Formosa algae gen. nov., sp. nov., a novel member of the family Flavobacteriaceae

According to the phylogenetic arrangement of taxa presented in the second edition of Bergey's Manual of Systematic Bacteriology, the family Flavobacteriaceae Reichenbach 1992 emend. Bernardet et al. 1996, 2002 is placed in the class Flavobacteria, one of three classes of the phylum Bacteroidetes (Bernardet et al., 1996, 2002; Garrity & Holt, 2001). Currently, the family Flavobacteriaceae comprises more than 20 genera (Flavobacterium, Aequorivita, Arenibacter, Bergeyella, Capnocytophaga, Cellulophaga, Chryseobacterium, Coenonia, Croceibacter, Empedobacter, Gelidibacter, Muricauda, Myroides, Ornithobacterium, Polaribacter, Psychroflexus, Psychroserpens, Riemerella, Salegentibacter, Tenacibaculum, Vitellibacter, Weeksella and Zobellia) of ubiquitous bacteria from an array of environments that often overlap phenotypically with members of other phyla (Gherna & Woese, 1992; Holmes, 1993; Segers et al., 1993; Bernardet et al., 1996; Nakagawa et al., 1997). It is believed that members of the Flavobacteria play an important role in the degradation of complex polysaccharides and other biomacromolecules (Bernardet et al., 2002).

Polyphasic characterization of four bacterial isolates representing one of the numerically dominant groups of the brown alga-degrading enrichment community was carried out. On the basis of the results obtained and presented in this paper, it is proposed that these organisms represent a novel species in a new genus, Formosa algae gen. nov., sp. nov., in the family Flavobacteriaceae.

Bacterial cultures, sampling and isolation procedure.
Brown alga (Fucus evanescens) samples were collected by scuba divers in mid-summer (July 1999) in Kraternaya Bay, Kuril Islands, Pacific Ocean, during the 23rd scientific expedition of the R/V Akademician Oparin. The set-up of the enrichment experiments and bacterial isolation were carried out as described elsewhere (Ivanova et al., 2002a, b) with the only modification being addition of protein inhibitor for endo-(1→3)-β-D-glucanases (Yermakova et al., 2002) to the enrichment culture (E. Ivanova, unpublished results). Cultures were maintained on marine agar 2216 (Difco) plates and medium B [0·2 % (w/v) Bacto peptone (Difco), 0·2 % (w/v) casein hydrolysate (Merck), 0·2 % (w/v) Bacto yeast extract (Difco), 0·1 % (w/v) glucose, 0·02 % (w/v) KH₂PO₄, 0·005 % (w/v) MgSO₄.7H₂O, 1·5 % (w/v) Bacto agar (Difco), 50 % (v/v) natural sea water and 50 % (v/v) distilled water at pH 7·57·8] and preserved in marine broth supplemented with 30 % glycerol at 80 °C.

Phenotypic characterization.
Unless otherwise indicated, the phenotypic properties used for characterization of Flavobacterium-related species were tested following established procedures (McMeekin et al., 1971; Smibert & Krieg, 1994; Ivanova et al., 1996; Bernardet et al., 2002). To test for spreading growth and gliding motility, strains were grown on medium B with a reduced peptone content (0·2 g l¹). Gliding motility was verified using phase-contrast microscopy (Nikon) of hanging drop preparations. The bathochromic shift test with 20 % (w/v) KOH was performed to detect flexirubin pigmentation (Fautz & Reichenbach, 1980).

Starch, casein and gelatin hydrolysis was tested by the methods of Smibert & Krieg (1994). Degradation of macromolecules was tested using medium B. Chitin (1 %, w/v), elastin (0·1 %, w/v) and alginate (sodium salt; 0·1 %, w/v) hydrolysis was determined by development of clear zones around colonies. Cellulose hydrolysis was tested both by using cellulose overlay plates (1 % carboxymethylcellulose) and by examining strips of filter paper in liquid cell cultures for dissolution (Smibert & Krieg, 1994). Oxidation of 95 carbon sources was tested using Biolog GN microplates (Rüger & Krambeck, 1994; Ivanova et al., 1998).

Growth at different temperatures (445 °C), NaCl concentrations (015 %) and pH (4·512·0, adjusted using HCl and NaOH) was measured by OD₆₆₀ after 24 h incubation in marine broth. Cultures were incubated on a rotary shaker at 160 r.p.m. for 24 h at 28 °C (445 °C for temperature experiments).

Susceptibility to antibiotics was tested by the disc-diffusion plate method using medium B agar and discs impregnated with following antibiotics: kanamycin (30 µg), ampicillin (10 µg), benzylpenicillin (10 µg), streptomycin (30 µg), gentamicin (10 µg), lincomycin (15 µg), neomycin (30 µg), polymyxin B (25 µg) and tetracycline (30 µg).

Atomic force microscopic (AFM) imaging.
AFM characterization was carried out on a TopoMetrix Explorer (ThermoMicroscopes) in non-contact mode using either a 2 µm liquid scanner (0·8 µm z range) or a 100 µm liquid scanner (10 µm z range). Silicon cantilevers with a spring constant of 42 N m¹ and resonant frequency of 320 kHz were used and all imaging was performed in ethanol. All samples were prepared on freshly cleaved mica.

Pigment characterization.
The strains were grown on medium B for 2 days at 28 °C. Pigments were extracted using methanol. Absorption spectra were determined between 250 and 700 nm using a Specord M40 spectrophotometer.

Polar lipids.
For lipid analyses followed by fatty acid methyl ester generation, cells were grown as described earlier (Svetashev et al., 1995). Lipids were extracted according to Bligh & Dyer (1953). Two-dimensional micro-TLC of polar lipids was carried out using the method of Svetashev & Vaskovsky (1972), with chloroform/methanol/benzene/28 % NH₄OH (65 : 30 : 10 : 6, by vol.) for the first dimension and chloroform/methanol/benzene/acetone/acetic acid/water (70 : 30 : 10 : 5 : 4 : 1, by vol.) for the second dimension (Vaskovsky & Terekhova, 1979). Non-specific detection of lipids on the TLC was performed with a 10 % solution of H₂SO₄ in methanol at 180 °C. Specific reagents used for phospholipids were as described by Vaskovsky et al. (1975); 2 % (w/v) ninhydrin in acetone was used for amino-containing lipids. The phosphorus content in phospholipid spots was determined using micromethods described by Vaskovsky et al. (1975).

Fatty acid methyl ester analysis.
Analyses of fatty acid methyl esters were carried out on a Shimadzu GC-14A GC with an FID using both a non-polar SPB-5 fused-silica column (30 mx0·25 mm i.d.) at 210 °C and a polar Supelcowax-10 fused-silica column (30 mx0·25 mm i.d.) at 200 °C. The FID was operated at 240 °C. Helium was used as the carrier gas (Carreau & Dubacq, 1978; Christie, 1988). Catalytic hydrogenation of fatty acid methyl esters was carried out as described by Appelquist (1972).

Genetic analysis.
DNA was isolated following the method of Marmur (1961) and the DNA G+C content was determined by the thermal denaturation method of Marmur & Doty (1962). DNADNA hybridization was performed spectrophotometrically and initial renaturation rates were recorded as described by De Ley et al. (1970).

Phylogenetic analysis.
The 16S rRNA gene was amplified and sequenced by MIDI Labs (Newark, DE, USA). Briefly, primers used for amplification corresponded to Escherichia coli positions 5 and 1540. Amplification products were purified using Microcon 100 molecular mass cut-off membranes (Millipore) and checked for quality and quantity on an agarose gel. Cycle sequencing of the 16S rRNA amplification products was carried out using AmpliTaq ES DNA polymerase and Rhodamine dye terminators. Samples were electrophoresed on an ABI Prism 377 DNA sequencer.

Related sequences were selected according to previous phylogenetic analyses of a database of 62 000 previously aligned bacterial 16S rRNA gene sequences and BLAST searches against the latest release of the EBI (European Bioinformatic Institute). In a preliminary analysis, 150 sequences were selected according to the result of a BLAST query. An initial tree was built that allowed 27 closely related sequences to be selected from reference strains when available. When several sequences were available for a type species, the sequence with the fewest ambiguities was selected. Phylogenetic trees were constructed using three different methods (bioNJ, maximum-likelihood and maximum-parsimony). For the bioNJ analysis, distance matrices were calculated using Kimura's two-parameter correction. bioNJ analysis was performed according to Gascuel (1997). Maximum-likelihood and maximum-parsimony were from PHYLIP (Felsenstein, 1985, 1993). Phylogenetic trees were drawn using NJPLOT (Perrière & Gouy, 1996). Because of the genetic distance between the 27 strains, homoplasy was detected between aligned sequences. These domains were not used for phylogenetic analysis, and the tree shown was obtained using positions 91208, 212827 and 8351444 of the KMM 3553^T sequence. The topology shown is that of the bootstrap tree, as it has been demonstrated that this topology is often better than that of a simple neighbour-joining or maximum-parsimony analysis (Berry & Gascuel, 1996). There is no scale bar in Fig. 2 as this would be meaningless because the distances are corrected (see above) and this is a bootstrap tree.

(46K): Fig. 2. Phylogenetic position of Formosa algae KMM 3553T according to 16S rRNA gene sequence analysis.

Phenotypic characteristics of the isolates
The novel organisms, isolated from a microbial community formed during degradation of the thallus of a brown alga (Fucus evanescens), were subjected to detailed characterization. The isolates were Gram-negative and aerobic, although anaerobic growth was observed due to fermentation of D-glucose by anaerobic respiration of nitrate. Cells were gliding rods, slightly pointed, 0·40·9 µm in diameter and 0·81·8 µm long, according to AFM investigations (Fig. 1ac). AFM imaging enables high-resolution imaging of the native bacterial cell surface in three dimensions, without staining or shadowing, by mechanically scanning a tip mounted on a flexible cantilever over the sample surface, and allows accurate investigation of cellular morphology. Gliding was probably facilitated by production of slime, as shown by AFM (Fig. 1b, c). All strains showed remarkable resistance to the 10 antibiotics tested. Other phenotypic characteristics are given in Tables 1 and 3 and the species description.

(27K): Fig. 1. AFM of cells of Formosa algae KMM 3553T. (a) Deflection image of cells deposited on mica. (b) Deflection image and line profile of a cell freshly deposited on mica (30 min); the two curves correspond to the longitudinal and transverse profiles of the cell. Note the occurrence of noise along the edges of the bacterium. Closer investigation revealed a fine layer of conditioning film (extracellular polymeric substances), secreted by the bacterium to promote gliding and/or attachment. (c) Deflection image and line profile of a cell deposited on mica and visualized after 2 days. (d) Deflection image and line profile of a deflated cell deposited on mica and visualized after 2 days.

Table 1. Major phenotypic characteristics of Formosa algae gen. nov., sp. nov. Values for other strains are the number positive/number tested. W, Weakly positive.

Polar lipid and fatty acid composition
Phosphatidylethanolamine (PE) was the only phospholipid identified in this group of isolates. Amino-containing lipids that were detected on TLC plates were ninhydrin-positive. The predominant cellular fatty acids were mainly branched-chain saturated and unsaturated, namely i15 : 0, ai15 : 0, 15 : 0, i15 : 1 and 15 : 1ω6 fatty acids (Table 2). In spite of the intraspecific heterogeneity of the fatty acid content of micro-organisms belonging to the family, high levels of i15 : 0 and i17 : 0 3-OH are often observed (Holmes et al., 1984; Holmes, 1993; Bernardet et al., 1996, 2002). The novel isolates contained a high proportion of branched-chain saturated and unsaturated cellular fatty acids, which is a characteristic feature of the family.

Table 2. Cellular fatty acid composition of Formosa algae gen. nov., sp. nov. Values are percentages of total fatty acids for strain KMM 3553T; ranges for the other three strains are given in parentheses.

DNA base composition and DNA relatedness
The DNA G+C content was in the range 34·034·4 mol% (thermal denaturation method). Levels of DNA relatedness between the isolates were 8998 %. Since the level of DNA similarity of the four strains was greater than 70 %, all of them were assigned to a single species (Wayne et al., 1987).

Phylogeny
To establish the precise phylogenetic affiliation of strain KMM 3553^T, the almost entire 16S rRNA gene sequence was determined. Phylogenetic analysis was carried out including all known taxa from the phylum Bacteroidetes. According to 16S rRNA gene sequence analysis, a consensus of all three methods (bioNJ, maximum-parsimony and maximum-likelihood) showed that KMM 3553^T does not form a robust clade with any recognized species and/or genus (Fig. 2). The level of 16S rDNA sequence similarity to members of the two most closely related genera phylogenetically, Psychroserpens and Gelidibacter, was almost equal, i.e. 94·2 % or less, suggesting that the strains isolated in this study represent a novel genus. The most closely related species, Psychroserpens burtonensis, had 94·2 % sequence similarity and 82 sequence differences, whereas the most closely related sequence, from an uncultured species from the toxic dinoflagellate Alexandrium tamarense (Groben et al., 2000), was 97·2 % similar and had 27 differences. Because the 16S rRNA gene sequence of KMM 3553^T does not form a clade with any recognized species and differs greatly from sequences of closely related species, phylogenetic analysis suggests that strain KMM 3553^T should be classified in a new genus.

Taken together, all of these features allow us to conclude that the isolates bear sufficient resemblance to species and genera included in the emended family Flavobacteriaceae (Bernardet et al., 2002) to allow their inclusion in the family. The novel isolates can be easily differentiated from previously described genera (Table 3). One of two Flavobacterium taxa described recently for marine bacteria related to the [Flexibacter] maritimus rRNA branch, Psychroserpens burtonensis (Bowman et al., 1997), has characteristic cell morphology, non-saccharolytic metabolism, requires Na⁺ ions for growth and has a low DNA G+C content. Strains of the second taxon, Gelidibacter algens, have different cell morphology, require Na⁺ ions for growth and have significant levels of characteristic ai15 : 1ω10c and ai17 : 1ω7c fatty acids. Consequently, it is proposed that the four strains isolated from a brown alga be classified as representatives of a new genus, as Formosa algae gen. nov., sp. nov.

Table 3. Differential characteristics of Formosa algae gen. nov., sp. nov. and other phylogenetically related halophilic genera in the family Flavobacteriaceae Genera/species: 1, Formosa; 2, Aequorivita; 3, Cellulophaga; 4, [Cytophaga] latercula; 5, [Cytophaga] marinoflava; 6, [Flexibacter] tractuosus; 7, Psychroflexus; 8, Salegentibacter; 9, Gelidibacter; 10, Polaribacter; 11, Psychroserpens; 12, Tenacibaculum; 13, Zobellia. Data from Bernardet et al. (1996), Bowman et al. (1997, 1998), Bowman & Nichols (2002), Barbeyron et al. (2001), Vandamme et al. (1999), McGuire et al. (1987), Suzuki et al. (2001), Macián et al. (2002) and this study. , Negative; +, positive; V, variable; ND, not determined; ±, previously reported results differ.

Description of Formosa gen. nov.
Formosa (For.mo'sa. L. fem. n. Formosa beautiful, finely formed).

Rod-shaped cells with slightly irregular sides and pointed ends, approx. 0·81·8x0·40·9 µm. Gram-negative. Does not form endospores or resting stages. Does not accumulate poly-β-hydroxybutyrate as an intracellular reserve product and does not have an arginine dihydrolase system. Aerobic. Anaerobic growth occurs by fermentation of D-glucose by anaerobic respiration of nitrate. Chemo-organotroph. Cytochrome oxidase-negative, catalase-positive. Phylogenetically, Formosa is a member of the family Flavobacteriaceae, class Flavobacteria, phylum Bacteroidetes. The type species is Formosa algae.

Description of Formosa algae sp. nov.
Formosa algae (al'gae. L. fem. gen. n. algae of an alga, pertaining to the source of isolation, brown algae).

Description is as for the genus plus the following. Exhibits gliding motility. Colonies are circular, 13 mm in diameter, low-convex on solid media containing high nutrient concentrations. Colonies are light yellow in colour. Produces carotenoid pigments with absorbance peaks at 455 and 480 nm. Flexirubin pigments are absent. No growth detected at 4 or 37 °C; optimum growth at 23 °C. Grows at pH 5·010·0; optimum growth at pH 8·08·5. Alkalitolerant. Grows in 06 % NaCl. Does not decompose agar or chitin. Starch and gelatin are hydrolysed weakly. Utilizes Tween 40, D-galactose, gentiobiose, α-D-glucose, mono-succinate, citric acid, D-glucuronic acid, succinamic acid, succinic acid, alaninamide, glycyl L-aspartic acid, hydroxy-L-proline, L-ornithine, L-pyroglutamic acid, urocanic acid, thymidine, 2-aminoethanol and glycerol. Does not utilize α-cyclodextrin, dextrin, glycogen, Tween 80, N-acetyl D-glucosamine, adonitol, D-cellobiose, L-arabinose, D-arabitol, i-erythritol, D-fructose, L-fucose, m-inositol, α-D-lactose, lactulose, maltose, D-mannitol, D-mannose, D-melibiose, methyl β-D-glucoside, D-psicose, D-raffinose, L-rhamnose, D-sorbitol, sucrose, D-trehalose, turanose, xylitol, acetic acid, cis-aconitic acid, formic acid, D-galactonic acid lactone, D-galacturonic acid, D-glucosaminic acid, D-glucuronic acid, α-hydroxybutyric acid, β-hydroxybutyric acid, γ-hydroxybutyric acid, p-hydroxyphenylacetic acid, itaconic acid, α-ketoglutaric acid, α-ketobutyric acid, α-ketovaleric acid, DL-lactic acid, malonic acid, propionic acid, quinic acid, D-saccharic acid, sebacic acid, bromosuccinic acid, glucuronamide, D-alanine, L-alanine, L-alanyl glycine, L-asparagine, L-aspartic acid, L-glutamic acid, L-histidine, L-leucine, L-ornithine, L-phenylalanine, L-proline, D-serine, L-serine, L-threonine, DL-carnitine, γ-aminobutyric acid, inosine, uridine, phenylethylamine, 2,3-butanediol, DL-α-glycerol phosphate, α-D-glucose 1-phosphate or D-glucose 6-phosphate according to the Biolog system. Not susceptible to ampicillin, lincomycin, benzylpenicillin, kanamycin, oleandomycin, tetracycline, neomycin, streptomycin, gentamicin or polymyxin B. PE is the only phospholipid identified. Sphingophospholipids are absent. The predominant cellular fatty acids are odd-numbered: 15 : 0, i15 : 0, ai15 : 0, i15 : 1 and 15 : 1(n-6). DNA G+C content is 34·034·4 mol%.

The type strain is KMM 3553^T (=CIP 107684^T); its DNA G+C content is 34·0 mol%.

The authors are grateful to two anonymous referees and Dr A. M. Lysenko for genetic analysis. This study was partially supported by funds from the Defence Advanced Research Projects Agency (DARPA) grant number N66001-00-18952, the Russian Foundation for Basic Research (RFBR) 02-04-49517 and grant 03-19 from the Ministry for Industry, Science and Technology of Russian Federation.