Comparative genomic analysis of Myroides odoratimimus isolates

Abstract Myroides odoratimimus is an important nosocomial pathogen. Management of M. odoratimimus infection is difficult owing to the multidrug resistance and the unknown pathogenesis mechanisms. Based on our previous genomic sequencing data of M. odoratimimus PR63039 (isolated from a patient with the urinary tract infection), in this study, we further performed comparative genomic analysis for 10 selected Myroides strains. Our results showed that these Myroides genome contexts were very similar and phylogenetically related. Various prophages were identified in the four clinical isolate genomes, which possibly contributed to the genome evolution among the Myroides strains. CRISPR elements were only detected in the two clinical (PR63039 and CCUG10230) isolates and two environmental (CCUG12700 and H1bi) strains. With more stringent cutoff parameters in CARD analysis, the four clinical M. odoratimimus contained roughly equal antibiotic resistance genes, indicating their similar antibiotic resistance profiles. The three clinical (CCUG10230, CCUG12901, CIP101113) and three environmental (CCUG12700, L41, H1bi) M. odoratimimus strains were speculated to carry the indistinguishable virulent factors (VFs), which may involve in the similar pathogenesis mechanism. Moreover, some VFs might confer to the high capacity of dissemination, attacking tissue cells and induction of autoimmune complications. Our results facilitate the research of antibiotic resistance and the development of therapeutic regimens for the M. odoratimimus infections.

M. odoratimimus infections are life-threatening due to its multidrug resistance and unknown pathogenicity (as summarized in Hu et al., 2016). In our previous report (Hu et al., 2017), to some extent,

| Genome sequences
In the NCBI Genome RefSeq Assembly Database, only nine genomic sequences of M. odoratimimus were found (Table 1). They included four clinically pathogenic strains, a human-associated strain, and four environmental isolates. Only PR63039 genome (Hu et al., 2017) was complete. Strain PR63039 (Hu et al., 2017) and CCUG12901 were isolated from the urine of patients with postinjury urinary tract infection. CCUG10230 and CIP101113 were isolated from skin wounds. Human-associated strain CCUG39352 was collected and sequenced by Shandong University. M. odoratimimus H1 bi, L41, CCUG 12700, and CCUG 3837 are environmental isolates. For the phylogenetic tree analysis of Myroides genomes, another humanassociated strain Myroides sp. A21 (CP010327) (Burghartz et al., 2015) with highly homologous 16S rRNA gene sequence to strain PR63039 (coverage 100%, identity 100%, 1,388 bp) (GenBank No. KR349266) was also included. Myroides sp. A21 was isolated from the urethral catheter of a patient without symptoms of a urinary tract infection, had extensive drug resistance; its full genomic sequence was available.

| Softwares and databases used for comparative genomics analysis
The analyses of whole-genome phylogenetic tree, circular genome mapping, insertion sequence elements (IS), multiple genome alignment, prophage, CRISPR, antibiotic resistance genes, and VF genes in the M. odoratimimus genomes were performed with the softwares and databases listed in Table 2. We should mention that, for identifying the resistance genes using CARD Resistance Gene Identifier (RGI) software (Jia et al., 2017;Mcarthur et al., 2013)

| Phylogenetic analysis of 10 Myroides genomes
Whole-genome phylogenetic tree of the 10 Myroides genomes was created ( Figure 1)

| Genomic variants among four clinically pathogenic M. odoratimimus strains
We compared the genomes of three clinically pathogenic M. odoratimimus strains (CCUG12901, CCUG10230, CIP101113) with the clinically pathogenic PR63039 genome as the reference (Table 2). Many highly variable regions were found ( Figure 2). Specifically, the following regions on the above three genomes were absent or had low identity with our strain PR63039: from 150 to 250 kb, 700 to 780 kb, 1,650 to 1,700 kb, 2,300 to 2,450 kb, 2,680 to 2,720 kb, 3,370 to 3,530 kb, 3,720 to 3,800 kb, 4,110 to 4,200 kb, 4,250 to 4,270 kb, and 4,350 to 4,560 kb. Interestingly, the region from 1,650 to 1,700 kb was predicted to be located in one prophage locus of CCUG-12901 genome.
Circular map of the genome comparisons indicated that there were a number of conserved or diverged genome segments among the genomes of these four clinically pathogenic M. odoratimimus.
In addition, the abovementioned variable regions were partially accompanied by several insertion elements, which might assist the integration of resistance-and pathogenesis-related genes and facilitate the transfer of drug resistance and pathogenic genes among M. odoratimimus strains. Furthermore, IS elements may enhance drug resistance and virulence by promoting gene expression (Heritier, Poirel, & Nordmann, 2006;Higgins, Dammhayn, Hackel, & Seifert, 2010).

| Synteny analysis among four clinically pathogenic and three environmental M. odoratimimus strains
Genome alignments can identify evolutionary traits. To study the genome synteny and rearrangements in four clinically pathogenic (PR63039, CCUG10230, CCUG12901, and CIP101113) and three environmental (CCUG12700, L41, and H1bi) M. odoratimimus bacteria (Table 2), the genome alignment software progressive MAUVE (Darling, Mau, & Perna, 2010) was used ( Figure S1). The synteny between the PR63039 genome and Myroides sp. A21 was approximately 83.7% (Hu et al., 2017). The genome arrangement of these four clinically pathogenic isolates mimics each other. Similarly, the genome context and arrangement in the three environmental strains exhibited great similarity. However, the genome synteny between the clinically pathogenic and environmental isolates was relatively low.
The alignment of the four genomes of clinically pathogenic isolates showed that their genome rearrangements were similar although there were inversions in some regions. Moreover, the chromosomal alignments of CCUG12901 and CIP101113 were nearly identical with large segments of high similarity. There were some white areas not aligned well because they might contain elements specific to a particular genome. Overall, the four clinically pathogenic strain genomes were similar although the genome synteny in the latter three (CCUG10230, CCUG12901, and CIP101113) was more related than that in PR63039.
CIP101113 contained only one prophage, but it was larger and more complete than any other prophages. It consisted of hypothetical proteins, phage-like proteins, attachment sites, tail shafts, and proteases.
Among these predicted prophages, attachment sites and proteases only existed in CIP101113 prophage. In bacterial genomes, integrases are useful markers for mobile DNA elements, such as prophages, integrative plasmids, and pathogenicity islands (Liu et al., 2015). However, no integrase was identified in these predicted prophages.

| CRISPR prediction in the genomes of four clinically pathogenic and three environmental M. odoratimimus strains
CRISPR is well known to contribute to the antibiotic resistance and prevent the foreign virulence genes from invading into pathogens. CRISPRs were identified in the genomes of CCUG12901, CIP101113, and L41 (Table S1).

| Comparative analysis of antibiotic resistance genes in the genomes of four clinically pathogenic M. odoratimimus strains
With CARD RGI software (Jia et al., 2017;Mcarthur et al., 2013), all the genomes of four clinically pathogenic M. odoratimimus strains PR63039, CCUG10230, CCUG12901, and CIP101113 were predicted to contain a number of genes related to antibiotic resistance, including the β-lactam resistance gene, fluoroquinolone resistance gene, antibiotic target replacement protein, antibiotic inactivation enzyme, triclosan resistance gene, diaminopyrimidine resistance gene, phenicol resistance gene, elfamycin resistance gene, and efflux pumps conferring antibiotic resistance (Table 3).
Overall, more resistance genes were predicted in the fully se-

M. odoratimimus
All these four clinically pathogenic strains (PR63039, CCUG12901, We could not further correlate these predicted antibiotic resistance gene profiles to the antibiotic susceptibility of the other three clinically pathogenic strains (CCUG12901, CCUG10230, CIP101113) due to the lack of these related data. However, the conservation of the antibiotic resistance genes among the clinically pathogenic M. odoratimimus strains indicated that these predicted antibiotic resistance gene profiles potentially provide the guidance for treating M. odoratimimus infections later.

| Pathogenicity of clinically pathogenic and environmental M. odoratimimus indicated by the predicted virulence factors
The VFs in M. odoratimimus were identified using VFDB protein Set B database which curates the experiment-verified and predicted virulence factor genes (   (Gaddy et al., 2012). Thus, bauE should be a survival factor of
By interacting with host surface nucleolin, the bacterial surface EF-Tu (elongation factor Tu), a GTP-binding protein involved in protein translation in Francisella tularensis, a highly infectious intracellular gram-negative bacterium, plays a crucial role in its invasion to host tissues (Barel et al., 2008). It is also an adhesion/invasion factor secreted by microbes during infection by bacteria (like Helicobacter pylori) (Chiu, Wang, Tsai, Lei, & Liao, 2017) and fungi (Marcos et al., 2016) through binding (Mycobacterium avium subsp. paratuberculosis) with fibronectin on host cells (Viale et al., 2014). Superoxide dismutases (SODs) protect the bacteria from oxidative damage by converting endogenously generated superoxide radicals into hydrogen peroxide and oxygen, are indispensable for intraphagocytic viability for pathogenic bacteria (Dhar, Gupta, & Virdi, 2013), SodB is required for colonization of Helicobacter pylori in the stomach (Tsugawa et al., 2015).  (Ghazaei, 2017).
Hsp60 is also involved in Clostridium difficile attachment to host cells (Hennequin et al., 2001), and the strong proinflammatory reaction (IL8) of monocytic cells induced by Helicobacter pylori (Lin et al., 2005). Chlamydia pneumonia Hsp60 can help to spread Chlamydial infection of blood monocytes to vascular wall cells (Rupp et al., 2005), and increase the pathogenesis and severity of Chlamydia infectioncorrelated atherosclerosis because of sequence homology between bacterial and human Hsp60 (mitochondria in endothelial cells) and subsequent induction of a strong autologous humoral and cellular immune responses (Kalayoglu et al., 2000;Mehta et al., 2005).
The presence of ureA, ureB, ureG in M. odoratimimus implies that this bacterium might be pathogenic in human stomach. Urease is a principal virulence factor of human gastric bacterium Helicobacter pylori (Stingl et al., 2008). It has oligomeric Ni 2+ -containing heterodimer of UreA and UreB subunits involved in converting gastric juice urea into NH 3 in bacterial periplasm which maintains an optimal pH, inner membrane potential and proton motive force, being critical for colonization within the human stomach (Sachs, Weeks, Melchers, & Scott, 2003 The surface enolase (eno) of bacteria, a glycolytic pathway enzyme, can bind human plasminogen and convert it into active plasmin (Cork et al., 2009) to facilitate bacterial adherence to host cells and destruction of host tissues through plasmin degrading intercellular junctions and extracellular matrix components (Attali, Durmort, Vernet, & Di Guilmi, 2008), like cellulitis (Bachmeyer et al., 2008), necrotizing fasciitis (Crum-Cianflone, Matson, & Ballon-Landa, 2014), and make the bacterial infection life-threatening (Cork et al., 2009). The containing eno in M. odoratimimus genome might explain the high death rate of patients infected by M. odoratimimus (as summarized in Hu et al., 2016).
In brief, M. odoratimimus not only possesses common virulence factors, like using bauE gene to compete the iron with host, general LPS synthesis genes, adherence factors (DnaK, Hsp60), but also can survive intracellularly (katA, clpP, EF-Tu, and sodB), even in human stomach (ureA, ureB, ureG), but also disseminate easily, destroy human tissues, induce autoimmune diseases. So, the M. odoratimimus is a life-threatening pathogen as reported (summarized in Hu et al., 2016).

| CON CLUS ION
The genomic analysis demonstrated that these M. odoratimimus isolates are closely related. Our analyses provided some insights in bacterial pathogenicity and antibiotic resistance mechanisms of M. odoratimimus and contribute to future development of the therapeutic regimens in M. odoratimimus infections.

CO M PLI A N CE WITH E TH I C A L S TA N DA R DS
This article does not contain any studies with human participants or animals performed by any of the authors.

CO N FLI C T O F I NTE R E S T
None declared.