Marinobacter salarius sp. nov. and Marinobacter similis sp. nov., Isolated from Sea Water

Two non-pigmented, motile, Gram-negative marine bacteria designated R9SW1T and A3d10T were isolated from sea water samples collected from Chazhma Bay, Gulf of Peter the Great, Sea of Japan, Pacific Ocean, Russia and St. Kilda Beach, Port Phillip Bay, the Tasman Sea, Pacific Ocean, respectively. Both organisms were found to grow between 4°C and 40°C, between pH 6 to 9, and are moderately halophilic, tolerating up to 20% (w/v) NaCl. Both strains were found to be able to degrade Tween 40 and 80, but only strain R9SW1T was found to be able to degrade starch. The major fatty acids were characteristic for the genus Marinobacter including C16:0, C16:1 ω7c, C18:1 ω9c and C18:1 ω7c. The G+C content of the DNA for strains R9SW1T and A3d10T were determined to be 57.1 mol% and 57.6 mol%, respectively. The two new strains share 97.6% of their 16S rRNA gene sequences, with 82.3% similarity in the average nucleotide identity (ANI), 19.8% similarity in the in silico genome-to-genome distance (GGD), 68.1% similarity in the average amino acid identity (AAI) of all conserved protein-coding genes, and 31 of the Karlin's genomic signature dissimilarity. A phylogenetic analysis showed that R9SW1T clusters with M. algicola DG893T sharing 99.40%, and A3d10T clusters with M. sediminum R65T sharing 99.53% of 16S rRNA gene sequence similarities. The results of the genomic and polyphasic taxonomic study, including genomic, genetic, phenotypic, chemotaxonomic and phylogenetic analyses based on the 16S rRNA, gyrB and rpoD gene sequence similarities, the analysis of the protein profiles generated using MALDI-TOF mass spectrometry, and DNA-DNA relatedness data, indicated that strains R9SW1T and A3d10T represent two novel species of the genus Marinobacter. The names Marinobacter salarius sp. nov., with the type strain R9SW1T ( =  LMG 27497T  =  JCM 19399T  =  CIP 110588T  =  KMM 7502T) and Marinobacter similis sp. nov., with the type strain A3d10T ( =  JCM 19398T  =  CIP 110589T  =  KMM 7501T), are proposed.


Introduction
The genus Marinobacter (family Alteromonadaceae, order Alteromonadales, class Gammaproteobacteria) was created by Gauthier et al. for a hydrocarbon degrading bacterium. At the time of writing, the genus comprises 33 validly described species, http://www.bacterio.net/marinobacter.html [1], which accommodates Gram-negative, chemoheterotrophic and halophilic, rodshaped bacteria [2,3]. The important role played by Marinobacter spp. in metabolizing hydrocarbons has long been noted, with M. hydrocarbonoclasticus [2], M. aquaeolei [4,5], M. maritimus [6], and M. algicola [7] having been characterized as being able to utilise aromatic and aliphatic hydrocarbons as their sole carbon and energy sources. It was also shown that bacteria of the genus Marinobacter are one of the dominant bacterial community groups constantly recovered from hydrocarbon polluted sites [8][9][10]. For example, it was recently demonstrated that M. vinifirmus was able to effectively degrade toluene, benzene, ethylbenzene, and p-xylene [11].
The objectives of this study were to classify two newly isolated marine bacteria; strain R9SW1 T , which was derived from a water sample collected from Chazhma Bay (Gulf of Peter the Great, Sea of Japan, Pacific Ocean) during taxonomic studies of microbial communities developed in sea water contaminated by radionuclides [12]; and strain A3d10 T , which was isolated from Port Philip Bay (the Tasman Sea, Pacific Ocean) during the course of polymer biodegradation studies [13]. The comparative taxonomic investigations of these bacteria, together with their close relatives, revealed their distinct taxonomic standing. This suggests that strain R9SW1 T and strain A3d10 T represent two novel species of the genus Marinobacter.

Materials and Methods
Isolation procedures, bacterial strains, and growth conditions Strain R9SW1 T was isolated from a sea water sample collected from Chazhma Bay in the Sea of Japan, Pacific Ocean, in 2000. Water sample collection was within the research program funded by the Federal Agency for Science of the Ministry of Education and Science of the Russian Federation, grant 2-2. 16 and by the Russian Foundation for Basic Research and grant 'Molecular and Cell Biology' from the Presidium of the Russian Academy of Sciences, grant 02-04-48211". The specific location of the studies (GPS coordinates) was 42u539380 N 132u229020 E. The permit issued by the Department of Marine Expeditions, Ministry of Education and Science of the Russian Federation. Strain A3d10 T was isolated from a sea water sample collected one metre below the water surface in Port Philip Bay, the Tasman Sea, Pacific Ocean, in 2008. Sea water collected from St Kilda Beach which is a publicly accessible beach area in Melbourne, not part of any protected area of land or sea. Furthermore, the field studies did not involve endangered or protected species. The specific location of the studies (GPS coordinates) was 37u519500S 144u589550E. The sample handling and isolation procedures used were as previously described [12,13]. Samples were plated on marine agar 2216 (BD, USA) and incubated aerobically at approximately 22-25uC for 5, 7 or 10 days. The isolation and purification procedure has been described elsewhere [14,15]. Ten  16S rDNA, gyrB, rpoD sequencing and phylogenetic analysis Genomic DNAs were isolated using a Wizard Genomic DNA Purification Kit (Promega, USA) according to the manufacturer's specifications. The 16S rRNA gene sequences for strains R9SW1 T and A3d10 T were extracted from the whole genome sequences [16] while gyrB and rpoD genes were amplified using primers (see Supporting Information, Table S1 in File S1) that have been previously described [17,18]. The 16S rRNA gene sequences of validly described Marinobacter species were retrieved from GenBank and aligned using the CLUSTAL W program [19]. Evolutionary phylogenetic trees were constructed using the neighbour-joining (NJ) [20], maximum-likelihood (ML) [21] and maximum-parsimony (MP) [22] algorithms. Genetic distances were calculated using Kimura's two-parameter model [23] by using the MEGA 5 software [24]. The GenBank/EMBL/DDBJ accession numbers of 16S rRNA gene, gyrB, rpoD and whole genome sequences were presented as in Table 1.

MALDI-TOF MS analysis
The sample preparation and MALDI-TOF MS analysis was carried out according to the techniques described elsewhere [25]. Briefly, 5 mL of the cultures grown overnight were transferred into microcentrifuge tubes and subjected to ethanol and formic acid protein extraction. One mL aliquots of the supernatant were transferred onto the MALDI target plate and air dried at room temperature, followed by the addition of 1 mL of matrix solution, then air dried. Samples were then subjected to analysis using a Microflex MALDI-TOF mass spectrometer (Bruker Daltonik GmbH, Leipzig, Germany) equipped with a 60 Hz nitrogen laser. Spectra were recorded in the positive linear mode for the mass range of 2,000 to 20,000 Da at the maximum laser frequency. The raw spectra were then analysed using the MALDI Biotyper 3.0 software package (Bruker Daltonik GmbH, Bremen, Germany) under the default settings. Measurements were performed via the automatic mode, without any user intervention.

GC content and DNA-DNA hybridization
The GC content of strains R9SW1 T and A3d10 T was calculated on the basis of their whole genome sequences [16,26], and these  [27]. DNA-DNA hybridization was carried out in duplicate using a 26 saline sodium citrate (SSC) buffer with 5% formamide as described by De Ley et al. [28], with consideration of the modifications described by Huss et. al. (1983) [29], using a model Cary 100 Bio UV/VIS-spectrophotometer equipped with a Peltier-thermostatted 666 multi-cell changer and a temperature controller with an in-situ temperature probe (Varian).

Genome comparison and genomic signatures analyses
Complete genome sequences for only two validly described species of Marinobacter, M. hydrocarbonoclasticus ATCC 49840 T [30] and M. adhaerens HP15 T [31], which have previously been assembled, were used in this study for genomic analysis. The fully sequenced and assembled genomes of both these species were retrieved from GenBank, and compared to those of R9SW1 T and A3d10 T . Genome comparison between strains R9SW1 T , A3d10 T , M. adhaerens HP15 T and M. hydrocarbonoclasticus ATCC 49840 T was carried out using reciprocal BLAST analysis, according to the method described by Goris et al. [32]. A map of the percentage identity between each of M. adhaerens HP15 T , R9SW1 T and A3d10 T to the type species was generated using the BLAST Ring Image Generator (BRIG) software [33]. The in-silico genome-to-genome distance (GGD) between the four strains was also calculated using the genome-to-genome distance calculator 2.0 (GGDC) provided by DSMZ, http://ggdc.dsmz.de [34,35]. The average amino acid identity (AAI) of all conserved proteincoding genes was calculated as previously described [36]. The conserved genes between a pair of genomes were determined by whole-genome pairwise sequence comparison using the BLAST algorithm release 2.2.5 [37] using a minimum cut-off of 40% identity and 70% of the length of the query gene. The difference in genome signature between two individual sequences is expressed in terms of the Karlin's genomic signature dissimilarity (d*), which was calculated by dividing the genomic dinucleotide frequencies with the corresponding mononucleotide content using the equation described by Karlin et al. [38]. Phylogenomic relationship between the four strains were also elucidated using Mauve multiple alignment software (v2.3.1) [39] and ClonalFrame software v1.2 [40], with Alteromonas sp. DE [41] used as an outgroup.
Genotype to phenotype analyses of a few distinctive phenotypes were also carried using the whole genome sequences of strains R9SW1 T , A3d10 T , M. hydrocarbonoclasticus ATCC 49840 T and M. adhaerens HP15 T using the methods as previously described [42].

Physiological and biochemical analysis
Six reference type strains, along with strains R9SW1 T and A3d10 T , were used for the phenotypic and biochemical tests ( Table 2). The cell morphology and motility were determined using scanning electron and light microscopies. Gram stain reaction, catalase (5% H 2 O 2 ) and starch hydrolysis analyses were performed according to the method described by Smibert and Krieg (1994) [43]. Determination of the oxidase activity was performed using Bactident oxidase strips (Merck Millipore, Germany). The capacity of the strains to oxidize and to ferment D -glucose and lactose was carried out according to the method described by Smibert and Krieg (1994) [43], using a modified semi-solid medium containing: 9.4 g L 21 O/F medium (Oxoid, UK), 20 g L 21 Sea Salt (Sigma-Aldrich, USA) and 1% carbohydrate. The strains were incubated at 30uC and the results were obtained after 48 hours. The temperature and pH tolerance ranges were determined via marine agar growth tests subjected to different temperature (4, 10, 15, 20, 25, 30, 37, 40, 45 and 50uC) and pH (4, 6, 7, 8, 9 and 11, adjusting the pH with HCl and NaOH) conditions. The NaCl tolerance was determined using different concentrations of NaCl (0, 0.5, 1, 3, 6, 10, 15, 20 and 25%) in modified salinity agar (SA) containing: 5 g L 21 peptone, 1 g L 21 yeast extract, 0.1 g L 21 ferric citrate, 3.24 g L 21 magnesium sulphate (MgSO 4 ), 0.55 g L 21 dipotassium phosphate (K 2 HPO 4 ), 15 g L 21 agar, and the respective NaCl concentration, each at a pH of 7.660.2. Plates were incubated under optimal temperature conditions and the results were recorded daily for 7 days.
The ability of the strains to oxidise a range of organic substrates was investigated using a 96-well Biolog GN2 microplate (Biolog, USA), in triplicate. Inoculates were prepared by suspending culture that had been grown overnight into 3% (w/v) saline solution, then adjusting the density of the suspension to McFarland standard no. 1, followed by pipetting 150 mL aliquots of the suspension into each well. All the plates were incubated at 30uC and results were manually obtained after 24 h and 48 h. Enzymatic tests were performed using API ZYM test strips (bioMérieux, France) in two individual experiments. Inoculations were prepared by suspending culture that had been grown overnight into 3% (w/v) saline solution and adjusting the density to McFarland standard no. 5. A Microbact 24E Gram-negative identification system (Oxoid, UK) was also used to test other biochemical reactions, namely: lysine and ornithine decarboxylase; H 2 S production; glucose, mannitol and xylose fermentation; hydrolysis of o-nitrophenyl-b-D -galactopyranoside (ONPG); indole production; urea hydrolysis; acetoin production (Voges-Proskaüer reaction); citrate utilisation; production of indolepyruvate; gelatin liquefaction; malonate inhibition; inositol, sorbitol, rhamnose, sucrose, lactose, arabinose, adonitol, raffinose and salicin fermentation; and arginine dihydrolase. All tests were carried out according to the manufacturer's specifications unless otherwise stated.

Fatty acids analysis
Fatty acid (FA) methyl esters were prepared as described elsewhere [44]. The resulting fatty acid methyl esters were analysed using a Shimadzu GC-14A gas chromatograph with a flame ionization detector, using both a nonpolar SPB-5 fused-silica column (30 m60.25 mm i.d.) at 210uC and a polar Supelcowax-10 fused-silica column (30 m60.25 mm i.d.) at 200uC. Table 2. Differential characteristics between strains R9SW1 T , A3d10 T , their close phylogenetic neighbors and type species of the genus Marinobacter.

Results and Discussion
Analysis of the complete 16S rRNA gene sequences of strains R9SW1 T and A3d10 T revealed that both strains are grouped with species of the genus Marinobacter, with the sequence similarity between strains R9SW1 T , A3d10 T and all validly described Marinobacter species being in the range of 93.84-99.40% and 93.91-99.53%, respectively. The two new strains, R9SW1 T and A3d10 T shared 97.6% of their 16S rRNA gene sequences, however, phylogenetic analysis showed that they cluster separately forming two different clusters, one with M. algicola DG893 T and another with M. sediminum R65 T , where both clusters were supported by the bootstrap value of 99% and 100% in both the NJ and ML methods ( Figure 1A and Figure S1 in File S1). The highest 16S rRNA gene sequence similarity between strain R9SW1 T and M. algicola DG893 T was found to be 99.40% (M. algicola DG893 T ), whilst strain A3d10 T displays the highest 16S rRNA gene sequence similarity with M. sediminum R65 T (99.53%).
Due to the high 16S rRNA gene sequence similarity between strains R9SW1 T and M. algicola DG893 T , and between A3d10 T and M. sediminum R65 T , an extended phylogenetic analysis based on gyrB and rpoD genes was carried out. The use of housekeeping genes in phylogenetic analysis can be beneficial, in that it overcomes the possibility of the presence of nucleotide polymorphisms in the 16S rRNA gene [45,46]. Two genes, gyrB and rpoD, were selected, since they have been previously reported to be excellent marker genes and sufficient for the identification and classification of various groups of microorganism [25,[47][48][49]. M.
sediminum LMG 23833 T , M. salsuginis CIP 109893 T , M. algicola LMG 23835 T , M. adhaerens CIP 110141 T , and M. flavimaris CIP 108615 T were selected, as they are phylogenetically close to strains R9SW1 T and A3d10 T according to their 16S rRNA gene sequences. M. hydrocarbonoclasticus SP.17 T was also included as representing the type species of the genus. A phylogenetic analysis of the gyrB and rpoD gene sequence similarities reconfirmed the clustering of strain R9SW1 T with M. algicola LMG 23835 T , and strain A3d10 T with M. sediminum LMG 23833 T , both of which were supported with 100% bootstrap values (Figure 1(B) and (C)). The gyrB and rpoD sequence similarities for strains R9SW1 T , A3d10 T and their phylogenetically related species was also determined and found to be in the range of 77.8-94.3% (R9SW1 T , gyrB), 80.0-93.5% (A3d10 T , gyrB), and 78.6-93.8% (R9SW1 T , rpoD), 78.6-96.2% (A3d10 T , rpoD), respectively ( Table 3). The gene sequence similarity for gyrB and rpoD between the previously described sister species of Marinobacter, i.e., M. adhaerens CIP 110141 T and M. flavimaris CIP 108615 T was found to be 99.0% and 98.4% respectively (Table 3), which is higher than that found for strains R9SW1 T , A3d10 T and their respective closest phylogenetic relatives. The sequence similarities of the gyrB gene of 94.3% and 93.5% for strains R9SW1 T , A3d10 T with their closest relatives were also lower than the previously proposed gyrB sequence similarity cut-off value of 98.95% for genus Amycolatopsis [50] and 98.22% for genus Kribbella [51]. Also, the data reported for the two Vibrio species, V. gigantis LGP 13 T and V. crassostreae LGP 7 T , were 98% for gyrB and 97% for rpoD [52], which again showed higher similarity values than the gyrB and rpoD sequence similarities of strains R9SW1 T , A3d10 T and their closest relatives. The sequence similarities for gyrB and rpoD between strains R9SW1 T and A3d10 T were significantly lower than the values mentioned above, i.e., 81.6% for gyrB and 78.2% for rpoD, suggesting distinct standing of new strains on the species level.
In order to further assess the taxonomic affiliation of the two new bacteria, a comparative analysis of the total protein profiles was performed using MALDI-TOF mass spectrometry (Figure 2). The results are in agreement with the phylogenetic analyses, clearly indicating that strain R9SW1 T is clustering with M. algicola LMG 23835 T , and strain A3d10 T is clustering with M. sediminum LMG 23833 T with a critical distance level below 500. As suggested in the previously reported studies, clustering below the distance level of 500 can be considered as reliable clustering [53,54], which was also in agreement with the recent studies on Alteromonas spp., where the clustering within the distance level of 500 was shown to be able to differentiate the closely related Alteromonas species [25,55]. Hence, the results of this study confirmed the confident clustering of the two new isolates within other species of the genus Marinobacter. Also, the clusters of both strains R9SW1 T and A3d10 T with their nearest neighbour were stable, but exceeded the minimum differences between existing species, e.g., the distance level between species in both clusters were greater than those within a cluster that contained M. gudaonensis CIP 109534 T , M. adhaerens CIP 110141 T , M. salsuginis CIP 109893 T , and M. flavimaris CIP 108615 T ; so does the position of strains R9SW1 T and A3d10 T resulting in different clusters in the MALDI dendrogram, provide evidence of the distinctive standing of two new bacteria.
In order to confirm the separate species standing of these two strains, a DNA-DNA hybridization experiment was conducted. DNA-DNA relatedness between strain R9SW1 T and M. algicola LMG 23835 T was found to be 63.0561.85%, and between strain A3d10 T and M. sediminum LMG 23833 T was found to be 67.6061.3%. Both of these relatedness values are below the 70% cut-off value generally recommended for species differentiation [56]. Recently, information of whole genome sequences have been recommended to be integrated into bacterial systematics [57][58][59]. In this study, whole genome sequences of strains R9SW1 T , A3d10 T , M. adhaerens HP15 T and M. hydrocarbonoclasticus ATCC 49840 T were visually compared using BLAST ( Figure S2 in File S1) and the average nucleotide identity (ANI), genome-togenome distance (GGD), average amino acid identity (AAI), and the Karlin's genomic signature dissimilarity (d*) between the four strains were calculated, the results of which are presented in Table 4. Due to the lack of the availability of the assembled, whole genome sequences for validly named Marinobacter species, genomic signatures between strains R9SW1 T , A3d10 T and validly described Marinobacter species can only be performed using those of M. adhaerens HP15 T [31] and M. hydrocarbonoclasticus ATCC 49840 T [30]. As can be seen from the information presented in Table 4, the ANIs between the four strains were in the range of 82.3-83.3%, which is significantly lower than the suggested threshold range of 95-96% [58,60]; the GGDs were calculated to be in the range of 19.8-20.7% which is lower than the cur-off value of 70% [61]; the AAI and Karlin signature dissimilarity values for the four strains were in the range of 68.1-77.6% and 31-36 respectively, each of which fall outside the range to be consider as same species [42,61]; and thus again indicating that strains R9SW1 T and A3d10 T can be considered as two novel species of the genus Marinobacter. The distinct standing of strains R9SW1 T and A3d10 T can also be confirmed by the phylogenomic relationship analysis using the core proteome of the genomes from the four strains ( Figure 3). The major features of the genomes of strains R9SW1 T and A3d10 T were identified as described elsewhere [16]. Briefly, they are 4,616,532 bp and 3,975,896 bp in size, composed of 99 and 29 contigs, both have 3 rRNAs, and 44 and 46 tRNAs, for strains R9SW1 T and A3d10 T , respectively. The DNA G+C content of strains R9SW1 T and A3d10 T were found to be 57.1 and 57.6 mol%, respectively (Table 2), the values which are consistent with those of the genus Marinobacter.
Both bacteria were found to be Gram-negative, aerobic, motile by means of a single flagellum and rod-shaped with the size of 1.9-3.260.40-0.72 mm for strain R9SW1 T and 1.3-2.160.40-0.45 mm for strain A3d10 T ( Figure S3 in File S1). The catalase and oxidase tests were found to be positive, H 2 S and indole tests were found to be negative. It can be seen that strain R9SW1 T can be clearly differentiated from M. algicola LMG 23835 T by its inability to reduce nitrate and nitrite, its ability to utilise mono-methyl succinate and L -serine, its inability to utilise L -phenylalanine and the absence of lipase (C14); while strain A3d10 T can be clearly differentiated from M. sediminum LMG 23833 T by its inability to reduce nitrite, its ability to utilise glycogen, c-hydroxybutyric acid and L -glutamic acid, and its weak activities for valine arylamidase and cystine arylamidase. The major phenotypic difference between strains R9SW1 T and A3d10 T are nitrate reduction, hydrolysis of starch, fermentation of D -glucose, and their utilisation of dextrin, D -fructose, maltose, acetic acid, propionic acid, succinic acid, L -serine and glycerol. Other phenotypic characteristics which differentiate the two novel strains from each other and their closest phylogenetic neighbours are shown in Table 2, Table S2 in File S1, and in their respective species descriptions. Both strains were found to be sensitive to penicillin G (10 mg), chloramphenicol (30 mg), and ampicillin (10 mg), and resistant to streptomycin (10 mg) and tetracycline (30 mg). The fatty acid composition of strains R9SW1 T and A3d10 T are shown in Table S3 in File S1, where the predominant  fatty acids were identified as being C 16:0 , C 16:1 v7c, C 18:1 v9c and C 18:1 v7c. The genotype to phenotype analyses were also carried out based on the whole genome sequences of the four strains, the results of which are presented in Table 5. It can be seen that of the results of physiological and biochemical tests match when comparing the in silico results, however a few discrepancies are noted. A similar level of deviation previously reported in the case of Vibrio species and it was suggested that expression of certain genes may be restricted by stop codon, repressor genes, regulatory proteins, global regulators, genome coverage or sequencing errors [42].
In summary, the comparative genomic and phylogenetic analysis based on the full-length of 16S rRNA gene sequence similarities, pheno-and chemotaxonomic properties revealed that strains R9SW1 T and A3d10 T can be affiliated to the genus Marinobacter. A further dual-locus sequence analysis based on gyrB and rpoD gene sequence similarities, the comparative analysis of whole cells protein profiles based on MALDI-TOF mass spectrometry analysis, their phenotypic characteristics and their DNA-DNA hybridization values below 70% confirmed that strains R9SW1 T and A3d10 T should be classified as two novel species of the genus Marinobacter for which the name Marinobacter salarius sp. nov. and Marinobacter similis sp. nov. are proposed.
Description of Marinobacter similis sp. nov.

Supporting Information
File S1 Includes Figures S1-S3 and Tables S1-S3. Figure S1. Neighbour-joining phylogenetic tree showing the taxonomic position of strains R9SW1 T and A3d10 T according to their 16S rRNA gene sequences. Figure S2. BLAST genome ring (A) and comparison of all proteins in the genomes in terms of the similar composition of the gene families (B) between strains R9SW1 T , A3d10 T , M. adhaerens HP15 T and M. hydrocarbonoclasticus ATCC 49840 T . Figure S3. Scanning electron micrographs of strains (A) R9SW1 T and (B) A3d10 T . Table S1. Genes and the corresponding primer sequences used for the amplification and sequencing. Table S2. Phenotypic characteristics of strains R9SW1 T , A3d10 T and closely related type strains and type species of the genus Marinobacter. Table S3. Cellular fatty acids composition of strains R9SW1 T , A3d10 T and closely related type strains and type species of the genus Marinobacter. (DOCX)