patient (2–6). As of 2 November 2018, 55 mcr-harboring isolates, including 53 from human samples and 2 from animal sources, have been reported in the United States (7). Early reports demonstrated unusually high conjugation frequencies of the mcr-1-harboring plasmids in Escherichia coli (1, 8, 9). In this study, we sought to characterize the plasmid vectors and strain backgrounds for three E. coli isolates and one Salmonella enterica isolate harboring plasmid-mediated mcr-1 obtained by the CDC during public health responses.

Those four isolates represent all of the mcr-1 isolates available to us at the start of this project (Table 1). Isolate PA08 was identified through the Multidrug-Resistant Organism Repository and Surveillance Network at the Walter Reed Army Institute of Research. Isolate NY86 was obtained through the SENTRY antimicrobial surveillance program and isolates CT17 and CTse86 by sequencing for routine public health surveillance and response. Susceptibility to antimicrobial agents was determined for four clinical isolates and their conjugants by broth microdilution according to CLSI guidelines (10). Colistin resistance was defined by determination of epidemiological cutoff values of ≥4 μg/ml. We completed experiments mating mcr-1-positive isolates with the recipient E. coli J53 strain (NaN₃ resistant) by combining mid-log-phase cultures of the donor and recipient at a 1:10 donor-to-recipient ratio. Transconjugants were selected on plates with 1 μg of colistin/ml and 200 μg of NaN₃/ml. Conjugation frequency was expressed as conjugant CFU count per recipient CFU count. The stability of colistin resistance in conjugant and parent strains was evaluated by subculturing one colony of each onto nonselective and selective plates (1 μg/ml colistin), in parallel.

TABLE 1.

Characteristics of isolates harboring mcr-1-bearing plasmids^a

Isolate	State	Source	Species	Strain type (MLST)	Total no. of plasmids	mcr-1 plasmid (size in bp)	% GC (chromosome/ mcr-1 plasmid	mcr-1 plasmid incompatibility group	AR gene(s) identified on mcr-1 plasmid	Conjugation frequency (conjugant CFUs/ recipient CFUs
CT17	CT	Stool	E. coli	2732	5	pMCR-1-CT (33,304)	50.5/41.8	IncX4	mcr-1	1.5 × 10−2
CTse86	CT	Stool	S. enterica	11	2	pMCR-1-CTse (33,304)	52.2/41.8	IncX4	mcr-1	3.3 × 10−2
NY68	NY	Bile	E. coli	117	7	pMCR-1-NY (33,304)	50.8/41.8	IncX4	mcr-1	3.0 × 10−2
PA08	PA	Rectal swab	E. coli	457	1	pMCR-1-PA (276,880)	50.5/49.1	IncN	mcr-1 aac(3)-Iva, aph4-la, mphA, blaCTX-M-55, blaCTX-M-14, tetR, dfrA5, fosA, blaTEM-1, floR, strA, strB, sul2	3.9 × 10−8

^aAR, antimicrobial resistance; CT, Connecticut; NY, New York; PA, Pennsylvania.

Whole-genome sequencing (WGS) was performed with a MiSeq system (Illumina, San Diego, CA) and PacBio platform (Pacific Biosciences, Menlo Park, CA). Long-read sequence data were polished using Quiver (11) and corrected by pilon (12) utilizing Illumina reads. The PlasmidFinder database (13) was used for in silico detection and characterization of plasmid replicon sequences among the PacBio contigs. Insertion sequences were predicted using ISfinder (14).

All four isolates demonstrated elevated colistin MICs (4 to 8 μg/ml) and polymyxin B MICs (2 to 8 μg/ml) (see Table S1 in the supplemental material). All three E. coli isolates were multidrug resistant and had similar susceptibility patterns (Table S1). The S. enterica isolate was susceptible to 22 of 26 antimicrobial agents tested and showed intermediate resistance to cefazolin, ciprofloxacin, and cefoxitin. The transfer frequencies (conjugant/recipient CFU) for isolates CT17, CTse86, and NY68 were approximately 1.5 × 10⁻², 3.3 × 10⁻², and 3 × 10⁻², respectively; the PA08 isolate had a lower transfer frequency (3.9 × 10⁻⁸) (Table 1).

The conjugants were confirmed positive for mcr-1 by PCR. Analysis of HindIII-digested plasmid DNA from three randomly selected conjugants demonstrated restriction profiles similar to those seen with the parent strains (data not shown). Susceptibility testing demonstrated that all conjugants had elevated colistin and polymyxin B MICs (Table S1). The conjugants from CT17 and NY68 displayed greater susceptibility to the third-generation cephalosporins, such as cefotaxime, ceftazidime, and ceftriaxone, than the parent strains, while the PA08 conjugant from PA08 remained resistant. All conjugants and parent strains tested retained elevated colistin and polymyxin B MICs after 18 passages.

Each isolate belonged to a different sequence type (ST) (Table 1). PacBio sequencing confirmed that mcr-1 was located on a plasmid for all four isolates. The plasmids harboring mcr-1 from CT17 (pMCR-1-CT), CTse86 (pMCR-1-CTse), and NY68 (pMCR-1-NY) were all 33,304 bp, belonged to the IncX4 plasmid group, and harbored no additional antibiotic resistance genes. pMCR-1-CT and pMCR-1-CTse were identical; pMCR-1-NY differed from those plasmids by only two single nucleotide polymorphisms (SNPs). The PA08 isolate carried one plasmid (pMCR-1-PA) which was 276,880 bp in size and harbored additional 13 antibiotic resistance genes (Table 1).

With regard to mobility, the type IV secretion system (T4SS) virB gene cluster was identified in pMCR-1-CT, pMCR-1-CTse, and pMCR-1-NY whereas conjugation transfer genes (tra cluster) were identified in pMCR-1-PA. Three IncX4 plasmids carried insertion sequence IS26, while pMCR-1-PA carried insertion sequence ISApl1 and multiple IS26 copies. Potential toxin/antitoxin systems HipAB and HicAB, which function to ensure plasmid success in daughter cells through postsegregational killing, were identified in pMCR-1-PA and in all three IncX4 plasmids, respectively.

A NCBI blastn search using the three IncX4 plasmids found that they were almost identical (100% coverage and 99.9% nucleotide identity) to previously reported mcr-1-harboring IncX4 plasmids isolated from multiple pathogens in multiple countries (GenBank accession numbers KU743383, KU761327, CP015977, KX236309, and KX447768). In contrast, pMCR-1-PA shared only some homology with its best match, pSLK172-1 in an E. coli strain from China (NZ_CP017632; 59% coverage and 99% nucleotide identity).

Our data demonstrated that pMCR-1-CT, pMCR-1-CTse, and pMCR-1-NY were efficiently mobilized to the E. coli recipient strain in a manner consistent with previous findings of high conjugation frequency (10⁻¹ to 10⁻³) between E. coli strains (1, 8). However, the pMCR-1-PA plasmid had a lower conjugation frequency, possibly because of the difference in plasmid replicon types and/or because of its larger size.

Stability assays and WGS data demonstrated the pMCR-1-CT, pMCR-1-CTse, and pMCR-1-NY plasmids were stable and were nearly identical to each other as well as to other mcr-1-harboring IncX4 plasmids reported from multiple countries (15–19). The stability of the plasmids investigated here and the evidence of their international dissemination further support the idea that IncX4 plasmids may play a significant role in the spread of mcr-1 among Enterobacteriaceae (20–21). However, this work was not designed to identify reservoirs, transmission routes, or specific molecular mechanisms facilitating dissemination of strains carrying mcr-1-bearing IncX4 plasmids. Several studies have indicated that IncX4 is one of the most prevalent plasmid types found in E. coli. The IncX4 plasmid is self-transferable at frequencies (10⁻¹ to 10⁻⁴) that are higher than those seen with other plasmid types, such as IncFII plasmid (10⁻³ to 10⁻⁷) (22–23).

The pMCR-1-PA plasmid (IncN; 276,880 bp) differed from other mcr-1-bearing plasmids characterized in this study in its size and replicon type (Table 1; see also Fig. S1 in the supplemental material). In comparison to the three IncX4 plasmids, which carried only one antibiotic resistance gene (mcr-1), the pMCR-1-PA plasmid harbored 13 additional antibiotic resistance genes. pMCR-1-PA plasmid in this study is larger than the previously reported pMR0516mcr plasmid (225,707 bp; KX276657) (5) found in an E. coli isolate from the same patient. These two plasmids share complete homology except for a 50-kb segment (Fig. S1) not found in the original pMR0516mcr plasmid. The original report indicated that the isolate harbored two plasmids, pMR0516mcr and a second (47-kb) plasmid (5). Our analysis suggests that the two plasmids might represent a single plasmid (pMCR-1-PA) based on PacBio and Illumina sequence data as well as on plasmid digestion profiles (data not shown).

Both CT isolates were from patients who had visited the Dominican Republic (DR) shortly before their positive cultures (6 and data not shown). NY68 was isolated from a patient who also had traveled to the DR prior to the positive culture, although the travel took place in 2013 and the specimen was collected in 2015 (24). Thus, three patients had connections to the DR and the corresponding plasmids were highly similar, even though the organisms differed. Hence, the results seen with those patients may represent travel-associated mcr-1 acquisition. The PA patient did not have a history of travel to DR or an epidemiologic link to the other case patients. However, this patient had an extensive history of health care exposure, which aligns with the finding that the mcr-harboring plasmid from this patient’s isolate carried 13 additional antibiotic resistance genes, whereas the mcr plasmid seen in the cases with travel history did not. Therefore, health care acquisition was more likely for the PA isolate, whereas community acquisition was more likely for the other cases.

mcr-1 plasmid-mediated colistin resistance in the United States is an important public health concern because of the potential for horizontal transmission of the colistin resistance mechanism through a variety of plasmid types associated with diverse bacterial genera. We describe the mobility and genetic organization of four mcr-1-bearing plasmids isolated from U.S. patients. Continued surveillance is needed, not only to understand the diversity of these isolates and the mcr-harboring plasmids but also to implement rapid detection of this mechanism as part of the containment strategy to prevent transmission and spread of antimicrobial resistance.

Data availability.

Whole-genome sequences for strains CT17 (2016C-3936C1), CTse86 (2017K-0021), NY68 (MDR-56), and PA08 (2016C-3878) can be found under BioSample accession numbers SAMN06159501, SAMN06183523, SAMN06344815, and SAMN05468123, respectively. pMCR-1-CT, pMCR-1-CTSe, pMCR-1-NY, and pMCR-1-PA have been deposited in GenBank under accession numbers CP018773, CP030795, CP019908.1, and CP029748.