Rhizobium rosettiformans sp. nov., isolated from a hexachlorocyclohexane dump site, and reclassification of Blastobacter aggregatus Hirsch and Müller 1986 as Rhizobium aggregatum comb. nov.

In the present study, the objective was to classify a bacterial strain, W3^T, which was isolated from a water sample drawn from a hand pump situated in a hexachlorocyclohexane (HCH) dumpsite. The water sample was plated on Luria–Bertani (LB) agar. A cream-coloured colony that appeared within 48 h of incubation at 28 °C was picked and purified by repeated streaking on LB agar. Phylogenetic trees reconstructed using 16S rRNA, atpD and recA gene sequences showed a close relationship of strain W3^T with Blastobacter aggregatus IFAM 1003^T in the Rhizobium cluster. Thus, the taxonomic position of B. aggregatus was also reassessed.

Species of the genus Blastobacter Zavarzin 1961, belonging to the order Rhizobiales and family Bradyrhizobiaceae, are rod-shaped, commonly aquatic (especially freshwater), budding, rosette-forming and do not fix nitrogen (Rothe et al., 1987; Sly & Hugenholtz, 2005; Balachandar et al., 2007). The genus is in fact polyphyletic, with a high degree of heterogeneity with respect to its phenotypic (Trotsenko et al., 1989; Sittig & Hirsch, 1992), chemotaxonomic and molecular characteristics (Rothe et al., 1987; Willems & Collins, 1992; Hugenholtz et al., 1994). Two of the five species of the genus Blastobacter with validly published names have been transferred subsequently to other genera: Blastobacter natatorius to Blastomonas (Sly & Cahill, 1997) and Blastobacter denitrificans to Bradyrhizobium (van Berkum et al., 2006), although the new combination ‘Bradyrhizobium denitrificans’ has not been validly published. B. aggregatus IFAM 1003^T showed a close relationship with Rhizobium species on the basis of 16S rRNA gene cataloguing (Rothe et al., 1987), and the need to revise the taxonomy of genus Blastobacter has also been suggested (Sly & Cahill, 1997; Doronina & Trotsenko, 2003). On the other hand, species of the genus Rhizobium Frank 1889, belonging to the class Alphaproteobacteria, order Rhizobiales, family Rhizobiaceae, are rod-shaped and include free-living species as well as those that form nitrogen-fixing symbioses with roots of leguminous plants. These symbioses are of considerable importance in agriculture and reduce the need for chemical fertilizers. Rhizobia are commonly found in soil, but have also been isolated from water (Zurdo-Piñeiro et al., 2004), non-legume plants (Peng et al., 2008) and bioreactors (Hunter et al., 2007; Quan et al., 2005). Fatty acids common to all species of the genus Rhizobium include 16 : 0, 18 : 0, summed feature 2 (one or more of 12 : 0 aldehyde, iso-16 : 1 I and 14 : 0 3-OH), summed feature 3 (16 : 1ω7c and/or 16 : 1ω6c) and summed feature 8 (18 : 1ω7c and/or 18 : 1ω6c) (Tighe et al., 2000).

The reference strains used in this study were obtained from the Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH (DSMZ) (Blastobacter aggregatus DSM 1111^T, Rhizobium daejeonense DSM 17795^T and Rhizobium radiobacter DSM 30147^T) and the Laboratorium voor Microbiologie, Universiteit Gent (Rhizobium selenitireducens LMG 24075^T).

Gram staining was performed using a Gram stain kit (HiMedia). Colony morphology was examined on LB agar plates under a light microscope (Myotic 200) after incubation at 28 °C for 48 h. Cells in the early exponential growth phase were used for detecting flagella using transmission electron microscopy as described by Kumar et al. (2008).

Genomic DNA was isolated as described by Sambrook et al. (1989). A nearly complete (1406 bp) 16S rRNA gene sequence of strain W3^T was obtained according to the method described by Kumar et al. (2008). For phylogenetic analyses, 16S rRNA gene sequences of strains of species with validly published names that are closely related to strain W3^T and B. aggregatus IFAM 1003^T (Fig. 1) were retrieved from the GenBank database at NCBI (; Altschul et al., 1997). The selected sequences were aligned using the clustal_x software, version 1.81b (Thompson et al., 1997), and terminal nucleotides not common to all sequences were trimmed. The evolutionary distance matrix was calculated using the distance model of Jukes & Cantor (1969) within the treeconw software package, version 1.3b (Van de Peer & De Wachter, 1994). A phylogenetic tree (Fig. 1) was reconstructed using the neighbour-joining method of Saitou & Nei (1987) and the resultant tree topology was evaluated using bootstrap analysis of 1000 replications. Similar tree topologies were obtained using Kimura’s two-parameter model in treeconw and a DNA parsimony method using phylip, version 3.5 (Felsenstein, 1993) (not shown). The results indicated that strain W3^T and B. aggregatus IFAM 1003^T grouped with representatives of the genus Rhizobium, with clear separation of B. aggregatus IFAM 1003^T from other species of the genus Blastobacter with validly published names. Strain W3^T shared 97.8 % 16S rRNA gene sequence similarity with R. selenitireducens B1^T, followed by R. daejeonense L61^T (97.7 %), R. radiobacter ATCC 19358^T (97.5 %) and B. aggregatus IFAM 1003^T (97.2 %). Likewise, B. aggregatus IFAM 1003^T shared 95.6 % 16S rRNA gene sequence similarity with R. daejeonense L61^T and Blastobacter capsulatus IFAM 1005^T, followed by R. selenitireducens B1^T (95.3 %) and R. radiobacter ATCC 19358^T (95.0 %).

To clarify further the taxonomic status of strain W3^T and B. aggregatus DSM 1111^T, two housekeeping genes, atpD and recA, were amplified and sequenced as described by Gaunt et al. (2001). Partial atpD and recA gene sequences of strain W3^T, B. aggregatus DSM 1111^T, R. selenitireducens LMG 24075^T, R. daejeonense DSM 17795^T and R. radiobacter DSM 30147^T were obtained. Phylogenetic trees based on atpD (Fig. 2) and recA (Fig. 3) gene sequences were reconstructed by using the neighbour-joining method of Saitou & Nei (1987). Strain W3^T and B. aggregatus DSM 1111^T exhibited 88.6–94.5 % atpD gene sequence similarity and 84.5–89.2 % recA gene sequence similarity to the reference Rhizobium strains used in this study.

To sum up, phylogenetic analyses based on ATP synthase β (atpD) and DNA recombinase A (recA) gene sequences corroborated the 16S rRNA gene sequence-based phylogeny, thereby providing strong evidence that strain W3^T and B. aggregatus DSM 1111^T belong to the genus Rhizobium.

Symbiotic properties are important characteristics of rhizobia; hence the presence of symbiotic genes in strain W3^T and B. aggregatus DSM 1111^T was examined by amplifying nifH (Poly et al., 2001), nodA (Haukka et al., 1998), nodC (Laguerre et al., 2001) and nodD (Rivas et al., 2002). A partial nifH gene sequence (344 bp) from strain W3^T showed highest sequence similarity with that of Azohydromonas lata IAM 12599^T (93.8 %) followed by Sinorhizobium sp. TJ170 (91.8 %) and Azohydromonas australica IAM 12664^T (91.0 %). However, no nifH gene could be detected in B. aggregatus DSM 1111^T. None of the three nod genes could be detected in strain W3^T or B. aggregatus DSM 1111^T. Fatty acid methyl ester analysis of strain W3^T, B. aggregatus DSM 1111^T, R. selenitireducens LMG 24075^T, R. daejeonense DSM 17795^T and R. radiobacter DSM 30147^T was performed at Disha Institute of Biotechnology Pvt. Ltd (Ahmedabad, India) using the method described by Miller (1982) and Kuykendall et al. (1988). The major fatty acid in strain W3^T and B. aggregatus DSM 1111^T was summed feature 8 (18 : 1ω7c and/or 18 : 1ω6c). Overall, the fatty acid profiles of strain W3^T and B. aggregatus DSM 1111^T (Supplementary Table S1, available in IJSEM Online) were in accordance with those of members of the genus Rhizobium (Hunter et al., 2007; Tighe et al., 2000), further supporting the placement of strain W3^T and B. aggregatus DSM 1111^T in the genus Rhizobium.

DNA–DNA hybridization is considered to be a standard method to affirm species delineation (Wayne et al., 1987). Hence, DNA–DNA hybridization studies were conducted between strain W3^T, B. aggregatus DSM 1111^T, R. selenitireducens LMG 24075^T, R. daejeonense DSM 17795^T and R. radiobacter DSM 30147^T as described by Tourova & Antonov (1987) and Bala et al. (2004). DNA (10 µg) from each strain was transferred onto a positively charged nylon membrane (Hybond-N; Amersham) using a dot-blot apparatus (Bio-Rad). The membrane was air-dried and cross-linked and the DNA probe for each strain was labelled with [α-³²P]ATP (BRIT) using a nick-translation kit (Amersham Pharmacia). Hybridization was performed overnight at 65 °C. After hybridization, the filter was washed with SSC and SDS to remove unbound probe. The amount of probe bound to the DNA was estimated using a β-scintillation counter (Perkin Elmer) and the hybridization values obtained were expressed as percentages of the amount of probe bound relative to the homologous reaction. Since low 16S rRNA gene sequence similarities (<97 %) were found between B. aggregatus IFAM 1003^T and type strains of other species of the genera Rhizobium and Blastobacter with validly published names, DNA–DNA hybridization studies were not performed between B. aggregatus DSM 1111^T and closely related type strains. The DNA–DNA relatedness of strain W3^T with R. selenitireducens LMG 24075^T, R. daejeonense DSM 17795^T, R. radiobacter DSM 30147^T and B. aggregatus DSM 1111^T was 42, 34, 30 and 34 %, respectively. These values are clearly below the recommended threshold value for delineation of genospecies (Stackebrandt & Goebel, 1994), indicating that strain W3^T represents a novel species of the genus Rhizobium.

The DNA G+C contents of strain W3^T, R. selenitireducens LMG 24075^T, R. daejeonense DSM 17795^T, R. radiobacter DSM 30147^T and B. aggregatus DSM 1111^T were estimated using the procedure of Gonzalez & Saiz-Jimenez (2002) (Table 1). The DNA G+C contents of strain W3^T and B. aggregatus DSM 1111^T were found to be 62.3 and 62.7 mol%, respectively, well within the range reported for other Rhizobium species (Jordan, 1984).

All phenotypic tests for strain W3^T and B. aggregatus DSM 1111^T were performed at 28 °C unless stated otherwise. The ability of the strains to grow at 4–45 °C, pH 4–12 and 0–5 % (w/v) NaCl was determined as described by Arden Jones et al. (1979). Strain W3^T showed rich growth on LB agar and tryptic soy agar (TSA) but slow growth was seen on nutrient agar (NA), whereas B. aggregatus DSM 1111^T showed good growth on LB agar but slow growth on TSA and NA. Motility of the two cultures was checked on motility agar (Farmer, 1999). Like B. aggregatus DSM 1111^T, strain W3^T was found to form rosettes (Supplementary Fig. S1). The two strains were tested for oxidase activity using Oxidase discs (HiMedia). Catalase activity was determined by bubble production in a 3 % (v/v) hydrogen peroxide solution (Merck) as described by McCarthy & Cross (1984). The ability of the strains to produce melanin was checked as described by Cubo et al. (1988). Urease activity was tested using the method of Christensen (1946). Utilization of different carbohydrates was detected by using the HiCarbohydrate kit (HiMedia). Tests for indole production, nitrate reduction and the Voges–Proskauer reaction were performed as described by Smibert & Krieg (1994) and Prakash et al. (2007). Hydrolysis of gelatin, starch and aesculin was assessed as described by Cowan & Steel (1965). Citrate utilization was tested on Simmons’ citrate agar (Simmons, 1926). Cellulase activity was checked on CEA medium (0.7 % CM-cellulose, 0.3 % yeast extract, 2.5 % agar) after 5 days of incubation at 28 °C (García-Fraile et al., 2007). The ability of the strains to reduce methylene blue was tested as described by Gao et al. (1994). The ability of the strains to produce H₂S was tested using TSI medium (l⁻¹: 3 g beef extract, 3 g yeast extract, 20 g meat peptic digest, 1 g glucose, 10 g lactose, 10 g sucrose, 0.3 g ferric citrate, 5 g NaCl, 0.3 g sodium sulfate, 12 g agar and 0.025 g phenol red; pH 7.4). Differential results of biochemical tests with the closest relatives of strain W3^T and B. aggregatus DSM 1111^T are given in Table 1.

While strain W3^T was found to be positive for utilization of citrate, reduction of nitrate and methylene blue, B. aggregatus DSM 1111^T was negative for these tests. Both strains were found to be positive for hydrolysis of aesculin and activities of catalase, oxidase and urease. Based on the results obtained, it is proposed that strain W3^T represents a novel species of the genus Rhizobium, for which the name Rhizobium rosettiformans sp. nov. is proposed. It is also proposed that Blastobacter aggregatus be transferred to the genus Rhizobium as Rhizobium aggregatum comb. nov.