Multiplex PCR for detection of plasmid-mediated colistin resistance determinants, mcr-1, mcr-2, mcr-3, mcr-4 and mcr-5 for surveillance purposes

Background and aim Plasmid-mediated colistin resistance mechanisms have been identified worldwide in the past years. A multiplex polymerase chain reaction (PCR) protocol for detection of all currently known transferable colistin resistance genes (mcr-1 to mcr-5, and variants) in Enterobacteriaceae was developed for surveillance or research purposes. Methods: We designed four new primer pairs to amplify mcr-1, mcr-2, mcr-3 and mcr-4 gene products and used the originally described primers for mcr-5 to obtain a stepwise separation of ca 200 bp between amplicons. The primer pairs and amplification conditions allow for single or multiple detection of all currently described mcr genes and their variants present in Enterobacteriaceae. The protocol was validated testing 49 European Escherichia coli and Salmonella isolates of animal origin. Results: Multiplex PCR results in bovine and porcine isolates from Spain, Germany, France and Italy showed full concordance with whole genome sequence data. The method was able to detect mcr-1, mcr-3 and mcr-4 as singletons or in different combinations as they were present in the test isolates. One new mcr-4 variant, mcr-4.3, was also identified. Conclusions: This method allows rapid identification of mcr-positive bacteria and overcomes the challenges of phenotypic detection of colistin resistance. The multiplex PCR should be particularly interesting in settings or laboratories with limited resources for performing genetic analysis as it provides information on the mechanism of colistin resistance without requiring genome sequencing.


Introduction
Colistin belongs to the antimicrobial class designated polymyxins which originates from the organism Paenibacillus polymyxa. This class consists of polymyxins A, B, C, D and E, of which only colistin (polymyxin E) and polymyxin B are used in clinical practice [1]. Colistin has been widely used in veterinary medicine in Asian, European and North American countries but human use was restricted due to its neuro-and nephrotoxicity [2][3][4]. In the past decade, emergence of multidrug-resistant Gram-negative bacteria (such as Enterobacteriaceae, Pseudomonas aeruginosa and Acinetobacter baumannii) led to an increase of colistin administration as a last resort antibiotic for human infections [5]. Prior to the detection of the plasmid-mediated resistance gene mcr-1, resistance to colistin in Enterobacteriaceaehad only involved chromosomal mutations, including mutations in the pmrA/ pmrB two-component system that regulate synthesis and structure of the lipopolysaccharide [6].
Global surveillance of plasmid-mediated colistin resistance is hindered by the technical difficulties of phenotypic tests owed to particular interactions of the drug with reagents or materials and resulting from its specific molecular structure. The main problems associated with phenotypic testing are the (i) irregular diffusion of colistin in agar media, (ii) interaction with cations in media and (iii) possible adsorption of the antibiotic to certain laboratory materials, which make the agreement of results between replicated testing and between laboratories difficult. Disk diffusion, agar dilution and Etest strips methods cannot be used to perform colistin susceptibility testing, and there is yet not enough information available regarding automated methods to reach a significant conclusion about their sensitivity [1,7]. Thus to date, only determination of minimum inhibitory concentration (MIC) by broth microdilution is recommended in technical guidance documents by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) and the Clinical and Laboratory Standards Institute (CLSI) [8,9].
The first plasmid-mediated colistin resistance gene was identified in an Escherichia coli strain from a pig in China and also found to be present in other Escherichia coli and Klebsiella pneumoniae isolates in samples from both animal and human origin, collected between 2011 and 2014 [4]. The mcr-1 gene (KP347127) encodes a phosphoetanolamine transferase that alters the lipid A in the lipopolysaccharide of the bacterial outer membrane. This reduces the attachment of colistin and therefore prevents cell lysis. Although this mechanism has only recently been discovered there are reports of isolates recovered in 1980 harbouring the mcr-1 gene, but most cases refer to isolates collected since 2009 [10]. After the first detection, additional four mcr genes were described: mcr-2 (LT598652) was found in E. coli from animal sources in Belgium [11], mcr-3 (KY924928) in E. coli from pigs in China [12], mcr-4 (MF543359) in E. coli and Salmonella enterica serovar Typhimurium from pigs in Italy, Spain and Belgium [13] and mcr-5 (KY807921) in Salmonella Paratyphi B dTa + from poultry in Germany [14]. Variants of mcr-1, mcr-3 and mcr-4 [15][16][17][18] as well as co-occurrence of mcr-1 and mcr-3 have been observed in Enterobacteriaceae [19,20].
Even though mcr genes have been more prevalent in animal and food isolates than in human strains, multiple cases of human infection by strains harbouring mcr-1, mcr-3 or mcr-4 have been observed [15,18,21]. Worldwide, over 100 Enterobacteriaceae strains harbouring mcr have been recovered from human hosts, mostly E. coli strains, in patients originating from, or who had recently travelled to, Asian countries [15]. This includes the first mcr-1 description by Liu et al. who identified 16 mcr-1-positive E. coli and K. pneumoniae clinical strains [4]. More recently, mcr-1 and mcr-1.2 were described in one E. coli and one K. pneumoniae strain from infected humans in Denmark [22] and Italy [23], respectively. Three additional E. coli strains carrying the mcr-1 gene were found in faecal samples of Dutch travellers [24] and a study of pilgrims travelling to Mecca revealed the gene in 10 E. coli and one K. pneumoniae strains [25]. These findings represent a snapshot of the cases so far described of mcr-1 i n Enterobacteriaceae of human origin and illustrate the global spread of bacteria carrying mcr-1. In Denmark, one E. coli and 10 Salmonella clinical strains collected between 2009 and 2017 were found to harbour mcr-3, six of which were present in people reporting previous travel to Thailand or Vietnam [17,19]. In Italy, mcr-4-positive Salmonella strains have been isolated from two humans with gastroenteritis [18].
The discovery of plasmid-mediated colistin resistance has raised international concern as the spread of this resistance might hamper the efficacy of colistin in humans, although the full clinical relevance has not yet been determined. Plasmid-mediated resistance genes have the potential to be further transmitted and spread globally, including to and within regions with already high levels of antimicrobial resistance where multidrugresistant bacteria represent a public health concern. This can lead to the emergence of colistin-resistance in already multidrug-resistant bacteria, for which colistin represents a last-resort antibiotic, and compromise antibiotic treatment. Thus, human infections for which therapeutic options have been depleted could occur in nosocomial settings.

Figure
Multiplex PCR for detection of mcr-1, mcr -2, mcr-3,  Considering (i) the global concerns, (ii) the unreliability of most phenotypic methods for colistin susceptibility testing for surveillance purposes, and (iii) the limited availability of whole genome sequencing (WGS) technology in some settings or laboratories with a lack of adequate resources, we aimed to develop and validate a method that enables screening of relevant isolates in a cheap, rapid and efficient way to identify those possibly harbouring plasmid-mediated colistin resistance genes and meriting further characterisation. Here, we describe a multiplex PCR protocol that detects mcr-1, mcr-2, mcr-3, mcr-4 and mcr-5 genes rapidly and reliably for surveillance purposes and epidemiological research, particularly in food, animal and environmental samples.

The multiplex PCR method development
To allow fast and simultaneous detection of all the currently described mcr genes and their variants and improve visualisation by gel electrophoresis, we focused on conventional block multiplex PCR technique. It was chosen because of its simplicity that enables it to be used in all laboratories and settings with limited laboratory resources, limited or no access to WGS, and in situations where a rapid answer is desired.
The original reference strains for mcr-2 (E. coli KP37) and mcr-5 (Salmonella Paratyphi B dTa + 13-SA01718), and strains available at the Technical University of Denmark in Lyngby for mcr-1 (E. coli 2012-60-1176-27), mcr-3 (E. coli2013-SQ352) and mcr-4 (E. coli DH5α with the entire mcr-4 gene cloned in pCR2 vector) were used as positive controls in the multiplex PCR (Table 1). DNA templates were obtained from overnight agar cultures by thermal cell lysis. Each PCR reaction consisted of 12.5 µL DreamTaq Green PCR Master Mix (Thermo Fisher Scientific, Waltham, Massachusetts, United States), 5.5 µL of nuclease-free water, 0.5 µL of each of the 10 primers solutions (10 µM), and 2 µL DNA lysate. Running conditions were: 1 cycle of denaturation at 94 °C for 15 min, followed by 25 cycles of denaturation at 94 °C for 30 s, annealing at 58 °C for 90 s and elongation at 72 °C for 60 s, and a final cycle of elongation at 72 °C for 10 min. The amplification was visualised by electrophoresis using 1.5% agarose gel at 130V followed by staining in ethidium-bromide.

Table 2d
Multiplex PCR for detection of mcr-1, mcr-2, mcr-3, mcr-4 and mcr-5 compared with WGS results, applied to a collection of Escherichia coli and Salmonella spp. isolates from food animals and food, European Union Reference Laboratory for Antimicrobial Resistance in the context of animal health and food safety, Spain, Germany, France, Italy, 2016 (n = 49) An internal amplification control was not used due to the incompatibility with DreamTaq Green PCR Master Mix, which contains a DNA polymerase synthesised in E. coli and thus yields amplicons when using 16S rRNA primers.

Test isolates for validation of the multiplex PCR
To validate the multiplex PCR, 49 isolates from four European countries were used. These isolates were collected within the framework of the European Monitoring of Antimicrobial Resistance pertaining to the Commission Implementing Decision 2013/652/EU [26]. They were analysed at the European Union Reference Laboratory for Antimicrobial Resistance (EURL-AR) in the context of animal health and food safety, in accordance with the EURL-AR/European Food Safety Authority (EFSA) confirmatory testing, that consists of MIC determination by broth microdilution and WGS by Illumina Hi-Seq (Illumina, Inc., San Diego, California, United States) (European Nucleotide Archive project PRJEB21546). The confirmatory testing represents a random snapshot of submitted unusual phenotypes within each participating country based on selective criteria from EFSA.

Results
Isolates collected in 2016 from Spain (n = 19), Germany (n = 16), France (n = 10) and Italy (n = 4) were selected for this study due to the diverse colistin resistance profiles found by WGS in these countries. Forty-two isolates were E. colifrom porcine (n = 20) and bovine (n = 22) sources (including caecum and meat samples) and seven isolates were Salmonella spp. from porcine (n = 5) and bovine (n = 2) sources.
The multiplex PCR protocol specifically amplified the fragments of the five mcr genes. Amplicon size ranged from 320 bp (mcr-1) to 1,644 bp (mcr-5) (Figure).
No interference between primer pairs and no unspecific products were observed. Analysing the validation set, the method detected mcr-1, mcr-3 and mcr-4 as singletons and in different combinations ( Table 2).

Validation test and comparison with WGS data
The validation test set revealed eight colistin-resistant E. coli isolates (MIC 4 -8 mg/L), five from bovine and three from porcine origin, harbouring the mcr-1 gene or one of its variants. An additional E. coli isolate from a pig phenotypically classified as borderline susceptible (colistin MIC of 2 mg/L), on repeated testing at two laboratories also harboured the mcr-1 gene.  (Table 2).

Discussion
The multiplex PCR reaction showed 100% specificity (no unpredicted amplification products) and 100% sensitivity (all of the predicted products were visualised in agarose gel) for the validation set, according to WGS data. Another observation of interest was the detection of two pmrA/pmrB point mutations in one colistin-resistant isolate, which reveals a need to further investigate and characterise this two-component system in order to accurately predict which mutations lead to phenotypic colistin resistance. Taking into account that the mutations found in this isolate are not consistently associated with a resistant phenotype we cannot confirm, without further analysis, if they represent the underlying mechanism of colistin resistance for this isolate [27,28].
Although the presented multiplex PCR does not allow identification of the specific variants and detects nonfunctional genes, it still provides valuable information for surveillance purposes as well as for research. The results of the multiplex PCR also allow the selection of isolates for further studies based on genomic epidemiology, focusing on characterisation of mcr genes, their mobilisation mechanisms and the plasmids they are harboured on. While the multiplex PCR has not yet been validated on human strains, it is expected to provide the same sensitivity and specificity as for animal isolates. The vast majority of mcr genes are located on transferable plasmids that are not restricted to reservoirs or hosts. Similarly, we expect that this multiplex PCR design will maintain its usefulness should new gene variants among mcr-1 to mcr-5 be detected in Enterobacteriaceae, taking into account that the chance of variations in the primer annealing positions should be small and no such variations have been observed so far. A great advantage of this method is the ability to detect mcr genes in phenotypically susceptible bacteria, allowing it to be used as a screening technique for isolates recovered from human and animal sources. We believe future efforts should include the screening of large pools of borderline susceptible Enterobacteriaceae in order to understand if, and how, the epidemiological cut-off value for colistin should be reviewed.
Although an internal amplification control was not used in this study, it should be possible to use a different reaction mix containing another DNA polymerase in order to allow the addition of 16S rRNA primers to obtain an internal reaction control. We encourage researchers to adapt the PCR reaction to their specific needs.
This screening method can provide a useful and accurate description of colistin resistance in Enterobacte riaceaecollections, particularly when used in connection with MIC susceptibility testing. The protocol and standard operating procedure are freely available at the EURL-AR website (https://www.eurl-ar.eu/). The intended use for this multiplex PCR is as part of the surveillance workflow, to screen colistin-resistant and borderline susceptible isolates after initial MIC determination by the standard broth dilution method. Colistin MIC determination allows classification of isolates as susceptible, borderline susceptible or resistant and, subsequently, the isolates determined borderline susceptible and resistant (MIC ≥ 2 mg/L) can be subjected to the multiplex PCR to provide further insight into the underlying resistance mechanism and potentially reveal the presence of the known mcr genes. For an initial surveillance analysis we do not recommend to perform the multiplex PCR before MIC determination, since it could lead to overlooking colistin resistant isolates with other underlying mechanisms of resistance. The method can, however, be used on its own if the research is focused, for example, on plasmid dissemination or mcr genes' epidemiology. Advanced investigations could be conducted according to available procedures and surveillance/research goals, and would ideally involve WGS of the bacterial genome, even though it may not be easily accessible in some laboratories or countries, in particular, low and middle income countries (LMIC). Colistin resistant bacteria lacking any of the currently known mcr genes should be screened for other mechanisms of resistance, such as mutations in the pmrA/pmrBsystem, new mcr genes or variants, or even presently unknown mechanisms.
Future prospects include validation of the multiplex PCR for human clinical strains, to assure that the protocol follows the principles of the One Health perspective and can be used to directly screen human, veterinary and food bacteria. Other prospects are to clone mcr-1, mcr-2, mcr-3 and mcr-5 amplicons into TA-cloning vectors (Thermo Fisher Scientific) in order to obtain non-pathogenic and easy to handle control strains, and be able to confirm and update the protocol continuously as new variants are being detected.
In conclusion, we developed a multiplex PCR that rapidly detects all mcr genes described in Enterobacteriaceae to date. The multiplex PCR method is critical for epidemiological surveillance and antimicrobial resistance investigation, especially in settings where laboratory resources and access to WGS are limited, and where plasmid-mediated colistin resistance may threaten the effectiveness of available antimicrobials.

Disclaimer
Beatriz Guerra is employed with the European Food Safety Authority (EFSA) in its BIOCONTAM Unit that provides scientific and administrative support to EFSA's scientific activities. The positions and opinions presented in this article are those of the authors alone and are not intended to represent the views or scientific works of EFSA.

* Erratum
Due to a technical error, the name of Cristina De Frutos Escobar was incorrectly displayed in the list of authors. The mistake was corrected on 12 Feb 2018.