Physiological and genomic features of a novel violacein-producing bacterium isolated from surface seawater

Strains JW1T and JW3, isolated from surface seawater of the Arabian Sea, were subjected to polyphasic taxonomic analysis. Cells of both strains were Gram-stain-negative, aerobic, and rod-shaped. They formed violet pigment and produced violacein. On the basis of 16S rRNA gene sequence analysis, strains JW1T and JW3 showed high 16S rRNA gene sequence similarity with Pseudoalteromonas byunsanensis JCM12483T (98.2%), P. shioyasakiensis SE3T (97.8%), P. arabiensis JCM 17292T (97.3%), and P. gelatinilytica NH153T (97.1%). The 16S rRNA gene sequence similarity between JW1T and JW3 was 100%. Phylogenetic analyses revealed that both strains fell within the cluster of the genus Pseudoalteromonas and represented an independent lineage. The average nucleotide identity and in silico DNA-DNA hybridization values between JW1T and type strains of the closely related Pseudoalteromonas species were 70.9–83.3% and 20.0–26.4%, respectively. The sole respiratory quinone in both strains is ubiquinone 8 (Q-8). The principal fatty acids are summed feature 3 (C16:1ω7c and/or iso-C15:0 2OH), C18:1ω7c, and C16:0. The major polar lipids are phosphatidylethanolamine, phosphatidylglycerol, one unidentified glycolipid, one unidentified aminolipid, and one unidentified phospholipid. The DNA G+C content was 43.3 mol%. Differential phylogenetic distinctiveness, chemotaxonomic differences, and phenotypic properties indicated that strains JW1T and JW3 could be differentiated from the Pseudoalteromonas species with validly published names. Therefore, it is proposed that strains JW1T and JW3 represent a novel species of the genus Pseudoalteromonas, for which the name Pseudoalteromonas amylolytica sp. nov. (type strain, JW1T = CGMCC 1.15681T = KCTC 52406T = MCCC 1K02162T) is proposed.


Introduction
The genus Pseudoalteromonas, the type genus of the family Pseudoalteromonadaceae [1], was proposed by Gauthier et al. (1995) [2]. Initially, the genus Pseudoalteromonas was differentiated from the genus Alteromonas based on the phylogenetic analysis of 16S rRNA gene sequences [2]. Currently, the genus Pseudoalteromonas consists of 43 species with validly a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 published names (http://www.bacterio.net/p/pseudoalteromonas.html). Members of the genus Pseudoalteromonas are widespread in nature and have a great adaptability to marine environments, such as coastal, open, and deep seawaters, sediments, marine invertebrates, fish, and algae [3]. The genus Pseudoalteromonas is Gram-negative, aerobic or facultatively anaerobic, and rod-shaped, it requires Na + ions for growth, usually does not denitrify, and possesses ubiquinone-8 (Q8) as major respiratory quinone [3].
Some Pseudoalteromonas species produce a variety of primary and secondary metabolites, including antibiotics [2], exopolymers [4,5], hydrolytic enzymes [6,7], and pigments [2,8]. Violacein is a natural indolocarbazole compound formed by condensation of two molecules of tryptophan [9]. It is a potential pharmaceutical agent owing to its extensive biological properties, such as antibacterial, antiviral, antioxidant, and antitumor activities [10]. Pseudoalteromonas luteoviolacea has been reported to produce violacein [11]. Here, we present a polyphasic study describing two novel violacein-producing strains, both of which were isolated from surface water of the Arabian Sea.

Organisms and culture conditions
Strains JW1 T and JW3 were isolated from the surface seawater collected from the Arabian Sea (E67˚N24˚). The seawater samples were stored at 4˚C until use. Natural seawater agar (pH 7.2-7.4) supplemented with 0.05% peptone (w/v; BD, Sparks, MD, USA) and 0.01% yeast extract (w/v; BD) was used for isolation. The seawater samples were diluted using the standard ten-fold dilution plating technique and spread on natural seawater agar. After ten days of aerobic incubation at 30˚C, two violet colonies, designated as JW1 T and JW3, were picked from different samples and purified by repeated restreaking. The purity was confirmed by the uniformity of cell morphology. The reference strains P. byunsanensis JCM 12483 T , P. shioyasakiensis JCM 18891 T , and P. arabiensis JCM 17292 T were obtained from the JCM (Japan Collection of Microorganisms). The reference strain P. gelatinilytica NH153 T was available in our lab [12]. Unless otherwise stated, the two strains were routinely cultured in marine broth 2216 (MB; BD) or on marine agar 2216 (MA; BD) at 30˚C and stored at -80˚C with 30% (v/v) glycerol.

16S rRNA gene and genome sequence determination
The 16S rRNA gene was amplified and analyzed as described previously [13]. PCR products were cloned into the vector pMD 19-T (TaKaRa, Dalian, China) and then sequenced to determine the almost-complete sequence of the 16S rRNA gene. High-quality genomic DNA was extracted with the AxyPrep™ Bacterial Genomic DNA Miniprep Kit (Axygen Scientific, Inc., Union City, CA, USA). The genomes of strains JW1 T , JW3, and P. byunsanensis JCM 12483 T were sequenced using the Solexa paired-end sequencing technology with the Illumina HiSeq 2000 platform (Anoroad Gene Technology Co. Ltd, Beijing, China). One paired-end library was constructed with 500-bp insert size. The sequencing generated approx. 1 Gb of clean data (approx. 500-fold genome coverage). De novo assembly of the reads was carried out using SOAPdenovo (version 2.0.1) [14]. Assembly k-mer was tested from 57 to 64 for seeking the optimal value, using the abyss-pe script. Assembly quality was estimated using MUMmer [15]. Completeness of the genome sequence was addressed using the bioinformatics tool CheckM (http://ecogenomics.github.io/CheckM/) [16]. The complete sequence of the 16S rRNA gene was annotated via the RNAmmer 1.2 Server [17] and was compared with related sequences of reference organisms using the EzTaxon-e service [18].

Phylogenetic status
Phylogenetic analysis was carried out using ARB (release 6.0.2) [19] and the All-Species Living Tree Project database (LTPs123, September 2015) [20]. The 16S rRNA gene sequences of strains JW1 T and JW3 were aligned with SILVA Incremental Aligner (SINA, version 1.2.11) (http://www.arb-silva.de) [21]. The alignment sequences were imported into the LTPs database and implemented in ARB. On the basis of the obtained All-Species Living Tree and the EzTaxon-e results, 24 species were selected and sequence data were aligned with ClustalW [22]. Phylogenetic trees were reconstructed using the MEGA 5 program package [23], using Algicola sagamiensis B-10-31 T as the outgroup, by neighbor-joining [24], maximum-parsimony [25], and maximum-likelihood methods [26]. Tree topology was evaluated by bootstrap analysis using 1000 resample datasets. Kimura two-parameter model [27] was used to calculate evolutionary distances and reconstruct phylogeny (neighbor-joining and maximum-likelihood methods).
The genomes of 20 type strains of Pseudoalteromonas species were retrieved from the Gen-Bank database (S1 Fig). Six housekeeping genes, atpD (beta subunit for ATP synthase), gyrB (DNA gyrase beta subunit), mreB (rod shape-determining protein), recA (RNA recombinase alpha subunit), rpoD (RNA polymerase), and topA (DNA topoisomerase I) were used for multilocus sequence analysis (MLSA). The concatenated sequence of six single genes was obtained from the genome and subjected to maximum-likelihood phylogenetic analyses [26] with MEGA 5.

Phenotypic characteristics
Cell morphology, size, and motility were examined using confocal laser scanning microscopy (TCS SP5; Leica) and transmission electron microscopy (JEM-1230; JEOL). The hanging-drop method was used for motility testing. Cell morphology and ultrastructure were observed using transmission electron micrographs.
Gram reaction, oxidase and catalase activities, and hydrolysis of starch and Tween-20, -40, and -80 were tested according to [28]. Violacein was extracted according to [29] and the absorbance of the violet pigment was monitored from 350 nm to 1000 nm using a UV/visible spectrophotometer (DU800, Beckman Coulter). The molecular weight of violacein was deduced by LC-MS analysis according to [29]. Chromatography was carried out on a Agilent 1200. Compound separation was achieved on an analytical column (Extend-C18, 3.5 μm, 2.1 × 100 mm; Agilent). Analysis of the pigment by eletrospray ionization mass spectrometry was conducted with Finnigan LCQ DECA XP MAX mass spectrometer (Thermo Electron Corp., USA). The flow rate of the pigment solution was 15 μL/min. The utilization of carbon substrates as sole carbon and energy sources was tested in BM [30] supplemented with filter-sterilized complex nutrients (yeast extract, peptone and tryptone, 0.2%, w/v), sugars (0.2%, w/v), alcohols (0.2%, w/v), organic acids (0.1%, w/v), or amino acids (0.1%, w/v). Yeast extract (0.01%, w/v) was added as a growth factor. Acid production was evaluated using marine oxidation-fermentation medium supplemented with 1% filter-sterilized sugars [31]. API 20NE and API 20E tests (bio-Mérieux) were used according to the manufacturer's instructions to determine additional physiological and biochemical characteristics. Strips were inoculated with a heavy bacterial suspension (MacFarland 5 standard) in AUX medium supplemented with 2% (w/v) sea salts (Sigma) [32]. API 20 NE and API 20 E strips were read after 48 h. Sensitivity to antimicrobial agents was determined with a two-layer plate method according to Wu et al. (2015) [33]. Four reference strains, P. byunsanensis JCM 12483 T , P. shioyasakiensis JCM 18891 T , P. arabiensis JCM 17292 T , and P. gelatinilytica NH153 T were used as controls in the above tests.

Chemotaxonomic characteristics
The cellular fatty acids of strains JW1 T , JW3, and the reference strains were determined under identical conditions in parallel. The quadrant streak method was used for inoculation and cellular fatty-acid methyl esters were obtained from cells grown on MA at 30˚C for 16 h from quadrant 3 (late exponential phase). Whole cell fatty acids were analyzed using the Microbial Identification System (MIDI Inc.) according to the manufacturer's instructions. Isoprenoid quinones were extracted and purified by two-dimensional thin-layer chromatography (TLC) and then analyzed by LC-MS (Agilent 1200 and Thermo Finnigan LCQ DECA XP MAX mass spectrometer) [34]. Total lipids were extracted and separated by two-dimensional TLC [35] on silica gel 60 F254 plates (Merck). Four types of spray reagent were used to detect the corresponding lipids, including molybdophosphoric acid for total lipids, α-naphthol reagent for glycolipids, ninhydrin reagent for lipids containing free aminolipids, and molybdenum blue for phosphorus-containing lipids [36].

Average nucleotide identities and genome analysis
The average nucleotide identity (ANI) was calculated using the OrthoANIu algorithm by ChunLab's online Average Nucleotide Identity calculator [37]. In silico DNA-DNA hybridization (DDH) values were calculated by GGDC [38].
rRNA genes were identified using the RNAmmer 1.2 Server [17] and tRNA genes were searched with the tRNAscan-SE 2.0 online server [39]. Gene prediction and functional annotation were carried out using the Rapid Annotation using Subsystem Technology (RAST) server online [40]. Selected predicted genes were classified using RPSBLAST against the COG database [41]. Translated genes were assigned to KEGG pathway using KEGG Automatic Annotation Server (KAAS) with the BBH method [42,43]. Orthologous cluster analyses were carried out using OrthoMCL [44]. Shared and unique orthologous clusters were established with inhouse shell scripts. CRISPR structures in the genomes were predicted with the CRISPRFinder program online (http://crispr.i2bc.paris-saclay.fr/Server/).

16S rRNA gene sequence similarities and phylogenetic analysis
The 16S rRNA gene sequences of strains JW1 T and JW3 (1527 nt) were obtained. According to EzTaxon as well as ClustalW results of 16S rRNA gene sequence comparison to representative bacteria with validly published names, strains JW1 T and JW3 showed high 16S rRNA gene sequence similarity to P. byunsanensis JCM12483 T (98.2%), P. shioyasakiensis SE3 T (97.8%), P. arabiensis JCM 17292 T (97.3%), and P. gelatinilytica NH153 T (97.1%), and exhibited less than 97.0% 16S rRNA gene sequence similarity with the type strains of other Pseudoalteromonas species. The 16S rRNA gene sequence similarity between strains JW1 T and JW3 was 100%.
The All-Species Living Tree indicated that the genus Pseudoalteromonas forms a monophyletic clade and strains JW1 T and JW3 fall within the cluster comprising the Pseudoalteromonas species. The topologies of neighbor-joining, maximum-likelihood, and maximum-parsimony phylogenetic trees based on the 16S rRNA gene also supported the notion that strains JW1 T and JW3 formed a stable lineage, with a high bootstrap value of 100%, and a distinct lineage from P. byunsanensis (bootstrap value 91%) (Fig 1). Phylogenetic trees based on concatenated sequences of the six housekeeping genes atpD, gyrB, mreB, recA, rpoD, and topA confirmed that the two strains formed a clade with P. byunsanensis as well as P. citrea and could not be associated with any of the recognized species in the genus Pseudoalteromonas (S1 Fig). Phylogenetic analysis indicated JW1 T and JW3 form another, independent lineage and might represent a novel member of the genus Pseudoalteromonas.

Chemotaxonomic analysis
Chemotaxonomic data supported the results of the phylogenetic analysis. The sole respiratory quinone found in strains JW1 T and JW3 was Q8, in line with all members of the genus Pseudoalteromonas [3]. Fatty-acid analysis revealed that summed feature 3, C 18:1 ω7c, and C 16:0 were the major fatty acids in JW1 T and JW3 and the references strains (Table 2). JW1 T and JW3 possessed phosphatidylethanolamine and phosphatidylglycerol as the major polar lipids, similar to the reference strains. In addition, JW1 T and JW3 possessed three unidentified glycolipids (GL2-GL4), one unidentified aminolipid (AL2), and one unidentified phospholipid (PL1) as moderate-to-minor polar lipids, which were similar to those of P. byunsanensis JCM 12483 T (S4 Fig). Moreover, JW1 T and JW3 possessed aminolipid, glycolipid, and phospholipid, all of which were detected in the four reference strains (S4 Fig) [12].
Chemotaxonomic data of JW1 T , JW3, and their relatives also showed some clear differences in fatty acid composition and polar lipid profile. The percentage of C 16:0 of strains JW1 T and JW3 (18.4% and 16.5%, respectively) was lower than that of the reference strains (21.0-27.1%). JW1 T and JW3 contained C 18:1 ω6c (4.1% and 4.6%, respectively), which were not detected in the reference strains (Table 2). One unidentified aminolipid (AL1) was present in both strain Table 1. Differential phenotypic characteristics of strains JW1 T , JW3, and closely related members of the genus Pseudoalteromonas. Strains/species: 1, strain JW1 T ; 2, strain JW3; 3, P. byunsanensis JCM 12483 T ; 4, P. shioyasakiensis JCM 18891 T ; 5, P. arabiensis JCM 17292 T ; 6, P. gelatinilytica NH153 T . All data were obtained in this study and under identical conditions, unless indicated otherwise. In silico DNA-DNA relatedness The DNA G+C content of strains JW1 T and JW3 calculated from the genome sequence was 43.3 mol%, a value in the range reported for members of the genus Pseudoalteromonas, i.e. 38-48 mol% [46]. JW1 T and the reference strains exhibited ANI values of 70.9-83.3% (Table 1).
These ANI values were far below the threshold of species boundary (94-96%) [47], indicating low taxonomic relatedness between JW1 T and the reference strains. The recommended results (formula 2) of the in silico DDH analysis revealed that JW1 T and the reference strains shared 20.0-26.4% DNA relatedness ( Table 1). The values were below 70%, indicating that the strains should be assigned to different genomic species [48]. In addition, the ANI and in silico DDH values between JW1 T and JW3 were 99.9%. These values were above the species boundary (94-96% for ANI values and 70% for in silico DDH values), suggesting that JW1 T and JW3 represent an identical genospecies.

Genomic features
General features of JW1 T and JW3 are displayed in Table 3 and S1 Table. The bioinformatics tool CheckM indicated that the genome completeness was 100% for both JW1 T and JW3, with a contamination percentage of 0.4% and 0.5%, respectively. Genome sequence completeness !95%, with 5% contamination, is considered to indicate an excellent reference genome for deeper analyses [16]. The genome size of the two strains and their related Pseudoalteromonas species varied from 4.5 Mb to 4.8 Mb. This variation can be attributed partially to the draft nature of the sequence. COG assignments were similar for all genomes. Twenty-two COG classes were annotated in the genomes of JW1 T , JW3, P. byunsanensis JCM 12483 T , and P. arabiensis JCM 17292 T , and 23 COG classes (with an extra Cytoskeleton class) were detected in the genomes of P. shioyasakiensis JCM 18891 T and P. gelatinilytica NH153 T (S1 Table). Furthermore, 13 and 6 unique orthologous clusters were found in the genomes of JW1 T and JW3, respectively ( Table 3). The two strains shared 640 unique orthologous clusters. Genome analysis of JW1 T , JW3, and the reference strains indicated the presence of genes encoding clustered regularly interspaced short palindromic repeats (CRISPRs ; Table 3). CRISPRs, in association with CRISPR-associated (Cas) proteins, make up the immune system that confers resistance to foreign genetic elements, such as plasmids and phages [49]. The CRISPRs/Cas system is being used for gene editing. Colonies of JW1 T and JW3 were violet and the two strains produced violacein. The biosynthesis of violacein begins with L -tryptophan and is successively catalyzed by enzymes VioA, B, E, D, and C, all of which are encoded by the vioABCDE operon [50]. Strains JW1 T , JW3, and P. byunsanensis JCM 12483 T produced violacein and possessed vioABCDE operon, while P. arabiensis JCM 17292 T , P. shioyasakiensis JCM 18891 T and P. gelatinilytica NH153 T did not (Table 3 and S3 Fig).
The genomes of strains JW1 T and JW3 were annotated and analyzed to identify the major metabolic pathways of carbon, nitrogen, sulfur, and phosphorus based on key genes. JW1 T and JW3 can use organic carbon sources (refer to the species description). They harbor key genes of the Entener-Doudoroff pathway, glycolysis pathway, pentose phosphate pathway, and tricarboxylic acid cycle. The genomes of JW1 T and JW3 possess an ammonium transporter gene, but they lack genes involved in nitrate reduction, nitrite reduction, nitrogen fixation, nitrification, or anaerobic ammonium oxidation. Thus, both strains can utilize only reduced nitrogen. Genes encoding urease and urea transporter were not detected, suggesting that JW1 T and JW3 are incapable of utilizing urea as a C or N source. Both genomes possess a variety of sulfate permease genes involved in assimilatory SO 4 2+ reduction. Sulfate can be reduced to sulfide and is subsequently incorporated into amino acids (e.g. cysteine). The genomes of JW1 T and JW3 harbor genes encoding low-affinity inorganic phosphate transporter and sodium-dependent phosphate transporter. The presence of alkaline phosphatase genes in both genomes indicates that JW1 T and JW3 are capable of utilizing both inorganic and organic forms of phosphorus.

Conclusion
Strains JW1 T and JW3 possess some properties, particularly chemotaxonomic characteristics, that species of the genus Pseudoalteromonas all share. On the other hand, JW1 T and JW3 could be distinguished from the type strains of their closely related species on the basis of phenotypic differences (e.g., color, NaCl, pH and temperature ranges and optima, arginine dihydrolase, citrate utilization, nitrate reduction, tryptophan deaminase, urease, Voges-Proskauer reaction, carbohydrate utilization, and acid production, Table 1). In addition, strains JW1 T and JW3 produce violacein, a potential pharmaceutical agent. On the basis of phylogenetic, genome, and chemotaxonomic data, as well as phenotypic characteristics, strains JW1 T and JW3 represent a novel species of the genus Pseudoalteromonas, for which the name Pseudoalteromonas amylolytica sp. nov. is proposed.
Supporting information S1 Table. COG annotations