Rhizobium tibeticum sp. nov., a symbiotic bacterium isolated from Trigonella archiducis-nicolai (Širj.) Vassilcz.

The genus Rhizobium was first described by Frank (1889) to accommodate all symbiotic nitrogen-fixing bacteria associated with legumes. After a series of revisions, some earlier described species have been moved into Bradyrhizobium (Jordan, 1982), Sinorhizobium (de Lajudie et al., 1994) and Mesorhizobium (Jarvis et al., 1997), while many novel species and genera of rhizobia within the Alphaproteobacteria and Betaproteobacteria have been reported (NZ Rhizobia, 2008). The current genus Rhizobium includes about 30 species of fast-growing, acid-producing rhizobia, which can be differentiated from Sinorhizobium species on the basis of 16S rRNA gene sequence-based phylogeny. With the exception of Rhizobium cellulosilyticus (Garcia-Fraile et al., 2007), most of them can induce root nodules on a certain range of legume species.

In a survey of rhizobial resources in Tibet, 12 strains of root-nodule bacteria were isolated with yeast mannitol agar (YMA) medium according to the method of Vincent (1970) from root nodules of Trigonella archiducis-nicolai (Širj.) Vassilcz., an ephemeral legume growing in grasslands at altitudes of 3700–4000 m. Six of them were identified as Sinorhizobium meliloti, and six as Rhizobium strains (Hou et al., 2009). Among the Rhizobium strains, CCBAU 85039^T and CCBAU 85027 were classified as a small group by numerical taxonomy, amplified 16S rRNA gene restriction analysis, amplified 16S–23S intergenic spacer restriction analysis and 16S rRNA gene phylogeny (Hou et al., 2009). To verify the taxonomic position of these two strains, further characterization was performed in the present study.

To clarify the phylogenetic relationships among the two Tibetan strains and defined Rhizobium species, the 16S rRNA gene, atpD, glnII, recA, nifH, nodA and the 16S–23S rRNA intergenic spacer were sequenced. In addition, SDS-PAGE of whole-cell proteins, DNA–DNA hybridization, phenotypic characterization, determination of the cellular fatty acid composition and cross-nodulation tests were performed.

In addition to the sequence from strain CCBAU 85039^T obtained previously (Hou et al., 2009), the 16S rRNA gene of strain CCBAU 85027 was sequenced in this study by using the same methods. The sequence of the 16S rRNA gene of strain CCBAU 85027 was identical to that of CCBAU 85039^T, and they formed a unique phylogenetic subbranch in the genus Rhizobium in a neighbour-joining phylogenetic tree (Fig. 1⇓), in which all Rhizobium species reported to date were included. The unique position of the two novel strains was supported by a phylogenetic tree reconstructed with the maximum-parsimony method (Kämpfer et al., 2003) (not shown).

Using genomic DNA extracted from CCBAU 85039^T by the method of Terefework et al. (2001) as a template and the PCR protocols of Vinuesa et al. (2005), fragments of the genes atpD (ATP synthase subunit beta; 500 bp), recA (recombinase A protein; 500 bp) and glnII (glutamine synthetase II; 660 bp) were amplified with the corresponding primer pairs atpD255F and atpD782R, recA41F and recA640R and glnII12F and glnII689R. These genes have been used previously to evaluate the phylogeny and classification of rhizobia (Gaunt et al., 2001; Turner & Young, 2000). The PCR products were sequenced directly as reported by van Berkum et al. (1996). The acquired sequences and related sequences obtained from GenBank were aligned by using the mega3.1 program (Kumar et al., 2004). In phylogenetic trees constructed by using the neighbour-joining method and bootstrapped with 1000 replicates, strain CCBAU 85039^T formed a distinct lineage in the genus Rhizobium (Supplementary Fig. S1, available in IJSEM Online). The highest sequence similarities were 87.8 % for recA, 92.2 % for atpD and 89.0 % for glnII to Rhizobium leguminosarum ATCC 14482, Rhizobium tropici USDA 9030 and Rhizobium etli CFN 42^T, respectively. These three most closely related species are all able to nodulate common bean (Martínez-Romero et al., 1991; Ramírez-Bahena et al., 2008; Segovia et al., 1993) in both tropical and temperate regions.

Using the method and primers described previously (Tan et al., 2001), the 16S–23S intergenic spacer was also sequenced for CCBAU 85039^T, and the strain formed a lineage distantly related to R. leguminosarum PEVF08 (92.5 %) in the phylogenetic tree (Supplementary Fig. S2). The low similarities between CCBAU 85039^T and the reference strains for other Rhizobium species indicated that this strain represented a species distinct from those currently defined, although the bootstrap values were low for the branch.

Symbiotic genes (nif and nod) may have evolutionary histories that are different from those of housekeeping genes in rhizobia, and they have been used to estimate the symbiotic properties of rhizobia (Kalita et al., 2006; Laranjo et al., 2008; van Berkum et al., 2007). Fragments of the nifH (nitrogenase reductase; about 680 bp) and nodA (N-acyltransferase; about 600 bp) genes were amplified from CCBAU 85039^T and sequenced directly, with primers nifH-1 and nifH-2 (Eardly et al., 1992) and nodA-1 and nodA-2 (Haukka et al., 1998), respectively. In contrast to the phylogeny of the 16S rRNA gene, the nifH and nodC sequences of CCBAU 85039^T were both most similar to those of the Medicago-nodulating Rhizobium mongolense USDA 1844^T, with 93.3 and 91.4 % similarity, respectively.

The DNA G+C contents of strains CCBAU 85027 and CCBAU 85039^T were 59.5 and 59.7 mol%, respectively, measured with the thermal denaturation method of De Ley (1970) and using DNA from Escherichia coli DH5α as standard. These values are within the range reported previously for Rhizobium (56.9–60.9 mol%). DNA–DNA relatedness between strains CCBAU 85027 and CCBAU 85039, estimated with the spectrophotometric method (De Ley et al., 1970), was 88.2 %, indicating that they represent the same genomic species. Less than 22 % DNA–DNA relatedness was detected between the representative strain CCBAU 85039^T and the type strains of the most related species R. etli (CFN 42^T), R. leguminosarum (USDA 2370^T) and R. mongolense (USDA 1844^T), indicating that the former represents a genomic species that differs from these species.

Phenotypic features of the two novel strains were determined as described previously (Gao et al., 1994) in comparison with the type strains of almost all defined Rhizobium species, including the utilization of sole carbon and nitrogen sources, resistance to antibiotics (5–300 μg ml⁻¹), tolerance of NaCl (1–5 %, w/v), ranges of temperature and pH (4.5–12) for growth and other physiological and biochemical tests. Some distinctive features of the two novel strains and the type strains of several species representing different subgroups in Rhizobium are presented in Table 1⇓, and the complete set of characteristics is available in Supplementary Table S1. The phenotypic characteristics showed that the two Tibetan strains could be differentiated from defined Rhizobium species.

To characterize the two Tibetan strains further, whole-cell proteins were extracted and analysed with SDS-PAGE as described previously (Tan et al., 1997), together with some reference strains of Rhizobium species. Using the Gelcompar II software package, the similarity between every pair of samples was calculated with the Dice coefficient and used in the construction of a UPGMA dendrogram. This analysis showed that the two Tibetan strains had identical electrophoretic protein patterns. Considering that these two strains also had 94 % similarity in numerical analysis of phenotypic characteristics, it could be concluded that the two strains are very similar. According to the protein patterns, these two strains were most similar (about 80 % similarity) to the reference strains of R. etli and R. leguminosarum (Supplementary Fig. S3), which was consistent with the phylogenetic relationships estimated from 16S rRNA gene sequences. Since the similarity of protein patterns among strains within a rhizobial species is always greater than 80 % (Gu et al., 2007; Lin et al., 2007; Liu et al., 2007; Yan et al., 2007), the results of this analysis also imply that the two Tibetan strains represent a species that is distinct from the Rhizobium species represented by the reference strains.

Profiles of cellular fatty acids have been used to discriminate species of Rhizobium and to describe novel bacterial species (de Lajudie et al., 1998; Quan et al., 2005; Tighe et al., 2000). In this study, fatty acids were analysed for the two Tibetan strains and reference strains for defined Rhizobium strains. After 3 days of incubation at 28 °C on YMA, the cells were harvested and fatty acids were extracted and identified by the standard method of the Microbial Identification System (MIDI, Inc.). Differences were observed between the two novel strains in the presence of 9 : 0, 12 : 0, 17 : 0 cyclo, 18 : 1 2-OH, 11-methyl 18 : 1ω7c, 19 : 0 cyclo ω8c and summed feature 4 (iso-17 : 1 I and/or anteiso-17 : 1 B), but the two strains shared another 10 compounds, demonstrating that they were different strains and not clones. The cellular fatty acid compositions of the Tibetan strains and reference strains for species representing distinct subgroups of Rhizobium in the 16S rRNA gene phylogeny are presented in Table 2⇓, and the complete dataset is available in Supplementary Table S2. It seems that the fatty acids 16 : 0, 16 : 0 3-OH, 18 : 0, summed feature 2 (one or more of 12 : 0 aldehyde, unknown ECL 10.928, iso-16 : 1 I and 14 : 0 3-OH) and summed feature 8 (18 : 1ω7c and/or 18 : 1ω6c) were common to all Rhizobium species, although the proportions varied in different species. The presence and abundance of other components were species or strain dependent. The two Tibetan strains showed fatty acid patterns that differed from those of the type strains of other Rhizobium species.

Nodulation and nitrogen-fixation abilities are important characteristics of the genus Rhizobium, and host range is an important feature for the description of novel rhizobial species (Graham et al., 1991). In the present study, cross-nodulation tests indicated that strains CCBAU 85027 and CCBAU 85039^T could nodulate Trigonella archiducis-nicolai, Medicago lupulina, Medicago sativa, Melilotus officinalis, Phaseolus vulgaris and Trigonella foenum-graecum under laboratory conditions.

Based on the results described above, it is clear that the two Tibetan strains represent a novel lineage of Rhizobium and are distinct from defined species by genomic analyses and phenotypic features. Therefore, we propose to accommodate them in a novel species, Rhizobium tibeticum sp. nov.