The antibiotic resistance and pathogenicity of a multidrug‐resistant Elizabethkingia anophelis isolate

Abstract Elizabethkingia anophelis 12012‐2 PRCM was isolated from a patient with multiple organ dysfunction syndrome and lower respiratory tract infection in China. Minimum inhibitory concentration (MIC) analysis demonstrated that it was resistant to 20 antibiotics including trimethoprim/sulfamethoxazole and ciprofloxacin, which were effective for the elimination of other Elizabethkingia infections. To investigate multidrug resistance and pathogenicity mechanisms, we analyzed genome features of 12012‐2 PRCM and compared them to the other Elizabethkingia species. The draft genome size was 4.02 Mb with a GC content of 32%, comparable to that of other E. anophelis strains. Phylogenetic analysis showed that E. anophelis 12012‐2 PRCM formed a sister group with E. anophelis 502, distinct from clades formed by other clinical and environmental E. anophelis isolates. E. anophelis 12012‐2 PRCM contained multiple copies of β‐lactamase genes as well as genes predicted to function in antimicrobial efflux. It also contained 92 genes that were potentially involved in virulence, disease, and defense, and were associated with resistance and pathogenicity. Comparative genomic analysis showed high homology among three clinical and two environmental E. anophelis strains having a variety of similar antibiotic resistance and virulence factor genes, and similar genomic structure. Applications of this analysis will contribute to understanding the antibiotic resistance and pathogenic mechanisms of E. anophelis infections, which will assist in the management of infections as it increases in prevalence.

The unknown pathogenesis mechanisms, multidrug resistance mechanisms, and misclassifications as other bacteria complicate management of E. anophelis infections (Frank et al., 2013;Hu, Jiang, Zhang et al., 2017;Lau et al., 2016Lau et al., , 2015. Routine phenotypic and biochemical tests often fail to distinguish them from other bacteria; moreover, E. anophelis has been frequently misdiagnosed as E. meningoseptica (previously known as Chryseobacterium meningosepticum) with automated microbial identification systems (Kampfer et al., 2011;Lau et al., 2016Lau et al., , 2015Nicholson et al., 2016;Teo et al., 2013). Often, molecular methods (i.e., the 16SrRNA sequencing, MALDI-TOF MS) fail to resolve different Elizabethkingia species (Breurec et al., 2016;Han et al., 2017). Empirical treatments are difficult because of multiple drug resistance and lack of drug susceptibility testing standards for these bacteria. Particularly, our knowledge of the antibiotic resistance spectra and the resistance mechanisms remain limited in E. anophelis because it is a relatively newly discovered bacterium. The pathogenesis mechanisms in Elizabethkingia remain unclear. Strains isolated during the outbreak in Wisconsin harbored a mutation in the MutY gene which is involved in DNA repair (Perrin et al., 2017), but the relevance of it to virulence is unknown.
Our E. anophelis strain, 12012-2 PRCM, was isolated from a patient with multiple organ dysfunction syndrome (MODS) (Hu, Jiang, Zhang et al., 2017). This isolate was not susceptible to any selected antibiotics, demonstrating it was a multidrug-resistant strain. Therefore, the aim of this study was to investigate drug resistance and pathogenesis mechanisms. We performed genome sequencing for E. anophelis 12012-2PRCM and conducted a comparative genomic analysis to those in other environmental and clinical isolates.
Our results contribute to the management of Elizabethkingia infection and the better understanding the pathogenicity of E. anophelis.

| DNA extraction and antimicrobial susceptibility testing
A multidrug-resistant E. anophelis strain, designated 12012-2PRCM, was isolated from an 82-year-old male patient presenting with MODS and lower respiratory tract infection (Hu, Jiang, Zhang et al., 2017). Antimicrobial susceptibility testing (AST), bacteria culturing, and genomic DNA extraction were done as previously performed .

| Whole-genome sequencing, assembly, and annotation for E. anophelis 12012-2PRCM
Genome sequencing was done with the MiSeq instrument (Illumina, Inc., San Diego, CA) using 500 bp library preparations. Raw data processing and genome assembly were performed by the SOAPdenovo 2.04-r240 version (Li et al., 2010). After assembly, we obtained a 402,331,983-bp genome containing 83 contigs and 76 scaffolds. It was deposited into GenBank (LPXG00000000). The genome annotation was done with RAST (Aziz et al., 2008;Overbeek et al., 2014).
Alignment of five E. anophelis genomes, including the strain

12012-2PRCM
The assembly of strain 12012-2PRCM sequence data generated 83 scaffolds. It had a genome of 4.02 M bp with an average GC content of 35.5%. E. anophelis 12012-2PRCM had 3,680 genes including 3,554 protein-encoding genes, 82 pseudogenes, and 42 tRNAs ( Table 1). The RAST showed that E. anophelis 12012-2PRCM genome had 27 subsystems that consisted of 87 categories (Figure 1). At least 330, 275, 268, and 121 CDSs were assigned to the "amino acid and derivatives," "carbohydrate metabolism," "protein metabolism," and "RNA metabolism" categories, respectively. Moreover, the "virulence, disease and defense" category contained 92 CDSs that were involved in resistance to antibiotics and toxic compounds, indicating that this strain was possibly resistant to multiple antibiotics (also see below).
Resistance to tetracycline, trimethoprim/sulfamethoxazole, and ciprofloxacin raised a serious concern because these drugs have been

| Resistome analysis
Antibiotic resistance genes were predicted by searching the CARD database (Jia et al., 2017;McArthur et al., 2013). At least eight classes of antibiotic resistance genes were found in E. anophelis 12012-2 PRCM (Table 2).
Elizabethkingia bacteria are well known to be highly resistant Besides the mutational gyrA, the fluoroquinolone-resistant genes, rpsJ and tetB (48), were discovered in strain 12012-2PRCM, which may also contribute to the resistance to fluoroquinolones.
All five E. anophelis strains contained many catB genes or cat variants (Table 2), which usually play a role in the composition of gene cassette or integron, and confer to the ability of antibiotic resistance.

| Comparative analysis of the virulence factor genes in E. anophelis strains
The homologs of the virulence factors (VFs) in E. anophelis iso- in Mycobacterium tuberculosis pathogenesis course (Pym et al., 2001). IlpA, a membrane-bound lipoprotein, has been known to function as an adhesion factor in Vibrio vulnificus. It helps the adhesion to human immune cells through its C-terminal domain.
Consequentially, it induces cytokine production, which plays an important role in V. vulnificus infection (Goo, Han, Kim, Lee, & Park, 2007;Lee et al., 2010). One can assume the same physiological roles in 12012-2-PRCM due to their good amino acid sequence homology. The presence of IlpA in our strain 12012-2 htpB, and DnaK may be involved in defense or invasion during the course of pathogenesis, already discussed in our previous report (Hu et al. 2018). In addition, it is worth noting that E. anophelis isolates from mosquitoes also shared these conserved virulence factors. However, their potential for pathogenicity in humans have not been investigated.
In our strain 12012-2 PRCM, only one prophage was identified. It had nine CDs located at 47,038 bp-56041 bp (9 kb). The strain CSID 3000521207 also contained one 7.8-kb prophage extending from 2,136,491 bp to 2,144,356 bp. NUHP1 was predicted to carry four prophages (8.3 kb, 7.8 kb, 7.9 kb, and 7.2 kb, respectively) ( Figure A1). Strains Ag1 and R26 shared three prophages (8.9 kb, 7.2 kb, and 6.2 kb, respectively), although the prophages were located on different sites in two of the genomes ( Figure A1), implying that genome rearrangements existed. Of interest, our strain 12012-2 PRCM shared one prophage of Ag1 and R26 while prophage of CSID 3000521207 was similar to the one in NUHP1 ( Figure A1), demonstrating that prophages in E. anophelis species were conserved. However, among these predicted prophages, many elements were lost. For example, a significant component integrase (a marker for mobile DNA elements and participating in bacteria pathopoiesis (Liu et al., 2015) was not predicted in any of the above prophages. In our strain 12012-2 PRCM, using the database ICEberg 2.0, a putative ICE region (location: 2,558,736 to 2,565,836 bp) was identified. In this mobile genetic element, both relaxase and integrase (TIGR02249) were predicted ( Figure A2). The CSID 3000521207, one present representative isolate of the outbreak in Wisconsin, had the integrative and conjugative element ICEEa1 (Perrin et al., 2017). Also, the Anopheles mosquito strains Ag1 and R26 contained ICEEaIII (Xu et al., 2018). More detailed analysis of ICEs will clarify pathogenesis and drug resistance mechanisms of E. anophelis.

| Synteny analysis of five E. anophelis strains
The selected E. anophelis genomes had some chromosomal rearrangements with some inversions ( Figure A3) and syntenic rearrangements. However, the genome arrangement of the three clinical isolates mimicked each other. Instead, the clinical and environmental isolates showed less similarity ( Figure A3).
The other selected genomes had no CRISPR. The defense of the invasions of foreign genetic elements such as plasmids, transposons, or phages may require both restriction modification systems (RMs) and CRISPRs in Elizabethkingia. However, the detailed mechanisms need to be further investigated.

| CON CLUS ION
Genomic analysis provided partial insight on the antibiotic resistance and pathogenicity mechanisms of clinical multidrug-resistant E.
anophelis isolates. This could prove useful information in the development of future therapeutic regimens to eliminate the infections caused by the emerging pathogen E. anophelis.

ACK N OWLED G EM ENTS
Our sincere thanks are due to Department of Quanzhou First Hospital, Fujian, China for providing the E. anophelis strain. We express gratitude towards the Beijing Novogene Bioinformatics Technology Co., Ltd, BNNT for providing technical support.

CO N FLI C T O F I NTE R E S T S
The authors declare that they have no conflict of interest.

E TH I C S S TATEM ENT
A informed consent was obtained from the patient's relatives to retrieve and analyze this bacterial isolate. The research was approved by the Ethics Committee of Quanzhou First Hospital.

DATA ACCE SS I B I LIT Y
All genomic data are available through the NCBI (https://www. ncbi.nlm.nih.gov) using the corresponding accession numbers provided. F I G U R E A 3 Alignment of Elizabethkingia anophelis12012-2PRCM, NUHP1, CSID3000521207, Ag1 andR26 withthe progressive MAUVEsoftware. Colored blocks: a region of the genome sequence which was assumed to be homologous and internally free from genomic rearrangement. Regions outside blocks: no homology among these genomes. Completely white areas: not aligned and possibly containing specific sequence elements to a certain genome