Bradyrhizobiumelkanii, Bradyrhizobiumyuanmingense and Bradyrhizobiumjaponicum are the main rhizobia associated with Vignaunguiculata and Vignaradiata in the subtropical region of China

Cowpea ( Vigna unguiculata ) and mung bean ( Vigna radiata ) are important legume crops yet their rhizobia have not been well characterized. In the present study, 62 rhizobial strains isolated from the root nodules of these plants grown in the subtropical region of China were analyzed via a polyphasic approach. The results showed that 90% of the analyzed strains belonged to or were related to Bradyrhizobium japonicum , Bradyrhizobium liaoningense , Bradyrhizobium yuanmingense and Bradyrhizobium elkanii , while the remaining represented Rhizobium leguminosarum , Rhizobium etli and Sinorhizobium fredii . Diverse nifH and nodC genes were found in these strains and their symbiotic genes were mainly coevolved with the housekeeping genes, indicating that the symbiotic genes were mainly maintained by vertical transfer in the studied rhizobial populations.


Introduction
Cowpea (Vigna unguiculata) and mung bean (Vigna radiata) are important legumes cultivated in China; each can fix nitrogen through nodule symbiosis with rhizobia. Their drought tolerance, nitrogen-fixation capacity and shade tolerance make these plants important components in intercropping with maize or sorghum. Cowpea is native to Africa, but has been cultivated in China as a vegetable and grain crop and as a herbal medicine for centuries since it was first recorded in the literature some 500 years ago. Mung bean is native to India, but has been cultivated for around 3000 years in China. The seeds of mung bean are commonly used as a cooling food in Chinese cuisine and as a herbal medicine.
Previous studies have classified cowpea and mung bean rhizobia as cowpea miscellany belonging to the genus Bradyrhizobium (Jordan, 1982), but the species status of these taxa was unclear. Several Vigna rhizobial strains have been classified recently as representing Bradyrhizobium spp. and Rhizobium spp. based on 16S rRNA gene sequence phylogeny (Wolde-Meskel et al., 2005;Germano et al., 2006;Yokoyama et al., 2006). However, little information is available regarding the rhizobia associated with cowpea and mung bean in Chinese soils.
Given the importance of cowpea and mung bean in sustainable agriculture and the lack of data on their rhizobia, we decided to collect and characterize the rhizobia naturally associated with these legumes grown in the subtropical region of China. This region has a humid climate and acid soils in which rhizobial populations may be different from those isolated in other regions. For example, Sinorhizobium fredii was predominant in soybean nodules (Camacho et al., 2002) and Bradyrhizobium was isolated from clover (Liu et al., 2007) in Hubei, a subtropical province of China. The aim of the present study was to establish the diversity of rhizobia naturally associated with cowpea and mung bean in Chinese soils.

Isolates and strains
Sixty-two new isolates and 28 reference strains of rhizobia were used (Table 1). The Vigna rhizobia were isolated using a routine method and yeast mannitol agar (YMA) (Vincent, 1970) from root nodules collected in the fields of nine subtropical provinces in China. Nodulation of each strain on its original host was confirmed by nodulation tests (Vincent, 1970). The nitrogen-fixing ability of the isolates was verified after 1 month based on the presence of red color (leghemoglobin) inside the nodules. All strains were maintained on YMA medium and incubated at 28 1C.
Amplified 16S rDNA restriction analysis (ARDRA) and restriction fragment length polymorphism (RFLP) of 16S--23S rRNA gene internal transcribed spacer (ITS) Primers P1 and P6 and DNA extracted from each strain were used to amplify the 16S rRNA gene as described previously (Tan et al., 1997). The PCR products were digested separately with MspI, HinfI, HaeIII and AluI (Laguerre et al., 1996). The ITS fragments were amplified with primers FGPS1490 and FGPL132 0 according to the procedure of Laguerre et al. (1996) and digested separately with MspI, HaeIII and AluI. The rRNA gene and ITS restriction fragments were separated by electrophoresis and visualized as described (Laguerre et al., 1996). The RFLP profiles were analyzed using the GELCOMPAR II software package. In the cluster analysis of RFLP patterns, the Dice similarity coefficient and unweighted pair group method with arithmetic averages (UPGMA) were used. Strains sharing identical RFLP patterns were designated as the same rRNA type in ARDRA or ITS type in ITS-RFLP.
Sequencing and phylogenetic analyses of 16S rRNA, nifH and nodC genes Amplification of the 16S rRNA gene was performed as described above. nifH gene fragments were amplified with primers nifH40F and nifH817R according to the procedure of Vinuesa et al. (2005a). The nodC gene was amplified using primers NodCfor540 and NodCrev1160 according to the protocol described by Sarita et al. (2005). The PCR products were purified and directly sequenced as described by Vinuesa et al. (2005a). The resulting sequences, together with the related sequences obtained from the GenBank database, were aligned using CLUSTAL W (Thompson et al., 1994). Phylogenetic trees were constructed using the neighbor-joining method with the kimura two-parameter model and were bootstrapped based on 1000 replicates using the programs in the MEGA3.1 package (Kumar et al., 2004).

Phenotypic characterization and numerical taxonomy
One hundred and twenty-nine phenotypic features, including utilization of carbon and nitrogen sources, growth ranges of pH and temperature, resistance to antibiotics and dyes, tolerance to NaCl, and some enzyme activities, were analyzed (Gao et al., 1994). The phenotypic characters were used to estimate the simple matching similarity coefficient (S sm ) of each strain pair, which was subsequently used in cluster analysis with UPGMA (Sneath & Sokal, 1973).

Isolation and nodulation of bacteria
A total of 62 isolates were obtained, including eight fastgrowing, acid-producing bacteria that produced colonies >2 mm in diameter after 2-3 days of incubation, and 54 slow-growing, alkali-producing bacteria with colonies 1 mm in diameter after 5-7 days of incubation ( Table  1). The nodulation results demonstrated that most of the isolates formed nitrogen-fixing nodules on their hosts of origin (supplementary Table S1).  S1). The eight fast-growing isolates were found in five rRNA types identical or similar to Rhizobium etli, Rhizobium leguminosarum and S. fredii. The 54 slow-growing isolates were found in three rRNA types, which were respectively identical to the reference strains for Bradyrhizobium elkanii, Bradyrhizobium japonicum-Bradyrhizobium liaoningense and Bradyrhizobium yuanmingense.

ITS PCR--RFLP
In this analysis, 29 ITS types were identified among the 62 isolates (Table 1), and these were grouped into 11 clusters at a similarity level of 63% (Table 1, supplementary Fig. S2). At this level of similarity, most of the reference strains were divided into different clusters corresponding to their species, but the reference strains of four Sinorhizobium species were found in the same cluster (cluster 9). These grouping results were consistent with the rRNA types in most cases, except for strains in rRNA types 20 and 21. Several ITS clusters were found in rRNA type 20, and ITS clusters 1 and 2 were found in both rRNA types.

16S rRNA gene phylogeny
For each rRNA type, 1-4 representative isolates were used in 16S rRNA gene sequence analysis. In the reconstructed phylogenetic tree (supplementary Fig. S3), these isolates were respectively grouped together with reference strains of R. etli, R. leguminosarum, S. fredii, B. elkanii, B. japonicum and B. yuanmingense.

Numerical taxonomy
A subset of 50 isolates and nine reference strains were examined by phenotypic characterization. Seventeen features were the same for the test strains and only the 112 variable features were used in cluster analysis, and all the strains were divided into eight phena corresponding to species in general and four single strains at a similarity level of 80% (Table 1, Fig. 1). The eight fast-growing isolates and reference strains for R. etli, R. leguminosarum and S. fredii were respectively found in phena 1, 2 and 3 (Table 1). Phena 4-8 were members of the genus Bradyrhizobium. Phenon 5 had three isolates of rRNA types 20 and 21. Phenon 6 contained eight isolates in rRNA type 20 and two B. japonicum reference strains. Phenon 7 had 14 isolates that were further divided into two subgroups: 7a with 10 isolates of rRNA type 21 and the type strain of B. yuanmingense; and 7b with four isolates of rRNA type 20. Fifteen isolates in rRNA type 19 were found in phenon 8 together with B. elkanii USDA 76 T and in phenon 4.

Phylogeny of nifH and nodC gene sequences
In the present study, nifH and nodC gene fragments were amplified from several representative isolates. In the nifH and nodC phylogenetic trees (Fig. 2), the isolates were grouped according to their genera and to species in most cases, although two groups were found in the B. japonicum strains. The exceptions were B. elkanii isolates CCBAU 51159 and CCBAU 23011, which were in the B. yuanmingense and B. japonicum groups, respectively, in the nifH tree, and B. japonicum CCBAU 51377, which was in the B. yuanmingense group in the nodC tree.

Phylogeny of housekeeping genes
According to the grouping results of the previous analyses, representative strains were used in sequencing the atpD, glnII and recA genes. The phylogenetic relationships obtained from these three genes (Fig. 3 for recA, supplementary Figs S4 and S5 for atpD and glnII) were generally similar to each other and were congruent with the groupings in the other analyses. Seven phylogenetic groups corresponding to B. elkanii, B. japonicum, B. liaoningense, B. yuanmingense, R. etli, R. leguminosarum and S. fredii were defined among the Vigna rhizobia. Five strains (marked in bold type in Fig. 3) showed unstable relationships that were closer to B. japonicum in the glnII tree or between B. liaoningense and B. yuanmingense, to B. liaoningense in the atpD tree, with bootstrap values of 42-66, while two of them were grouped with B. liaoningense with a bootstrap value of 51% and three were in the B. japonicum group.

Discussion
Previously, cowpea and mung bean rhizobia have not been systematically studied and their taxonomic positions were not clear, although several rhizobial strains associated with these two legumes grown in tropics, including the center of origin for cowpea (Wolde-Meskel et al., 2005) and other regions (Germano et al., 2006;Yokoyama et al., 2006), have been studied based on 16S rRNA and nod gene sequencing. The present study revealed that Bradyrhizobium spp. occupied 90% of the cowpea and mung bean nodules in the subtropical region of China and the remaining proportion was fast-growing rhizobia. Currently, a polyphasic approach including 16S-23S rRNA gene ITS-RFLP, numerical taxonomy, DNA-DNA hybridization experiments, multilocus sequence analysis (Willems et al., 2003;Martens et al., 2008), ARDRA and 16S rRNA gene sequence analyses has been used to define rhizobial species. According to the consensus of the results from different analyses, Vigna isolates could be assigned to seven species (Table 1). All the isolates in rRNA type 21 were identified as representing B. yuanmingense. All the isolates in rRNA type 19 were defined as representing B. elkanii species based on the results of numerical taxonomy, ITS-RFLP, multilocus sequence analysis and 16S rRNA gene sequence phylogeny. The eight fast-growing isolates were defined as representing R. leguminosarum, R. etli and S. fredii. In the present study, the S. fredii and Sinorhizobium xinjiangense strains were very similar in all the sequence analyses and they might represent the same species as reported recently (Lloret et al., 2007;Martens et al., 2008). Previously, both R. etli (Hernandez-Lucas et al., 1995) and S. fredii (Pueppke & Broughton, 1999) have been reported to nodulate and fix nitrogen in V. unguiculata under laboratory conditions. The results of the current study verified that these three rhizobial   species were also microsymbionts of V. unguiculata in fields, but their minor proportion indicated that they were not preferable rhizobia for that host.
As described previously, species definition within the genus Bradyrhizobium is difficult and B. japonicum and B. liaoningense are very closely related species (Willems et al., 2001(Willems et al., , 2003. In the present study, four isolates could be defined as representing B. liaoningense based on consensus numerical taxonomy (phenon 7b), ITS-RFLP (group 5) and multilocus sequence analysis. Three isolates could be confirmed as representing B. japonicum based on numerical taxonomy and multilocus sequence analysis. Although 11 isolates were listed as B. japonicum-related strains ( Vigna has been reported to be one of the most promiscuous plants, and its nodulation with Bradyrhizobium spp. and S. fredii has been recorded (Thies et al., 1991;Pueppke & Broughton, 1999). However, only B. elkanii-related rhizobia were isolated from mung bean and B. japonicum-related strains were isolated previously from six wild Vigna species in Thailand (Yokoyama et al., 2006). Although cowpea and mung bean did not originate in China, the diversity observed among our isolates was high and comparable with that detected in rhizobia isolated from other legumes in their sites of origin, such as Lima bean rhizobia (Ormeño-Orrillo et al., 2006). The great diversity of cowpea and mung bean rhizobia in the subtropical region of China might be related to the large sampling size and sites used in our study. Furthermore, the sharing of ITS types 5, 12, 18, 19 and 20 by the rhizobia isolated from both hosts demonstrated that the two plants have attracted rhizobia from common gene pools.

Supplementary material
The following supplementary material is available for this article: Table S1. Nodulation results of Vigna rhizobia. This material is available as part of the online article from: http://www.blackwell-synergy.com/doi/abs/10.1111/j. 1574-6968.2008.01169.x (This link will take you to the article abstract).
Please note: Blackwell Publishing are not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.