Clonal Spread of Escherichia coli ST93 Carrying mcr-1-Harboring IncN1-IncHI2/ST3 Plasmid Among Companion Animals, China

The purpose of this study was to investigate the occurrence of plasmid-mediated colistin resistance gene mcr-1 in Enterobacteriaceae isolates from companion animals in Guangzhou, China. Enterobacteriaceae isolated from 180 samples collected from cats and dogs were screened for mcr-1 by PCR and sequencing. MCR-1-producing isolates were further characterized by multilocus sequence typing and pulsed-field gel electrophoresis (PFGE). Plasmid characterization was performed by conjugation, replicon typing, S1-PFGE, and Southern blot hybridization. Plasmid pHN6DS2 as a representative IncN1-IncHI2/ST3 plasmid from ST93 E. coli was fully sequenced. pHN6DS2-like plasmids were screened by PCR-mapping and sequencing. The mcr-1 gene was detected in 6.25% (8/128) Escherichia coli isolates, of which, five belonged to E. coli ST93 and had identical PFGE patterns, resistance profiles and resistance genes. mcr-1 genes were located on ∼244.4 kb plasmids (n = 6), ∼70 kb plasmids, and ∼60 kb plasmids, respectively. Among them, five mcr-1-carrying plasmids were successfully transferred to recipient by conjugation experiments, and were classified as IncN1-IncHI2/ST3 (∼244.4 kb, n = 4, all obtained from E. coli ST93), and IncI2 (∼70 kb, n = 1), respectively. Plasmid pHN6DS2 contained a typical IncHI2-type backbone, with IncN1 segment (ΔrepA-Iterons I-gshB-ΔIS1294) inserted into the multiresistance region, and was similar to other mcr-1-carrying IncHI2/ST3 plasmids from Enterobacteriaceae isolates of various origins in China. The remaining five mcr-1-bearing plasmids with sizes of ∼244.4 kb were identified to be pHN6DS2-like plasmids. In conclusion, clonal spread of ST93 E. coli isolates was occurred in companion animals in Guangzhou, China.


INTRODUCTION
Colistin has been a last-resort treatment option in human medicine for infections caused by multiresistant Gram-negative bacteria (Kaye et al., 2016). Resistance to colistin had only been involved with chromosomal mutations until the identification of plasmid-mediated colistin resistance gene mcr-1 from a porcine Escherichia coli isolate in China in 2015 . The emergence and dissemination of mcr-1 is a significant global concern and poses a serious threat to clinical treatment. Since the discovery of mcr-1, it has been increasingly reported in Enterobacteriaceae from humans, animals, food products, and the environment worldwide, particularly in China (Jeannot et al., 2017;Wang et al., 2018). To date, mcr-1 has been identified on various plasmid types, with IncI2, IncHI2, and IncX4 being the major carriers, and ISApl1 is involved in mcr-1 mobilization between DNA molecules (e.g., plasmid, chromosome) (Matamoros et al., 2017;Li et al., 2018;Wang et al., 2018). Previous studies have demonstrated high mcr-1 prevalence in E. coli isolates from food-producing animals (pigs and chickens) and meat (pork and chicken meat) in China Wu et al., 2018). However, mcr-1 has been rarely reported in companion animals, though 8.7% Enterobacteriaceae isolates were identified to carry mcr-1 among companion animals in Beijing, China, meanwhile, mcr-1positive E. coli isolates may transfer between companion animals and close contactors in a pet store in Guangzhou, China (Sun et al., 2016;Zhang et al., 2016;Lei et al., 2017). Thus, in this study, we investigated the prevalence and characterization of mcr-1 in Enterobacteriaceae isolates from companion animals in Guangzhou, China, to provide insights into the spread of mcr-1 in companion animals.

Sample Collection and mcr-1 Detection
During July to August 2016, 180 samples were collected from cats and dogs at four animal hospitals located in four districts in Guangzhou, China, including 68 feces samples from healthy animals, 112 samples (feces, urine, eye secretion, ear exudates, nasal secretion, and skin) from diseased animals ( Table 1). Samples were incubated in LB broth for 16∼24 h and then inoculated on the MacConkey agar. One isolate per sample was selected and identified by MALDI-TOF mass spectrometry or 16S rRNA sequencing (Supplementary Table S1). In all isolates, mcr-1 was detected by PCR and sequencing (Supplementary Table S1).

Molecular Typing
The genetic diversity of mcr-1-positive E. coli isolates was characterized by multilocus sequence typing (MLST) 1 . Five mcr-1-carrying ST93 E. coli isolates in this study and the mcr-1positive ST93 E. coli strain PET01, that was previously obtained from a cat in Guangzhou, China  were further analyzed by pulsed-field gel electrophoresis (PFGE) (Gautom, 1997).

Conjugation/Transformation Experiments and Plasmid Characterization
Conjugation experiments were carried out using streptomycinresistant E. coli C600 as the recipient strain as previously described (Chen et al., 2007). Transconjugants were selected using 2 mg/L colistin and 3,000 mg/L streptomycin. Transfer frequencies were calculated as the number of 1 http://enterobase.warwick.ac.uk/species/index/ecoli transconjugants per recipient, experiments were performed in triplicate. Transformation was conducted by heat-shock and electroporation using E. coli strain DH5α as the recipient strain, and selected by 2 mg/L colistin. The presence of mcr-1 in the transconjugants was confirmed by PCR and sequencing. Transconjugants with a single mcr-1-carrying plasmid, verified by S1-PFGE (Barton et al., 1995) and Southern blot hybridization, were selected for further study. The location of mcr-1 in the original isolates which failed to obtain transconjugants/transformants was determined by S1-PFGE and Southern blot hybridization. All the transconjugants were characterized by PCR-based replicon typing and IncI2 and IncX4 plasmids were screened according to previously described protocols (Carattoli et al., 2005;Johnson et al., 2012;Lv et al., 2013). IncHI2 plasmids were further characterized by plasmid double locus sequence typing (García-Fernández and Carattoli, 2010). The genetic structure of mcr-1 was determined by PCR mapping and sequencing in five transconjugants and three original isolates without transconjugants/transformants (Supplementary Table S2).

Antimicrobial Susceptibility Testing
The original mcr-1-positive E. coli isolates, the recipient strain C600, and transconjugants were tested for their susceptibility to ampicillin, cefotaxime, imipenem, gentamycin, amikacin, tetracycline, chloramphenicol, florfenicol, ciprofloxacin, sulfamethoxazole/trimethoprim, colistin, and fosfomycin by the agar dilution method or the broth microdilution method (limited to colistin). Antimicrobial susceptibility tests were performed and interpreted according to M100, 28th edition of the CLSI (Wayne, PA, United States) (Clinical Laboratory Standards Institute [CLSI], 2018). Colistin ( > 2 mg/L), and florfenicol ( > 16 mg/L) were interpreted according to the clinical breakpoints or epidemiological cutoff values of EUCAST. 2 The E. coli strain ATCC 25922 was used for quality control. The mutations within gyrA and parC were detected in ciprofloxacin-resistant mcr-1-positive E. coli isolates (Supplementary Table S1). Other resistance genes, including bla CTX−M , floR, rmtB, oqxAB, and fosA3 were screened in original mcr-1-positive isolates and their transconjugants using the primers listed in Supplementary Table S1.

Plasmid Sequencing
Plasmid pHN6DS2, as a representative IncN1-IncHI2/ST3 plasmid from ST93 E. coli isolate, was selected to extract from the transconjugant using QIAGEN R Plasmid Midi Kit (Qiagen, Hilden, Germany) and sequenced by Illumina Miseq technology (Illumina, San Diego, CA, United States). Sequence reads were assembled into contigs with SOAPdenovo version 2.04. Nine contigs of pHN6DS2 were assembled into the complete plasmid sequence with PCR amplification and Sanger sequencing (Supplementary Table S3) using related mcr-1-carrying plasmids as references by BLAST 3 . Analysis and annotation of plasmid pHN6DS2 were performed using the RAST server (Aziz et al.,

Nucleotide Sequence Accession Number
The nucleotide sequences of plasmid pHN6DS2 has been deposited in the GenBank database under the accession number MH459020.
Since colistin is not applied to companion animals in China, pet food containing chicken meat might be one source of mcr-1    (Lei et al., 2017). Close contact to food-producing animals in local hog and poultry markets, as well as to humans, might also be the potential origins. Furthermore, the widely use of cephalosporins, aminoglycosides, and fluoroquinolones in companion animal medicine (data not shown) could allow for the co-selection of isolates harboring mcr-1, as well as bla CTX−M and fosA3, conferring resistance to crucial clinical antibiotics.

Molecular Typing
Eight mcr-1-positive E. coli isolates were assigned to ST93 (n = 5), ST1011, ST3285, and a new ST, respectively ( Table 2). ST93 has been sporadically described as avian and human extra-intestinal pathogenic or diarrhoeagenic E. coli in humans, animals, and food products worldwide (Chen et al., 2014;Maluta et al., 2014;Vogt et al., 2014), and particularly it has been previously detected as mcr-1 carriers from a pig in Laos (Olaitan et al., 2015(Olaitan et al., , 2016, from a cat in Guangzhou, China , and from a patient in Finland (Gröndahl-Yli-Hannuksela et al., 2018). The five mcr-1-carrying ST93 E. coli isolates were obtained from both intestinal and extraintestinal sites from two animal hospitals located within a distance of 7 km. They showed indistinguishable PFGE patterns which differed from previously described mcr-1-harboring ST93 E. coli isolate PET01 from a cat in Guangzhou ; Figure 1), indicating that clonal spread of mcr-1-harboring E. coli had occurred among companion animals within two hospitals in Guangzhou. The observation that they had identical antimicrobial susceptibility profiles, resistance genes, and mutations within gyrA and parC may further support this hypothesis ( Table 2). However, small numbers of samples were collected from four animal hospitals in this study, thus limiting this hypothesis. The prevalence and dissemination mechanisms of mcr-1 in companion animals in Guangzhou should be further investigated by using large scale samples from more animal hospitals. Though rare, it is possible for these two hospitals to exchange animal patients. The possibility of acquisition of mcr-1-harboring ST93 E. coli from a common ancestor could not be ruled out. Although horizontal transfer mediated by mobile elements such as insertion sequence and plasmids has been the major reason for mcr-1 worldwide dissemination, clonal spread of mcr-1-harboring strains, such as E. coli ST93 in the present study, Salmonella Typhimurium ST34 in pigs Yi et al., 2017) might be another reason accounting for mcr-1 transmission. Most importantly, the potential of mcr-1 transmission mediated by MCR-1-producing clones from companion animals to humans through close contact should not be underestimated, which might have already occurred in China by E. coli ST354 and ST101 clones Lei et al., 2017).

Characterization of mcr-1-Carrying Plasmids
Five strains successfully transferred mcr-1 to E. coli C600 at frequencies of 10 −2 to 10 −4 transconjugants/recipient (Supplementary Table S5), and the remaining three strains failed to transfer mcr-1 to E. coli C600 or DH5α by conjugation or transformation. S1-PFGE and Southern hybridization indicated that mcr-1 was located on ∼244.4 kb plasmids (n = 6), ∼60 kb plasmids (n = 1), or ∼70 kb plasmid (n = 1) ( Table 2). Additionally, five mcr-1-harboring transconjugants with single plasmid were classified as IncN1-IncHI2/ST3 (∼244.4 kb, n = 4) which were all obtained from ST93 E. coli isolates, and IncI2 (∼70 kb, n = 1) ( Table 2 and Supplementary Figure S1), which agree with previous observation that IncHI2 and IncI2 plasmids have been the major vectors for mcr-1 global dissemination (Matamoros et al., 2017;Wang et al., 2018). Furthermore, the transconjugants showed elevated MICs for colistin (1∼2 mg/L; 8-16-fold) compared with the recipient E. coli C600. In addition, co-transfer of resistance to ampicillin, cefotaxime, gentamycin, chloramphenicol, florfenciol, sulfamethoxazole/trimethoprim, and fosfomycin was observed in four transconjugants with IncN1-IncHI2/ST3 plasmid from ST93 E. coli isolates, resistance genes bla CTX−M−14 , floR and fosA3 were also co-transferred with mcr-1 ( Table 2). The co-transfer of bla CTX−M−64 with mcr-1 on an IncI2 plasmid in the remaining transconjugant caused resistance to ampicillin and cefotaxime ( Table 2). The presence of other resistance genes co-located on the same plasmid allows for the selection of mcr-1 under pressure posed by other agents, thus facilitating mcr-1 transmission. It has been hypothesized that mcr-1 is initially captured and mobilized by the composite transposon Tn6330 (ISApl1-mcr-1-pap2-ISApl1), followed by the loss of ISApl1 over time, leading to the formation of mcr-1 in diverse genetic structures, with the structure mcr-1-pap2 being dominant, followed by the structure ISApl1-mcr-1-pap2 (Snesrud et al., 2018;Wang et al., 2018). The genetic structure of mcr-1 in our study was determined by PCR mapping. We did not observe the complete Tn6330, but the presence of ISApl1 upstream was common, identified in six transconjugants or original isolates with ∼244.4 kb mcr-1carrying plasmids, the structure mcr-1-pap2 was also identified (n = 2) ( Table 2). Our results further support that mobile elements (ISApl1, IncHI2, and IncI2 plasmids) play an important role in the mobilization and dissemination of mcr-1 in E. coli from different sources.

Plasmid Sequencing and Comparative Analysis
Plasmid pHN6DS2 had a size of 253, 783 bp, and was organized similarly to other IncHI2 plasmids, containing regions for functions of replication, multiresistance, conjugal transfer, maintenance, and stability (Supplementary Figure S2).
These data suggested that similar mcr-1-carrying IncHI2/ST3 plasmids, after acquiring, losing or reorganizing various regions, could spread among Enterobacteriaceae species in livestock, humans, vegetables, and the environment, particularly in different regions in China. The presence of pHN6DS2-like plasmids further supported this hypothesis and highlighted the potential of pHN6DS2-like plasmid to become an efficient vehicle for mcr-1 dissemination between distinct organisms or regions.

CONCLUSION
In conclusion, the spread of mcr-1 in companion animals in the present study might be mainly attributed to clonal dissemination of E. coli ST93 isolates within two hospitals, associated with IncN1-IncHI2/ST3 plasmids. Although the origin of mcr-1 in companion animals is unclear, it is possible for mcr-1-positive clones or plasmids to transfer from companion animals to humans through close contact, thus the dissemination of mcr-1 among companion animals needs continued vigilance.

ETHICS STATEMENT
This study was carried out in accordance with the recommendation of ethical guidelines of South China Agricultural University. Individual written informed consent for the use of samples was obtained from all the animal owners.