Multilocus sequence analysis (MLSA) of Bradyrhizobium strains: revealing high diversity of tropical diazotrophic symbiotic bacteria

Abstract

Symbiotic association of several genera of bacteria collectively called as rhizobia and plants belonging to the family Leguminosae (=Fabaceae) results in the process of biological nitrogen fixation, playing a key role in global N cycling, and also bringing relevant contributions to the agriculture. Bradyrhizobium is considered as the ancestral of all nitrogen-fixing rhizobial species, probably originated in the tropics. The genus encompasses a variety of diverse bacteria, but the diversity captured in the analysis of the 16S rRNA is often low. In this study, we analyzed twelve Bradyrhizobium strains selected from previous studies performed by our group for showing high genetic diversity in relation to the described species. In addition to the 16S rRNA, five housekeeping genes (recA, atpD, glnII, gyrB and rpoB) were analyzed in the MLSA (multilocus sequence analysis) approach. Analysis of each gene and of the concatenated housekeeping genes captured a considerably higher level of genetic diversity, with indication of putative new species. The results highlight the high genetic variability associated with Bradyrhizobium microsymbionts of a variety of legumes. In addition, the MLSA approach has proved to represent a rapid and reliable method to be employed in phylogenetic and taxonomic studies, speeding the identification of the still poorly known diversity of nitrogen-fixing rhizobia in the tropics.

INTRODUCTION

Species of the large family Leguminosae (Fabaceae in the USA) occupy a broad-range of terrestrial biomes, with several of them being capable of establishing symbioses with bacteria collectively called rhizobia, starting the process of fixing atmospheric nitrogen (N₂). Diazotrophic bacteria were first isolated from root nodules and characterized in 1888 by Martinus Willem Beijerinck (2) as Bacillus radicicola, but then reclassified into the genus Rhizobium, with the taxonomy based on nodulation with certain host plants, establishing the "cross-inoculation group" concept (11). Taxonomical classification remained unaltered until the early 1980s, when numerical taxonomy considering morphological, physiological and genetic patterns led to the positioning of some strains into a new genus, Bradyrhizobium, with a unique defined species, B. japonicum (18). Bradyrhizobium includes slow growers that produce alkaline reaction in culture medium with mannitol as carbon source, while Rhizobium contains fast growing acid producers (18, 19).

A few years later, high diversity among Bradyrhizobium strains led to the description of a new species named Bradyrhizobium elkanii (23); however, the development of several molecular techniques helped to identify high genetic diversity among Bradyrhizobium strains isolated from a wide range of leguminous plants (e.g., 13, 28, 29, 46). Nowadays, there are nine defined Bradyrhizobium species (7), but certainly many more should be described in the next years.

Sequencing of the 16S rRNA has become the method of choice for tracing bacterial phylogenies (12, 50, 52), but in several genera including Bradyrhizobium variability in the 16S rRNA is often low and may not reflect the diversity detected by other morpho-physiological and genetic properties (e.g., 13, 28, 42, 46, 51). To detect higher diversity, other ribosomal genes or regions have been chosen, as the 23S rRNA and the 16S-23S rRNA intergenic transcribed spacer (ITS), since they evolve at a faster rate than the 16S rRNA, thus adding valuable information to the analysis (e.g., 29, 41, 42, 46, 51). However, proximity of ribosomal genes might not properly indicate the correct phylogeny in the case of horizontal gene transfer (43).

The multilocus sequence analysis (MLSA) method has been increasingly used in phylogeny and taxonomy studies. The method consists of the analysis of several conserved housekeeping genes dispersed in at least 100 kb of the genome (4, 6, 15, 25). Successful definitions of phylogenetic groups of rhizobia have also been achieved with the use of the MLSA (29, 31, 36).

Tropical rhizobia represent a key component for the sustainability of tropical soils, and the few results obtained so far clearly indicate that diversity is largely underestimated within the Bradyrhizobium genus (13, 28, 29). Therefore in this study the MLSA approach was applied to achieve a better phylogenetic resolution of twelve strains showing high genetic diversity of the 16S rRNA gene in relation to the described Bradyrhizobium species. The objective is to delineate strategies which may help to more promptly define diversity of tropical Bradyrhizobium.

MATERIALS AND METHODS

Strains

Twelve Bradyrhizobium strains were used in this study and are listed in Table 1. The strains were chosen from previous studies from our group (3, 13, 28, 29, 37), based on the high level of genetic diversity observed in comparison to the described species of Bradyrhizobium. The strains have been isolated from members of two subfamilies and five tribes of the family Leguminosae, and originated from four countries located in different continents: Australia, Brazil, Malaysia and Zimbabwe (Table 1). The strains were purified on yeast extract-mannitol agar (YMA) medium (45) containing Congo red (0.00125%) and stocks were prepared on YMA and kept at -80°C (under 30% of glycerol) for long-term storage and at 4°C as source cultures. The strains are deposited at the ''Culture Collection of Diazotrophic and PGPR Bacteria" of Embrapa Soja (http://www.bmrc.lncc.br) and at the Brazilian official culture collection "SEMIA" of rhizobial strains (IBP World Catalogue of Rhizobium Collections n°443 in the WFCC World Data Center on Microorganisms).

Table 1.

Information about the Bradyrhizobium strains from the Embrapa Soybean culture collection used in this study

SEMIA number	Other designationsa	Origin of the strain	Country of origin	Host speciesb	Subfamilyb	Tribeb	Previous studiesc
656	SEMIA original, CNPSo 988	FEPAGRO	Brazil	Neonotonia wightii (Wight & Arn.) Lackey	Papilionoideae	Phaseoleae	a, b, d
662	CB 188, CNPSo 990	CSIRO	Australia	Vigna unguiculata (L.) Walp	Papilionoideae	Phaseoleae	a, b, d
696	CB 627, CNPSo 993	CSIRO	Australia	Desmodium uncinatum (Jacq.) DC	Papilionoideae	Demodieae	a, b, d, e
6002	CB 756, TAL 309, RCR 3824, CNPSo 1092	CSIRO	Zimbabwe	Vigna unguiculata (L.) Walp	Papilionoideae	Phaseoleae	a, b, d
6028	TAL 569, SPRL 472, MAR 472, CNPSo 1094	NIFTAL	Zimbabwe	Desmodium uncinatum (Jacq.)DC	Papilionoideae	Demodieae	a, b, d
6053	TAL 827, UMKL 28, CNPSo 1095	NIFTAL	Malaysia	Clitoria ternatea L.	Papilionoideae	Phaseoleae	a, b, d, e
6144	SMS 400, USDA 3187, MAR 11, CNPSo 1109	IAC	Zimbabwe	Arachis hypogaea L.	Papilionoideae	Aeschynomeneae	a, b, d
6145	BR 2001, CNPSo 1110	Embrapa Agrobiologia	Brazil	Crotalaria juncea L.	Papilionoideae	Crotalarieae	a, b, d
6148	SMS 303, CNPSo 1112	IAC	Brazil	Neonotonia wightii (Wight & Arn.) Lackey	Papilionoideae	Phaseoleae	a, b, d
6154	BR 446, CNPSo 1117	Embrapa Agrobiologia	Brazil	Stylosanthes spp.	Papilionoideae	Aeschynomeneae	a, c, e
6160	BR 5610, CNPSo 1123	Embrapa Agrobiologia	Brazil	Albizia lebbeck (L.) Benth.	Mimosoideae	Ingeae	a, b, d
6395	BR 4301, CNPSo 1161	Embrapa Agrobiologia	Brazil	Calliandra houstoniana (Mill.) Standl.	Mimosoideae	Ingeae	c

^a Culture collections: BR (Brazil, Embrapa Agrobiologia, Seropédica, Brazil); CB (Commonwealth Scientific and Industrial ^a Research Organization – CSIRO, Canberra, Australia); CNPSo (Centro Nacional de Pesquisa de Soja, Brazil); MAR (Marondera, Grasslands Rhizobium Collection, Soil Productivity Research Laboratory, Marondera, Zimbabwe; also called SPRL); SEMIA (Seção de Microbiologia Agrícola, FEPAGRO, Porto Alegre, Brazil); SMS (Seção de Microbiologia do Solo, IAC, Campinas, Brazil); TAL (NifTAL, Nitrogen Fixation by Tropical Agricultural Legumes Project, University of Hawaii, Paia, USA); USDA (United States Department of Agriculture, Beltsville, USA);^b Taxonomy based on ILDIS (www.ildis.org);^c Previous studies with the strains: (a) Germano et al. (2006); (b) Menna et al. (2006); (c) Binde et al. (2009); (d) Menna et al. (2009); (e) Roma Neto et al. (2010).

DNA extraction and sequencing analysis of the 16S rRNA

The DNA samples were extracted from bacterial batch cultures grown in YM broth until late exponential phase (10⁹ cells mL^-1) as described before (20). The DNAs were submitted to the amplification with primers for the 16S rRNA, recA, atpD, glnII, gyrB and rpoB genes, as listed in Table 2. The PCR products were purified with the PureLink^TM PCR Purification Kit (Invitrogen), and the reactions for the sequencing analysis contained purified PCR products of each bacterium culture (80 ng per reaction), 3 μL of dye (DYEnamic ET terminator reagent premix for the MEGA BACE) and 3 pmol of each primer (Table 2). The conditions of PCR amplification were as described by Menna et al. (28). The sequencing was performed on a MEGA BACE 1000 (Amersham Biosciences) capillary sequencer according to the manufacturer's instructions.

Table 2.

Primers and DNA amplification conditions used in this study

Primer	Sequence (5`- 3`)a	Target gene (position) †	PCR cycling	Reference
TSrecAf	CAACTGCMYTGCGTATCGTCGAAGG	recA (8-32)	2 min 95°C, 35 X (45s 95°C, 30s 58°C, 1,5 min 72°C and 7 min 72°C.	Stepkowski et al. (2005)
TSrecAr	CGGATCTGGTTGATGAAGATCACCATG	recA (620-594)
TSatpDf	TCTGGTCCGYGGCCAGGAAG	atpD (189-208)	2 min 95°C, 35 X (45s 95°C, 30s 58°C, 1,5 min 72°C and 7 min 72°C.	Stepkowski et al. (2005)
TSatpDr	CGACACTTCCGARCCSGCCTG	atpD (804-784)
TSglnIIf	AAGCTCGAGTACATCTGGCTCGACGG	glnII (13-38	2 min 95°C, 35 X (45s 95°C, 30s 58°C, 1,5 min 72°C and 7 min 72°C.	Stepkowski et al. (2005)
TSglnIIr	SGAGCCGTTCCAGTCGGTGTCG	glnII (681-660)
gyrB343F	TTCGACCAGAAYTCCTAYAAGG	gyrB (343-364)	5 min 95°C, 5X (2 min 94°C, 2 min 58°C, 1 min 72°C) 28 X (30s 94°C, 1 min 58°C, 1 min 72°C and 5 min 72°C.	Martens et al. (2008)
gyrB1043R	AGCTTGTCCTTSGTCTGCG	gyrB (1061-1043)
rpoB83F	CCTSATCGAGGTTCACAGAAGGC	rpoB (1081-1061)	5 min 95°C, 3X (2 min 94°C, 2 min 58°C, 1 min 72°C) 30 X (30s 94°C, 1 min 58°C, 1 min 72°C and 5 min 72°C.	Martens et al. (2008)
rpoB1061R	AGCGTGTTGCGGATATAGGCG	rpoB (83-103)
fD1	AGAGTTTGATCCTGGCTCAG	16S rRNA (9-29)	2 min 95°C, 30 X (15s 94°C, 45s 93°C, 45s 55°C, 2 min 72°C and 5 min 72°C.	Weisburg et al. (1991)
rD1	CTTAAGGAGGTGATCCAGCC	16S rRNA (1474-1494)

^aMixtures of bases used at certain positions are given as: K, T or G; S, G or C; Y, C or T; R, A or G; M, A or C

^†Position of the primer in the corresponding sequence of Bradyrhizobium japonicum USDA 110

Cluster analyses

The 16S rRNA, recA, atpD, glnII, gyrB and rpoB sequences generated were analyzed with the programs Phred (8, 9), Phrap (http://www.phrap.org) and Consed (16). The consensus sequences obtained and confirmed in the 5´ and 3´ directions were submitted to the GenBank database and received the accession numbers listed in Table 3. Some genes have been previously sequenced by our group, but were re-sequenced in this study; as the identities were confirmed in 100% of the nucleotide bases, the original accession numbers were maintained (Table 3). Sequences for other reference/type strains were retrieved from the GenBank database and are also listed in Table 3. Caulobacter crescentus strain CB 15 (genome, AE005673), was used as an outgroup.

Table 3.

GenBank/EMBL/DDBJ accession numbers for the sequences of the Bradyrhizobium strains used in this study and of the reference/type strains

Strain	1 6S rRNA	recA	atpD	glnII	gyrB	rpoB
Bradyrhizobium strains
SEMIA 656	AY904732a	FJ391146c	FJ390946c	FJ391026c	HQ634882e	HQ634901e
SEMIA 662	AY904734a	HQ634894e	HQ634871e	HQ634877e	HQ634883e	HQ634902e
SEMIA 696	AY904736a	HQ634895e		GQ160506d	HQ634884e	HQ634903e
SEMIA 6002	AY904743a	HQ634896e	HQ634872e	HQ634878e	HQ634885e	HQ634904e
SEMIA 6028	AY904744a	FJ391159c	FJ390959c	FJ391039c	HQ634886e	HQ634905e
SEMIA 6053	AY904745a	FJ391160c	FJ390960c	FJ391040c	HQ634887e	HQ634906e
SEMIA 6144	AY904750a	HQ634897e	HQ634873e	HQ634879e	HQ634888e	HQ634907e
SEMIA 6145	AY904751a	HQ634898e	HQ634874e	HQ634880e	HQ634889e	HQ634908e
SEMIA 6148	AY904753a	FJ391168c	FJ390968c	FJ391048c	HQ634890e	HQ634909e
SEMIA 6154	FJ025100f	HQ634899e	HQ634875e	GQ160500d	HQ634891e	HQ634910e
SEMIA 6160	AY904762a	FJ391171c	FJ390971c	FJ391051c	HQ634892e	HQ634911e
SEMIA 6395	FJ025101f	HQ634900e	HQ634876e	HQ634881e	HQ634893e	HQ634912e
Reference/type strains
B. betae LMG 21987T	AY372184f	AB353734f	FM253129f	AB353733f	FM253217f	FM253260f
B. jicamae PAC 68T	AY624134f	HM047133f	FJ428211f	FJ428204f	HQ873309 f	HQ587647 f
B. pachyrhizi PAC 48T	AY624135f	HM047130f	FJ428208f	FJ428201f	HQ873310 f	HQ587648 f
B. canariense LMG 22265T	AJ558025f	FM253177f	AY386739f	AY386765f	FM253220f	FM253263f
B. yuanmingense LMG 21827T	AF193818f	AM168343f	AY386760f	AY386780f	FM253226f	FM253269f
B. liaoningense LMG 18230T	AF208513f	AY591564f	AY386752f	AY386775f	FM253223f	FM253266f
B. elkanii USDA 76 T	U35000f	AY591568f	AY386758f	AY599117f	AM418800f	AM295348f
B. japonicum USDA 6T	X66024f	AM182158f	AM168320f	AF169582f	AM418801f	AM295349f
B. iriomotense EK05T	AB300992f	AB300996f	AB300994f	AB300995f	AB300997f	HQ587646 f
Rhizobium tropici CIAT 899T	EU488752f	EU488815f	AM418789f	EU488791f	AM418836f	AM295354f
Rhizobium etli CFN 42T	EU488751f	EU488824f	+NC_007761f	EU488776f	+NC_007761f	+NC_007761f
Rhizobium rhizogenes ATCC 11325T	AY945955f	AM182126f	AM418786f	FJ816281f	AM418833f	AM295353f
Mesorhizobium loti LMG 6125T	X67229f	AM182156f	AM946552f	AF169581f	EU273810f	+A000012.4f
Mesorhizobium huakuii USDA 4779T	D13431f	AJ294370f	AJ294394f	AF169588f	AM076344f	FJ393283f

^aFrom the study by Menna et al. (2006)

^bFrom the study by Binde et al. (2009)

^cFrom the study by Menna et al. (2009)

^dFrom the study by Roma Neto et al. (2010)

^eFrom this study

^fFrom the GenBank (www.ncbi.nlm.nih.gov)

⁺Genome

The MLSA analysis was performed considering only the complete aligned sequences (size among parenthesis) obtained for the Bradyrhizobium strains and for the type/reference strains retrieved from GenBank: 16S rDNA (1,347 bp), recA (293 bp), atpD (395 bp), glnII (442 bp), gyrB (434 bp) and rpoB (395 bp).

All sequences obtained in this study or retrieved from GenBank were analyzed individually and concatenated using the MEGA (Molecular Evolutionary Genetics Analysis) software version 4.0 with the default parameters, K2P distance model (21), and the Neighbor-Joining algorithm (38). Statistical support for tree nodes was evaluated by bootstrap (10) analyses with 1000 samplings (17).

RESULTS

Diversity in the 16S rRNA

The phylogenetic tree built with the 16S rRNA sequences split the Bradyrhizobium strains in two large groups, with final bootstrap supports for each group of 70 and 94%, respectively (Fig. 1). The first group (G-I) comprised eight SEMIA strains, all grouped with B. elkanii and also with B. pachyrhizi and B. jicamae. The second group (G-II) included SEMIAs 6395, 656, 6002 and 6144 showing closer relation with B. iriomotense, in addition to reference/type strains of B. betae, B. canariense, B. yuanmingense, B. liaoningense and B. japonicum. Fig. 1 also highlights that the clustering analysis based on the 16S rRNA gene did not define clear positions for the SEMIA strains.

Figure 1.

Phylogenetic relationships of Bradyrhizobium strains from this study and of reference/type rhizobial strains based on the 16S rRNA. Phylogeny was inferred using the Neighbor-Joining method. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1,000 replicates) are shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. All positions containing gaps and missing data were eliminated from the dataset (Complete deletion option). Phylogenetic analyses were conducted in MEGA4.

Diversity in the atpD, glnII, gyrB, recA and rpoB genes

The additional housekeeping genes selected to refine the phylogeny analysis in this study are highly conserved among bacteria of the order Rhizobiales, are dispersed in the genome of B. japonicum strain USDA 110 and encode important proteins. For each housekeeping gene, phylogenetic trees were constructed and resulted in distinct groups (Fig. 2).

Figure 2.

Phylogenetic relationships of Bradyrhizobium strains from this study and of reference/type rhizobial strains based on the (A) recA, (B) atpD, (C) glnII, (D) gyrB and (E) rpoB genes. Method and parameters of analysis were as described for Figure 1.

When compared to the 16S rRNA, higher variability was detected in the analysis of the recA gene, and strains also fit into two groups (Fig. 2A). All four strains previously positioned in G-II of the 16S rRNA were closer to B. iriomotense, but with the appearance of at least one subgroup including SEMIA 6002 and SEMIA 6144 strains. This first group of strains had a high bootstrap support, of 99%. Subgroups were even more evident with the strains previously positioned in G-I of the 16S rRNA, with the delineation of four subgroups based on the recA gene: one including strains SEMIAs 6160, 662 and B. pachyrhizi, the second with SEMIA 696 and B. elkanii and two new subgroups including exclusively SEMIA strains, one with SEMIAs 6148 and 6154 and the other with SEMIAs 6028, 6053 and 6145. These subgroups had bootstrap supports ranging from 81 to 99% (Fig. 2A).

Greater variability was also observed with atpD gene (Fig. 2B), when compared to the 16S rRNA. The tree built with the atpD gene resulted in the definition of three main groups (G-I, G-II and G-III), with a final bootstrap support of 81%. Although the DNA of strain SEMIA 696 amplified with the atpD primers, sequencing of the fragment failed, therefore the strain was not included in the analysis. In G-I, strains were split in two subgroups, the first clustering four strains with higher resemblance with B. elkanii, and the second with three strains grouping with B. pachyrhizi (Fig. 2B). The other four SEMIA strains from this study were positioned in G-III of the atpD tree, with SEMIA 6395 showing higher similarity with B. betae, while the other strains were grouped with B. liaoningense (Fig. 2B).

The tree built with the glnII gene also resulted in two groups. In G-I four SEMIA strains were clustered with B. pachyrhizi, SEMIA 696 was clustered with B. elkanii, followed by the inclusion of SEMIA 662 and B. jicamae (Fig. 2C). In G-II of the glnII, it is worth mentioning the clustering of strains SEMIAs 6002 and 6144 with a bootstrap support of 97% (Fig. 2C), also confirmed with high support in the previous trees of recA (Fig. 2A) and atpD (Fig. 2B), but not well defined in the 16S rRNA (Fig. 1).

In the trees built with the gyrB and rpoB genes the strains were also split in two groups, with the formation of a third group in the gyrB tree, which clustered only the type strains of B. betae and B. canariense. In general, a better definition of the strains positioned in G-I of the 16S rRNA (Fig. 1) was not achieved with the analysis of the gyrB (Fig. 2D) and rpoB (Fig. 2E) genes. The other four SEMIAs analyzed were positioned in G-II in both trees. In G-II of gyrB, the SEMIA 6395 clustered with B. iriomotense, while SEMIAs 656, 6002 and 6144 were closer to B. yuanmingense. In the rpoB tree, these four SEMIAs were clustered in two well defined subgroups (656-6395 and 6002-6144).

It is also interesting that, excepting for the atpD gene, B. pachyrhizi, B. jicamae and B. elkanii were positioned in the same group in the analysis of the 16S rRNA, recA, glnII, gyrB and rpoB (Fig. 1 and Figs. 2A to 2E). On the other hand, definition of the other Bradyrhizobium species considered in this study was not completely clear when considering those genes (Fig. 1 and Figs. 2A to 2E).

From the analysis of the 16S rRNA (Fig. 1) and of five housekeeping genes (Fig. 2), the division of Bradyrhizobium in two main groups was clear, with the appearance of a third group not completely well defined only in the trees built with the atpD and gyrB genes. Each SEMIA strain used in our study was always positioned in the same great group in all six trees.

Concatenated analysis of recA, atpD, glnII, gyrB and rpoB

All five sequences of the housekeeping genes were concatenated to gain a better understanding of the strains; SEMIA 696 was not included. A concatenated sequence with 1,959 bp was obtained and 2,028 sites were analysed, resulting in 1,496 conserved, 463 variable and 304 parsimony-informative sites (Table 4).

Table 4.

Sequence information obtained in this study. Twelve strains were analysed, together with nine type and reference strains, as described in Methods.

Locus	Strains analysed (n)	Nucleotides (%)				Frequency T/C/A/G (%)
		Conserved	Variable	Parsimony-informative	Total*
16S rRNA	21	1,266 (91,5)	86 (6,2)	52 (3,7)	1,347/1,383	20.4/24.0/24.7/30.9
atpD	20	318 (77,6)	77 (18,8)	57 (13,9)	395/410	15.7/33.3/19.6/31.3
glnII	21	334 (72,8)	108 (23,5)	77 (16,8)	442/459	16.2/32.7/19.9/31.2
gyrB	21	338 (74,8)	96 (21,2)	56 (12,4)	434/452	21.7/29.9/16.5/32.0
recA	21	217 (72,6)	76 (25,4)	52 (17,4)	293/299	15.8/32.3/17.3/34.7
rpoB	21	287 (70,8)	108 (26,7)	63 (15,5)	395/405	17.4/32.8/18.7/31.1
Concatenated genes	20	1,496 (73,7)	463 (22,8)	304(15)	1,959/2,028	17.5/32.2/18.4/31.9

^*Mean number of nucleotides amplified/number of sites analysed, including gaps.

The tree built with the concatenated genes resulted in two great groups, with a bootstrap support of 100% (Fig. 3). G-I assembled SEMIAs 656, 6395, 6002 and 6144 together with type/reference strains of B. yuanmingense, B. liaoningense, B. iriomotense, B. japonicum, B. betae and B. canariense. Within G-I, subclusters that were not well defined in the 16S rRNA (Fig. 1) tree were now shown (Fig. 3), including the pairs of strains SEMIAs 656-6395 and SEMIAs 6002-6144, with bootstrap supports of 95 and 100%, respectively. The results highlight that these strains deserve further studies, as they may represent new species. The concatenated tree has also detected higher diversity of the strains occupying G-II, with the SEMIAs 6160, 6028, 6053 and 6145 showing similarity with the type strain of B. pachyrhizi. The pair of strains 6148-6154 could also represent a new species. Finally, it is important to mention that in all trees Bradyrhizobium was clearly apart from the other rhizobial genera.

Figure 3.

Evolutionary tree inferred using the Neighbor-Joining method for 22 strains based on concatenated genes (recA, atpD, glnII, gyrB, rpoB). The percentage of replicate trees in which the associated strains clustered together in the bootstrap test (1,000 replicates) are shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. Codon positions included were 1st+2nd+3rd+noncoding. All positions containing gaps and missing data were eliminated from the dataset (Complete deletion option). Phylogenetic analyses were conducted in MEGA4.

Another important observation is that clustering of strains by means of both 16S rRNA and housekeeping genes showed no relation with the host plant. Strains clustered in G-I of the MLSA were isolated from subfamilies Papilionoideae (SEMIAs 656, 6002 and 6144) and Mimosoideae (SEMIA 6395), and G-II also included strains from both subfamilies. Other examples refer to isolates from Vigna, clustered in both G-I (SEMIA 6002) and G-II (SEMIA 662), while isolates from tribe Ingeae were clustered in both G-I (SEMIA 6395) and G-II (SEMIA 6160) of the MLSA (Fig. 3).

DISCUSSION

Bradyrhizobium is an intriguing genus of bacteria encompassing a number of interesting features. First, the genus has been considered as the ancestor of all rhizobia (24, 32, 35, 49), and it has been isolated from a variety of legumes distributed worldwide. However, the great majority of the reports on diversity and genetics of diazotrophic symbiotic bacteria has been performed with fast-growing rhizobia, thus studies with Bradyrhizobium may reveal new insights into the evolution of rhizobia. The second important feature is that Bradyrhizobium has probably originated in the tropical region (24, 32), and indeed bradyrhizobia seem to represent the majority of the isolates from leguminous trees in Brazilian tropical forests (30). As it has been pointed out since the pioneer studies of ribosomal genes, apparently there are many more varieties of rhizobia in tropical and subtropical than in temperate regions (33, 46); therefore, studies on the genetic diversity of Bradyrhizobium may expose a high level of genetic diversity. Finally, the third important feature relies on the reports of high rates of nitrogen fixation related to bacteria belonging to the genus Bradyrhizobium (14), and excellent examples comprise the symbioses with cowpea (Vigna spp.) and soybean (Glycine max). Indeed, the SEMIA strains used in our study are highly effective in fixing nitrogen with their host plants.

Previous studies from our group employing analyses of the 16S rRNA, 23S rRNA, ITS and housekeeping genes have indicated an unexpected genetic diversity of Bradyrhizobium (13, 28, 29). In our study the use of five housekeeping genes in the MLSA approach highlighted a far higher diversity in comparison to the single analysis with the 16S rRNA, clearly indicating putative new species.

DNA-DNA hybridization is still required to define new species (12), but arguments against its obligatory use have been raised, including: high cost and intensive work (5, 44), existence of more accurate approaches (22), doubts about its adequacy (1). MLSA has then been proposed as a more accessible tool for assessing phylogeny and taxonomy of prokaryotes (4, 6, 15, 25). Also in this context, the use of at least four housekeeping genes for the phylogenetic analysis and taxonomic classification of bradyrhizobia has been proposed (29, 34, 39, 40, 48), and was confirmed as a successful approach in our study.

It is also interesting that host specificity was not related to genetic clustering, confirming previous reports from our group (3, 13, 28, 29, 37), and indicating that other genes must be searched aiming at getting a better understanding of the evolution of the symbioses, probably nodulation and nitrogen fixation genes.

Biological nitrogen fixation plays a key role in the environment and agriculture sustainability. The results from our study highlight the high genetic variability associated with Bradyrhizobium microsymbionts of a variety of legumes. Our results confirm that the MLSA approach can represent an important, effective, fast and low-cost strategy to reveal the still poorly known diversity of Bradyrhizobium and certainly other nitrogen-fixing rhizobial species (26, 29, 36, 40, 47). In our study MLSA clearly contributed to a better phylogeny definition, as well as to the identification of new subgroups indicative of new species.