Insights from the draft genome into the pathogenicity of a clinical isolate of Elizabethkingia meningoseptica Em3

Elizabethkingia meningoseptica is an emerging, healthcare-associated pathogen causing a high mortality rate in immunocompromised patients. We report the draft genome sequence of E. meningoseptica Em3, isolated from sputum from a patient with multiple underlying diseases. The genome has a length of 4,037,922 bp, a GC-content 36.4%, and 3673 predicted protein-coding sequences. Average nucleotide identity analysis (>95%) assigned the bacterium to the species E. meningoseptica. Genome analysis showed presence of the curli formation and assembly operon and a gene encoding hemagglutinins, indicating ability to form biofilm. In vitro biofilm assays demonstrated that E. meningoseptica Em3 formed more biofilm than E. anophelis Ag1 and E. miricola Emi3, both lacking the curli operon. A gene encoding thiol-activated cholesterol-dependent cytolysin in E. meningoseptica Em3 (potentially involved in lysing host immune cells) was also absent in E. anophelis Ag1 and E. miricola Emi3. Strain Em3 showed α-hemolysin activity on blood agar medium, congruent with presence of hemolysin and cytolysin genes. Furthermore, presence of heme uptake and utilization genes demonstrated adaptations for bloodstream infections. Strain Em3 contained 12 genes conferring resistance to β-lactams, including β-lactamases class A, class B, and metallo-β-lactamases. Results of comparative genomic analysis here provide insights into the evolution of E. meningoseptica Em3 as a pathogen. Electronic supplementary material The online version of this article (10.1186/s40793-017-0269-8) contains supplementary material, which is available to authorized users.

Infections by E. meningoseptica are difficult to treat with antimicrobial agents due to multiple drug resistance [4]. Tetracycline, chloramphenicol, and β-lactams have been used to treat patients [10], but increasingly clinical isolates lack susceptibility to these antibiotics [11]. Analysis of the resistome in the related bacterium E. miricola revealed multiple drug resistance genes [12]. Some antibiotics effective against Gram-positive bacteria such as vancomycin, quinolones, tigecycline, and rifampin have been used for treating E. meningoseptica-infected patients, though the mechanism of action remains unclear [12,13]. Also, the effectiveness of these antibiotics varied; many patients resolved infection but isolates showed high MICs in vitro, thus the relationship between MICs and clinical response was obscure [14]. Further genome analyses will elucidate the breadth of antibiotic susceptibility and resistance mechanisms in Elizabethkingia spp.
Differentiation of Elizabethkingia species using routine morphological and biochemical tests is difficult in clinical laboratories [14]. Comparison of 16S rRNA identity does not provide sufficient resolution to identify and separate these closely-related Elizabethkingia species [2,14]. Characterization of Elizabethkingia species by MALDI-TOF mass spectrometry would facilitate it if species reference spectra were added to the database [14]. A limitation is that MALDI-TOF mass spectrometry is not available in many smaller clinical microbiology laboratories. Whole genome analysis facilitates the development of molecular diagnosis tools (such as single nucleotide polymorphisms) that can be potentially useful for small laboratories. In this study, we sequenced, annotated and analyzed a clinical E. meningoseptica genome, with the aim of providing a better understanding of antibiotic resistance and pathogenesis mechanisms in this pathogen, and of unveiling useful biosystematic molecular markers.

Organism information
Classification and features E. meningoseptica Em3 (Fig. 1) was isolated from a sputum sample from a patient with multiple underlying diseases and on life support. E. meningoseptica Em3 is Gram-negative, non-motile and non-spore-forming ( Fig. 1 and Table 1). A taxonomic analysis was performed by comparing the 16S rRNA gene sequence to those in the GenBank (Fig. 2). The phylogenetic tree based on the 16S rRNA gene sequences indicated that strain Em3 was clustered within a branch containing other E. meningoseptica and departing from the clusters E. anophelis and E. miricola in the genus Elizabethkingia (Fig. 2). We further calculated the ANI and DDH values among the representative Elizabethkingia (Table 2). Our results showed that strain Em3 belongs to E. meningoseptica because of the high ANI (>95%, cutoff for species differentiation) and DDH (>70%, cutoff for species differentiation) values between strain Em3 and E. meningoseptica ATCC 13253 T [15].
The motility was tested on semi-TSA. The cells of strain Em3 are straight and rods and have a diameter of 0.7 μm and length of 24.0 μm. Strain Em3 grew on TSA, producing yellow pigment (Fig. 1). This bacterium also grew well on SBA with greyish discoloration around the colonies, showing it had the α-hemolytic activity (Fig. 1). E. meningoseptica Em3 did not grow on MacConkey agar, a finding consistent with strain-dependent growth on this medium; e.g., E. meningoseptica CCUG 214 T grew on MacConkey agar whereas other hospitalassociated E. meningoseptica strains did not [2]. Of those Fig. 1 Demonstration of cell growth, pigment production and micrograph. a Em3 growing on SBA medium; b Demonstration of Em3 grown on SBA medium (control, up part) and MacConkey agar (low part). c Pigment production in Em3 grown on TSA agar. d Scanning electron microscopy image of Em3 strains growing on MacConkey agar, lactose was not utilized [2]. The optimal growth temperature for strain Em3 was 37°C (Table 1). Carbon source, nitrogen source utilization and osmotic tolerance were assayed by incubating cells in Biolog GEN III microplates at 37°C overnight (CA, USA). The results showed that E. meningoseptica Em3 did not tolerate 4% NaCl. E. meningoseptica Em3 utilized several carbon sources, including D-maltose, D-trehalose, D-gentibiose, Dmelibiose, D-glucose, D-mammose, D-fructose, D-fucose, D-mannitol and D-glycerol. The ability to use D-melibiose can differentiate E. meningoseptica from E. anophelis and E. miricola [16]. The inability to grow on cellobiose or citrate was consistent with previous reports [16]. Moreover, E. meningoseptica Em3 utilized D-serine, L-alanine, Laspartic acid, L-glutamic acid, L-histidine and L-serine when tested on Biolog GEN III microplates.

Extended feature descriptions
Phylogenetic analysis (Additional file 1: Figure S1) was further conducted by using 19 genomes with 1181 core genes per genome (22,439 in total). As expected, E. meningoseptica Em3 grouped together with the selected E. meningoseptica species and separated from the clusters E. anophelis, E. endophytica and E. miricola, a finding similar to the phylogenetic analysis based on 16 s rRNA sequences. Further, both trees ( Fig. 1 and Additional file 1: Figure S1) show that species E. anophelis and E. endophytica are not separated well, which is consistent with previous reports [17].

Genome project history
The genome of E. meningoseptica Em3 was selected for whole genome sequencing because of its association Evidence codes -IDA: Inferred from Direct Assay; TAS: Traceable Author Statement (i.e., a direct report exists in the literature); NAS: Non-traceable Author Statement (i.e., not directly observed for the living, isolated sample, but based on a generally accepted property for the species, or anecdotal evidence). These evidence codes are from the Gene Ontology project [41] with pulmonary disease. Comparison of strain Em3 genome with other Elizabethkingia species may provide insights into the molecular basis of pathogenicity and metabolic features of this strain. The high-quality draft genome sequence was completed on August 1, 2016 and was deposited to GenBank as a Whole Genome Shotgun project under accession number MDTY00000000 and the Genome OnLine Database with ID Gp0172366 (Table 3).

Growth conditions and genomic DNA preparation
For genomic DNA isolation, E. meningoseptica Em3 (CL16-200185, Bureau of Laboratories, Michigan Department of Health and Human Services) culture was grown overnight Fig. 2 Phylogenetic tree displays the position of E. meningoseptica Em3 (shown in bold) relative to the other type strains of Elizabethkingia based on 16S rRNA. The phylogenetic tree was constructed by MEGA v. 7.0.14 using the Neighbor-Joining method [42]. The percentage of replicate trees where the associated taxa clustered together in the bootstrap test (500 replicates) is indicated next to the branches. The branch lengths are scaled to the same units as those of the evolutionary distances for inferring the phylogenetic tree. The accession numbers for 16 s rRNA sequences are listed in the parenthesis following selected bacteria:  The Illumina data were assembled into contiguous sequences using SPAdes version 3.9.0 [19], then short contigs (<400 bp) were filtered out. The 11 contigs identified in this strain were therefore submitted to the NCBI database as a Whole Genome Shotgun project.

Genome annotation
Annotation of the 11 contigs was first done through the NCBI Prokaryotic Genome Automatic Annotation Pipeline [20]. The predicted CDSs were next translated and analyzed against the NCBI non-redundant database, iPfam, TIGRfam, InterPro, KEGG and COG. The RAST system was used to check the annotated sequences [21,22]. Additional gene prediction and manual revision was performed by using the IMG/MER platform. E. meningoseptica Em3 genome is available in IMG (genome ID = 2,703,719,242).

Genome properties
The draft genome sequence is 4,037,922 bp long, 36.37% G + C rich and contains 11 scaffolds (Table 4). Of 3729 genes predicted, 3673 encoded proteins and 56 were RNAs. 2585 (69.32%) were assigned a putative function, while the other 1088 (30.68%) were designated as hypothetical proteins. The distribution of coding genes into general COG functional categories analyzed by IMG is listed in Table 5. Collectively, the genome features were similar to those in other sequenced E. meningoseptica (Additional file 2: Table S1).

Insights from the genome sequence
Elizabethkingia bacteria cause sepsis, bacteremia, meningitis or respiratory tract infections in hospitalized patients, indicating that they have the ability to colonize host tissues, suppress the host immune response, and disrupt erythrocytes to obtain nutrients and propagate in the host bloodstream [1,13,14]. Genome analysis showed that E. meningoseptica Em3 carried a gene (BFF93_RS1398) encoding a hemagglutinin protein.
Hemagglutinins as adhesins are required for virulence in bacterial pathogens [23]. Hemagglutins facilitate infection via adherence to epithelial cell lines from the human respiratory tract in Bordetella pertussis [24]. Darvish et al. showed that filamentous hemagglutinin adhesins were crucial for bacterial attachment and subsequent cell accumulation on target substrates [25]. An in vitro biofilm assay showed that, compared to the   (Fig. 3). Furthermore, E. meningoseptica Em3 had better ability to form biofilm than did E. miricola Emi3. The capacity for strain Em3 to form biofilm was further exemplified by discovery of an operon involved in curli biosynthesis and assembly (BFF93_RS03755, BFF93_RS03760, BFF93_RS03765, BFF93_03725 and BFF93_RS03775). In vitro studies demonstrated that curli fibers participated in bacterial adhesion to target cell surfaces, cell aggregation, as well as biofilm formation [26,27]. Moreover, some studies showed that curli mediated host cell attachment and invasion in vivo [28]. Curli were involved in inducing the host inflammatory response [29]. It should be noted that this curli synthesis operon is present in E. meningoseptica while it is absent in E. miricola. Further experiments are warranted to test if the curli gene cluster contributed to biofilm formation in strain Em3 because biofilm formation may involve other genes. A cytolysin encoding gene (BFF93_RS16990) was found in the strain Em3 genome, whose product belonged to a thiol-activated, CDC family [30]. CDC as a virulence factor is widely distributed among Gram-positive, opportunistic pathogens [31]. For example, Streptococcus pyogenes utilized pore-forming CDC to translocate a protein into eukaryotic cells [32]. Disruption of expression of a hemolysin (CDC) gene in the intracellular pathogen Listeria monocytogenes reduced virulence in mice, showing that CDC was critical for full virulence [33]. Furthermore, perfringolysin, a CDC toxin, has cytotoxicity and leukostasis activities, allowing the cells to escape from macrophage phagosomes during Clostridium perfringens gas gangrene [34]. Only a few CDCs have been found in Gram-negative bacteria [31], and this is the first report of CDC genes in E. meningoseptica. It should also be noted that this cytolysin gene is located immediately downstream of hmuY, which together comprise part of an iron metabolism gene cluster. Such gene organization was only seen in E. meningoseptica. This CDC protein sequence in strain Em3 shared 87%, 83% and 81% identity to that in E. meningoseptica ATCC 13253, E. meningoseptica B2D and E. meningoseptica NBRC 12535, respectively. It is interesting that it did not have close identity to that in E. meningoseptica FMS-007 (48%) and E. meningoseptica 502 (48%); it was absent in an E. meningoseptica strain associated with endophthalmitis [35]. Similarly, it was not conserved in E. anophelis (identity ranging from 0 to 50%) and absent in Besides a CDC gene in strain Em3 genome, we found that there was a gene encoding the hemolysin with a CBS domain (BFF93_RS14485). Hemolysin can be possibly secreted and involved in lysis of the erythrocytes [35]. The predicted amino acid sequence was conserved in most E. meningoseptica strains (> 90%). Further examination of hemin-degrading/transporter/utilization proteins led to a discovery of the gene cluster including SAM-dependent methyltransferases (BFF93_RS02055), iron ABC transporter (BFF93_RS02045), hemin-degrading protein (hmuS, BFF93_RS02060), hemin importer ATP-binding protein (BFF93_RS02050) and iron-regulated protein (BFF93_ RS02065). Furthermore, there was a gene encoding a hemin receptor (BFF93_RS03140).
Elizabethkiniga infections can be fatal in immunecompromised patients if appropriate antibiotic therapy is delayed or the antimicrobial treatment is not properly provided [9,14]. However, Elizabethkiniga spp. are multi-drug resistant [4,13]. The prediction results by CARD and RAST (Table 6) showed that there are at least 31 genes involved in antibiotic resistance including antibiotic inactivation enzymes and related efflux pumps in E. meningoseptica Em3. Many of them are possibly involved in mupirocin, vancomycin, β-lactam, aminocoumarin, elfamycin, isoniazid, tetracycline and fluoroquinolone resistance ( Table 6). Several drugs used to treat Elizabethkinigainfected patients in the past are not effective anymore [4], which agrees with recent resistome assays in clinical E. meningoseptica isolates [12]. Genes associated with resistance to β-lactams, aminoglycosides, tetracycline, vancomycin, and chloramphenicol, reported here in strain Em3, are present in most of the studied Elizabethkingia spp. (Table 6). Remarkably, at least 12 β-lactam resistance genes encoding MBL fold metallo-hydrolases, metallo-βlactamases and β-lactamases (class A and B) were found in E. meningoseptica Em3 genome (Table 6). Alternatively, antibiotics such as ciprofloxacin, minocycline, trimethoprimsulfamethoxazole, rifampin and novobiocin may remain effective due to absence of relevant antibiotic resistance genes in Elizabethkingia sp. [36]. Therefore, a combination of antimicrobial tests and resistome analysis, combined with rapid identification of infections, will contribute to efficient management for E. meningoseptica infections in the future.

Conclusions
The draft genome sequence of E. meningoseptica Em3 isolated from a sputum sample in a patient was sequenced, annotated and described. We found that E. meningoseptica Em3 had novel genes encoding thiol-activated cholesteroldependent cytolysin, curli and heme metabolism related proteins, showing that E. meningoseptica Em3 may be a causative agent. Our results also indicated that E. meningoseptica might be resistant to β-lactam antibiotics due to the production of diverse MBLs and β-lactamases. Furthermore, these β-lactamase encoding genes were also found in other Elizabethkingia species, indicating that Elizabethkingia species were important reservoirs of novel β-lactamase genes. Comparative genomics is a crucial approach in the discovery of novel virulence determinants in Elizabethkingia species. Genome-based approaches contribute to develop novel genetic markers for future molecular diagnosis of Elizabethkingia infections. Fig. 3 In vitro biofilm assay in the selected Elizabethkingia sp. The cells were first cultured by shaking in TSB at 37°C overnight. The cell density was adjusted to the same OD at 600 nm (0.1). 200 μl of cells were placed on 96-well plates for 24 h. The biofilm assay was carried out using crystal blue staining [49]. Values are mean values for single measurements from eight independent cultures. The error bars are standard deviations. The statistical test was the Student's t-test. The asterisk indicates a significant difference compared to biofilm formation in E. anophelis Ag1 (p < 0.05). "+" or "-" indicates the presence or absence of genes in the selected Elizabethkingia