Bradyrhizobiumottawaense sp. nov., a symbiotic nitrogen fixing bacterium from root nodules of soybeans in Canada

Soybean [Glycine max (L.) Merrill] is an economically important grain legume that can form a nitrogen-fixing association with species of soil bacteria belonging to the genusBradyrhizobium. In a previous study (Tang et al., 2012), populations of symbiotic bacteria associated with soybeans at two field sites in eastern Canada were characterized on the basis of multilocus sequence analysis (MLSA) of six protein encoding (housekeeping) genes. Phylogenetic analyses resulted in the identification of several novel lineages within the genusBradyrhizobium. One of these novel lineages was encountered at only one of the two field sites and consisted of a group of 16 strains. In this work, we further characterize these strains using a variety of genotypic and phenotypic methods, and, based on the results, a novel species, Bradyrhizobium ottawaense sp. nov. is proposed.

The 16 novel strains of symbiotic bacteria (Table S1, available in the online Supplementary Material) were obtained from root nodules of soybeans grown at a field site in Ottawa, Ontario (Tang et al., 2012). Bacteria were grown on yeast-extract mannitol (YEM) agar medium (Vincent, 1970) and pure cultures maintained in 20 % (w/v) glycerol at −80 °C. Bacterial cells were Gram-stain-negative (Powers, 1995), and, based on the Schaeffer–Fulton staining method (Hussey & Zayaitz, 2012), were non-spore-forming. Colonies on YEM agar were mucoid, beige, translucent, circular and measured <1 mm in diameter after 7 days at 28 °C. Cell morphology was investigated using a transmission electron microscope (H-7000; Hitachi). Bacteria were cultured in stationary YEM broth for 2 days at 28 °C and stained with 1 % phosphotungstic acid (pH 7.0) (Hayat & Miller, 1990). All tested strains (OO99^T, OM9 and OO85) had rod-shaped cells, subpolar flagella and a cell size (Fig. S1) that is consistent with the characteristics of the genusBradyrhizobium(Garrity et al., 2005). Cell motility was demonstrated using the semisolid medium puncture method (Shields & Cathcart, 2012) with bacteria grown on semisolid YEM agar medium. Production of an alkaline reaction on YEM agar containing bromothymol blue after 21 days at 28 °C (Bromfield et al., 2010) and mean generation times (12–13 h) of bacterial cultures in YEM broth (Itakura et al., 2008; Wittwer, 2014) (Table 1) were also typical of the genusBradyrhizobium(Garrity et al., 2005; Vincent, 1970).

Almost full-length 16S rRNA gene sequences were generated using primers 16Sa and 16Sb (van Berkum & Fuhrmann, 2000) and sequence alignment was carried out using the Infernal secondary-structure-based aligner implemented in the Ribosomal Database Project program version 11.1 (Cole et al., 2014). Partial sequences of housekeeping (atpD, glnII, recA, gyrB, rpoB and dnaK) and symbiotic (nodC and nifH) genes were generated for the 16 novel strains of the genusBradyrhizobiumas well as for reference taxa not available in public databases. Preparation of genomic DNA, amplification, primers, nucleotide sequencing, sequence alignment and editing was as described previously (Tang et al., 2012). The dnaK sequence of ‘Bradyrhizobium retamae’ Ro19 required designing a second forward sequencing primer (FdnaK_SC1: GAGCAGCAGATCCGGATTCA) in order to obtain the 3′ end of the sequence. GenBank accession numbers of nucleotide sequences are given in Tables S1 and S2.

Bayesian phylogenetic analyses were carried out using MrBayes version 3.2.1 with default priors (Altekar et al. 2004). For each dataset, two concurrent analyses with four chains (three heated and one cold) were run for ten million generations with sampling every 2000 generations. Convergence was judged satisfactory when the average sd of split frequencies fell below 0.01 and the potential scale reduction factor statistics were approaching 1.0 (Ronquist et al., 2012). For each dataset, trees from the first 25 % of sampled generations were removed as burn in and a majority rule consensus tree was estimated based on pooled post-burn-in trees.

Best-fit substitution models were selected using the Bayesian information criterion implemented in jModelTest version 2 (Darriba et al., 2012). Maximum-likelihood (ML) phylogenetic analyses (Guindon et al., 2010) were carried out as previously described using 1000 non-parametric bootstrap replications to assess support (Tang et al., 2012). In all instances, tree topologies from Bayesian and ML analyses were similar; for brevity only the Bayesian trees are shown.

Consistent with a previous report (Wang et al., 2013), two major groups of species of the genusBradyrhizobiumwere evident in phylogenetic trees of 16S rRNA gene sequences: one represented byBradyrhizobium japonicumand the other byBradyrhizobium elkanii(Fig. S2). All 16 novel strains had identical 16S rRNA gene sequences and were placed in the phylogenetic group represented byB. japonicum. The sequence similarities of novel strains varied between 98.44 (Bradyrhizobium denitrificans) and 99.79 % (Bradyrhizobium liaoningense) relative to type strains of the 15 species in theB. japonicumgroup (Table S3).

MLSA of at least five housekeeping genes is used as a reliable method to define phylogenetic relationships and to identify novel lineages within the genusBradyrhizobium(Rivas et al., 2009; Tang et al., 2012). The Bayesian phylogenetic tree of six concatenated housekeeping gene sequences (length 3210 bp) for the 16 novel strains and reference taxa is shown in Fig. 1. Consistent with the results of Tang et al., (2012), the novel strains comprised three multilocus sequence types (STs) (represented by strains OO99^T, OO85 and OM9) that were placed with high confidence (>95 % posterior probability) in a lineage that was distinct from named species of the genusBradyrhizobium. Relative to OO99^T, strains OO85 and OM9 had sequence similarities for the six concatenated housekeeping genes of >99.60 % whereas those for 19 named species varied between 88.35 and 95.42 % (Tables S3).

Fig. 1.

Bayesian phylogenetic tree of atpD–glnII–recA–gyrB–rpoB–dnaK concatenated gene sequences (3210 bp) for Bradyrhizobium ottawaense sp. nov. and reference taxa of the genusBradyrhizobium(GTR+G+I substitution model). Only posterior probabilities ≥95 % are shown. Values in parentheses represent the number of strains in each of the three STs of B. ottawaense. Bar, expected substitutions per site.

To further analyse genetic differences between novel strains, we generated random amplified polymorphic DNA (RAPD) fingerprints for representative strains OO99^T, OM9 and OO85 and reference taxa using four random primers (P1, P2, P3 and P5) and amplification methods described by Sikora et al. (2002). An example of the fingerprint profiles generated by one of the primers (P1) is shown in Fig. S3. A dendrogram based on the combined character matrix of fingerprint profiles generated by the four primers was reconstructed using upgma and the Dice coefficient implemented in GelCompare II software version 5.10 (Applied Maths). The three representative strains were readily distinguished and were placed in a single cluster separate from reference taxa (Fig. S4), consistent with their assignment to distinct STs, based on MLSA of protein encoding genes.

DNA–DNA hybridization experiments were performed as described by Willems et al., (2001) using three representative strains (OO99^T, OM9 and OO85) and relatives,Bradyrhizobium yuanmingenseLMG 21827^T andB. liaoningenseLMG 18230^T. DNA–DNA hybridization values (Table S4) among the three novel strains varied between 73 % and 88 % whereas values for these strains hybridized withB. liaoningenseLMG 18230^T andB. yuanmingenseLMG 21827^T were between 23 % and 38 %. These data are consistent with the value of 70 % DNA–DNA relatedness that is considered the threshold for novel species definition Wayne et al., (1987). The G+C mol% content of DNA for the three strains, determined by HPLC (Mesbah et al., 1989), varied between 62.3 and 63.5 % which is within the range of DNA G+C content for the genusBradyrhizobium.

Analysis of accessory genes encoding symbiotic functions (nod and nif) can provide useful information on the host range and specificity of symbiotic bacteria (Laguerre et al., 2001). The phylogenetic tree of partial nodC sequences of novel strains and reference taxa is shown in Fig. S5. All 16 novel strains had identical nodC sequences and were placed in a group with five named species of the genusBradyrhizobiumthat are symbiotic with soybeans. Sequence similarities of novel strains relative to these five species were ≥99.86 % (Table S3). Analyses of nifH gene sequences (Fig. S6 and Table S3) provided almost identical results.

Plant tests were carried out using Leonard jars as described by Tang et al., (2012). The results (Table S5) show that representative strain, OO99^T, elicited effective nitrogen fixing nodules (fix+) on Glycine max, Glycine soja and Macroptilium atropurpureum, partially effective nodules (fix±) on Desmodium canadense and Vigna unguiculata, and ineffective nodules (fix-) on Amphicarpaea bracteata and Phaseolus vulgaris.

For analysis of fatty acids, strains were grown on YEM agar at 28 °C. Bacteria were harvested and fatty acids extracted as described by Sasser (1990). Fatty acid identification was done using the Sherlock Microbial Identification System (MIDI) version 6.0 and the RTSBA6 database. Fatty acid profiles of three novel strains and reference taxa are shown in Table S6. The fatty acid profile of OO99^T was characteristic of the genusBradyrhizobiumwith fatty acids 16 : 0 (12.83 %) and summed feature 8 (76.99 %) predominating (Tighe et al., 2000). Five fatty acids (16 : 0, 16 : 1ω5c, 18 : 0 and summed feature 8) were common to all strains tested whereas seven (12 : 0, 14 : 0, 17 : 1ω8c, 18 : 1ω5c, 19 : 0 cyclo ω8c and summed feature 3) were detected in only some strains.

Phenotypic characteristics were further investigated using a variety of tests. Utilization of carbon and nitrogen sources was tested using YNB and YCB (Becton Dickinson) basal liquid media as described by Chahboune et al., (2011) except that bromothymol blue was omitted. Cultures were incubated at 28 °C for 20 days on a rotary shaker when bacterial growth (turbidity) was recorded visually and using a Spectronic 21 UV spectrophotometer (Milton Roy) at 660 nm.

Tests of intrinsic antibiotic resistance were done according to Bromfield et al. (2010) using YEM agar medium amended with filter-sterilized solutions of the following antibiotics (Sigma Aldrich) at final concentration (µg ml⁻¹): erythromycin (0, 50, 100, 200), penicillin G sodium salt (0, 10, 25), kanamycin sulphate (0, 25, 50, 100), tetracycline (0, 5, 10, 20), chloramphenicol (0, 25, 50), cefuroxime sodium salt (0, 5, 15, 30). Bacterial growth was recorded after 7 days at 28 °C. Tests for bacterial growth on YEM agar at 10, 28 and 37 °C, at pH 5 and pH 10 and in the presence of 1 % (w/v) NaCl were done as described by Xu et al., (1995). The combination of phenotypic characteristics listed in Table 1 could be used to differentiate novel strains from recognized species of the genusBradyrhizobium.

Based on data for genotypic and phenotypic analyses presented in this study, we propose that the novel group of 16 strains represent a novel species, named Bradyrhizobium ottawaense sp. nov.