Tunisian Multicenter Study on the Prevalence of Colistin Resistance in Clinical Isolates of Gram Negative Bacilli: Emergence of Escherichia coli Harbouring the mcr-1 Gene

Background: Actually, no data on the prevalence of plasmid colistin resistance in Tunisia are available among clinical bacteria. Objectives: This study aimed to investigate the current epidemiology of colistin resistance and the spread of the mcr gene in clinical Gram-negative bacteria (GNB) isolated from six Tunisian university hospitals. Methods: A total of 836 GNB strains were inoculated on COL-R agar plates with selective screening agar for the isolation of GNB resistant to colistin. For the selected isolates, mcr genes, beta-lactamases associated-resistance genes and molecular characterisation were screened by PCRs and sequencing. Results: Colistin-resistance was detected in 5.02% (42/836) of the isolates and colistin-resistant isolates harboured an ESBL (blaCTX-M-15) and/or a carbapenemase (blaOXA-48, blaVIM) encoding gene in 45.2% of the cases. The mcr-1 gene was detected in four E. coli isolates (0.59%) causing urinary tract infections and all these isolates also contained the blaTEM-1 gene. The blaCTX-M-15 gene was detected in three isolates that also carried the IncY and IncFIB replicons. The genetic environment surrounding the mcr-carrying plasmid indicated the presence of pap-2 gene upstream mcr-1 resistance marker with unusual missing of ISApl1 insertion sequence. The Conclusions: This study reports the first description of the mcr-1 gene among clinical E. coli isolates in Tunisia and provides an incentive to conduct routine colistin susceptibility testing in GNB clinical isolates.


Introduction
Colistin has been considered one of the last-resort antibiotics, for the treatment of serious infections caused by extensively drug resistant Gram negative bacteria (GNB) mainly carbapenem-resistant Enterobacteriaceae (CRE). In recent years, multiple studies have described the rapid increase in the prevalence of colistin resistance among Enterobacteriaceae, Acinetobacter spp. and Pseudomonas spp. [1,2]. In 2015, Liu et al. identified the mobile colistin resistance gene, mcr-1, primarily in Enterobacteriaceae, mainly Escherichia coli and K. pneumoniae [3]. Since then, several reports have indicated that mcr-1 has silently spread worldwide since 1980. To date, nine different plasmid-encoded colistin resistance genes have been described (mcr-1 through mcr-9, that have been isolated from bacteria in human, animal and environmental samples [4]. Of particular concern is the dissemination of the mcr gene in CRE and/or extended-spectrum-beta-lactamase (ESBL)-producing GNB, potentially leading to pan-drug-resistant isolates [3,5]. Until the discovery of mcr-1, all reported polymyxin resistance mechanisms were chromosomally mediated, due to mutation and regulatory changes [3], and had never been reported to be transmitted horizontally [3].
With the emergence of mcr determinants in bacteria from animals and humans, as well as the continuous and global use of colistin in both clinical and non-clinical settings, the surveillance of mcr variants in bacteria from clinical, veterinary and environmental sources is needed. Recently, there have been numerous studies on the emergence of the mcr-1 gene in humans, animals, food and environment worldwide [2][3][4][5]. In Tunisia, several studies confirmed the presence of this gene in bacteria isolated from healthy and sick animals [6][7][8][9], but to the best of our knowledge, there has not been a report of infection caused by a pathogen harbouring an mcr gene. [1,10,11]. Besides, colistin susceptibility testing was routinely performed only for carbapenem resistant GNB, which may contribute to the unawareness of resistance.
The aim of this work was to investigate the prevalence of colistin-resistance and to determine the occurrence of the mcr genes in clinical GNB isolates collected from human samples in six Tunisian university hospitals.

Study Design
During a two-month period (March-April 2019), a multicenter-prospective study including all GNB strains was conducted in six different university hospitals.

Sample Processing and Microbial Study
A total of 836 GNB strains, consecutively isolated from clinical samples, were collected from several samples (mainly from urine, blood, sputum and pus). All specimens were treated according to standard microbiological procedures (REMIC) [12]. Only one isolate per patient was included and isolates from colonization screening studies were not considered. Species intrinsically resistant to colistin were excluded. Isolates were firstly identified based on colony morphology and Gram stain. Then all colonies were tested for oxidase by adding bacterial inoculum on the cotton tipped swab already impregnated by the N,N,N,N tetramethyl-paraphenylenediamine reagent. According to the results of the Gram stain and oxidase test, the API 20E and API NE systems (bioMérieux, Marcy l'Etoile, France) were used for biochemical identification of Enterobacteriaceae and non-Enterobacteriaceae, respectively. API strips test containing 20 wells with dehydrated substrates to detect enzymatic activity. A bacterial suspension was used to rehydrate strip wells before incubation. All results from the tests were compiled to obtain a profile number, which is then compared with profile numbers in the API-Web software to determine the identification of the bacterial species. E. coli ATCC 25922 and P. aeruginosa ATCC27853 were used as control strains.
All participants followed the same protocol for the initial colistin resistance (COL-R) GNB screening. A bacterial suspension of 0.5 McFarland was prepared from isolated colonies on trypticase soy agar. Then a loopful (10 µL) of each GNB suspension was streaked onto a selective agar medium CHROMagar™ COL-APSE (CHROMagar, Paris, France) and incubated overnight at 37 • C. The results were defined by observing different coloured colonies, using the manufacturer's interpretation criteria for identifying COL-R GNB. To avoid the inoculum effect problem, only colonies that have grown in the second dial were taken.
All the selected isolates were stored in brain heart infusion broth with 20% glycerol at −80 • C during the data collection period. Then, they were sent to the National Reference Laboratory at CNH for further investigations (colistin susceptibility testing, mcr-screening and genotyping, etc).

Screening for mcr Genes and Determination of Its Genetic Environment
A multiplex PCR assay was used to detect the mobile colistin resistance genes in all colistin resistant isolates, mcr-1 to mcr-9, using already described protocols and primers (Supplementary Materials) [4,14]. The genetic environment of the mcr-1 gene was determined by PCR mapping using specific primers as previously described (Supplementary Materials) [6,15]. Indeed, To screen for the upstream presence of ISApl1, mcr-1-positive strains were examined with PCR using primers ISApl1-mcr-F and ISApl1-mcr-R.

Characterization of Isolates Harbouring mcr Gene Conjugation Assays and Plasmid Replicon Type
The transfer of the mcr gene was investigated by conjugation assays using E. coli J53 as the recipient strain. MH agar plates (Bio-Rad) containing sodium azide (200 mg/L) supplemented with colistin sulfate (2 mg/L) and/or cefotaxime (2 mg/L) was used to isolate transconjugants. PCR-based replicon typing method (PBRT) was used to determine the plasmid incompatibility groups (Inc), among donor and transconjugants strains (Supplementary Materials) [18].
Detection of E. coli Phylogenetic Groups Phylogenetic groups were determined by multiplex PCR targeting four genes (chuA, yjaA, TspE4.C2, and arpA) allowing the isolates to be classified into seven main phylogenetic groups (A, B1, B2, C, D, E, and F) and cryptic clade I as described previously (Supplementary Materials) [20].

Ethical Approval
The strains investigated in this study were issued from bacteriological diagnostic of pathogenic specimens routinely received by the different microbiology departments participating in this study. Therefore ethical approval and consent were not required.
The mcr-1 gene was detected in four E. coli isolates with colistin MIC ranging from 16 mg/L to 32 mg/L. They were responsible for urinary tract infections, communityacquired in three of four cases (Table 3). All mcr-1 positive isolates (n = 4) co-harboured bla TEM-1 , while three of them (E2, E3 and E4) were ESBL producers co-harbouring the bla CTX-M-15 gene.
The conjugation assay was positive for only one (E1) of the four E. coli isolates carrying mcr-1 despite repeated attempts. The presence of the mcr-1 and bla TEM-1 genes was confirmed in the E. coli J53 transconjugant by standard PCR. Other antibiotic-resistant phenotypes (fluoroquinolones and tetracyclin) could not be co-transferred with colistin. Colistin MIC of the transconjugant showed a 4-fold decrease in the MICs in comparison with the donor E. coli « E1 » (16 mg/L). The analysis of the transconjugant revealed that the mcr-1 plasmid was untypeable by PBRT. The genetic environment surrounding this plasmid was mcr-1-pap2. PCR results showed missing ISApl1 in the upstream region of mcr-1.
The three other mcr-1 positive E. coli isolates (E2, E3 and E4) carried IncY and IncFIB replicons. Their phylogenetic analysis revealed that they belonged to the phylogroup D. The remaining mcr-1-carrying E. coli was untypeable.

Discussion
The present study described the prevalence of colistin resistance in clinical GNB isolates from six different university hospitals in Tunisia as well as the emergence of the mcr gene. Colistin resistance was found in 5.02% of the isolates. Colistin susceptibility is typically tested in multidrug resistant clinical isolates for therapeutic reasons. Thus, until now the prevalence of colistin resistance among clinical isolates still remains underestimated. In our study, only 45.2% of colistin-resistant isolates harboured ESBL and/or carbapenemase genes.
This study reported, for the first time, the emergence of human clinical E. coli isolates harbouring the mcr-1 gene causing urinary tract infections in Tunisia with a high prevalence of 10% (4/42) among colistin-resistant GNB compared with prevalence rates reported from other studies [3,21]. To date, in our country, multiple studies have reported the dissemination of colistin resistance among GNB, mainly among K. pneumoniae causing nosocomial infections [1,11], which was mostly linked to alterations identified within the mgrB gene but not related to plasmid-encoded colistin resistance genes mcr-1 to mcr-9 [1,10,11]. Previous Tunisian studies have described the dissemination and the high prevalence of E. coli strains harbouring mcr-1 in animals, including healthy food-producing animals and diseased animals [6][7][8]. Overall, with the extensive use of colistin in hospitals, colistin resistance has already spread worldwide [2]. Besides, there are also reports of colistin resistance in humans who have not previously received this drug. Colistin use in animals suggests that animals may be an important source of transmission of colistin resistance to humans [2].
According to the medical record review about patients who acquired E. coli harbouring the mcr-1 gene, the onset of patient symptoms occurred before the consultation of emergency and urology wards. These data suggested the community origin of the urinary tract infection caused by isolates harbouring the mcr-1 gene. In order to support this hypothesis concluded during this short surveillance period, further studies are needed to evaluate the spread of plasmid colistin resistance outside hospitals and control their transmission.
The four mcr-1 positive E. coli isolates were MDR and carried the bla TEM-1 gene, while three of them were ESBL positive, co-harbouring the bla CTXM-15 gene. The co-existence of mcr-1 with ESBL has been frequently reported [9,21]. One study suggests that the co-occurrence of mcr-1 and bla CTXM-15 is perhaps due to intricate genetic actions taken under antibiotic pressure due to the integration of mcr-1 into the bacterial chromosome in some strains. Most of the mcr-1 positive strains have been proven to be carrying different beta-lactamase genes including pAmpC, ESBL, and carbapenemase genes [6,9], resulting in the emergence of veritable "superbugs" which can threaten human health [5].
In this study, the detection of the mcr-1 gene among E. coli « E1 » isolates susceptible to third-generation cephalosporins and carbapenems highlights the difficulty of surveillance of plasmid mediated resistance to colistin. In fact, colistin susceptibility testing is restricted to carbapenem-resistant isolates in our country suggesting the potential role of such isolates in the silent dissemination of the mcr-1 gene among other GNB.
The genetic environment of the mcr-1 gene lacked the ISApl1 element, avoiding the threat of IS-mediated transmission. As previously described, ISApl1 was presumably involved in the initial transposition of the mcr-1 element and then lost for stabilization of mcr-1 on plasmids.
Colistin resistance was successfully transferred to the E. coli J53 strain via a conjugation experiment, suggesting that the mcr-1 gene is located on a transferable plasmid. This result confirms the finding that the mcr-1 gene is mobilized on plasmids that have spread to different Enterobacteriaceae [5].

Conclusions
In conclusion, colistin resistance was detected in 5% of the clinical Gram-negative bacteria collected from six Tunisian university hospitals. The main species found were P. aeruginosa, E. coli and K. pneumoniae. Colistin-resistant was associated with other resistance markers including ESBL (bla CTX-M-15 ) and or carbapenemases (bla OXA-48 , bla VIM ). We detected the emergence of sporadic mcr-1-positive E. coli strains in patients with urinary tract infections in Tunisia. Furthermore, the co-existence of colistin-resistant and ESBL genes; in this study suggests their possible spread to other human and environmental pathogens. Unfortunately, colistin susceptibility testing for non-carbapenem-resistant Enterobacteriaceae is not a routine procedure in diagnostic laboratories. Therefore, continuous colistin resistance surveillance studies are necessary, and the prescription of this antibiotic should be controlled.