Detection of colistin resistance among multidrug-resistant Klebsiella pneumoniae and Escherichia coli clinical isolates in Turkey

In this study investigation of plasmid-mediated mcr 1-5 resistance genes was performed among multidrug-resistant (MDR) colistin sensitive and resistant Klebsiella pneumoniae and Escherichia coli strains isolated in our laboratory. We aimed to evaluate automated system (Vitek-2), broth micro-dilution (BMD) reference method and chromogenic media performance. Totally 94 MDR K. pneumoniae and six E. coli isolates were included in the study. CHROMID ® Colistin R agar (COLR) (bioMerieux, France) was used to determine the colistin resistance by chromogenic method. Standard PCR ampli ﬁ cation was performed using speci ﬁ c primers to screen the plasmid-mediated mcr 1-5 genes. Sixty-one isolates were resistant to colistin and 39 were susceptible with reference BMD. The essential and categorical agreement of Vitek-2 was determined as 100 and 99%. The sensitivity of COLR medium was 100%, the speci ﬁ city was 97.5%. In our study mcr-1 was detected in eight isolates, while other mcr genes were not detected. Due to the high sensitivity and speci ﬁ city of the COLR medium, it can be used in routine diagnostics for the detection of colistin resistance. In our study we detected 8% prevalence of mcr-1 among MDR strains however, two mcr-1 positive isolates were found sensitive to colistin by


INTRODUCTION
Colistin is a polycationic antibiotic isolated from Paenibacillus polymyxa subsp.colistinus, effective against Gram-negative bacteria [1].Use of colistin was abandoned since the 1970s due to its nephrotoxic side effect.The dramatic increase of multidrug-resistant (MDR) Gramnegative bacterial infections, especially the emergence of carbapenem resistance, revived the use of colistin as a last option antibiotic.
Colistin resistance can be intrinsic or acquired.The most common cause of resistance development is mutation in genes encoding the regulatory system responsible for lipid A synthesis.As a result of mutation, negative charge of outer membrane and colistin binding are reduced.Mutations in the PmrA-PmrB or PhoP-PhoQ regulatory system are chromosomally encoded resistance mechanisms [2].
Plasmid-mediated colistin resistance encoded by mcr-1 gene, was identified in an Echerichia coli isolate in China, in 2015.Despite there is no clinical use of colistin until 2017 in China, identification of mcr-1 and after a while the identification of variants of the mcr-1 and the other mcr 2-8 genes from human, animal and environment isolates, has showed that colistin resistance is a global public health problem.Because plasmid-mediated colistin resistance can spread horizontally, colistin sensitivity should be determined quickly and reliably [3].
Polymyxin susceptibility tests are technically problematic for many reasons.Disk diffusion and gradient strip test methods have poor performance due to poor agar diffusion of large and cationic colistin molecule and these methods yield inconsistent results [4].The binding of polymyxins to plastic surfaces (pipette tip, titration plate, etc.) is another technical reason that makes it difficult to detect colistin sensitivity [3].Although broth microdilution (BMD) method is the recommended technique, it is limited in routine laboratory practice due to the difficulty and the time it takes [5,6].There are several commercially available selective culture media [CHROMagar COL-APSE, Super Polymyxin media, CHRO-MID ® Colistin R Agar (COLR)] to identify and rapidly detect the polymyxin resistant bacteria from both culture and stool samples.In our study, it was aimed to examine the usability of the culture medium in routine laboratory studies for screening and qualitative colistin resistance detection by comparing the COLR medium and BMD results.
In this study, investigation of plasmid-mediated mcr 1-5 resistance genes was performed in MDR colistin sensitive/ resistant strains.It was also aimed to evaluate the performance of commercial chromogenic media produced for ease of use in routine detection of colistin resistance together with the automated system (Vitek-2) and reference BMD results.

MATERIAL AND METHODS
Ninety-four multidrug resistant Klebsiella pneumoniae and six E. coli strains isolated from various clinical specimens (37% blood, 35% urine, 9% wound, 7% tracheal aspirate, 12% other) between 2017 and 2019 were included in our study.The identification and antibiotic susceptibility tests of the strains were performed using MALDI-TOF MS (bio-M erieux, France) and Vitek-2 (bioM erieux, France) system.

BMD reference method
Reference BMD test was performed using polystyrene microplate and colistin sulfate salt (Sigma-Aldrich C4461, USA) according to ISO-standard (20776-1) recommendations.A stock solution was made by dissolving 25.6 mg of colistin in 10 mL sterile distilled water (dH 2 O) and from this a 1/10 diluted solution containing 256 mg/L colistin was prepared from the stock solution for the use.Minimum inhibitory concentration (MIC) was studied in the 0.125-128 mg/L dilution range.A suspension suitable for 0.5 McFarland turbidity (1.5-2 3 10 8 colony forming units; cfu/mL) was prepared in sterile saline water by using bacterial colonies from fresh culture by direct colony suspension method.The bacterial suspension was diluted using CAMHB (Mueller-Hinton II Broth Cation-Adjusted Becton-Dickinson, 212322, France) and distributed to a final bacterial concentration of 5 3 10 5 in the wells.BMD plates were incubated at 35-37 8C for 18-24 h.After incubation, the results were evaluated independently of each other in a double eye control.Quality control was performed with colistin susceptible ATCC 25922 E. coli and colistin resistant NCTC 13846 (mcr-1 positive) E. coli strains.Strains with MIC values ≤2 mg/L were considered susceptible and strains with MIC >2 mg/L were considered resistant in accordance with the EUCAST version 10.0 recommendations [5].The essential agreement (EA) (MIC results within ±1 dilution) and categorical agreement (CA) (number of overlapping (S,R) results) between Vitek-2 and reference method; the rates of major errors (ME) (susceptible strains with the reference method that are detected as false resistant) and very major errors (VME) (resistant strains with the reference method that are detected as false susceptible) were calculated.Results were evaluated according to ISO criteria (EA and CA 90%; ME and VME <3%) [6].

CHROMID ® Colistin R Agar
For the CHROMID ® Colistin R Agar (COLR) (bioMerieux, France) qualitative resistance detection procedure, 0.5 McFarland (1.5-2 3 10 8 CFU/mL) bacterial suspension was prepared with sterile saline from pure colonies grown in selective/non-chromogenic medium in line with the company recommendations.Ten mL of sample from the 10 6 CFU/mL bacterial suspension prepared by diluting 1/100, was inoculated on CHROMID ® Colistin R Agar.After 18-24 h of incubation, culture medium were evaluated for bacterial growth.E. coli was seen as pink-burgundy and K. pneumoniae as blue-green colonies on the COLR medium (Fig. 1).For the medium quality control, NCTC 13846 E. coli strain (mcr-1 positive) and colistin susceptible ATCC 25922 E. coli strain were used in accordance with the manufacturer's recommendation.

Detection of mcr genes by PCR
DNA isolation of strains was performed using bacterial DNA isolation kit (Canvax Biotech, Spain) for mcr 1-5 screening by molecular method.PCR amplification was performed using the primers and PCR conditions listed in Table 1.Amplicons obtained after PCR were detected in 2% agarose (Canvax Agapure Agarose LE) gel electrophoresis using 50-1,000 bp (Canvax BrightMAX) DNA ladder.Big-Dye PCR (initial denaturation: 96 8C 1 min; 253; 96 8C 10 s, 50 8C 5 s, 60 8C 4 min) using amplicon purification and Canvax Clean-Easy purification kit were performed on PCR products with positive results.PCR products were purified using Canvax DNA purification SPRI magnetic beads kit.Positive results were confirmed by sequencing with the ABI PRISM ® 310 Genetic Analyzer (Applied Biosystems, USA).
Ethics Committee Approval: Ethics Committee approval was not taken as the study was performed from the collection material.

RESULTS
Sixty one of the strains were found to be resistant to colistin and 39 were susceptible with the reference method.All 61 colistin resistant isolates were identified as K. pneumoniae.All six E. coli isolates were colistin susceptible.The resistance rates of the isolates for carbapenems was 96%, for cephalosporins 100%, for fluoroquinolones 100% and for aminoglycosides was 93%.MIC values of the quality control strains were found in the expected range (ATCC 25922 MIC: 0.5 mg/L, NCTC 13846 MIC: 4 mg/L).The essential and categorical agreement of our Vitek-2 results and the reference method was determined as 100 and 99%, respectively.While no ME was detected with Vitek-2, a VME was detected in one strain (1.6%).The results are summarized in Table 2.
All 61 strains that were resistant to colistin with reference BMD grew on COLR medium.Colistin susceptible strains (n 5 39) did not grow on the COLR medium except one.The MIC value of the mcr-1 positive strain with false positivity in the medium was found to be 2 mg/L (sensitive) with the reference method.The sensitivity of the COLR medium was 100% and its specificity was 97.5%.The results are summarized in Table 2.
In our study plasmid-mediated mcr 1-5 genes were investigated by PCR method, the expected (309 bp) band size for mcr-1 gene region was obtained in 10 isolates (Fig. 2); no positivity was detected for other mcr (mcr 2-5) genes.The amplicons with positive results were confirmed by sequence analysis for eight isolates, and the sequence verification could not be made in two isolates with weak bands.Six positive isolates were resistant to colistin with reference BMD, and two isolates were susceptible to colistin.The results are summarized in Table 2.

DISCUSSION
The increasing use of colistin worldwide has led to an increase in polymyxin resistance, especially in countries where carbapenem resistant Enterobacterales strains are endemic [7,8].
In order to phenotypically determine colistin resistance with alternative methods, rapid tests such as Polymyxin NP and the usability of various selective media in routine studies are being investigated extensively [9][10][11].We found the results are highly compatible with the reference method in our study where we investigated the qualitative detection of   colistin resistance rates using the selective chromogenic screening medium COLR agar.Fernandez et al. in their screening study with 59 Enterobacterales, 20 of which were mcr-1 positive, found the specificity and sensitivity of the COLR medium as 100 and 88.1%, respectively and commented that the medium could be used to screen colistin resistance for Enterobacterales, including strains carrying mcr-1, despite false negative results in an E. coli and a K. pneumoniae strains [9].Girlich et al. compared performance of Superpolymyxin medium (ELITechGroup) and COLR medium, found the sensitivity (86.8%) and specificity (97.9-100%) of both media to be high and concluded that both media could be used as a useful method for screening colistin resistance in routine studies [10].Roche et al. using 46 colistin resistant, 20 colistin susceptible strains and 61 stool samples, determined the sensitivity of the COLR medium as 95% and the specificity as 100% in their study and commented that the culture medium provides an advantage compared to traditional methods in the differentiation of species known to carry mcr and in the evaluation of mixed cultures [11].Vitek-2 system is routinely used in our laboratory to detect antibiotic sensitivity results.Very different rates are reported regarding the performance of automated systems in detecting colistin sensitivity.Pfennigwerth et al. have compared the detection of colistin resistance with six different methods in 206 K. pneumoniae strains.While Vitek-2 system didn't detect any ME, in 12 strains VME's were detected; EA and CA were 81.7-94.2%[12].In our study, the EA and CA (>90%) between the Vitek-2 and reference method, ME and VME (<3%) results also were found suitable according to ISO criteria.Despite the high agreement of the results we obtained with the Vitek-2 system with the reference method, there are also publications stating that the VME rates of Vitek-2 systems are unacceptable (VME: 36-57.4%)[4].
In a study conducted by Liu et al. in 2016, they reported that the mcr-1 gene encoded by the plasmid was responsible for colistin resistance in an E. coli strain isolated from animal samples [13].Since the first definition of the mcr gene region around the world, seven different members (mcr 2-8) of the mcr family have been identified [14,15].In our study, only plasmid-mediated colistin resistance genes (mcr 1-5) were investigated in isolates, most of which are multidrug resistant strains (94% K. pneumoniae, 6% E. coli) isolated from clinical samples.The expected (309 bp) band size was obtained for the mcr-1 gene region in 10 isolates, and amplicons detected positive were confirmed by sequence analysis for 8 isolates (8%).No positivity was detected for other mcr genes (mcr 2-5).In addition to our colistin resistant isolates with mcr-1 (n 5 6), the detection of mcr-1 in two strains found to be susceptible to colistin indicates the presence of non-functional mcr genes, as emphasized by other researchers and shows that the mcr genes should be screened epidemiologically in colistin resistant samples as well as in colistin sensitive samples [3] € Ozhelvacı in his study that included 40 colistin resistant K. pneumoniae strains isolated from various clinics, could not detect mcr-1 gene, but they detected mcr-2 gene in two isolates [19].In a study investigating mcr 1-2 gene regions in 2049 Gram-negative bacterial strains isolated from urine of patients with urinary infections in Switzerland, an E. coli strain was found to be mcr-1 positive, while no strain with the mcr-2 gene was found.The strain found to be mcr-1 positive in the study was found to be susceptible to colistin, and it was interpreted that the spread of the mcr gene could be hidden [20].
Zhong et al. found the mcr-1 positivity with the standard PCR method in 2.1% of 144 E. coli isolates isolated from patients with bloodstream infection.All strains found to be mcr-1 positive (n 5 3) were also found resistant to colistin using the reference BMD method [21].Quan et al. in 2066 strains (1495 E. coli and 571 K. pneumoniae) isolated from bloodstream infections, mcr-1 positivity was found as 1%.While 20 of the 21 mcr-1 positive isolates were found to be colistin resistant (MIC: 4-32 mg/L) with BMD, one strain was found to be colistin susceptible (MIC: 0.06 mg/L) and this was interpreted that it may be due to the dysfunction of the gene in some species [22].Arabacı et al. have investigated the mcr-1 gene in 57 K. pneumoniae strains isolated from blood cultures and the mcr-1 was detected in three isolates (5.7%); all mcr-1 positive strains were found resistant by BMD method [23].In a study of Lee et al. conducted in 2019 in which the presence of mcr 1-5 genes in different bacterial groups was investigated, mcr-1 was detected in two (9.1%) of 22 colistin resistant K. pneumoniae isolates and in one of the two E. coli isolates.From these three isolates, two were found resistant and one sensitive (MIC 2 mg/L) with BMD [24].Zhang et al. have investigated mcr 1-5 genes in human vaginal samples and they detected high percentage of mcr-4 (12.7% mcr-4, 1.5% mcr-2, 1.5% mcr-3, 0.7% mcr-1, 0.7% mcr-5).They also found that animal and human mcr genes in their cities are identical with phylogenetic studies [25].In a comprehensive study investigating colistin mcr 1-9 genes in carbapenem resistant clinical isolates collected between 2014 and 2019 in China, mcr-1 positivity in E. coli strains was determined as 2.1%; It was also determined that it increased to 6.3% after 2017 [26].In the study where other mcr variants were not detected, no interpretation was made for isolates mcr-1 positive that were found to be susceptible to colistin.
As a result, the remarkable rate of mcr-1 positivity detected in our study, which consisted of clinical isolates, although in a limited number, is an important epidemiological finding.In the future, the situation in our country will be revealed in real terms only with large series and multidisciplinary contribution and thus healthier results can be obtained regarding the spread, and source of resistance and measures can be taken.In addition, due to the high sensitivity and specificity of the COLR medium we tested, it has been concluded that it can be useful in the practical and rapid detection of colistin resistant strains and can be used in routine studies.

Table 1 .
Primers and PCR conditions

Table 2 .
Summary results of 94 K. pneumoniae and 6 E. coli strains