Prevalence of Colistin Resistance Gene mcr-1 and Absence of mcr-2 in Escherichia coli Isolated from Healthy Food-Producing Animals in Japan

ABSTRACT

We screened mcr-1 and mcr-2 genes in 9,306 Escherichia coli strains isolated from healthy animals in the Japanese Veterinary Antimicrobial Resistance Monitoring (JVARM) system. mcr-1 was detected in 39 strains (5, 20, and 14 strains isolated from cattle, swine, and broilers, respectively), whereas mcr-2 was not detected. mcr-2 was also not detected with the investigation sequence homology search against our curated GenEpid-J database.

TEXT

Colistin has been used as a veterinary drug for the treatment of Gram-negative gastrointestinal infections and as a feed additive to promote healthy development in food-producing animals for more than 5 decades. Colistin is currently considered the last-resort antibiotic for the treatment of infections caused by multidrug-resistant Gram-negative bacteria worldwide in human medicine. The resistance mechanism to colistin was only known to involve chromosomal mutations. However, in 2015, Liu et al. reported the first plasmid-mediated colistin resistance gene, mcr-1, in Enterobacteriaceae isolated from food-producing animals, retail meat, and humans in China (1).

In a previous study, after analyzing data from the Japanese Veterinary Antimicrobial Resistance Monitoring (JVARM) system, we showed that only two isolates were mcr-1 positive in 90 colistin-resistant Escherichia coli isolates (MIC, ≥8 mg/liter), and the plasmids carrying mcr-1 from animal isolates were identified in the GenEpid-J database (2). However, other studies have shown that E. coli strains with an MIC of ≥2 mg/liter had mcr-1 genes (3, 4). Kusumoto et al. reported that more than 13% of the isolates were mcr-1 positive, with MICs of ≥4 mg/liter in swine-pathogenic E. coli (5). Furthermore, Xavier et al. reported the presence of another plasmid-mediated colistin resistance gene, mcr-2, which encodes the MCR-2 protein and has 76.7% nucleotide identity to mcr-1 (6). In response to these findings, we screened the mcr-1 and mcr-2 genes in E. coli isolates with colistin MICs of ≥2 mg/liter, part of which were colistin-susceptible isolates according to the breakpoint of the European Committee on Antimicrobial Susceptibility Testing (EUCAST).

We screened 9,306 E. coli isolates isolated from healthy animals from JVARM (3,134 isolates from cattle, 2,052 isolates from swine, 2,017 isolates from broilers, and 2,103 isolates from layers) between 2000 and 2014. MICs of colistin were determined by the agar dilution method in the isolates from 2000 to 2009 and broth dilution method in the isolates from 2010 to 2014, according to the recommendation of the Clinical and Laboratory Standards Institute (7). A total of 732 (7.9%) isolates showed an MIC of ≥2 mg/liter for colistin and were examined for the presence of mcr-1 and mcr-2 by PCR, as described by Liu et al. (1) and Xavier et al. (6), respectively. Additionally, we searched a plasmid carrying mcr-2 against the GenEpid-J database. By November 2015, 1,747 plasmid sequences from 671 Gram-negative bacteria were analyzed using next-generation sequencing and curated in the GenEpid-J database; these sequences represented 431, 184, and 56 isolates from patients admitted to hospitals, animals, and the environment, respectively.

A total of 39 isolates, including 12 isolates of MIC of 2 mg/liter, were mcr-1 positive, corresponding to 5.3% (39/732) of the isolates of colistin MIC of ≥2 mg/liter and 0.42% (39/9,306) of the total collected isolates. The numbers of mcr-1-positive isolates from cattle, swine, and broilers were 5, 20, and 14, respectively, and overall, the number of positive isolates increased slightly over the years (Fig. 1). By using the broth mating method (8), 9 (3 cattle, 3 swine, and 3 broilers) of 39 (23.0%) isolates were transferred to E. coli DH5α. All the transconjugants had a size of roughly 60 kb and incompatibility type of IncI2 plasmid by using pulsed-field gel electrophoresis (PFGE) after S1 nuclease digestion (9), a PCR-based replicon-typing assay (10, 11), and hybridization (12). These results suggested that they were similar to the plasmid reported by Liu et al. in China (1). However, the number of mcr-1-positive isolates was limited in healthy animals, and none of the layer isolates were mcr-1 positive. From 2000 to 2014, colistin-resistant isolates (MIC, >2 mg/liter by EUCAST criterion) were consistently low, and even after taking into account isolates with an MIC of 2 mg/liter, the proportion of colistin-resistant and reduced-susceptibility isolates in E. coli did not increase since the first mcr-1 detection in 2008 (Fig. 1).

FIG 1.

Proportion of Escherichia coli isolates that showed different MICs to colistin, and mcr-1-positive strain numbers in E. coli isolates from healthy animals between 2000 and 2014 in the Japanese Veterinary Antimicrobial Resistance Monitoring system. Filled black circle and solid line shows proportion of isolates with a colistin MIC of ≥2 mg/liter. Filled black circle and dashed line shows proportion of isolates with a colistin MIC of ≥4 mg/liter. Filled black square and dashed line shows proportion of isolates with a colistin MIC of ≥8 mg/liter. Filled gray circle and solid line shows proportion of isolates with a colistin MIC of ≥16 mg/liter. The dark gray bar shows the number of mcr-1-positive isolates from broilers, the hatched bar shows the number of mcr-1-positive isolates from swine, and the light gray bar shows the number of mcr-1-positive isolates from cattle.

Compared with the swine-pathogenic E. coli strains isolated from diseased swine in Japan (5), the mcr-1-positive (healthy swine, 20/2,052 [0.97%] versus diseased swine, 90/684 [13.16%]) and resistance (≥4 mg/liter; healthy swine, 92/2,052 [4.48%] versus diseased swine, 309/684 [45%]) rates of isolates from healthy animals were much lower. Colistin is the common treatment option of swine diarrheal diseases in Japan. The difference in the selection pressure of colistin between diseased and healthy swine may explain the discrepancy in the prevalence of mcr-1 between E. coli isolated from diseased and healthy swine.

The prevalence of mcr-2 in swine colistin-resistant E. coli in Belgium was higher than that of mcr-1 (6), necessitating an immediate introduction of mcr-2 screening in ongoing molecular epidemiological surveillance of colistin-resistant Gram-negative pathogens; however, mcr-2 was not detected in this investigation by using Xavier's method. Moreover, we could not detect the mcr-2 gene in the GenEpid-J database. It is speculated that the distribution of mcr-2 between European countries and Japan is different and might be related to geography or differences in veterinary practices between these regions, but further research is needed.

In conclusion, the prevalence of mcr-1 in E. coli from healthy animals increased slightly over the years, but the prevalence remained very low, and mcr-2 was not detected in this study. The proportion of strains with an MIC of ≥2 mg/liter colistin has not increased since the first mcr-1 detection. In Japan, there have been no reports of mcr-1-positive bacteria isolated from food and humans. However, continuous surveillance and monitoring and ensuring the prudent use of antibiotics in veterinary medicine are essential to prevent or reduce the transfer of resistant bacteria or resistance determinants within animal populations and the environment, and between animals and humans.