|Home | About | Journals | Submit | Contact Us | Français|
Lotus species are forage legumes with potential as pastures in low-fertility and environmentally constrained soils, owing to their high persistence and yield under those conditions. The aim of this work was the characterization of phenetic and genetic diversity of salt-tolerant bacteria able to establish efficient symbiosis with Lotus spp. A total of 180 isolates able to nodulate Lotus corniculatus and Lotus tenuis from two locations in Granada, Spain, were characterized. Molecular identification of the isolates was performed by repetitive extragenic palindromic PCR (REP-PCR) and 16S rRNA, atpD, and recA gene sequence analyses, showing the presence of bacteria related to different species of the genus Mesorhizobium: Mesorhizobium tarimense/Mesorhizobium tianshanense, Mesorhizobium chacoense/Mesorhizobium albiziae, and the recently described species, Mesorhizobium alhagi. No Mesorhizobium loti-like bacteria were found, although most isolates carried nodC and nifH symbiotic genes closely related to those of M. loti, considered the type species of bacteria nodulating Lotus, and other Lotus rhizobia. A significant portion of the isolates showed both high salt tolerance and good symbiotic performance with L. corniculatus, and many behaved like salt-dependent bacteria, showing faster growth and better symbiotic performance when media were supplemented with Na or Ca salts.
Legumes can establish nitrogen-fixing associations with Gram-negative soil bacteria collectively known as rhizobia. Although the symbiotic relationships among rhizobia and many legume species of agricultural importance have been intensively studied, relatively little is known about the symbiotic bacteria of certain plant genera. Lotus is a genus of legumes that includes 125 to 130 species of herbs and small shrubs, mainly distributed in the Northern Hemisphere. Several Lotus species, particularly Lotus corniculatus, Lotus uliginosus, and Lotus tenuis, are used as pasture forage worldwide and are included by phylogenetic studies in the same clade as the model legume Lotus japonicus (4). Until recently, bacteria nodulating Lotus included both intermediate-growing (mesorhizobia) and slow-growing bacteria (12, 16). The mesorhizobia can form effective symbioses with certain Lotus spp. (group I, e.g., L. corniculatus, L. tenuis, or L. japonicus) but form tumor-like structures that do not contain bacteria on L. uliginosus, Lotus subbiflorus, and Lotus angustissimus (group II Lotus spp.) (21, 24). On the other hand, slow-growing strains are usually efficient with Lotus group II species but form no nodules or form inefficient nodules in group I species (12). However, there are rare exceptions to this rule, like strain NZP2037, that can form effective symbioses with both groups of Lotus spp. (23, 25, 28). Furthermore, fast-growing Ensifer meliloti bv. lancerottense strains have been shown to be the symbionts of Lotus lancerottensis but are unable to fix nitrogen with either group I or group II Lotus spp. (19).
No apparent relationship exists between the phylogenetic position of Lotus spp. and the type of rhizobia associated. For instance, L. uliginosus and L. angustissimus, which are efficiently nodulated by the bradyrhizobia, are clustered in the same clade as L. corniculatus, L. tenuis, and L. japonicus (clade B) (4), species associated with mesorhizobia. In contrast L. subbiflorus, usually associated with the same rhizobia as L. uliginosus, is clustered in a different clade.
The narrow-host-range rhizobia associated with L. corniculatus and other Lotus species were initially classified as Rhizobium loti (13). Later, when the genus Mesorhizobium was created, R. loti was reclassified as Mesorhizobium loti (14), which is considered the type species. Besides the expected differences between the moderate- and the slow-growing Lotus rhizobia, large variabilities in nitrogen-fixing effectiveness (23) as well as in total DNA-DNA hybridization (3, 6) and phylogeny (5, 40) have been shown among the “meso-growing” rhizobia strains classified as M. loti, indicating that they do not form a homogeneous group. Indeed, one of the best-characterized strains of M. loti, strain MAFF303099, has been reclassified as Mesorhizobium huakuii biovar loti (35). In fact, diverse rhizobia have recently been reported to establish symbiosis with Lotus group I species. For instance, bacteria belonging to the newly described species Mesorhizobium gobiense and Mesorhizobium tarimense, were isolated from Lotus frondosus and L. tenuis in China (10). Also, rhizobia assigned to different genera (Rhizobium, Mesorhizobium, Agrobacterium, and Aminobacter) have recently been reported as symbionts of L. tenuis in the Salado River Basin in Argentina (7). While these recent reports indicate that bacteria nodulating Lotus spp. are diverse, their symbiotic genes are rather homogeneous. In fact, most isolates from Argentina and China, regardless their taxonomic assignment, had symbiotic genes closely related to M. loti (7, 10).
Soil salinity is a serious and expanding threat to agricultural productivity. Improving crop productivity in saline soils requires selection of well-adapted plant genotypes and, in the case of legumes, highly efficient rhizobial partners adapted to soil conditions. As part of the Euro-South American cooperation project LOTASSA (http://www.lotassa.com/), and aiming to isolate and select for salt-tolerant bacteria able to establish efficient symbiosis with forage Lotus spp., we explored the diversity of Lotus rhizobia in two different locations of Granada province, Spain, where the presence of native Lotus spp. had previously been reported (30).
Soil analyses were carried out at the analytic laboratory of the Consejería de Agricultura y Pesca of the Junta de Andalucía in Atarfe, Granada, Spain, in accordance with the methodology suggested by Junta de Andalucía order 5/12/1975 and referring to that published by the Spanish Ministry of Agriculture and Fishery (MAPA) (20). Soil samples were air-dried, crushed to pass through a 2-mm sieve, and then ground to <60-μm-diameter particles. Texture was determined by the pipette method, and pH (in water) was measured with a glass electrode using a 1:2.5 sample/water ratio. Total organic matter content was determined using the Walkley-Black wet-dichromate-oxidation method, and total N was determined by the macro-Kjeldahl digestion method (20).
Soil and/or nodule samples were collected in a semiarid region located in the Guadiciano-Bastetano district of the northeastern part of Granada province, Spain, from two different locations, called El Margen and Galera, separated some 20 km from each other, where the presence of native Lotus spp. had previously been reported (30). In El Margen (universal transverse mercator [UTM] coordinates: X 0537333, Y 4166640), in an agriculturally undisturbed site (code M), next to a salty pool that was dry at the time of sampling (April 2006), we found several native Lotus corniculatus subsp. crassifolius plants from which a few nodulated plants as well as root-surrounding soil were collected. Adjacent to this site (less than 10 m away), there was an agricultural field (site code P) from which several bulk soil samples were also collected. In Galera (UTM coordinates: X 0538009, Y 4177093), several patches of wild Lotus maritimus were found growing on the bank of the river, and plants and root-surrounding soil were collected (site code G). Ten samples of soil per site were collected with a T collector (30 cm long and 3 cm wide). Soil samples were mixed to generate a single one for each location and site. Ten grams of soil was resuspended in 100 ml of sterile distilled water and used to inoculate seedlings of trap plants L. uliginosus LE627, Lotus tenuis cv. Pampa INTA, and Lotus corniculatus cv. San Gabriel. Nodules were carefully detached from roots, surface sterilized with a 0.25% solution of HgCl2, washed extensively with sterile water, and crushed. A drop sample was immediately plated on yeast extract-mannitol (YEM) agar medium (38), tryptone-yeast extract-calcium chloride (TY) agar medium (38), or TY agar medium supplemented with 0.25% (17 mM) of CaCl2 (the total content of CaCl2 was 0.3%, or 20.4 mM) or 0.1% (5.84 mM) SO4Ca and incubated at 28°C until growth. The isolates and their characteristics are listed in Tables S1, S2, and S3 in the supplemental material.
Tolerance to sodium chloride (NaCl) and calcium chloride (CaCl2) was tested by placing drops of serially diluted cell suspensions (containing approximately 106 to 103 cells) of each strain on YEM agar or TY agar plates supplemented with increasing amounts of NaCl (0 to 0.5 M) or CaCl2 (0 to 1%). Plates were incubated at 28°C, and growth was visually determined and periodically recorded for 10 days, as described previously (22). To better understand the growth requirements, several NaCl-dependent isolates and one Ca-dependent isolate were grown for 3 to 4 days in TY medium supplemented with 0.25% (17 mM) CaCl2 and diluted in fresh TY medium and TY medium supplemented with NaCl (150 to 300 mM) or with 150 mM NaCl plus 0.25% CaCl2. The growth was monitored every 2 h with a BioScreen C apparatus (LabSystems, Helsinki, Finland). Strains unable to grow with 150 mM NaCl were considered salt sensitive (SS), those that grew with up to 150 mM NaCl were considered to have low salt tolerance (LST), those that grew with up to 300 mM NaCl were considered to have moderate salt tolerance (MST), and those that grew with 400 mM or higher concentrations of NaCl were considered to have high salt tolerance (HST).
pH tolerance was tested similarly to salt tolerance, but the pH of YEM or TY agar medium was adjusted to pH 6 or pH 8 and buffered with MES (morpholineethanesulfonic acid [5 mM]; Sigma Chemical Co., St. Louis, MO). When necessary, TY medium was supplemented with 0.2% (13.6 mM) CaCl2. The data regarding salt and pH tolerance were subjected to statistical one-way analysis of variance (ANOVA), followed by Duncan's test. Since the distribution of means did not follow a normal distribution and the number of isolates from each location was not the same, a nonparametric alternative Kruskal-Wallis test was used. Statistical significance was judged at P values of <0.05 using the program Statistica 6.1 (StatSoft, Tulsa, OK).
The host range and effectiveness of rhizobial isolates were tested on L. uliginosus LE627, L. tenuis cv. Pampa INTA and L. corniculatus cv. San Gabriel. Seeds from Lotus species were soaked in concentrated H2SO4 for 5 min, rinsed extensively with sterile distilled water, and then surface sterilized after immersion in 0.5% sodium hypochlorite for 5 min. The seeds were germinated on 0.8% water agar plates for 1 or 2 days in the dark at room temperature.
Lotus seedlings were transferred to 100-cm2 plates containing 50 ml Rigaud and Puppo nutrient agar (27) and roots inoculated with 20 μl bacterial suspensions containing approximately 106 cells. Plates were covered with opaque paper to prevent direct root exposure to light. Plants were grown under controlled conditions in a growth chamber, with a 16/8-h light/dark photoperiod at 23/18°C for 46 days. Apparent nitrogen-fixing efficiency was visually determined by comparing the sizes and aspects of aerial parts of plants with those inoculated with commercial strain NZP2037 or NZP2309.
Bacterial genomic DNAs were isolated and purified by following the method described earlier (7). Repetitive extragenic palindromic PCR (REP-PCR) genomic fingerprints were generated with primers REP1R and REP2I, as previously reported (37). Comparative analysis of electrophoretic REP fingerprints was performed with InfoQuest FP from Bio-Rad using Pearson's product-moment correlation analysis. Similarity matrices were clustered using the unweighted-pair group method with averages (UPGMA) algorithm (31) and the highest level of overall similarity. Gel normalization, background subtraction, and zone definition were performed as previously described (26).
For 16S rRNA (rrs) gene amplification, primers 41f and 1488r were used, with cycling parameters and reactions as described previously (11). Products were digested with endonucleases HinfI and MspI, provided by Fermentas, and the restriction fragment length polymorphism (RFLP) patterns resolved by 3% (wt/vol) agarose gel electrophoresis (Pronadisa, Spain) for 2 h at 80 mV. Comparative analysis of the restriction patterns was performed with InfoQuest FP from Bio-Rad as described above.
Amplifications of nodC and nifH gene fragments were done using primers nodCFI, nodCF2, and nodCFu for nodC and primers nifHI and nifHF for nifH in accordance with procedures previously described (18). Amplification of the atpD and recA genes was done using primers atpD273, atpD294, and atpD771 for atpD and recA6 and recA555 for recA, as described previously (9).
PCR products were purified from agarose gels after electrophoresis, using a QUIAEX kit (Qiagen). Sequencing was performed with an ABI373 automated sequencer. DNA sequence edition and analyses were performed with the Chromas 1.45 (Technelysium Pty., Ltd.) software package.
Multiple nucleotide sequence alignments of the 16S rRNA, nifH, nodC, atpD, and recA genes were generated using Clustal W (34) and optimized manually. Phylogenetic and molecular evolutionary analyses were conducted using Phylip (phylogeny inference package) and MEGA version 4 (32) and performing bootstrap (BS) (8) and neighbor-joining (NJ) analyses (29), with distances estimated using the Jukes-Cantor and Kimura two-parameter tests, with 1,000 bootstrap replications. Maximum composite likelihood (33) analysis was performed using the two different software implementations.
The sequences have been submitted to the EMBL database under accession numbers FM203301 to FM203307 and FN256290 for 16S rRNA genes, FM203308 to FM203313 for atpD, FN556458 to FN556463 for recA, FM203314 to FM203325 and FN397444 to FN397446 for nodC, and FM203326 to FM203337 and FN397447 to FN397449 for nifH gene sequences.
This study aimed to isolate and characterize salt-tolerant bacteria able to establish symbiotic associations with forage Lotus spp. in two locations in the province of Granada, Spain, where the presence of native Lotus spp. had previously been reported (30). Soils from the first location, El Margen, were classified as gypsyric regosols with a silty-loam texture affected with calcium sulfate precipitations (gypsum), composed of 10.96% sand, 62.34% silt, and 26.7% clay, with a pH in water of 8.0 and a carbonate content of 31 to 36%. The two soils (M [noncultivated or native] and P [cultivated]) sampled in this location were similar in texture and composition but differed in electrical conductivity (4.5 mS/cm for the native soil and 2.2 for the cultivated one). Also, small differences were found between the cultivated and native soils from this location, particularly in organic matter (5.41% for the native soil and 7.16% for the cultivated soil) and total nitrogen (0.31% for the native soil and 0.43% for the cultivated one). Soil from the second location of isolation, Galera (G), was classified as calcaric fluvisol, with a silty texture (43.31% sand, 32.24% silt, and 24.45% clay), a conductivity value of 1.1 mS/cm, a pH in water of 8.2, a carbonate content of 50%, 2.46% organic matter, and 0.14% total N.
Bacteria were isolated mostly from nodules of trap plants grown in the laboratory but also from native, field-grown nodulated plants. L. uliginosus trap plants inoculated with each of the three soil samples formed many nodule-like structures from which no bacteria could be isolated, indicating the absence of group II Lotus rhizobia in these soils. In contrast, L. tenuis and L. corniculatus plant roots inoculated with any soil sample were always abundantly nodulated. YEM medium was the growth medium of choice for bacterial isolation, and whenever available, 30 nodules from each plant-soil combination were randomly taken and individually crushed. From Galera (G), we obtained totals of 28 L. corniculatus isolates (CGS isolates), 28 isolates from L. tenuis (GGS isolates), and 22 isolates from field L. maritimus root nodules (GA isolates). From the cultivated soil in El Margen (P), we could recover bacteria from only 15 out of 30 nodules from L. corniculatus and from 17 out of 30 nodules from L. tenuis trap plants, named CPS and GPS, respectively. Surprisingly, no rhizobia could be isolated from the noncultivated soil in El Margen (M) when YEM was the medium for bacterial growth. When TY medium was used, bacteria could be isolated from only 2 nodules from native L. corniculatus subsp. crassifolius plants (M isolates) and from only 8 nodules from L. corniculatus (CMS isolates) or L. tenuis (GMS isolates) trap plants. The success of bacterial isolation was very significantly enhanced when the TY medium was supplemented with 0.25% CaCl2, and thus, we could recover 31 isolates out of 31 nodules from L. tenuis and 29 isolates out of 30 nodules from L. corniculatus plants inoculated with the noncultivated soil from El Margen. A total of 180 isolates were used for further characterization.
REP-PCR was performed to discriminate siblings among isolates and to obtain a first picture of existing diversity. High-resolution REP-PCR fingerprints of the 180 isolates were generated. Isolate CPS20 did not produce a PCR amplification profile, while the remaining 179 isolates were resolved in 120 distinctive profiles after Pearson/UPGMA analysis with a conservative threshold level of 98%. Sibling isolates were disregarded for further analyses.
REP-PCR analysis of the 120 isolates revealed a relatively high level of genetic diversity among isolates, with degrees of relatedness ranging from 20% to 97% and REP-PCR patterns ranging from very complex to very simple (see Fig. S1 in the supplemental material). Although certain subgroups of isolates were exclusive or predominant in only one of the sampled locations, we could find closely related isolates in both locations, which is not surprising, given their proximity (20 km apart). All isolates with distinctive REP-PCR profiles were used for further phenotypic and genetic characterizations.
Isolates from Galera ranged from salt sensitive (SS) to moderately salt tolerant (MST), with 50% (23/46) of the isolates unable to grow on YEM medium supplemented with 150 mM NaCl (SS), 7% (3/46) that tolerated up to 150 mM NaCl (LST), 41.3% (19/46) that tolerated up to 300 mM NaCl (MST), and only one isolate that tolerated up to 500 mM NaCl (HST) (Fig. (Fig.11 A; see also Table S1 in the supplemental material). Of the isolates from the cultivated soil El Margen (P), 54.2% (13/24) tolerated up to 300 mM NaCl (MST), 37.5% (9/24) tolerated up to 150 mM NaCl (LST), and 8.3% (2/24) of the isolates were considered SS (Fig. (Fig.1A;1A; see also Table S2 in the supplemental material). On the other hand, 78% (39/50) of isolates from the noncultivated soil of El Margen (M) were HST, able to grow on TY medium supplemented with 400 mM NaCl; 20% (10/50) tolerated up to 300 mM NaCl (MST); and 2% (1/50) of the isolates tolerated up to 150 mM NaCl (LST) (Fig. (Fig.1A;1A; see also Table S3 in the supplemental material). None of the isolates from soil M were considered SS. As a conclusion, 50% of isolates from soil G but only 8.3% from P were considered salt sensitive, unable to grow with 150 mM NaCl, in contrast to isolates from El Margen native soil, which were all able to grow with 150 mM or higher concentrations of NaCl. In fact, 78% of the isolates from soil M were considered HST (grew with up to 400 mM) while only 2.2% from Galera tolerated salt concentrations higher than 300 mM. A statistically significant correlation between NaCl tolerance or sensitivity and soil of origin was found (P < 0.05) (Fig. (Fig.1A1A).
Concerning Ca2Cl tolerance tests, tolerance distribution among soils was similar to NaCl tolerance distribution. Thus, most bacteria from native soil from El Margen (M), but only one isolate from Galera, grew in TY medium supplemented with 1% Ca2Cl (see Tables S1, S2, and S3 in the supplemental material), and significant statistical differences were also found between the two soils. Interestingly, almost 92% of the isolates from El Margen (M) behaved like salt-dependent bacteria, requiring a salt-enriched medium for optimal growth, and therefore needed extra addition of CaCl2 and/or NaCl. Some 8% were Ca dependent, i.e., were unable to grow on TY medium supplemented with NaCl but grew when an extra 0.25% CaCl2 was added (see Table S4 in the supplemental material). We could determine that most salt-dependent isolates showed growth on TY agar only after very prolonged incubation and that in liquid cultures the doubling times were significantly higher in media without extra salt supplementation. For instance, strain CMSS27 had a generation time of 20 h in liquid TY medium, which was reduced to 8 h in TY medium with 0.25% CaCl2 and to 6 h in TY medium supplemented with both 0.25% CaCl2 and 150 mM NaCl. Although REP-PCR analysis established distinctive fingerprints, with a relatively high degree of relatedness among isolates from both locations, El Margen and Galera, there were clear physiological differences among isolates from the two locations. Soil from El Margen contained certain bacterial types which could be classified as salt-dependent rhizobia, which were difficult to isolate in standard media because of a requirement for NaCl and/or CaCl2 for optimal growth (see Table S4 in the supplemental material). Despite this behavior, none of the isolates seemed to tolerate salt concentrations above 400 mM. The behavior of these isolates resembled that of the category 2 group of rhizobia nodulating mesquite in the Sonoran Desert (15), which showed increased grow rates with 100 mM and even 300 mM NaCl. The higher salinity of soils from El Margen could explain why most rhizobial strains were more tolerant to salt and so well adapted to these conditions that 92% of the isolates from noncultivated soil needed moderate saline conditions to grow.
All isolates were able to grow at neutral pH, but a contrasting situation between salt tolerance and pH tolerance was observed. Most isolates from Galera tolerated a rather wide pH range, since 70% (30/43) of the isolates were able to grow at both pH 6 and pH 8 (see Table S1 in the supplemental material). In contrast, 50% (12/24) (see Table S2 in the supplemental material) of the isolates from the agricultural soil of El Margen (P) and only 37.5% (18/48) (see Table S3 in the supplemental material) of isolates from El Margen native soil (M) were able to grow at such range of pH. Despite these differences, there was no statistically significant correlation between pH range tolerance and soil of origin (P < 0.05) (Fig. (Fig.1B1B).
Almost-full-length 16S rRNA genes from the 121 isolates showing unique REP fingerprints, plus the reference strain M. loti NZP2311, were amplified. Gel electrophoresis of undigested PCR products revealed that all strains produced a single band of about 1500 bp as expected for the 16S rRNA genes of most members of the Rhizobiaceae (17). PCR products were separately digested with MspI and HinfI, which were selected as the best endonucleases for distinguishing various mesorhizobial species (7). Four different 16S rRNA gene RFLP groups were distinguished (see Tables S1, S2, and S3 in the supplemental material). Ribogroup I included 92% of the population analyzed and was very similar to ribogroup II, which included only the type strain M. loti NZP2213. Ribogroups III and IV represented 8% of the isolates.
Partial sequences of amplified 16S rRNA, atpD, and recA genes from representative isolates of each ribogroup were obtained. The sequences of the amplified 16S rRNA genes of ribogroup I isolates GAC133P, GAC138P, and GA13P were identical and had a single-nucleotide mismatch with regard to that of GA12P. Those 16S rRNA gene sequences displayed 100% identity with corresponding sequences from M. tarimenseT and Mesorhizobium tianshanense strain ST2-Setubal and a single mismatch (99.93% sequence identity) with regard to strain M. tianshanenseT USDA 3592 and therefore were grouped in a well-supported branch with these Mesorhizobium species (Fig. (Fig.22).
The atpD genes of isolates GA138P and GA133P shared 98.5% sequence identity and had M. tarimense as the closest relative. Sequences from GAC133P and GAC138P were located in a separated branch inside a poorly resolved cluster that included the M. tarimense, M. mediterraneum, M. temperatum, M. gobiense, and M. tianshanense type strains (Fig. (Fig.33 A). recA phylogeny provided similar results, albeit GA138P appeared closer to M. tarimense and M. gobiense, whereas GA133P was in the same subbranch (99% identity) as M. tianshanense (Fig. (Fig.3B).3B). Thus, ribogroup I isolates seem closely related to M. tarimense, a recently described species that includes bacteria isolated from Xinjiang desert soils (China) which have Lotus frondosus or L. tenuis as host plants but can also nodulate other legumes, like L. corniculatus, Glycyrrhiza uralensis, Oxytropis glabra, and Robinia pseudoacacia (10). As in this study, M. tarimense strains were originally isolated from arid environments, although M. tarimense strains were described as rather salt-sensitive bacteria since they did not grow on YEM medium in the presence of 1% (approximately 180 mM) NaCl. M. tianshanenseT strains were also originally isolated from an arid region of China and were also described as bacteria with low salt tolerance (1). Among our ribotype I isolates, a low portion of bacteria were salt sensitive, unable to grow in the presence of 150 mM NaCl. However, many others, particularly those from El Margen, tolerated up to 400 mM NaCl. The reported optimum pH for M. tarimense and M. tianshanenseT was between 6 and 8 (1, 10), but we found great variability regarding the pH tolerance of our ribogroup I isolates.
Ribotype III was represented by only five isolates with moderate salt tolerance and an inability to grow at pH 6. These strains were recovered from El Margen and Galera soils when either L. corniculatus or L. tenuis was used as a trap plant. The 16S rRNA gene sequences of strains CGS22 and CPS13 were 100% identical but had no clear match with any Mesorhizobium type species (Fig. (Fig.2).2). Over the total sequence obtained (1,368 nucleotides [nt]), the most closely related strain was M. chacoenseT (10 mismatches, with 99.3% identity) (36). However, in the phylogenetic tree shown in Fig. Fig.2,2, where only 1,291 nt were considered, CGS22 and CPS13 were closer to M. albiziaeT (39). The 16S rRNA gene sequences of M. albiziae and M. chacoense are more divergent at the 5′ end, which would explain this discordance.
The atpD gene sequences from isolates CPS13 and CGS22 shared 96% identity and clustered in a well-resolved branch with M. albiziae and M. chacoense (Fig. (Fig.3A).3A). Alignments over the complete sequence showed that atpD from isolate CGS22 had 94.2% similarity with M. chacoense and 94.6% similarity with M. albiziae, while that of strain CPS13 shared 96.4% similarity with M. chacoense and 95.5% similarity with M. albiziae. The recA gene sequences of CPS13 and CGS22 shared 98.8% identity and identities ranging 89.3 to 90.7% with both M. albiziae and M. chacoense. In the recA tree, these two isolates were also clustered with M. chacoense and M. albiziae. Although both CPS13 and CGS22 appeared closer to M. albiziae, the BS values below 50 did not support this result (Fig. (Fig.3B).3B). Thus, further studies would be necessary to determine if ribotype III isolates are phylogenetically closer to M. albiziae or to M. chacoense.
The 16S rRNA gene sequences of strains CMSS27 and CMSS1 (ribogroup IV) had the highest similarities (99.9% and 100%, respectively) with Mesorhizobium alhagi (Fig. (Fig.2),2), a species that so far includes 11 strains isolated from Alhagi sparsifolia in China (2). The results for the recA phylogenetic analysis of strains CMSS27 and CMSS1 correlated well with the 16S phylogenies and placed these two isolates very close to M. alhagi (Fig. (Fig.3B).3B). In contrast, the atpD genes from isolates CMSS27 and CMSS1 shared 95.8% sequence similarity, and their closest relative was Aminobacter aminovorans STM2150T and A. aminovorans strain 135, isolated from L. tenuis in Argentina (7). The atpD sequences from these two isolates displayed 92.7% and 93.2% identity with that of A. aminovorans. Among Mesorhizobium species, the M. chacoense atpD gene sequence was the most similar to CMSS27 and CMSS1, displaying only 82.15% and 80.4% sequence identity, respectively. The clustering of CMSS27 and CMSS1 with A. aminovorans evidenced a discordant phylogeny between the atpD gene and the rrs and recA genes (Fig. (Fig.22 and and3).3). In the description of M. alhagi, several housekeeping genes were used as phylogenetic markers, but atpD was not included and therefore was not available for this study (2).
The results showed strong relationships between all isolates and the genus Mesorhizobium (Fig. (Fig.2).2). Therefore, three clearly differentiated 16S RFLP groups were identified: ribogroup I was closely related to the species M. tarimense and M. tianshanense, ribogroup III was related to either M. albiziae or M. chacoense, and ribogroup IV was closely related to M. alhagi. It is worth noting that ribogroup IV isolates, like CMSS27, were recovered only from L. corniculatus trap plant nodules inoculated with the noncultivated soil from El Margen (M) but that ribotypes I and III were recovered from either field-grown or trap plant hosts and found in the two locations explored in this work.
A great majority of the isolates were able to nodulate both L. corniculatus cv. San Gabriel and L. tenuis cv. Pampa INTA, but none of them formed nitrogen-fixing nodules on L. uliginosus LE627, suggesting that these isolates are narrow-host-range Lotus rhizobia.
Seventy-two percent of the isolates from L. tenuis and L. corniculatus trap plants were efficient in nitrogen fixation, 13% were moderately efficient, 13% were inefficient, and 2% were unable to nodulate L. corniculatus cv. San Gabriel (see Tables S1, S2, and S3 in the supplemental material). Likewise, bacteria isolated from L. maritimus and L. corniculatus subsp. crassifolius field nodules were infective and efficient on L. corniculatus plants, with the exceptions of GA16P, GA25P, and GA27P, which did not form nodules, and strains M75P and 133P, which were only moderate fixers compared to the commercial strains used as controls. Five isolates induced nodules in L. uliginosus but did not fix nitrogen. As an unusual phenomenon, at least 59% (27/46) of the salt-dependent bacteria needed an extra addition of 0.012% of CaSO4 to the plant growth medium to nodulate and/or to show efficient nitrogen fixation with L. corniculatus (see Table S4 in the supplemental material).
The sequences of the symbiotic genes nodC and nifH genes were obtained for some 15 isolates, representing all ribotypes, Lotus hosts, and soils of origin. In contrast to what was found for housekeeping genes, the level of diversity of symbiotic genes was not so high; the similarities among the nodC sequences of our isolates ranged from 91 to 100%. Isolate CMSS27 harbored the more distant nodC sequence, ranging 91% to 93% identity with the other isolates and displaying 93% identity with M. loti strains R7A, NZP2213, and MAF303099 and only 90% identity with the broad-host-range commercial strain M. loti NZP2037. As shown in Fig. Fig.4,4, nodC from CMSS27 was placed in a clade with previously reported Lotus symbionts, but in a well-differentiated branch, suggesting that it could represent a novel symbiotype, as probably does the broad-host-range strain NZP2037. The remaining isolates clustered together with nodC from narrow-host-range Lotus symbionts, including reference M. loti strains. Isolates GMSS11, CPS20, CPS13, and M75P (all from El Margen) were located in the same branch as Mesorhizobium loti reference strains R7A, NZP2213T, and MAFF303099 and M. tarimenseT (Fig. (Fig.4).4). The identities among the nodC genes included in the same branch as CGS2 ranged from 95% to 99%, but those among the nodC genes from the rest of the isolates from this work and from the reference M. loti strains were only 91% to 95%. In an intermediate position between the two narrow-host-range Lotus rhizobial branches was nodC from GGS10, together with nodC from BA151, a strain isolated from L. tenuis in the Salado river basin in Argentina (7).
The nifH genes from most of the isolates were clustered together with the nifH genes from reference M. loti and M. tarimense strains, with similarity values ranging from 97 to 99%. However, nifH from strains CMSS27 and GGS20 (sharing 99% identity) were located in a well-resolved branch together with the M. albiziae nifH gene (see Fig. S2 in the supplemental material). Noteworthy is the fact that several strains from ribogroup IV were not able to fix nitrogen in symbiosis with L. corniculatus, which could suggest that this legume species is not their primary host.
Although typical symbiotic genes of M. loti could be found among the Spanish isolates, no M. loti-like bacteria were identified. This resembles the case study reported in China (10) and also the report by Estrella and coworkers (7), who found that M. loti-like rhizobia were rather infrequent in Argentinian soils. A common situation among these two previous reports and the present study is that Lotus rhizobia were isolated from stressed environments. Thus, it is possible that M. loti strains may not be sufficiently adapted to certain environments where other bacteria represent the most-abundant Lotus symbionts. On the other hand, the presence of related symbiotic genes among distinct species in geographically unrelated locations adds further evidence to the capacity of Mesorhizobium symbiosis islands for horizontal transfer.
All our isolates appeared to be capable of nodulating Lotus spp. belonging to group I, like L. corniculatus and L. tenuis, but not species belonging to group II, like L. uliginosus. It seems that the soils explored did not contain bacteria compatible with the latter host, as no rhizobia could be isolated when L. uliginosus was used as a trap plant. A significant portion of the isolates showed significant salt tolerance and good symbiotic performance with L. corniculatus, and thus, they are promising candidates for further selection of effective inoculants for L. corniculatus and L. tenuis growing in saline environments.
This work was supported by the EU-INCO project LOTASSA and a Junta de Andalucia (Spain) LOTASSA support grant.
We thank M. Rebuffo and M. J. Estrella for supplying Lotus seeds.
Published ahead of print on 30 April 2010.
†Supplemental material for this article may be found at http://aem.asm.org/.