The Microbiological Characteristics of Carbapenem-Resistant Enterobacteriaceae Carrying the mcr-1 Gene

Objectives: This study aims to assess the prevalence of the mcr-1 gene among carbapenem-resistant Enterobacteriaceae (CRE) isolated from clinical specimens and to further investigate the clinical significance and microbiological characteristics of CRE carrying the mcr-1 gene. Methods: Four hundred and twenty-three CRE isolates were screened for the presence of the mcr-1 gene. After identification, their clinical significance, antibiotic susceptibility, and antibiotic resistance mechanisms including the ESBL gene, carbapenemase gene, outer membrane protein (OMP), and plasmid sequencing were assessed. Results: Only four (0.9%) isolates of carbapenem-resistant Escherichia coli (E. coli) were found to carry the mcr-1 gene and demonstrated different pulsed-field gel electrophoresis (PFGE) patterns and sequence types (ST). While one patient was considered as having mcr-1-positive carbapenem-resistant E. coli (CREC) colonization, the other three mcr-1-positive CREC-related infections were classified as nosocomial infections. Only amikacin and tigecycline showed good in vitro activity against these four isolates, and three of them had a minimum inhibitory concentration with colistin of ≥4 mg/L. In the colistin-susceptible isolate, mcr-1 was nonfunctional due to the insertion of another gene. In addition, all of the mcr-1-positive CREC contained various resistant genes, such as AmpCCMY, blaNDM, blaTEM, blaSHV, and blaCTX. In addition, one strain (EC1037) had loss of the OMP. Conclusions: The emergence of the mcr-1 gene among CRE, especially E. coli, remains worth our attention due to its resistance to most antibiotics, and a further national survey is warranted.


Introduction
In the era of increasing carbapenem resistance among Enterobacteriaceae, the treatment options against these drug-resistant pathogens has become limited. Currently, the mainstays of pharmacotherapy against carbapenem-resistant Enterobacteriaceae (CRE)-associated infections only include tigecycline, aminoglycosides, and several old antibiotics such as fosfomycin and colistin. mcr-1 is the first plasmid-mediated colistin resistance gene that can be transferred horizontally via plasmids [1]. Since the first identified Enterobacteriaceae carrying a plasmid-encoded colistin-resistance gene mcr-1 in China [2], more and more sites, including Europe [3,4], Canada [5], Vietnam [6], Hong Kong [7], and Taiwan [8], have reported the presence of mcr-1-positive isolates from both animals and humans. The mcr-1-positive Escherichia coli (E. coli) infections are found to be associated with male sex, immunosuppression, and the use of antibiotics, especially for fluoroquinolone and carbapenems. However, most mcr-1 associated studies are based on the surveillance of colistin-resistant bacteria, and the epidemiology of the mcr-1 gene among CRE isolates remains unknown [9]. As colistin is one of the limited therapeutic options against CRE, clinicians should be seriously concerned about the colistin resistance mediated by the mcr-1 gene. Therefore, we conducted this investigation to assess the prevalence of the mcr-1 gene among CRE isolated from clinical specimens and further investigate the microbiological characteristics of CRE carrying the mcr-1 gene.

Bacterial Isolates
This study was conducted at two medical centers, the Chi Mei Medical Center and the National Cheng Kung University Hospital in southern Taiwan. The bacterial species were confirmed using a VITEK 2 automated system (bioMérieux, Marcy l'Etoile, France) with a VITEK ® 2 GN ID card. These isolates were stored at −80 • C in Protect Bacterial Preservers (Technical Service Consultants Limited, Heywood, UK) before investigation. CRE was defined as resistance to any of four carbapenems (ertapenem, imipenem, doripenem, meropenem). All of the clinical specimens positive for CRE between April 2014 and June 2017 were screened for the presence of the mcr-1 gene as previously described [10].

Pulsed-Field Gel Electrophoresis (PFGE)
PFGE was performed with a CHEF DR II apparatus (Bio-Rad Laboratories, Hercules, CA, USA) as previously reported [19]. Briefly, DNA was digested by XbaI, and electrophoresis was performed in a 1% agarose gel. The bacteriophage lambda ladder pulsed-field grade (PFG) and low-range PFG molecular weight markers were loaded onto all gels. The similarities of the PFGE profiles of each strain were compared using a Dice coefficient at 1.0% of tolerance and 1.0% of optimization.

Southern Blotting of mcr-1
Southern blotting was performed using a semidry transfer system (Bio-Rad) and the mcr-1-containing plasmids were identified by hybridization with Dig-labeled mcr-1-specific probe generated by the PCR DIG Probe Synthesis Kit, and Detection Starter Kit II (Roche Applied Sciences, Mannheim, Germany) [21].

Plasmid Sequencing
Bacterial pellets in centrifugation tubes were resuspended in buffer. The cell wall was removed by enzymatic digestion in the presence of RNase. Cell lysis and chromosome removal were achieved by alkaline lysis, followed by acid aggregation and centrifugation. DNA in the supernatant was extracted using organic solvent and recovered by ethanol precipitation. Concentration of samples was determined by fluorescence quantification. Purified DNA was analyzed by 0.4% agarose gel electrophoresis. The Illumina MiSeq System (Illumina, San Diego, CA, USA) was used for plasmid sequencing. The derived reads were assembled using the CLC Genomics Workbench 5.51 (CLC bio, Aarhus, Denmark) [22].

Clinical Significance
Among 423 CRE isolates, including Klebsiella pneumoniae (K. pneumoniae) (n = 323), E. coli (n = 35), Enterobacter spp. (n = 35), Citrobacter spp. (n = 17), Serratia spp. (n = 5), and other bacterial species (n = 8), four E. coli were found to carry the mcr-1 gene. All carbapenem-resistant E. coli (CREC) exhibited different PFGE patterns ( Figure 1). The clinical significance of these four isolates is summarized in Table 1. The age range of the patients was between 54 years and 90 years. Two of the mcr-1-positive CREC were isolated from urine specimens, one from ascites, and one from a rectal swab. All of four patients had variable underlying immunocompromised conditions, such as malignancy and chronic kidney disease. In addition, two patients had undergone surgery within one month before the diagnosis of mcr-1-positive CREC. All of the patients had received prior broad-spectrum antibiotics including carbapenem, piperacillin/tazobactam, and third-generation cephalosporins. However, none of the patients had previously received colistin. While one patient was considered as having mcr-1-positive CREC colonization, the other three mcr-1-positive CREC-related infections were classified as nosocomial infections, including two episodes of catheter-associated urinary tract infection and one of peritonitis. Various antibiotic regimens were used for the three cases with infections caused by mcr-1-positive CREC. Two of three patients with mcr-1-positive CREC infection died in hospital. Only the one patient who had mcr-1-positive CREC peritonitis and the patient with mcr-1-positive CREC colonization survived. For cases 2 and 4 with survival, mcr-1-positive CREC were detected at two and three weeks, respectively. For cases 1 and 3 with mortality, mcr-1-positive CREC remained persistent for about one week till the deaths of the patients.

MICs
The MICs of these four isolates are shown in Table 2. For the four E. coli isolates, amikacin and tigecycline showed good in vitro activities. Each of the three E. coli isolates remained susceptible to fosfomycin and minocycline. In contrast, these four strains showed resistance against the other broad-spectrum antibiotics including cephalosporin, ciprofloxacin, and most carbapenems. Regarding colistin, one isolate (EC826) had an MIC level of only 0.25 mg/L and was classified as colistin wild-type (WT). The other three E. coli (EC516, EC868, and EC1037) were identified as colistin nonwild-type (NWT) (MICs ≥4 mg/L). Bold type indicates resistance.

Molecular Characteristics
Based on the MLST, these four strains belonged to different sequence types (ST), EC516 (ST617), EC826 (ST457), EC868 (ST10), and EC103 (ST1196). In this study, we used PCR first to screen and identify mcr-1 gene, and then we used S1-neculease PFGE and Southern blot of mcr-1 to help confirm the location of plasmid-mediated mcr-1 gene and the size of plasmid. Plasmid DNA was linearized by S1 nuclease on PFGE. An agarose gel of the S1 nuclease PFGE-based sizing of plasmids for the four isolates is shown (Figure 2A). Figure 2B shows the corresponding gel after Southern blotting, and the plasmids expressing mcr-1 were detected by hybridization of the Southern blot with a specific probe. The findings of the Southern blot indicated that the mcr-1 gene was found in the plasmids of EC516, EC826, and EC1037. However, it was not found in the plasmids of EC868 ( Figure 2B Table 3 lists the antibiotic resistance genes for the four mcr-1-positive CRE isolates. One strain (EC1037) had blaSHV, which is closely matched with SHV-5 with only one mutation at L31Q. Three of them carried blaCMY-2 genes and one had the blaTEM-1 gene. In addition, one E coli isolate (EC516) carried blaCTX-M-65 and blaNDM-9. None of them had IMP, VIM, KCP, and OXA48 genes. Regarding the outer membrane proteins, OmpC and OmpF were detected in three of the isolates. Some of them were wild type (EC516 and 868) and one (EC826) was an insertion type. Only one mutation G181D of OmpC was found in EC868 (Table 3). OmpC and OmpF were nondetectable in EC1037. Table 3. Antibiotic resistance mechanisms of mcr-1-positive Enterobacteriaceae isolates.

Sequencing Analyses of the Carbapenem-Resistance Plasmids
All of the sequencing analyses of carbapenem-resistance plasmids are shown in Figure 3. For E. coli EC516, its blamcr-1-encoding plasmid was p5CRE51-MCR-1, with a size of approximately 60.9 kb. For E. coli EC826, its blamcr-1-encoding plasmid was pHNGDF93, with a size of approximately 62.3 kb containing one insertion within mcr-1. For E. coli EC1037, its main blamcr-1-encoding plasmid was  Table 3 lists the antibiotic resistance genes for the four mcr-1-positive CRE isolates. One strain (EC1037) had bla SHV , which is closely matched with SHV-5 with only one mutation at L31Q. Three of them carried bla CMY-2 genes and one had the bla TEM-1 gene. In addition, one E coli isolate (EC516) carried bla CTX-M-65 and bla NDM-9 . None of them had IMP, VIM, KCP, and OXA48 genes. Regarding the outer membrane proteins, OmpC and OmpF were detected in three of the isolates. Some of them were wild type (EC516 and 868) and one (EC826) was an insertion type. Only one mutation G181D of OmpC was found in EC868 (Table 3). OmpC and OmpF were nondetectable in EC1037. Table 3. Antibiotic resistance mechanisms of mcr-1-positive Enterobacteriaceae isolates.

Sequencing Analyses of the Carbapenem-Resistance Plasmids
All of the sequencing analyses of carbapenem-resistance plasmids are shown in Figure 3. For E. coli EC516, its bla mcr-1 -encoding plasmid was p5CRE51-MCR-1, with a size of approximately 60.9 kb.
For E. coli EC826, its bla mcr-1 -encoding plasmid was pHNGDF93, with a size of approximately 62.3 kb containing one insertion within mcr-1. For E. coli EC1037, its main bla mcr-1 -encoding plasmid was pHNGDF93 with a size of approximately 63.6. Only one CRCE 868 isolate with mcr-1 inserted in the chromosome, not within the plasmid, was found. pHNGDF93 with a size of approximately 63.6. Only one CRCE 868 isolate with mcr-1 inserted in the chromosome, not within the plasmid, was found.

Discussion
In this multicenter study, the overall prevalence of mcr-1 among CRE was 0.9% (4/423), and all of the four CRE harboring mcr-1 were E. coli. Therefore, the prevalence of mcr-1 among 35 CREC in

Discussion
In this multicenter study, the overall prevalence of mcr-1 among CRE was 0.9% (4/423), and all of the four CRE harboring mcr-1 were E. coli. Therefore, the prevalence of mcr-1 among 35 CREC in this surveillance study was 11.4%. In addition, all of them exhibited different PFGE patterns and ST types, which indicated that they did not result from the spread of a single clone. In China, a national surveillance study [23] of 1105 CRE isolates showed that two CREC carried the mcr-1 gene and that the prevalence was less than 0.2%. Another surveillance study [24] at a single center in China revealed a similar finding in that only one E. coli strain among 1311 CRE isolates harbored mcr-1. Although all of these findings indicate that the prevalence of the mcr-1 gene among CRE isolates remains low, further regular surveillance investigation is still needed to assess the epidemiology of mcr-1 among CRE isolates. Moreover, the emergence of mcr-1 among CREC should be closely monitored.
In this study, all of the patients with mcr-1 CRE had various underlying diseases, and two of them had undergone abdominal surgery recently. Although all of them had received broad-spectrum antimicrobial agents, especially piperacillin/tazobactam before acquiring mcr-1 CRE, none of them had received colistin recently. The clinical manifestations of patients infected or colonized with mcr-1 CRE in this study were consistent with those of a previous report [8], and patients with multiple comorbidity and history of broad-spectrum antibiotic might be at risk of acquiring mcr-1-positive Enterobacteriaceae. In line with previous reports [9,25], our study shows that prior use of colistin may not be necessary In agreement with a previous study [24], all of mcr-1-positive CREC in this study displayed resistance against most of the antibiotics. Only minocycline, tigecycline, amikacin, and fosfomycin showed good in vitro activities against more than three CREC strains. This finding and previous reports [26,27] about CRE isolates suggested that combination therapy or some active agents, such as tigecycline, fosfomycin, and amikacin, may be appropriate options for antibiotic treatment. However, a further large-scale study is needed to assess the activity of antibiotics against mcr-1 gene harboring CRE.
In this study, all of mcr-1-positive CREC had resistance genes AmpC CMY , bla NDM , bla TEM , bla SHV , and bla CTX . In addition, some of them (EC516, EC826 and EC868) had the detectable outer membrane proteins OmpF and OmpC. However, one strain (EC1037) had a loss of the outer membrane protein. This phenomenon has been observed in previous studies in Taiwan [13,28,29]. The coexistence of plasmidic AmpC β-lactamase and outer membrane protein loss could be the main underlying mechanism for non-carbapenem-susceptible Enterobacteriaceae [12,24].
Generally, bacteria with the gene encoding plasmid-mediated colistin resistance (mcr-1) would be considered as colistin NWT. However, one recent study [9] showed that two (3%) of 76 mcr-1-positive E. coli remained susceptible to colistin. In this study, one isolate (EC826) was found to have mcr-1 during the screening, but its colistin MIC was only 0.25 mg/L. We found that another gene was inserted in mcr-1 in EC826. This finding may explain why the CRE with mcr-1 demonstrated a susceptible colistin MIC due to the malfunction of mcr-1. A similar finding was reported by Terveer et al., where mcr-1-positive E. coli was phenotypically susceptible to colistin with a MIC of <0.25mg/L, due to a 1329-bp transposon IS10R inserted into the mcr-1 gene [30]. However, a further large-scale study is warranted to clarify this issue.
Although only 35 (8.2%) clinical isolates in this surveillance of 423 CRE belonged to E. coli, all of mcr-1-positive CRE were exclusively E. coli. In contrast, K. pneumoniae (n = 323, 76.4%) was the most common CRE in this study, but none of them were found to carry the mcr-1 gene. Several large surveillance studies demonstrated that the prevalence of mcr-1 gene among E. coli isolates varied, at 0.25% (10/3902) in Italy [31], and 1% (76/5332) in China [9]. For ESBL E. coli, the prevalence of the mcr-1 gene was 3.5% (25/706) in China [25], but for CREC, only limited reports [23,24,32,33] showed the co-presence of the mcr-1 gene in human and animal isolates. For this study, the sample size of CREC is small, and thus may result in the high prevalence of mcr-1 gene among CREC isolates (11.4%, 4/35). A further large-scale study is needed to investigate the prevalence of mcr-1 gene among CREC isolates.
In conclusion, the emergence of the mcr-1 gene among CRE, especially E. coli in southern Taiwan, remains worth our attention due to its resistance to most antibiotics. Although the E. coli harboring mcr-1 gene is rarely susceptible to colistin, a further national survey is warranted.

Conflicts of Interest:
The authors declare no conflict of interest.