Identification of novel mobile colistin resistance gene mcr-10

ABSTRACT Mobile colistin resistance (mcr) genes represent an emerging challenge. Here we describe a novel mcr gene, mcr-10, on an IncFIA plasmid of an Enterobacter roggenkampii clinical strain. mcr-10 has the highest nucleotide identity (79.69%) with mcr-9 and encodes MCR-10 with 82.93% amino acids identical to MCR-9. mcr-10 confers 4-fold increase in colistin MIC (from 1 to 4 mg/L) when cloned into a colistin-susceptible E. roggenkampii strain. By screening GenBank, mcr-10 was found in various Enterobacteriaceae species of countries in four continents, suggesting that this gene has widely spread. MCR-10 shows 79.04% to 83.67% amino acid identity and highly conserved predicted protein structures with chromosomally encoded MCR-like phosphoethanolamine transferases (designated MCR-B here) of various Buttiauxella species. MCR-10, MCR-9 and MCR-B proteins may, therefore, originate from a common ancestor. mcr-10 was adjacent to a site-specific recombinase-encoding gene and was bracketed by IS903 and may be mobilized by site-specific recombination or composite transposon. Our results indicate that mcr-10 is a novel plasmid-borne colistin resistance gene and warrants immediate monitoring and further studies.


Introduction
Colistin is a last resort antimicrobial agent against carbapenem-resistant Gram-negative bacteria including Enterobacteriaceae but strains with acquired colistin resistance have also emerged worldwide [1]. Colistin resistance in the Enterobacteriaceae can be due to chromosomal mechanisms and plasmid-borne mobile colistin resistance genes (mcr). Since the report of the first mcr gene, mcr-1, in 2016 [2], a few other mcr genes including mcr-2 [3], mcr-3 [4], mcr-4 [5], mcr-5 [6], mcr-6 [7], mcr-7 [8], mcr-8 [9] and mcr-9 [10] have been described in Enterobacteriaceae and mcr-1 and mcr-4 have also been reported in Acinetobacter spp. [11,12] or Pseudomonas spp. (mcr-1 only) [12]. All of the MCR proteins are phosphoethanolamine (PEA) transferases [13]. These PEA transferases catalyse the attachment of PEA to lipopolysaccharides (LPS)-lipid A, lead to a reduction of the negative charge of LPS upon structural alteration of lipid A and therefore result in resistance to colistin [13]. Of note, the discovery of mcr-9 in colistin-susceptible strains suggests that strains carrying mcr genes may not exhibit colistin resistance phenotype due to low-level gene expression [10]. This promotes us to investigate the presence of mcr-like genes in a colistin-susceptible Enterobacter clinical strain, for which the minimum inhibitory concentration (MIC) of colistin was 2 mg/L, close to the 4 mg/L resistance breakpoint defined by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) (http://www.eucast.org/). We found a new mcr variant, designated mcr-10, in the strain and report the findings here.

Methods
The strain, in vitro susceptibility testing Strain 090065 (also called WCHER090065) was a clinical isolate recovered in 2016 at West China Hospital. This study was approved by the Ethical Committee of West China Hospital with waiving the inform consent. MICs of aztreonam, cefepime, ceftazidime, colistin, imipenem, meropenem, piperacillin-tazobactam, and tigecycline were determined using the broth microdilution method of the Clinical and Laboratory Standards Institute (CLSI) [14]. For colistin and tigecycline, the breakpoints defined by EUCAST (http://www.eucast. org/) were applied.

Genome sequencing and analysis
Strain 090065 was subjected to whole-genome sequencing using the HiSeq X10 (Illumina; San Diego, CA, USA) according to the manufacturer's instructions. Genomic DNA was prepared using the QIAamp DNA Mini Kit (Qiagen, Hilden, Germany). Generated reads were de novo assembled into contigs using SPAdes v3.13.0 [15] applying the careful and autocut-off modes. To determine the location of mcr-10, strain 090065 was also subjected to long-read wholegenome sequencing using a MinION Sequencer (Nanopore; Oxford, UK). The de novo hybrid assembly of both short (Illumina) and long reads was performed using Unicycler v0.4.3 [16] under conservative mode for increased accuracy. Pilon v1.22 [17] was used to correct complete circular contigs with Illumina reads for several rounds until no change was detected.

Nucleotide sequence accession numbers
The complete sequence of the chromosome and plasmids of strain 090065 has been deposited into Gen-Bank under the accession no. CP045064-CP045066. The sequence of mcr-10 has been deposited into Gen-Bank under the accession no. MN179494.

Precise species identification
For precise species identification, the pair-wise average nucleotide identity (ANI) between strain 090065 and type strains of Enterobacter species was determined using JSpeciesWS based on BLAST [19]. A ≥ 95-96% ANI cut-off was used to define a bacterial species [20].

Induction tests
To examine whether the expression of mcr-10 is inducible as reported for mcr-9 [10], strain 090065 was subjected to induction with IPTG (isopropyl-β-D-thiogalactopyranoside; BBI, Shanghai, China) or lactose (Meilun, Dalian, China) as described previously [21]. After induction, MIC of colistin for strain 090065 was determined in the absence or presence of 1 or 3 mmol/L lactose or 0.4 or 1 mmol/L IPTG.

Conjugation and electroporation experiments
Conjugation experiments were carried out in broth and on filters with the azide-resistant E. roggenkampii strain 120033 AizR (an azide-resistant variant of 120033) as the recipient at both 25°C and 37°C as described previously [22]. Potential transconjugants were selected on LB agar plates containing 2 mg/L colistin and 150 mg/L azide. Electroporation was performed with Escherichia coli DH5α as described previously [23]. Transformants were selected on Luria-Bertani agar plates containing 2 mg/L colistin. The presence of mcr-10 in the transformant was confirmed by PCR with primers 090065_up_SacI/090065_down_EcoRI and subsequent Sanger sequencing. MICs of aztreonam, ceftazidime, colistin, and meropenem were determined as described above.
Screening the presence of mcr-10 in GenBank We screened the presence of mcr-10 in sequences including complete or draft genome sequences deposited in GenBank by BLAST (https://blast.ncbi.nlm. nih.gov/Blast.cgi, accessed by 30 August 2019). Matches with >90% identity and >90% coverage were retrieved from GenBank.
Structure comparisons of MCR-10, MCR-9 and MCR-Bs The amino acid sequences of MCRs were retrieved from Bacterial Antimicrobial Resistance Reference Gene Database (BioProject no. PRJNA313047). MCR-B proteins from genus Buttiauxella were retrieved from their whole-genome assemblies. Along with the two alleles of MCR-10 found in this study (see below for details), the amino acid sequences of all genes (n = 75) were aligned using Prank v1.70427 [24] with 50 iterations first, followed by aligning corresponding nucleotide sequences using aligned amino acid sequences as the guide in the same program. The aligned nucleotide and amino acid sequences were fed into RAxML v8.2.12 [25] with a 1000-bootstrap test under GTRGAMMA and PROTGTR-GAMMA model, respectively, for inferring maximum-likelihood phylogenies.

Results
A novel mcr, mcr-10, was identified in strain 090065 of E. roggenkampii Strain 090065 belongs to E. roggenkampii, a recently described Enterobacter species [30] as it has 98.51% ANI value with the type strain of E. roggenkampii (strain DSM16690 T , Gen-Bank accession no. CP017184).
Strain 090065 has two known antimicrobial resistance genes, bla MIR-5 (a chromosomal ampC gene intrinsic to Enterobacter mediating resistance to aztreonam, 1st to 3rd cephalosporins, and penicillins) and fosA (mediating resistance to fosfomycin). Known mcr genes including mcr-1 to -9 were not identified in the draft genome sequence of strain 090065. However, a PEA transferase-encoding gene, which has 79.69% identity and 99% coverage with mcr-9.1 (GenBank accession no. NG_064792) [10], was identified. The PEA transferase encoded by the gene has 82.93% amino acid identity with MCR-9, suggesting that the transferase is a novel MCR-like protein.
To determine whether this novel mcr-like gene mediates colistin resistance or not, the gene was cloned on pBC SK (Stratagene, La Jolla, CA, USA) to construct pBC SK-mcr10, which was transferred into a colistinsusceptible E. roggenkampii clinical strain, named 120033. MIC of colistin against strain 120033 containing pBC SK-mcr10 and strain 120033 containing pBC SK alone was 4 and 1 mg/L, respectively. The four-fold increase in colistin MIC in the presence of mcr-10 suggests that this gene indeed mediates colistin resistance.

mcr-10 was carried by an IncFIA plasmid in strain 090065
To determine the location of mcr-10, strain 090065 was also subjected to long-read whole-genome sequencing using MinION. The hybrid assemblies of Illumina and MinION reads revealed that strain 090065 had a 4.86-Mb circular chromosome and two plasmids, p1_090065 (10,944-bp, replicon type undetermined) and pMCR10_090065 (71,775-bp, containing an IncFIA(HI1) replicon).
Despite repeated attempts of conjugation experiments, transconjugants containing pMCR10_090065 were not obtained, suggesting that pMCR10_090065 was not self-transmissible. Examining the complete sequence of pMCR10_090065 revealed that there was no conjugation module on this plasmid. E. coli transformant carrying mcr-10 was obtained by electroporation, confirming that mcr-10 is plasmidborne. mcr-10-carrying transformant was susceptible to colistin (MIC, 2 mg/L), meropenem (MIC, <1 mg/L), aztreonam (MIC, <1 mg/L), and ceftazidime (MIC, 4 mg/L). This suggests that the resistance to these agents seen in strain 090065 was not cotransferred with mcr-10.
mcr-10 is found in a few genera of the family Enterobacteriaceae and has a global distribution In GenBank, a total of 34 matches that have >90% identity and >90% coverage with mcr-10 were identified including 30 draft genome sequences (Table 1) and four complete plasmid sequences (Table 2). In addition, an incomplete mcr-10 was found in the draft genome sequence of an E. coli strain (accession no. LLYM01000000), which was truncated by insertion sequence IS3. The mcr-10-containing strains belonged to 13 species of 6 genera (Citrobacter, Enterobacter, Escherichia, Klebsiella, Kluyvera, and Raoultella) of the family Enterobacteriaceae but most (24/35) were of the genus Enterobacter. The strains were found in 11 countries (Australia, Canada, China, France, Germany, Japan, the Netherlands, Spain, Thailand, USA and Vietnam) of four continents. Most (n = 30) of the 35 strains were from human, mcr-10 was also present in strains from animal (dog, n = 1) and environment (water, n = 4). As only contigs of the 31 draft genome sequences are available, we were unable to reliably determine the location (chromosome or plasmid) of mcr-10 for these strains. Nonetheless, all of the four mcr-10-carrying plasmids contained one or two IncF replicons ( Table 2).
The mcr-10 genes of 33 strains encode MCR-10 that has an identical amino acid sequence with that in strain 090065. The remaining sequence of the truncated mcr-10 (accession no. LLYM01000000) is also identical to that of strain 090065. However, another MCR-10 variant with 15 amino acid substitutions (97.2% [524/539] amino acid identity) is found in plasmid pECC18A13-1 of Enterobacter sp. 18A13 (all call as DSM16690; Gen-Bank accession no. AP019635). The alignment of the two MCR-10 variants is shown in Figure S1 in the Supplementary file. This suggests that mcr-10 has been diverged.
mcr-10 may originate from a yet-to-identified species closely related to known Buttiauxella species Like MCR-9 [ Rectal swab Note: NA, not available. a In the four strains, mcr-10 is located on a plasmid ( Table 2). b mcr-10 in strain 18A13 encodes an MCR-10 variant with 15 amino acids different from MCR-10 encoded by mcr-10 genes in all other strains. c mcr-10 in strain AZ74 is truncated.
Comparison with known MCRs (MCR-1 to MCR-9) and MCR-B of various Buttiauxella species showed MCR-10 form a cluster with MCR-9 and MCR-B, which is well separated from other MCRs (Figure 1; a phylogenetic tree of mcr genes is shown as Figure S2 in the Supplementary file). Nonetheless, there are a  MCR-10, MCR-9 and MCR-Bs are highly similar at the structural level Three-dimensional (3D) models showed that both the membrane-anchored domain and the soluble catalytic domain of MCR-1 to -10 and MCR-Bs had high levels of conservation (Figure 2(A)). The N-terminal membrane-anchored domain and the C-terminal soluble catalytic domain of these MCR proteins were conserved in both amino acids and structural elements (Figure 2(B)).
The mobilization of mcr-10 may be due to sitespecific recombination On pMCR10_090065 and three other plasmids carrying mcr-10, mcr-10 was located at the immediate downstream of a XerC-type tyrosine recombinase-encoding gene, designated xerC here (Figure 3). It has been known that XerC-type tyrosine recombinases are able to mediate mobilization of adjacent genetic components including antimicrobial resistance genes via site-specific recombination [33,34]. In Enterobacter spp., an XerC-type tyrosine recombinase has been identified to mediate the mobilization of bla IMI , a carbapenemase gene [35]. Therefore, the mobilization of mcr-10 may also be mediated by the XerC-type tyrosine recombinase. As the downstream sequence of mcr-10 was truncated by insertion sequence ISEc36, it is therefore impossible to identify the recombination sites recognized by the XerC-type tyrosine recombinase. Of note, two copies of IS903 were located at upstream and downstream of xerC-mcr-10 on pMCR10_090065,  [29]. Secondary structure elements, a helixes, b sheets, and 3 10 -helixes (representing by η), are indicated. β-strands are rendered as arrows, and strict β-turns are shown as TT letters. Gene xerC (shown in orange) encodes a XerC-type tyrosine recombinase, which may mediate mobilization of adjacent genetic components via site-specific recombination. Δ represents truncated insertion sequences or transposons. Identical regions are highlighted by grey rectangles. On pMCR10_090065, two copies of IS903 are located at upstream and downstream of xerCmcr-10 (mcr-10 is shown in red) and the 9-bp abutting sequences are indicated. On pOZ171, there is an IS903 at upstream with the 9-bp abutting sequence being shown. However, there is no IS903 at downstream of xerC-mcr-10 but instead, transposon Tn1722 is presented. On pEC27-2, there is no IS903. A complete ISKpn26 is present at upstream of xerC-mcr-10 and an IS26, which is interrupted by the insertion of ISSen4, is at downstream. On an unnamed plasmid (accession no. CP023893), several open reading frames (orfs) without known function are present at upstream of xerC-mcr-10, while ISEc36 is present at downstream. On pECC18A13-1, the mcr-10 encodes an MCR-10 variant with 15 amino acids different from MCR-10 encoded by the other plasmids and is shown in dark red. No xerC is present, while truncated IS100 and truncated ISSen7 are located at upstream and downstream of the mcr-10, respectively. respectively, and could form a composite transposon. On insertion, IS903 generates 9-bp direct target repeats but the 9-bp sequences abutting the two IS903 were different, suggesting that the region bracketed by IS903 was not due to direct insertion. Nonetheless, the IS903-formed composite transposon has the potential to mediate mobilization of the intervening genetic components. In contrast, IS903 has also been found upstream of mcr-9 but other insertion sequences such as IS26 are present downstream instead [32].

Discussion
Enterobacteriaceae is the most common pathogen causing human infections [36]. Multi-drug resistant organisms such as carbapenem-resistant Enterobacteriaceae (CRE) have become a major global challenge [37]. Colistin is a last resort antimicrobial agent and one of the only options to treat serious infections caused by CRE. However, Enterobacteriaceae strains with acquired colistin resistance have emerged worldwide, which are significantly jeopardizing the efficacy of colistin [38]. Plasmid-borne mcr colistin resistance genes can be transmitted across Enterobacteriaceae species and may have spread to multiple continents, representing a particular threat to public health and clinical management in the whole world [39]. Monitoring the spread antimicrobial resistance is a core component of strategies for combating antimicrobial resistance [40]. Identification of new mcr genes can be used to improve the monitoring of plasmid-borne colistin resistance and may, therefore, help to develop effective control measures.
In this study, we identified a novel mcr gene in strain 090065 of Enterobacter roggenkampii. mcr-10 has the highest nucleotide identity (79.69%) with mcr-9 and encodes MCR-10 with 82.93% amino acids identical to MCR-9. mcr-10 confers 4-fold increase in colistin MIC (from 1 to 4 mg/L) when cloned into a colistinsusceptible E. roggenkampii strain. This suggests that mcr-10 was indeed a colistin resistance gene. Of note, strain 090065 carrying mcr-10 was susceptible to colistin (MIC, 2 mg/L), while the originally colistin-susceptible E. roggenkampii strain became resistant to colistin (MIC, 4 mg/L) after receiving mcr-10 by cloning. The discrepancy in colistin susceptibility is likely due to the expression as mcr-10 was carried by a large naturally occurring plasmid in strain 090065 but was cloned onto the small-size vector pBC SK in strain 120033. Nonetheless, such discrepancy warrants further studies. We revealed that mcr-10 was adjacent to sitespecific recombinase-encoding gene and was bracketed by IS903 on an IncFIA plasmid in strain 090065, indicating that mcr-10 has the potential to be mobile. We also found that mcr-10 has been present in a few genera of the family Enterobacteriaceae with a global distribution. Of note, the earliest match of mcr-10 in GenBank was plasmid pOZ172 of a Citrobacter freundii clinical strain, which was recovered in 1998 in Guangzhou, southern China, suggesting that mcr-10 has been mobilized by plasmids within Enterobacteriaceae for decades. We showed that MCR-10, MCR-9 and MCR-B proteins originated from a common ancestor. The large number and diffuse distribution of amino acid variations between MCR-10 and MCR-B proteins suggest that MCR-10 is not directly derived from these Buttiauxella species but may have originated from a yet-to-identified species that are closely related to known Buttiauxella species.
In conclusion, we identified and characterized a novel mcr gene and we also found that the gene has been spread globally under the radar. This suggests that the novel mcr gene is of significance for health. Our findings are essential for developing effective countermeasures including surveillance.