Mcr colistin resistance gene: a systematic review of current diagnostics and detection methods

Abstract Resistance to colistin, mediated by chromosomal mutations and more recently, by plasmid‐borne mcr genes, is increasingly being reported in bacterial isolates taken from humans, animals, farms, foods, and the environment. To easily identify and contain this quickly spreading menace, efficient diagnostics that are cheaper, faster, simpler, sensitive, and specific have become indispensable and urgently necessary. A thorough and systematic review of the literature available at Pubmed, ScienceDirect and Web of Science was thus undertaken to identify articles describing novel and efficient colistin resistance‐ and mcr gene‐detecting methods. From the final 23 studies included in this review, both phenotypic and molecular tests were found. The phenotypic tests consisted of novel culture media viz., SuperPolymyxin™, CHROMagar COL‐APSE and LBJMR media, commercial automated MIC‐determining instruments such as MICRONAUT‐S, Vitek 2, BD Phoenix, Sensititre and MicroScan, and novel assays such as Colistin MAC test, Colispot, rapid polymxin NP test (RPNP), alteration of Zeta potential, modified RPNP test, MICRONAUT‐MIC Strip, MIC Test Strip, UMIC System, and Sensitest™ Colistin. Molecular diagnostics consisted of the CT103XL microarray, eazyplex® SuperBug kit, and Taqman®/SYBR Green® real‐time PCR assays, with 100% sensitivity and specificity plus a shorter turnaround time (<3 hr). Based on the sensitivity, specificity, cost, required skill and turnaround time, the RPNP test and/or novel culture media is recommended for under‐resourced laboratories while the Multiplex PCR or Taqman®/SYBR Green® real‐time PCR assay alongside the RPNP or novel culture media is suggested for well‐resourced ones.

were used in managing carbapenem-resistant infections before the emergence of colistin resistance .
However, the emergence of colistin resistance, particularly in carbapenem-resistant strains, is compromising this combination regimen and has made tigecycline the sole last-resort antibiotic for managing carbapenem-and colistin-resistant MDRIs.
The transferable plasmid-mediated colistin resistance gene, mcr, and variants such as mcr-1. 1, -1.2, 1.3, 1.4, 1.5, 1.6, 1.7 and -1.8 and mcr-2, -3, -4, and -6, is a PEtN transferase enzyme that enzymatically transfers PEtN to lipid A (Abdul Momin et al., 2017;Esposito et al., 2017;Poirel, Jayol, Nordmann, et al., 2017) ( Figure 1). This, as described for the chromosomal mutations above, results in reduced anionic charges on lipid A, preventing electrostatic interactions with cationic polypeptide molecules such as polymyxins (colistin and polymyxin B), leading to colistin resistance (Esposito et al., 2017;Tendon, Poirel, & Nordmann, 2017). Addition of PEtN to the 4′ position of lipid A results in F I G U R E 1 Mechanism of mcr-mediated colistin resistance; adapted from . (a) Schematic representation for LPS-lipid A modification by MCR-2 in E. coli. In the cytoplasm, bacterial LPS-lipid A is synthesized using UDP-GlcNAc as the primer substrate. The fatty acid intermediates (C12 and C14) from the bacterial type II fatty acid synthesis (FAS II) pathway enter into the conservative 10-step route of lipid A synthesis involving nine enzymes (LpxA, LpxC, LpxD, LpxH, LpxB, LpxK, LpxL, LpxM, and KdtA). The nascent lipid A from the cytoplasm is translocated by the ABC transporter MsbA, a lipid flippase (35), across the inner membrane into the periplasm. The integral membrane protein MCR-2 is supposed to be localized on the periplasm side of inner membrane and catalyzes the chemical modification of the 2-keto-3-deoxyoctulosonic acid (Kdo2)-lipid A, giving Kdo2-PEA-4 = -lipid A. The modified form of Kdo2-lipid A, Kdo2-PEA-4 = -lipid A, then is exported by LptABCFG and LptDE into the outer leaflet of the outer membrane (36), thus reducing the negative membrane charge. That is the reason for the low/decreased affinity of bacterial surface to the cationic antibiotic polymyxin. (b) Chemical reaction in which MCR-2 catalyzes the modification of lipid A with 4 = -phosphatidylethanolamine. MCR-2 catalyzes the addition of phosphatidylethanolamine to position 4 = of lipid A, giving the final products of both PEA-4 = -lipid A and diacylglycerol a compound called PEtN-4′-lipid A (PEA-4′-lipid A) ( Figure 1).
The Clinical Laboratories Standard Institute (CLSI) and the European Committee on Antimicrobial Susceptibility Testing (EUCAST) have recommended the use of broth microdilution (BMD) as the standard testing protocol for determining colistin susceptibility among Gram-negative bacteria (Chew, La, Lin, & Teo, 2017). However, the relatively higher skill and difficulty associated with integrating the BMD into normal clinical routines have made other recently developed culture media and assays very relieving (Jayol, Nordmann, Brink, et al., 2017;Jayol, Nordmann, Lehours, et al., 2017;Nordmann, Jayol, & Poirel, 2016a,b). Notwithstanding, the BMD is used as the gold standard in testing the essential agreement (EA), categorical agreement (CA), major error (ME), and very major error (VME) of colistin minimum inhibitory concentration (MIC)measuring diagnostics (Abdul Momin et al., 2017;Nordmann et al., 2016a,b;Poirel, Larpin et al., 2017). CA refers to agreement in the interpretation of the MIC between the test compared to BMD, and EA occurs when an MIC result is within a twofold dilution from the BMD's. A ME occurs when the tested MIC is resistant while the BMD MIC is susceptible. VMEs occur when the evaluated method's MIC was susceptible while BMD MIC was resistant (Chew et al., 2017).
Although there are studies evaluating the relative efficiencies of the various commercial susceptibility-testing platforms and media, they are few and mostly undertaken with small sample sizes that do not express all known mcr types and variants (Jayol, Nordmann, Brink, et al., 2017;Jayol, Nordmann, Lehours, et al., 2017;Chew et al., 2017;Esposito et al., 2017;Carretto et al., 2018). As well, a standard and accepted protocol for screening, F I G U R E 2 Catalytic domain structure of mcr-1 enzyme; adapted from (Stojanoski et al., 2016). (a) Structure of the active-site phosphothreonine with associated zinc ions. The phosphothreonine (TPO285) is represented as a yellow-orange-red stick model and the zinc ions (ZN1, ZN2, ZN3, and ZN4) that surround the phosphothreonine are shown as slate blue spheres. The 2Fo−Fc simulated annealing difference map of the final refined model contoured at σ = 4.0 is shown as a gray mesh. ZN4 is also coordinated by Glu405 from a neighboring molecule in the crystal. The neighboring MCR-1 protein is colored white and labeled with the prefix #2. (b) Representation of the zinc ions identified in the active site of cMCR-1. Zinc ions are shown as slate blue spheres and active-site residues are represented in stick model. In yellow, is one MCR-1 (#1) molecule, and in white, is another MCR-1 (#2) molecule located adjacent to the first one. ZN4 from the second molecule is positioned at the interface and is shared by the two molecules. Structural water molecules are labeled and hydrogen bonds and zinc interactions are shown with dashed lines identifying and confirming colistin-resistant Gram-negative bacteria or mcr-producing Enterobacteriaceae in clinical routines is nonexistent.

| Purpose of this review
In the face of these challenges, this systematic review seeks to comprehensively describe all available polymyxin resistance-and mcr-detecting diagnostics in the context of their composition (for culture media), primers and cycling conditions (for multiplex and real-time PCR), sensitivities, specificities, turnaround time, skill, relative cost, EA, CA, ME, and VME. A flow diagram suggesting a standard protocol for screening, isolating, identifying and confirming colistin-resistant isolates ( Figure 3) is also included in this review, using the relative efficiencies, cost and required skill of the various diagnostics and detection methods as a guide.
Finally, none of the 13 published reviews on mcr-1 and colistin resistance addresses colistin resistance-and mcr-detecting diagnostics and detection methods, either narratively or systematically (Supplementary file S1), making this work the first, to my knowledge.

| Methods used
A systematic search of the literature was undertaken using the search words 'mcr-1′ and 'colistin resistance' on Pubmed, Science Direct and Web of Science on three different occasions. The dates filter was turned to between 2010 to May 20, 2018. All reviews, non-English articles, and papers not describing diagnostic or detection methods were subsequently discarded. The PRISMA guidelines were followed in searching, screening and including papers for this review (Figure 4).
The following data were extracted from the included articles: diagnostic tool or methods used, types and sample size of bacterial species used for the evaluation, sensitivity, specificity, EA, CA, ME, VME, turnaround time, media composition, real-time PCR cycling conditions, cycle threshold, product size, primers and probes used, color of media, and appearance of colonies on media (Tables 1-3).

| RE SULTS AND D ISCUSS I ON
A final list of 23 articles were included in this systematic review from the 3010 screened manuscripts (Figure 4) (Carretto et al., 2018) are novel commercial but manual tests (Table 3). E-test and BMD, which are older MIC-determining methods, were evaluated by two and single studies, respectively (Table 3). were also evaluated to determine their sensitivity, CA, EA, ME, and VME, using BMD as a gold standard (Table 3).
Thus, all the current colistin resistance-determining diagnostics/assays can be grouped into phenotypic and molecular tests, in which the phenotypic tests, except in a few cases discussed below, mainly determines the presence or absence of colistin resistance in Enterobacteriaceae and nonfermenting Gram-negative bacteria without establishing the underlying mechanism. On the other hand, all the molecular tests identified the presence of known mcr genes and variants, particularly mcr-1, in Enterobacteriaceae without necessarily determining phenotypic colistin resistance (Rebelo et al., 2018). The various phenotypic and molecular colistin (polymyxin) resistance-and mcr-detecting methods or assays described so far are comprehensively described below.

| Phenotypic tests: Screening media and MICdetermining tools
Simple and cheap (agar or broth) media that can easily but efficiently identify colistin-resistant Gram-negative bacteria while inhibiting nonresistant ones are crucial for surveillance purposes to early identify sources and carriers of these strains (Abdul Momin et al., 2017;Chew et al., 2017;Poirel, Joyal et al., 2017;Poirel, Larpin et al., 2017). But for the longer turnaround time of 18-24 hr required for these media, which is a major disadvantage, and lower mcr specificity and sensitivity compared to molecular methods, these screening media would be well-suited for less-resourced laboratories due to their lower costs and simple operating skill required (

BMD
BMD is currently the accepted and gold standard for colistin MIC determination and evaluation of the EA, CA, ME, and VME of other MIC-determining tools. Within the CLSI-EUCAST joint declaration document espousing the BMD as the method of choice for determining MICs, it has been recommended that nonpolystyrene-treated plates should be used (Jayol, Nordmann, Brink, et al., 2017;Jayol, Nordmann, Lehours, et al., 2017;Chew et al., 2017;Abdul Momin et al., 2017). This is because colistin can bind to or adsorb to plastics, thus reducing its concentration in the broth and ultimately affecting the MIC values; hence, colistin solutions should be stored in glass instead of plastics to maintain accurate concentrations (Jayol, Nordmann, Brink, et al., 2017;Jayol, Nordmann, Lehours, et al., 2017;Chew et al., 2017;Abdul Momin et al., 2017). Furthermore, it is recommended that sulfate salts of colistin instead of the colistimethate, which is used in human medicine, should be used in determining colistin MICs without adding polysorbate 80 (Jayol, Nordmann, Brink, et al., 2017;Jayol, Nordmann, Lehours, et al., 2017;Chew et al., 2017;Abdul Momin et al., 2017). However, agar dilution and disc diffusion methods were ruled out by the joint CLSI-EUCAST document because the larger molecular size of polymyxins  Bardet et al., (2017) A single study evaluated the ability of BMD to detect MCR-1positive Enterobacteriaceae where sensitivity to colistin and polymyxin B was, respectively, 71.4% and 81.0% at a breakpoint of ≤2 mg/L, or 90.5% and 85.7% at a cut-off of ≤1 mg/L (Chew et al., 2017) (Table 3). This study found that the MICs of polymyxin B and colistin were not interchangeable, although the results from selective culture media supplemented with either polymyxin B or colistin were found to be the same (Bardet et al., 2017;Nordmann et al., 2016a,b;Poirel, Larpin et al., 2017). It is thus obvious that BMD results cannot be relied on completely to detect mcr-positive isolates.
This is not surprising as mcr-1-positive strains have been found to be susceptible to colistin, and acquired colistin resistance is known to confer low-level colistin resistance (Chew et al., 2017). On the other hand, it suggests that decreasing the colistin MIC cut-off to ≤1 mg/L can increase the sensitivity of BMD and other commercial MICdetermining platforms to detect mcr-positive isolates (Chew et al., 2017). Thus, it is necessary to confirm mcr expression with molecular assays as BMD is not 100% MCR-sensitive.
Both the clinical significance of colistin-susceptible mcr-positive strains and a correlation between MICs and clinical outcome, as a guide to treatment, are still not well established. In addition, the clinical effect of heteroresistance is still unknown, especially when colistin is given as combination therapy (Chew et al., 2017). The BMD method however, cannot be adopted in routine clinical microbiology laboratories without some challenges. The method is seen by many as laborious and time consuming as it requires the manual preparation of antibiotic solutions, broths, etc., besides the 24-hr incubation time required to read results. The weighing of the powders requires precision, without which errors could be introduced (Jayol, Nordmann, Brink, et al., 2017;Jayol, Nordmann, Lehours, et al., 2017;Poirel, Larpin et al., 2017). These challenges make the other automated methods, none of which meets the CLSI's recommended performance standards for commercial AST systems (EA ≥ 90%, CA ≥ 90%, VME ≤ 1.5%, ME ≤ 3.0%) (Chew et al., 2017), and recently developed selective screening media that cannot determine MICs, more welcome and easily patronized (Nordmann et al., 2016a,b;Jayol, Nordmann, Brink, et al., 2017;Jayol, Nordmann, Lehours, et al., 2017).
Moreover, it has been already reported that the Vitek 2 has poor sensitivity in detecting heteroresistance (Jayol, Nordmann, Brink, et al., 2017;Jayol, Nordmann, Lehours, et al., 2017). A study using the agar dilution reference method to evaluate MicroScan found a CA of Thus, in terms of sensitivity, EA, CA, ME, and VME, Sensititre has so far been the most efficient followed by the BD Phoenix   Rebelo et al., (2018) and Phoenix 100™ instruments, whose results are interpreted by the BD Epicenter Software, have lower sensitivity for the mcr gene.
Moreover, strains with colistin MICs as high as 16-128 mg/L were not determined by the BD Phoenix instrument, which was possibly due to heteroresistance in those strains (Jayol, Nordmann, Brink, et al., 2017;Jayol, Nordmann, Lehours, et al., 2017). Further testing with larger samples comprising of most members of the Enterobacteriaceae and Gram-negative nonfermenters expressing all known colistin resistance mechanisms is necessary to confirm these preliminary findings, which were obtained with smaller sample sizes and nonrepresentative intrinsic and acquired colistin resistance isolates.  (Esposito et al., 2017). Through the inhibition of MCR-1 activity because of zinc chelation, the MCR-1-positive isolate is unable to maintain its resistance to colistin. Notably, the chelatorbased MCR-detecting tests were more efficient with E. coli than with other species such as K. pneumoniae (Coppi et al., 2017;Esposito et al., 2017). The remaining assays, which does not involve metal chelators, mainly detects colistin resistance, albeit they had very high MCR-1 sensitivity (Table 3). These assays are comprehensively described below.

Rapid Polymyxin NP test (RPNP) and commercial RPNP
The RPNP test is one of the novel colistin resistance-determining assays designed by Nordmann et al. (2016a,b)  A cut-of ≥8-fold increase in MIC was interpreted as mcr-positive while an MIC reduction of ≤2-fold was interpreted as mcr-negative. The authors used DA instead of EDTA because DA has a higher affinity for zinc. Moreover, when the authors used the disc diffusion method instead of the BMD with DA, they could not attain the same 100% sensitivity and specificity (Coppi et al., 2017). This is obviously due to the poorer diffusion of colistin through agar as already stated above. Similarly, Esposito et al. (2017) showed that the addition of EDTA to colistin discs yielded lower sensitivity and specificity, 96.7% and 89.6%, respectively, for mcr detection (Esposito et al., 2017).
Hence, disc diffusion-based tests with either DA or EDTA as a means to detect mcr-positive enterobacteria is not advised. This test has a turnaround time of 24 hr due to incubation. The CMR test was designed according to the EUCASTrecommended BMD method (Esposito et al., 2017;EUCAST, 2013), with the addition or nonaddition of 80μg/ml EDTA to wells containing 0.06-32 μg/ml (or to 512 μg/ml for intrinsic-resistant strains)

EDTA-based assays
colistin. However, CAMHB was not used in this assay as calcium and magnesium supplementation will chelate with EDTA and affect the test's outcome. In addition, calcium has been found to enhance the activity of putative PEtN transferases in E. coli (Esposito et al., 2017). nevertheless, false positives and negatives were recorded with this cut-off. The MCR-1 sensitivity and specificity of this method was, respectively, 96.7% and 83.3%, which makes it less efficient than the CDT method and more sensitive than BMD; it is thus inadvisable to use this test. Given the laborious nature of this test vis-a-vis that of the CDT, the CDT is more recommendable than the CMR assay.
The RPNP test was modified into the MRPNP test by the addition of two extra wells containing 80 μg/ml EDTA (without colistin) and 80 μg/ml EDTA plus 5 μg/ml colistin (Esposito et al., 2017).
All colistin-resistant isolates were positive for the PNP test, that is, changed color from orange to yellow, but only MCR-1-positive isolates were inhibited by EDTA, in that EDTA-containing wells resulted in no color change (orange) after incubating for 1-4 hr (Esposito et al., 2017). Thus, isolates with positive PNP results that showed no color change (orange) in the presence of EDTA were interpreted as MCR-1-positive while those with positive PNP results and color change (yellow) in the presence of EDTA were MCR-1 negative (Esposito et al., 2017). The sensitivity and specificity of this test was, respectively, 96.7% and 100.0% with a turnaround-time of <4 hr, which makes it better than the CDT and CMR tests in terms of time and efficiency in detecting MCR-1-positive E. coli.
The alteration of Zeta potential test bases on the resultant surface-membrane ionic charges of E. coli in the absence and presence of 80 μg/ml EDTA to detect MCR-1-producers. A ZetaPALS ZetaPotential Analyzer (Brookhaven Instruments Corporation, Holtsville, NY) was used to measure the particle size (diameter, mm) and Zeta potential ( was interpreted as MCR-1 positive; however, a false-negative result was obtained, possibly due to lower or no MCR-1 expression in that isolate (Esposito et al., 2017). The test had a sensitivity and specificity of 95.1% and 100.0%, respectively (Esposito et al., 2017), which makes it second to only the MRPNP test in terms of efficiency among the EDTA-based assays.
This study, for the first time, confirmed that colistin resistance resulted from reduction in surface anionic charges, which reduced colistin's binding affinity for lipid A (Esposito et al., 2017). It also showed that EDTA increased the anionic charges on the surface membrane of MCR-1-positive isolates such that the Zeta potential of MCR-1-positive strains became similar to that of colistin-susceptible ones. Thus, further Zeta potential alteration tests with a larger sample size and representation of all colistin resistance mechanisms is needed to confirm these findings as this test can be easily adopted and used in many well-resourced microbiology laboratories. colistin-resistant E. coli and can be adopted in human medicine due to its simplicity and low cost.

STC (Liofilchem, Italy) is a new commercial test kit for determining
colistin MICs for four isolates at a time. It comes with lyophilized colistin in seven-two-fold dilutions (0.25-16 μg/ml), with one additional well as growth control (Carretto et al., 2018). The test is similar to the BMD testing, albeit much simpler and limited to only colistin testing for a maximum of four test isolates. Following the manufacturer's as well as EUCAST and CLSI instructions, Carretto et al. (2018) evaluated the STC kit with 353 bacterial isolates and found it to have an EA of 96%, a CA of 98.9%, an ME of 0.92% and a VME of 1.46% (Table 3). Notably, an EA of 98.8% was recorded for MCR-1-positive isolates. However, a study by EUCAST with 75 isolates showed that STC had an EA of 88% with seven VMEs and one ME (Carretto et al., 2018). Thus, further evaluations with more isolates expressing diverse colistin resistance mechanisms is necessary to establish the relative efficiency of this kit in detecting colistin resistance.
The MIC results were read visually using turbidity, pinpoint colonies (Hafnia alvei) or buttons at the bottom of the wells. They also found the kit to be highly stable, reliable, and reproducible even at room temperature and varying temperatures. It was found to be highly reproducible (Carretto et al., 2018). Further, minimum bactericidal concentration could be also determined from the STC plate by spotting 1-10 μl from the wells unto CAMHA plates.

| Novel agar-based screening media: SuperPolymyxin ™ , CHROMagar COL-APSE and LBJMR media
There are currently three novel polymyxin (colistin/polymyxin B) resistance-detecting screening media (Tables 2 and 3) Some of them contain chromogenic compounds for species differentiation, but they all prevent swarming by Proteus spp. (Table 2). The components, sensitivity, specificity and limit of detection (LOD) of these media are described below.

SuperPolymyxin™
SuperPolymyxin, which is now marketed as a commercial patented product by ELITech Group solutions (www.elitechgroup. com/product/Superpolymyxin/) as SuperPolymyxin™, is the first screening medium developed to detect both intrinsic and acquired polymyxin-resistant Enterobacteriaceae isolates from clinical, environmental, food and fecal specimen (Nordmann et al., 2016a,b).
Higher colistin concentrations and presence of deoxycholates in earlier media inhibited strains with acquired resistance (MIC of 4-8 mg/L) due to lower colistin-resistance levels (Nordmann et al., 2016a,b;Bardet et al., 2017). Hence, the SuperPolmyxin was developed with 3.5 μg/ml colistin, EMB powder to selectively inhibit non-Gram-negative bacteria, 10 μg/ml daptomycin (because vancomycin potentiated colistin's activity against several Gramnegative bacteria) and 5 μg/ml amphotericin B ( Table 2). The stock solutions used for preparing the media could be stored at −20°C for a year (Nordmann et al., 2016a,b). The medium was evaluated with 88 Gram-negative bacteria and resulted in a sensitivity, specificity and LOD of 100%, 100% and 10 1 (10 1−2 for spiked stool samples) cfu/ml, respectively. Additional studies evaluating these two media are lacking and further tests are necessary to show the media with the best colistinresistance and MCR-detecting efficiency, particularly in identifying heteroresistant strains. It is, however, clear that the CHROMagar COL-APSE has a broader target spectrum than the SuperPolymyxin.

LBJMR
The Lucie-Bardet-Jean-Marc-Rolain medium is the most recent colistin-resistance screening medium to be developed for identifying colistin-resistant Enterobacteriaceae and Gram-negative nonfermenters as well as vancomycin-resistant Enterococci (VRE) from cultured bacteria and stool samples (Bardet et al., 2017).
Preliminary tests showed that purple agar base with glucose and bromocresol purple (Table 2) provided better results (sensitivity and specificity) than other combinations and media such as BD Cepacia medium and Columbia Colistin Nalidixic Acid agar+5% sheep blood (Bardet et al., 2017). Evaluation with 143 cultured bacterial isolates and 68 stool samples, followed by screening of 1052 stool samples from around the world, resulted in 100% sensitivity and specificity, with an LOD of 10 1 . The LBJMR medium inhibited Proteus spp. swarming 48 hr after incubation, and was as sensitive as SuperPolymyxin in detecting MCR-positive bacteria from stool samples and culture; however, it was more sensitive than SuperPolymyxin in identifying non-fermenters. Both Enterobacteriaceae and Enterococci appear as yellow colonies on the medium's purple background, but with different colony sizes (Bardet et al., 2017).
While the combination of daptomycin and colistin on EMB agar led to systematic inhibition of MCR-positive E. coli, particularly those with lower MICs, the addition of amphotericin B, vancomycin or daptomycin to LBJMR medium did not affect its sensitivity (Bardet et al., 2017). Rather, LBJMR medium detected low concentrations of pathogens in cystic fibrosis samples, including Burkholderia cepacia, than Cepacia medium (Bardet et al., 2017). Direct culturing of samples without prior decontamination is possible on the LBJMR medium. Primary cultures can also be directly analyzed, using PCR and AST on LBJMR without subculturing (Bardet et al., 2017). This medium still requires further multicentre studies and comparative evaluation with the SuperPolymyxin and CHROMagar-COL-APSE media.
However, its efficiency in detecting both colistin-and vancomycinresistant bacteria is very welcoming due to the clinical importance of colistin and vancomycin resistance.

| Molecular diagnostics
But for their cost and higher skill requirements, the shorter turnaround time and higher efficiency (100% sensitivity and specificity) of molecular diagnostics in detecting MCR-producing Enterobacteriaceae and mcr genes at very low concentrations in cultured bacteria as well as in clinical, fecal, environmental and food samples make them ideal tools (Tables 1 and 3). Microarray, LAMP, multiplex PCR and real-time PCR assays have so far been designed to directly or indirectly identify mcr genes in Enterobacteriaceae (Tables 1 and 3). These assays cannot confirm colistin resistance as colistin susceptible MCR-positive strains exist (Chew et al., 2017 be also tested on this instrument to ascertain its ability to identify mcr-1/-2 in diverse enterobacterial species. Each probe of the microarray consists of two arms with a universal primer binding site, target-specific gene sequence and a zip code for the hybridization.
Ligated and amplified probes were hybridized to the microarray, visualized with biotin-labeled primers and interpreted automatically with software . A major advantage to the microarray diagnostic technology is its potential to be upgraded with new or emerging resistance genes to detect more mcr gene types and variants. However, the cost and skilled involved will make it inaccessible to under-resourced laboratories.

| Loop-mediated isothermal amplification
A commercially available LAMP instrument called eazyplex ® SuperBug (Amplex Biosystems GmbH, Giessen, Germany) that detects mcr-1 from cultured bacteria with a turnaround time of ≤30 min was evaluated by Imirzalioglu et al. (2017) with 104 Enterobacteriaceae isolates: 67 were MCR-positive, 37 were MCR-negative and nine were intrinsically resistant to colistin (Imirzalioglu et al., 2017). The kit was 100% sensitive and specific.
The sample preparation was simple and can be used on the field (livestock farms, food processing plants or hospitals) for diagnostic purposes as the Genie ® II instrument is mobile and can last for 4 hr without power supply (Imirzalioglu et al., 2017). However, the kit's ability to directly detect mcr-1 from samples without preculture has not been assessed and it can only process six samples per hour; the additional cost for scaling up for more samples are unknown (Imirzalioglu et al., 2017). Given the mobility, efficiency and shorter turnaround time of this kit, it will be very advantageous to improve upon it by increasing the spectrum of mcr types and variants it can detect as well as enhance its ability to directly detect these genes in samples without culturing.

| Conventional, multiplex, & real-time PCR, and WGS
Conventional PCR and WGS are the gold-standards and first diagnostic tools used in identifying the mcr-1 gene from swine E. coli isolates Rebelo et al., 2018 Nijhuis et al. (2016) , to quantitatively and qualitatively detect mcr-1 genes in cultured bacteria and chicken feces. Using a total of 100 bacterial isolates from humans and animals, of which only 18 were mcr-1-positive (E. coli = 12, K. pneumoniae = 6), and 833 broiler fecal samples having five mcr-1positive strains, the assay proved to be 100% sensitive and specific with an LOD of 10 1 -10 8 DNA copies (Tables 1 and 3). This assay could directly identify mcr-1 genes from biological samples with high specificity due to the Taqman ® probes used. As explained above, the number and species diversity of the MCR-1-harboring strains used in this study were nonrepresentative. Hence, the assay needs to be subjected to further evaluations with larger mcr-containing samples and species.
Three SYBR ® Green-based real-time PCR assays have so far been designed to detect mcr-1, mcr-2 and/or mcr-3 in Enterobacteriaceae. Bontron, Poirel, and Nordmann (2016) first designed a SYBR ® Greenbased real-time PCR to detect only mcr-1 from cultured bacteria and directly from human and stool samples (Tables 1 and 3) (Bontron et al., 2016). Only eight of the 19 isolates used in this study were mcr-1-positive. The assay was 100% sensitive and specific, with an LOD of 10 2 cultured bacteria. This study thus needs further evaluation with a larger sample size and different enterobacterial species. Dona et al. (2017) also designed a SYBR ® Green real-time assay to identify mcr-1 from fecal and cultured samples (Dona et al., 2017). However, they had to first suspend the fecal samples in Luria Bertani (LB) enrichment broth overnight followed by subsequent plating on selective agar plates supplemented with 4 mg/L colistin to get 100% sensitivity and specificity. Using native stools directly without an enrichment step resulted in lower sensitivity, which could be due to the presence of PCR inhibitors, inadequate or few mcr-1-producing strains and/or the negative effect of longterm storage (at −80°C without a cryoprotectant) of stool samples (Dona et al., 2017). The assay was also used to test DNA directly extracted from native and enriched stool samples, with native nonenriched stool samples resulting in higher cycle thresholds (Ct) of between 34.37 and >40 while enriched stools resulted in lower Ct values of 21-23. A total of 88 stool samples from volunteers were used in evaluating this assay, which had an LOD of 10 1 DNA copies/reaction (Dona et al., 2017). Evidently, the enrichment step used in this assay substantially increased the sensitivity and reduced the Ct value of this assay and could be adopted in other assays. However, the enrichment step takes at least 12 hr, making it time-consuming. The assay's sensitivity in directly determining mcr-1 in human stools is thus limited, without an enrichment step, compared to other molecular tests.
Another SYBR ® Green-based real-time assay was designed by Li et al. (2017) to detect mcr-1, mcr-2, and mcr-3 genes in Enterobacteriaceae from clinical, soil, and fecal specimen. This is the second molecular real-time PCR assay designed to detect mcr-3 genes. The test was evaluated with 25 isolates plus mcr-1, -2, and -3-containing mutants, resulting in 100% sensitivity and specificity, with an LOD of 10 2 cultured bacteria. A copy of mcr-1 gene per 10 −6 16S rRNA copies could be detected. However, the assay could not detect all three mcr genes in a single reaction as is obtainable from a Taqman assay (Li et al., 2017). Further developments in this assay to increase the number of mcr gene types that can be detected and enable it to detect all mcr genes in a single reaction will make it one of the best molecular diagnostic tool available.
Two most recent molecular assays that can detect mcr-1, -2, -3, -4 and -5 genes were designed by Rebelo et al. (2018) and Lescat et al. (2018) using a multiplex PCR, agarose gel (1.5% and 2.5%, respectively) electrophoresis and ethidium bromide staining (Lescat et al., 2018;Rebelo et al., 2018). Although Rebelo et al.'s assay was designed to screen for mcr genes in E. coli and Salmonella spp. from animals (calves and pigs) in Europe, it can be extended to humans as a screening agent in well-resourced laboratories. The assays were 100% sensitive and specific, and, respectively, found mcr-1 and mcr-4, mcr-1 and mcr-3, as well as mcr-1 and mcr-5 genes in single E. coli and Salmonella enterica isolates in a single reaction, meaning that the assays can identify single and multiple mcr genes in single strains. Lescat et al. (2018)'s method had a shorter turnaround time (<2 hrs) and used internal controls, which were absent in that of Rebelo et al. (2018) (Lescat et al., 2018;Rebelo et al., 2018). An isolate with an MIC of 2 mg/L was found to harbor an mcr-1 gene, which suggests the need to revise the epidemiological cut-off for colistin in Enterobacteriaceae as well as introduce an intermediate resistance as suggested by Chew et al. (2017). This is necessary to identify MCR-positive but susceptible strains that will otherwise not be detected.

| CON CLUS ION
The rapid expansion and dissemination of the mcr gene across bacterial species and regional boundaries is a major cause for concern, underscoring the urgency for better, simpler and cheaper diagnostic tools that can quickly and effectively detect colistin-resistant bacteria. Among the currently available diagnostic tools, the RPNP test, which has a turnaround time of ≤2 hr, and/or the LBJMR, SuperPolymyxin, or CHROMagar COL-APSE medium will be ideal for under-resourced laboratories due to their lower cost as initial screening tools. This can be followed up with the colistin MAC or MRPNP tests, which, respectively, have turnaround times of 24 hr and <2 hr, to identify mcr producers. Colistin-resistant strains can be sent to well-resourced laboratories for further molecular tests if necessary (Figure 4). For well-resourced laboratories, the multiplex PCR assay, the Taqman or SYBR Green real-time PCR assays can be directly used on cultures or on biological and environmental samples alongside the RPNP test and/or LBJMR, SuperPolymyxin, or CHROMagar COL-APSE medium to simultaneously identify mcr producers and colistin-resistant isolates. Further species identification and typing can be undertaken with an API kit or MALDI-TOF MS and PCR or WGS, respectively ( Figure 4).
Going forward, further evaluations and modifications of available tests and methods should be undertaken to improve on the sensitivity, specificity, turnaround time, and costs. Moreover, periodic surveillance of hospitals, farms, foods, and the environment should be undertaken to quickly detect and contain colistin-resistant and mcr-producing strains from further dissemination. This is necessary to obtain the true prevalence of colistin resistance and mcr genes, and inform colistin stewardship, treatment guidelines, and protocols for colistin-resistant infections.

CO N FLI C T O F I NTE R E S T
The author has no conflict of interest or transparency to declare.