A method to detect Escherichia coli carrying the colistin-resistance genes mcr-1 and mcr-2 using a single real-time polymerase chain reaction and its application to chicken cecal and porcine fecal samples

Abstract

Colistin is one of the last-resort antibiotics for the treatment of multidrug-resistant infections in humans, but transmissible colistin-resistance genes have emerged in bacteria from animals. The rapid and sensitive detection among animals of colonization with bacteria carrying these genes is critical in helping to control further spread. Here we describe a method for broth enrichment of colistin-resistant Escherichia coli from animal fecal and cecal samples followed by real-time polymerase chain reaction (PCR) for the simultaneous detection of two of the main colistin-resistance genes, mcr-1 and mcr-2. The PCR uses a single set of nondegenerative primers, and mcr variants can be differentiated by melt-curve analysis. Overnight culture enrichment was effective for amplifying colistin-resistant E. coli, even when initially present in numbers as low as 10 bacteria per gram of sample. The mcr-1 and mcr-2 genes were not found in any of the Ontario swine and poultry samples investigated.

The recently described plasmid-mediated colistin-resistance gene mcr-1 (1) has been found around the world in numerous animal and human sources, particularly Escherichia coli (2–4). Although other forms of colistin resistance have been well-documented, this was the first transmissible colistin-resistance determinant to be characterized. Minor variants of mcr-1 have since been described, as have different genes conveying colistin resistance, notably mcr-2 (5), mcr-3 (6), mcr-4 (7), and mcr-5 (8). The mcr-1 and mcr-2 genes are closely related and therefore detectable with a single set of polymerase chain reaction (PCR) primers, whereas the other mcr genes are considerably different and require separate sets of primers. In addition, previous studies have shown that direct PCR testing of fecal samples to detect mcr genes may not be as sensitive as culture methods and that only enrichment with selective broth cultures before PCR provides sufficient sensitivity (9). Therefore, in order to obtain a test rapid and sensitive enough to identify animals colonized with or shedding E. coli carrying mcr-1 and mcr-2, we developed a broth-enrichment culture method for colistin-resistant E. coli from fecal and cecal material followed by a single real-time PCR for mcr-1 and mcr-2. We subsequently assessed the prevalence of these genes in porcine and chicken samples in the province of Ontario.

Because mcr-1 and mcr-2 are the most closely related colistin-resistance genes, at 78.8% sequence identity, we designed primers using the open-source software Primer3 (10) to amplify a fragment of both genes 115 base pairs (bp) long in a consensus region. All 3 of the more recently described colistin-resistance genes, mcr-3, mcr-4, and mcr-5, would not be amplified with the use of these primers, having less than 59% gene identity and no significant homology in the priming regions. Amplification of other published variants of mcr-1 (i.e., mcr-1.2 through mcr-1.7) would result in identical PCR products, as all of these variants’ nucleotide substitutions would be outside of the priming and amplicon regions (6). Although mcr-1 and mcr-2 differ by 5 bp in this amplified region, the net guanosine–cytosine content increases by only a single bp for mcr-1. Effectively, this should produce a change in melting temperature (ΔT_m) of 0.4°C, a difference discernible by a real-time PCR instrument given sufficient resolution.

Using primers umcrF 5′-ATGATGTCGATACCGCCAAATACCA-3′ and umcrR 5′-AGCTTATCCATCACGCCTTTTGAGT-3′ at 250 nM, 1 μL of boiled lysate (spiked samples and standards), and 19 μL of master mix, mcr-1 and mcr-2 were amplified in triplicate for 45 cycles of 95°C, 60°C, and 72°C for 15 s each, and melt curves were generated by continuous fluorescence capture between 65°C and 99°C (0.2°C/s). Results for each lysate were expressed as the average of these 3 technical replicates.

Known concentrations of an mcr-1-positive wild-type E. coli isolate obtained from a 2010 sample of Canadian lean ground beef (11) and an E. coli DH10B isolate transformed with a complete mcr-2 gene cloned into a pUC57 vector were prepared to determine the minimum detection limit for a real-time PCR assay. Triplicate cultures (biologic replicates) were incubated in Luria–Bertani broth (Becton Dickinson, Sparks, Maryland, USA) with shaking at 37°C for 16 h. Cultures were then serially diluted from 10⁸ to 10⁰ colony-forming units (CFU)/mL, and a lysate was prepared by incubating 1 mL of each culture at 100°C for 15 min.

To determine the effect of broth enrichment, approximately 1 g of 3 porcine fecal samples was added to 9 mL of E. coli (EC) broth (Becton Dickinson) supplemented with colistin, 1 μg/mL, and either spiked with 10-fold serial dilutions of mcr-1-positive E. coli (10¹ to 10⁶ CFU/10 mL) or not spiked, for the negative control; this was considered time 0. Each culture, in triplicate (biologic replicates), was incubated for 5 h (for same-day testing) and 16 h (for 2-day testing) as described. Serial dilutions were also made in parallel in EC broth alone (without feces) to assess any possible inhibitory effect of the fecal material on the PCR. A 1-mL lysate of each culture boiled for 15 min was prepared at time points 0, 5, and 16 h. Lysates were used in triplicate (technical replicates) as DNA templates for SYBR Green-based real-time PCR (Roche Applied Science, Mannheim, Germany) on a Roche LC480 II platform. This experiment was not done with the mcr-2 transformant since the cloned mcr-2 was not expressed and did not display a colistin-resistance phenotype.

Between 2015 and 2017, 40 composite fecal samples from swine, 50 cecal samples from swine, and 242 cecal samples from chickens were collected from farms in the province of Ontario. The composite fecal samples consisted of pools of freshly dropped feces from 5 pigs per farm at 8 farms. The cecal samples were collected at slaughter through the Canadian Integrated Program for Antimicrobial Resistance Surveillance (CIPARS) and pooled in sets of 10 (swine) or 5 (chicken). All the samples were screened by PCR after 16 h of culture in enriched broth as described with complete biologic duplication. Spiked fecal and cecal samples were used as positive controls in each PCR plate, and mock enrichments (no inoculum) were used as negative controls.

With serial dilutions of pure LB culture for both mcr-1 and mcr-2 isolates, the smallest mean concentrations detectable by PCR were approximately 54 CFU/mL [standard deviation (SD) ± 5.02 CFU/mL and 0.63 CFU/mL ± 0.0586 CFU/mL, respectively]. The plasmid copy number of the mcr-1 wild-type isolate is unknown; however, the high copy number (about 600 copies per cell) of the mcr-2 pUC57 vector (12) likely explains the difference in detection limit between the isolates carrying mcr-1 and mcr-2.

As shown in Figure 1, the lowest detectable concentration of E. coli from feces at the time of inoculation of the enrichment medium (time 0) was 1000 CFU/g (equivalent to 100 CFU/mL when diluted in the enrichment broth). In comparison, the detection limit in EC broth only (without fecal or cecal material) was a mean of 653 CFU (equivalent to 65 CFU/mL when suspended in EC broth). This demonstrates the negligible effects of fecal material and EC broth on the assay’s sensitivity. As little as 10 CFU in 10 mL containing 1 g of feces was already detectable 5 h after initial inoculation, and the results were improved after 16 h. The signal intensity was consistently higher after 16 h for samples with very small numbers of resistant bacteria. Although differences among samples were visible after 5 h of incubation, each enrichment, regardless of initial spiking concentration, reached a similar signal intensity after 16 h.

Figure 1.

Amplification of mcr-1 in porcine fecal samples spiked with dilutions of mcr-1-carrying Escherichia coli and incubated in an enrichment broth at 37°C for up to 16 h. The graph shows means and standard errors for 3 different fecal samples (each with 3 technical replicates), with additions of 10-fold dilutions of 10¹ to 10⁶ colony-forming units (CFU) at time 0, and measured at times points 0, 5, and 16 h. A standard curve with known concentrations of bacteria was used for quantification.

Using our set of primers, we were able to amplify and differentiate mcr-1 and mcr-2 with their respective control isolates. The average T_m was 81.46°C [95% confidence interval (CI): 81.42 to 81.50] for mcr-1 and 81.04°C (95% CI: 81.00 to 81.08) for mcr-2, which resulted in a significant ΔT_m of 0.42°C (P < 0.0001). We could also predict the variant according to the T_m with 100% accuracy using a blinded set of 10 mcr-1 cultures and 10 mcr-2 cultures (Figure 2).

Figure 2.

Melting peaks for 10 mcr-1 (black) and 10 mcr-2 (gray) biologic replicates for determination of the variant based on the thermal profile by real-time polymerase chain reaction. The vertical axis shows the negative first derivative of the fluorescence curve.

Finally, the mcr-1 and mcr-2 genes were not detected by PCR in any of the samples from swine and chickens after 16 h of enrichment culture.

In summary, we developed a simple and effective method for the detection of colistin-resistant E. coli in fecal and cecal samples with a demonstrable positive effect of incubation time on detection of small numbers of resistant bacteria. Although we used a more selective enrichment broth (EC broth instead of LB broth) the overall effect on the sensitivity of mcr detection was similar to what was observed by Donà et al (9) in another test combining enrichment and PCR for the detection of mcr-1. However, as those authors suggested, the combined detection of mcr-1 and mcr-2 in our test is certainly a welcome addition. The method also allows us to distinguish between 2 important mcr groups using a single set of primers. A higher-resolution melting dye such as EvaGreen would likely enhance the method’s precision (13); however, SYBR Green worked well in this study. Our method does not require a complex DNA extraction procedure and can be done quickly. The primers described here have the potential to be multiplexed with additional primers for the detection of other newly discovered colistin-resistance genes. Although we did not detect either of the targeted resistance determinants in our fecal and cecal samples from Ontario, we have the ability to screen large numbers of samples going forward. Possessing the ability to enrich and subsequently detect small numbers of mcr-carrying bacteria will become more and more important as the prevalence of bacterial resistance to colistin continues to increase in Canada and globally.