Study of mcr-1 Gene-Mediated Colistin Resistance in Enterobacteriaceae Isolated from Humans and Animals in Different Countries

In this study, we aim to characterize the genetic environment of the plasmid-mediated colistin resistance gene mcr-1 in 25 Escherichia coli and seven Klebsiella pneumoniae strains from different countries and continents. Multilocus sequence typing, conjugation experiments, plasmid typing, and the presence and location of the insertion sequence ISApl1 were investigated. Whole genome sequencing of four E. coli was performed to analyse the genetic environment of the mcr-1 gene. Colistin minimum inhibitory concentration of mcr-1 strains varied from 3 to 32 µg/mL. Six E. coli sequence types were detected: ST 4015, ST 3997, ST 10, ST 93, ST 48, and ST 648. IncHI2, IncI2, and IncP plasmid types were predominant and were unrelated to a specific country of origin. ISApl1 was found in 69% of analysed plasmids that were mainly around the mcr-1 gene. Analysis of four closed mcr-1 plasmids revealed the integration of mcr-1 into hotspots. We found that the spread of mcr-1 gene was due to the diffusion of a composite transposon and not to the diffusion of a specific plasmid or a specific bacterial clone. The ease with which the mcr-1 gene integrates into various regions facilitates its dissemination among bacteria and explains its large diffusion all over the world, both in animals and in humans.


Introduction
Antibiotic resistance is a major issue around the world.
This phenomenon has led clinicians to adapt treatment strategies and to use powerful, broad spectrum antibiotics, such as carbapenems against multi drug resistant Gram-negative bacteria. However, the recent emergence of carbapenemase-producing bacteria around the world [1] has obliged clinicians to turn, as a last resort, to colistin [2,3]. The re-use of colistin has led to the appearance of colistin resistance, which is mediated by complex chromosomal resistance mechanisms in human and animal isolates [4,5]. Recently, a transferable colistin resistance mechanism, due to the presence of mcr-1 genes and variants that code for a phosphoethanolamine transferase, which have been detected in all continents [5] on a plasmid, has been described [6][7][8][9][10][11]. Initially, the mcr-1 gene was carried on IncI2-type plasmid, but has also been found in other plasmid types, such as IncHI2, IncX4, and IncP [5,12,13]. Generally, the mcr-1 gene has been described as being associated with an open reading frame (ORF), encoding a protein that is similar to a PAP2 superfamily protein, following the mcr-1 gene with an insertion sequence ISApl1 downstream to it [6]. ISApl1 is an insertion sequence that belongs to the IS30 family of transposons and was initially described in Actinobacillus pleuropneumoniae [14]. The mcr-1 gene may be surrounded by two copies of ISApl1, leading to the formation of a composite transposon. This transposon Tn6330 (ISApl1-mcr-1-ORF-ISApl1) has been described as a composite transposon that is able to mobilize the mcr-1 gene [15,16]. However, to date, the integration of this transposon into plasmids, its stabilization, and its evolution have not been comprehensively described. Thus, the aim of our study was to conduct an epidemiological and molecular characterization of mcr-1 strains from different origins, and to study the genetic environment of the mcr-1 gene and to characterize four different plasmids mediating colistin resistance for a better understanding of the integration of the mcr-1 gene into plasmids.

Strains
Thirty-two mcr-1 positive strains, including 25 Escherichia coli and seven Klebsiella pneumoniae, from Laos, Thailand, France, and Algeria, as well as from Hajj pilgrims (returning from their pilgrimage from Saudi Arabia to France), were analysed [17][18][19][20][21][22]. Five isolates were from animals and 27 from humans ( Table 1). The minimum inhibitory concentration (MIC) of colistin was tested by microdilution in accordance with European Committee on Antimicrobial Susceptibility Testing (EUCAST) recommendations [23]. The presence of the mcr-1 gene was confirmed by real-time polymerase chain reaction (RT-PCR) [24] and multilocus sequence typing performed using the Warwick method [25] on E. coli strains and the Pasteur method [26] on K. pneumoniae strains. Resistance stability was analysed by subculture over the course of 30 passages of each strain and by checking for the presence of the mcr-1 gene every five passages. In the event of gene loss, the MIC of colistin was tested and the procedure was repeated to confirm the results.

Conjugation and Plasmid Analysis
Conjugation was tested with azide-resistant E. coli J53. Transconjugants were selected on MacConkey agar (Beckton Dickinson, Le Pont de Claix, France) added by 120 µg/mL sodium azide and 4 µg/mL colistin, as described [27]. In the event of unsuccessful conjugation, transformation using the electroporation method was performed [28]. Transformants bacteria were selected on Luria Bertani agar (Beckton Dickinson, Le Pont de Claix, France) supplemented with 4 µg/mL colistin (Sigma-Aldrich, Saint Louis, USA). The presence of the mcr-1 gene was tested by RT-PCR on transconjugant and transformant strains. Plasmid typing was carried out on positives. Nineteen different plasmid types, including IncI2 [29,30], and the presence of the ISApl1 insertion sequence both downstream and upstream of the mcr-1 gene were tested by conventional PCR [31].

Whole Genome Analysis
Four E. coli strains (1RC4, LH1, LH30, LH57), one isolated from Hajj pilgrims, and three from healthy individuals living in Laos (Table 1), were sequenced using the Next Generation Sequencing (NGS) Miseq (Illumina Inc., San Diego, CA, USA). We selected the strain E. coli LH1 because conjugation and transformation experiments failed and we suspected a chromosomal location of mcr-1 gene. The strain E. coli LH30 was chosen to understand the rapid loss of mcr-1 gene in the resistance stability experiment. We chose the strain E. coli LH57 because a chromosomal mechanism of colistin resistance was described in this strain. The strain E. coli 1RC4 was acquired during Hajj pilgrims, and we found the origin of this plasmid. Genomes were assembled using the A5 pipeline, annotated by RAST [32], resistance gene by ARG-ANNOT [33], plasmid presence by Plasmid finder software, and plasmid multilocus sequence typing by pMLST software [34]. The percentage of similarity between plasmids was calculated using a pairwise comparison of their Average Nucleotide Identity based on Blast (ANIb) [35] and using the Jspecies software [36].
A database containing all mcr-1 complete plasmids that were available in the National Center for Biotechnology Information (NCBI) database as of 14 February 2017 was created ( Table 2). Plasmid type was then determined using the Plasmid finder software, and in-silico analysis of tra genes was performed for the non-conjugative E. coli LH1 strain. The presence of insertion sequence ISApl1 in all plasmids, including those that were present in our created database, was checked. Our four complete mcr-1 plasmids were compared and visualized using CGView software [37]. These plasmids were named pLH30-mcr1, pLH57-mcr1, pLH1-mcr1, and p1RC4-mcr1, and were submitted to Genbank under accession numbers NKYL00000000, NKYM00000000, NKYJ00000000, and NKYK00000000, respectively. An analysis of the genetic environment of the mcr-1 gene was performed using blast X on all of the genes around the mcr-1 gene in the NCBI. The identity of insertion sequences was confirmed using the ISfinder software [38].

Strain Characteristics
The colistin MIC of strains harboring the mcr-1 gene varied from 3 to 32 µg/mL ( Table 1). The presence of known chromosomal mechanisms of colistin resistance was found in two E. coli strains due to the PhoQ mutation (E375K) [21]. Regarding three K. pneumoniae strains, resistance was due to mgrB stop in the first strain, a mgrB substitution in the second, and a PmrB mutation (T157P) in the last [22] (  Figure 1). Following the subculture of strains, the loss of the mcr-1 gene was observed in four K. pneumoniae strains after 25 passages. For E. coli strains, only LH30 lost its mcr-1 gene after 20 passages. Colistin susceptibility was restored in E. coli LH30 upon the loss of mcr-1 gene. However, this was not the case for K. pneumoniae strains, which continued to be colistin resistant after the loss of mcr-1 (Table 1).

Location of the mcr-1 Gene
The presence of plasmids was demonstrated in all 25 E. coli strains, of which 21 carried the mcr-1 gene on a conjugative plasmid. For K. pneumoniae strains, only two carried the mcr-1 gene on a conjugative plasmid ( Table 1). The predominance of IncHI2-type (44.5%), IncI2-type (40.7%), and IncP-type (14.8%) plasmids was observed on the whole strains. The majority of plasmids (77.8%) were isolated from humans, including 81.8% of IncI2-type, 75% of IncHI2-type, and 75% of IncP-type plasmids ( Table 1). The presence of insertion sequence ISApl1 around the mcr-1 gene was found in 56.25% of strains in the downstream position, and 12.5% in the downstream and upstream positions. Furthermore, 28.12% of strains did not have an ISApl1 insertion sequence that was close to the mcr-1 gene, and 3.12% had partial ISApl1 sequences in positions that were downstream and upstream of the mcr-1 gene.

Genome Analysis
Mcr-1 Database Fifty-nine mcr-1 complete plasmids were retrieved from the NCBI database. Taking into account the % GC content and the plasmid sizes, our database can be divided into three main groups. The plasmid size was 51.88 Kb in group 1, 244.81 Kb in group 2, and 6.39-97.56 Kb in group 3 ( Figure 2). Our created mcr-1 plasmid database enabled us to see that the closest mcr-1 plasmid for pLH30-mcr1 was KU743384, with 99.68% similarity, while pLH57-mcr1 was KX276657, with 99.11% similarity, pLH1-mcr1 was KX254341, with 99.91% similarity, and p1RC4-mcr1 was KU743384, with 99.98% similarity. All of these plasmids were isolated from E. coli strains. KU743384 was an IncHI2 plasmid that was isolated from E. coli ST68 in Saudi Arabia. KX276657 was an Inc F18:A-:B1 (IncN with Plasmid finder software) plasmid isolated from E. coli ST457 in the United States, while KX254341 was an IncHI2 plasmid also isolated from E. coli in China (Table 2). From our mcr-1 database, 49% of plasmids had ISApl1 around the mcr-1 gene, of which 75.9% were in a downstream position and 24.1% were transposons Tn6330 (Table 2). In eight plasmids, we also found a copy of ISApl1 sequences around the mcr-1 gene, as well as in other plasmid locations. For 51% of plasmids, no ISApl1 sequence was detected in the whole plasmid. The majority of plasmids were isolated from E. coli strains. Plasmid sizes ranged from 33 Kb to 369 Kb, and GC content varied from 41.84% to 50.67%. Eight types of plasmids were represented in the database, including 40.7% IncI2 and 25.4% IncX4 (Figure 2, Table 2).

Sequenced Plasmids
Concerning the four plasmids that we sequenced, plasmid sizes ranged from 219 Kb to 248 Kb, and GC% ranged from 46 to 48. The plasmids were all IncHI2 type. pLH30-mcr1 and pLH1-mcr1 belonged to ST 3, while p1RC4-mcr1 was ST 1 and pLH57-mcr1 had an unknown ST (Table 3). A single copy of the mcr-1 gene was found in all of our plasmids. Antibiotic resistance genes other than mcr-1 are shown in Table 3. On pLH1-mcr1, the ISApl1 sequence was found to be inserted into the traE gene, leading to a stop codon that inactivated it.

Genetic Integration of mcr-1 Gene
In silico analysis confirmed the presence of ISApl1 downstream and upstream of the mcr-1 gene in the pLH30-mcr1. The transposon was surrounded by two genes coding for hypothetical proteins, with one gene coding for a thiol:disulfide interchange protein DsbC to the left and one coding for a HNH endonuclease to the right of the mcr-1 gene. In p1RC4-mcr1, we observed an ISApl1 sequence downstream and upstream in opposite directions, as well as two truncated ORFs and a recombinase close to the mcr-1 gene. In pLH1-mcr1, a classic mcr-1 cassette (ISApl1-mcr-1-ORF) and truncated IS1 insertion sequences were found around the mcr-1 transposon. For pLH57-mcr1, we observed a truncated ISApl1 sequence downstream and upstream, and a resolvase near the mcr-1 transposon ( Figure 3).        Figure 3. Genetic environment of the mcr-1 gene in plasmids: pHNSHP45 (KP347127), pLH30-mcr1 (NKYL00000000), pLH57-mcr1 (NKYM00000000), pLH1-mcr1 (NKYJ00000000), and p1RC4-mcr1 (NKYK00000000). Comparison of the four complete mcr-1 plasmids was performed using blastP with CGview software. nickB refers to Nickel B, topB: topoisomerase B, DsbC: Thiol:disulfide interchange protein DsbC, HP: Hypothetical protein, HNH: HNH endonuclease, IS: Insertion sequence.

Discussion
The presence of the mcr-1 gene led to a relatively low colistin resistance between 2 and 8 µg/mL in isolates. For some strains, the high colistin MIC observed was due to the presence of additional chromosomal mechanisms of resistance (Table 1). Interestingly, we observed that K. pneumoniae strains with high MIC appeared to lose mcr-1 plasmids more easily. For these strains, the loss of the mcr-1 gene did not lead to a change in their MIC, and they remained resistant to colistin. Hence, the mcr-1 gene was not essential to their survival. In our study, the clonal dissemination of E. coli carrying the mcr-1 gene was diverse, with 17 different STs, although some STs appeared several times. These STs were not related to a common sample origin or to a specific country. The presence of the same ST 4015 in two strains that were isolated in a same village from a human and a pig was described as a possible case of animal/human transmission [21]. The presence of ST 648 in two different travellers and ST 3997 in two villagers would appear to be a case of inter-human transmission. ST 48 and specially ST 10 were widespread STs, and have often been described for mcr-1 in the literature [7]. Strains could carry different types of plasmids encoding the mcr-1 gene [12,64]. Hence, the spread of the mcr-1 gene could be unrelated to a specific clonal population.
Thus far, eight plasmid types carrying the mcr-1 gene have been described around the world [12]. Our study confirmed this by observing only three plasmid types that were randomly present in different countries, as well as in humans and animals. Mcr-1 database analysis confirms the global propagation of such plasmids, especially the smallest IncI2-and IncX4-type plasmids. The genetic environment of the mcr-1 gene was first described by the ISApl1-mcr-1-ORFcassette [6]. In this study, this combination was commonly found. The transposon Tn6330 (ISApl1-mcr-1-ORF-ISApl1), described as being responsible for mcr-1 gene transfer [15,58], was found in few strains, including E. coli LH30. This could be the reason for early mcr-1 gene loss in this strain. Interestingly, the presence of the ISApl1sequence was not always around the mcr-1 gene [13,45]. In pLH1-mcr1, ISApl1 was found downstream of the mcr-1 gene and inside the traE gene. This insertion sequence disrupts the traE gene, leading to the inactivation of this gene and affecting the capacity of the bacteria to transfer the plasmid through conjugation. This insertion site into the traE gene has already been described in such IncK2 plasmids in Switzerland, and has been associated with a lower frequency of conjugation [72,73].
A certain percentage of plasmids did not display an ISApl1 sequence around the mcr-1 gene [16,74]. This led to two hypotheses: the mcr-1 gene lost ISApl1 and stabilized it into a plasmid, or the mcr-1 gene was transferred by other insertion sequences. The presence of ISApl1 in other places in the plasmid supports the first hypothesis and confirmed its original role in mcr-1 gene transfer. Furthermore, in the composite transposon Tn2706, which is composed of 2 IS30, the loss of one IS was described as being favourable for the stabilization of genes in the new location [75].
In a recent study, ISApl1 was described as being a highly active IS and being able to transpose at a very high frequency in different nonspecific insertion sites [45]. For pLH1-mcr1, transposition was performed into a hotspot in the vicinity of a truncated IS1 insertion sequence. Moreover, a recombinase and a resolvase were found near the mcr-1 transposon of p1RC4-mcr1 and pLH57-mcr1. Resolvase or recombinase are nucleases involved in DNA recombination. Their presence was a sign of recombination hotspots that were favourable to a transposon insertion. For pLH30-mcr1, an HNH endonuclease was present near the mcr-1 transposon. Homing endonucleases were involved with lateral transfer as an intron to a homologous allele [76]. This could also be interpreted as a sign of a variable region suitable for insertion sequences.

Conclusions
Here, we show that worldwide dissemination of mcr-1 encoding gene was due to the spread of a transposon, which can be found in different plasmid types and/or bacterial chromosomes. Even if some STs were redundant, dissemination was not due to a specific clonal population or to a specific plasmid type. Initially, the transfer of the mcr-1 gene was due to Tn6330. Then, in order to stabilize in a new location, the transposon lost one or both ISApl1 sequences. It is likely that the evolution of the genetic environment of the mcr-1 gene could lead to a diversity of insertion sequences of the mcr-1 gene. This could raise the risk of a possible translocation into the chromosome. An emergence of a preponderant clone and the rapid dissemination of the mcr-1 gene in Gram-negative bacteria are possible. Hence, we consider that it is essential to continue to survey resistance to colistin in those strains.