Identification of Novel Mobilized Colistin Resistance Gene mcr-9 in a Multidrug-Resistant, Colistin-Susceptible Salmonella enterica Serotype Typhimurium Isolate

Colistin is a last-resort antibiotic that is used to treat severe infections caused by MDR and extensively drug-resistant (XDR) bacteria. The World Health Organization (WHO) has designated colistin as a “highest priority critically important antimicrobial for human medicine” (WHO, Critically Important Antimicrobials for Human Medicine, 5th revision, 2017, https://www.who.int/foodsafety/publications/antimicrobials-fifth/en/), as it is often one of the only therapies available for treating serious bacterial infections in critically ill patients. Plasmid-borne mcr genes that confer resistance to colistin pose a threat to public health at an international scale, as they can be transmitted via horizontal gene transfer and have the potential to spread globally. Therefore, the establishment of a complete reference of mcr genes that can be used to screen for plasmid-mediated colistin resistance is essential for developing effective control strategies.

In silico identification of mcr-9 in an MDR S. Typhimurium genome. MDR S. Typhimurium strain HUM_TYPH_WA_10_R9_3274 (NCBI RefSeq accession no. GCF_002091095.1) was isolated from a patient in Washington State in 2010 (10). It had previously been tested for resistance to a panel of 12 antimicrobials that did not include colistin (10). ABRicate version 0.8 (https://github.com/tseemann/abricate) identified 20 antimicrobial resistance (AMR) genes in the HUM_TYPH_WA_10_R9_3274 assembly using the ResFinder database (accessed 11 June 2018) (11) and minimum identity and coverage thresholds of 75 and 50% (10), respectively, none of which had been previously described to confer colistin resistance (see Table S1 in the supplemental material). Four plasmid replicons, including IncHI2 and IncHI2A, were detected with at least 80% identity and 60% coverage using ABRicate and PlasmidFinder (accessed 11 June 2018 [Table S1]) (12).
The amino acid sequence of mcr-9 most closely resembled those of mcr-3 and mcr-7 ( Fig. 1A; see Fig. S1 in the supplemental material). However, the S. Typhimurium isolate in which mcr-9 was detected was not resistant to colistin at the Ͼ2-mg/liter European Committee on Antimicrobial Susceptibility Testing (EUCAST [http://www.eucast.org]) breakpoint when a broth microdilution method was used to determine the colistin MIC (see Table S5 in the supplemental material).
Proteins encoded by mcr-1 to -9 revealed high levels of conservation for both the membrane-anchored domain and the soluble catalytic domain ( Fig. 2A). Interestingly, analyses of structural models of the nine Mcr homologues using the ESPript 3 server (20) showed that both amino acids and structural elements were conserved on the C-terminal catalytic domain, while only structural elements were conserved on the membrane-anchored N-terminal domain (Fig. 2B). FIG 1 (A) Comparison of mcr-9 to all previously described mcr homologues, based on amino acid sequence. The maximum likelihood phylogeny was constructed using RAxML version 8.2.12 with the amino acid sequences of novel mobilized colistin resistance gene mcr-9 (in blue) and all previously described mcr genes (mcr-1 to -8 [in black]). The phylogeny is rooted at the midpoint, with branch lengths reported in substitutions per site. Branch labels correspond to bootstrap support percentages out of 1,000 replicates. (B) Colistin killing assay of E. coli NEB5␣ harboring a pLIV2 empty vector (negative control), mcr-3 (positive control), or mcr-9, expressed under the control of the IPTG-controlled SPAC/lacOid promoter. Cells were grown in MH-II (Mueller-Hinton II) medium with IPTG to the mid-exponential phase. Colistin was added at concentrations of 0, 1, 2, 2.5, or 5 mg/liter, and the bacteria were incubated at 37°C for 1 h. The samples were diluted in phosphate-buffered saline (PBS) and plated on LB agar plates for the determination of CFU. Log CFU reduction was calculated by comparing CFU after each treatment to CFU levels obtained at 0 mg/liter colistin, using three independent biological replicates. Asterisks denote significant differences compared to empty vector treatment (P Ͻ 0.05 by Student's t test relative to the concentration's respective negative control after a Bonferroni correction). (C) Similarity matrix (composed of Dali Z-scores) of all previously described Mcr groups (Mcr-1 to -8) and Mcr-9, based on protein structure. The Dali server was used to perform all-against-all comparisons of 3D structural models based on all mcr homologues ( Fig. 2A); for this analysis, amino acid sequences of mcr-5.3 and mcr-8.2, which were not available in ResFinder, were additionally included from the National Database of Antibiotic Resistant Organisms (NDARO).
Novel Mobilized Colistin Resistance Gene mcr-9 ® Numerous genera of Enterobacteriaceae harbor mcr-9 on IncHI2 plasmids. blastp searches of mcr-9 against NCBI's nr database revealed that mcr-9 was present in multiple genera of Enterobacteriaceae (Table S3). The 10 highest-scoring hits in the nr database matched mcr-9 with at least 99% amino acid identity (including mcr-9 characterized here [Table S3 and Fig. S1A]); the amino acid identities of the remaining hits with high query coverage (Ͼ 90%) dropped below 88% identity (Table S3 and Fig. S1A). mcr-9 was detected in 335 genomes linked to NCBI identical protein groups (IPGs) associated with the 10 highest-scoring protein accession numbers (accessed 23 January 2019 [see Tables S3 and S6 in the supplemental material]). Analysis of the mcr-9 promoter region in 321 of these genomes (Text S1) showed conserved putative 70 family-dependent Ϫ35 and Ϫ10 regions and an inverted repeat (Fig. 2C). The con- Mcr-9, based on lipooligosaccharide phosphoethanolamine transferase EptA. Models were constructed using the Phyre2 server, and structures were viewed and edited using UCSF Chimera. Structural models show conservation of two EptA domains: transmembrane-anchored and soluble periplasmic domains. (B) Location of Mcr-9 secondary structure elements within the alignment of Mcr amino acid sequences, constructed using the ESPript 3 server. The top track denotes Mcr-9 secondary structure elements (alpha helixes and beta sheets). Green digits below the alignment denote cysteine residues forming a disulfide bridge (e.g., 1 forms a bridge with 1, 2 with 2, etc.). Within the amino acid sequence alignment itself, a strict identity (i.e., identical amino acid residue at a site) is denoted by a red box and a white character. A yellow box around an amino acid residue denotes similarity across groups, where groups were defined using the default "all" specification in ESPript 3 (ESPript 3 total score [TSc] Ͼ in-group threshold [ThIn]), while a residue in boldface denotes similarity within a group (ESPript 3 in-group score [ISc] Ͼ ThIn). (C) Organization of the mcr-9 locus in S.
Typhimurium. An unknown function cupin fold metalloprotein is encoded by the gene downstream of mcr-9 (unlabeled black arrow). The mcr-9 locus is flanked by two different terminal repeat sequences (IRR) from the IS5 (orange box) and IS6 (red box) families. The mcr-9 upstream region contains highly conserved putative Ϫ35 and Ϫ10 70 -dependent promoter elements (blue boxes and blue text). Moreover, the mcr-9 promoter region contains an inverted repeat motif (green box, green text, and sequence logo) that is conserved in more than 95% of 321 mcr-9 genes, as shown by the sequence logo (constructed using WebLogo) (24). served DNA motif in the mcr-9 promoter is likely a recognition sequence for a transcription regulator, suggesting that additional factors or induction/derepression conditions might be needed for full expression of wild-type mcr-9. Promoter variation (21) and testing conditions (22,23) have been shown to influence mcr expression and the colistin MIC, which may explain why the S. Typhimurium strain queried here was colistin susceptible under the tested conditions.
Of the 335 genomes in which mcr-9 was detected, 65 had at least one plasmid replicon (detected using ABRicate and PlasmidFinder as described above) present on the same contig as mcr-9; in 59 of these 65 genomes, IncHI2 and/or IncHI2A replicons were detected on the same contig as mcr-9 (Table S6). In 32 of the 37 closed genomes in which it was detected, mcr-9 was harbored on a plasmid (Table S6). These results indicate that mcr-9 has the potential to reduce susceptibility to colistin, up to and beyond the EUCAST breakpoint, and can be found extrachromosomally in multiple species of Enterobacteriaceae, making it a relevant threat to public health. Future studies querying the plasmids that harbor mcr-9 (e.g., transferability, stability, and copy number variation) will offer further insight into the potential role that mcr-9 plays in the dissemination of colistin resistance worldwide.
Accession number(s). The nucleotide and amino acid sequences of mcr-9 are available under NCBI reference sequence accession no. NZ_NAAN01000063.1 (NCBI protein accession no. WP_001572373.1).

ACKNOWLEDGMENTS
This material is based on work supported by the National Science Foundation (NSF) Graduate Research Fellowship Program under grant no. DGE-1650441, with additional funding provided by an NSF Graduate Research Opportunities Worldwide (GROW) grant through a partnership with the Swiss National Science Foundation (SNF).
We thank Julie Siler (Cornell University) for providing colistin resistance testing materials.