Homocitrate Synthase Genes of Two Wide-Host-Range Bradyrhizobium Strains are Differently Required for Symbiosis Depending on Host Plants

The nifV gene encodes homocitrate synthase, the enzyme that catalyzes the formation of homocitrate, which is essential for arranging the FeMo-cofactor in the catalytic center of nitrogenase. Some host plants, such as Lotus japonicus, supply homocitrate to their symbionts, in this case, Mesorhizobium loti, which lacks nifV. In contrast, Bradyrhizobium ORS285, a symbiont of Aeschynomene cross-inoculation (CI) groups 2 and 3, requires nifV for symbiosis with Aeschynomene species that belong to CI group 3, and some species belonging to CI group 2. However, it currently remains unclear whether rhizobial nifV is required for symbiosis with Aeschynomene species belonging to CI group 1 or with other legumes. We generated nifV-disruption (ΔnifV) mutants of two wide-host-range rhizobia, Bradyrhizobium SUTN9-2 and DOA9, to investigate whether they require nifV for symbiosis. Both ΔnifV mutant strains showed significantly less nitrogenase activity in a free-living state than the respective wild-type strains. The symbiotic phenotypes of SUTN9-2, DOA9, and their ΔnifV mutants were examined with four legumes, Aeschynomene americana, Stylosanthes hamata, Indigofera tinctoria, and Desmodium tortuosum. nifV was required for the efficient symbiosis of SUTN9-2 with A. americana (CI group 1), but not for that of DOA9. SUTN9-2 established symbiosis with all three other legumes; nifV was required for symbiosis with I. tinctoria and D. tortuosum. These results suggest that, in addition to Aeschynomene CI groups 2 and 3, CI group 1 and several other legumes require the rhizobial nifV for symbiosis.

Symbiosis between legumes and rhizobia for nitrogen fixation has been extensively studied for its potential contribution to sustainable agriculture. Rhizobia produce a nodulation factor (Nod-factor, NF) that is known to be important for establishing symbiosis with the host leguminous plant (12). NF is a lipochitooligosaccharide, and its sugar length and modifications are specific to rhizobial species (9). When host plants recognize a NF produced by a compatible rhizobium, they initiate a nodulation program (2,20,21). Although photosynthetic Bradyrhizobium strains lack canonical nod-genes for NF biosynthesis, they are able to establish symbiotic nitrogen fixation with Aeschynomene plants, which is independent of NF (8).
Rhizobia produce a molybdenum nitrogenase for nitrogen fixation. This nitrogenase consists of component 1 (MoFe protein) and component 2 (Fe protein). Component 1 is a heterodimer of NifD and NifK, and contains an iron-molybdenum cofactor (FeMo-co) and P-cluster. Component 2 is a homodimer of NifH and contains an iron-sulfur cluster. Other nif genes related to the biosynthesis and assembly of the FeMo-co, P-cluster, and iron-sulfur cluster are also required for nitrogenase maturation. One of the nitrogenase maturation genes, nifV, encodes homocitrate synthase, which catalyzes the condensation of acetyl-CoA and 2-oxoglutarate; its product, homocitrate, is essential for the positioning of FeMo-co in the catalytic center of nitrogenase (11).
Mesorhizobium loti, which lacks nifV, utilizes homocitrate supplied by its host plant, Lotus japonicus, during symbiosis (10). However, some rhizobia, such as Bradyrhizobium ORS285, Bradyrhizobium ORS278, and Azorhizobium caulinodans, which have their own nifV, may fix nitrogen in their free-living state (25). Strain ORS285, a symbiont of Aeschynomene CI groups 2 and 3, requires the nifV gene for nitrogen fixation in both free-living and symbiotic states with NF-independent Aeschynomene species (CI group 3) (18). To establish symbiosis with NF-dependent Aeschynomene species (CI group 2), the ORS285 requirement for nifV depends on the host plant (18). However, in other rhizobia/legume symbioses, including those with plants in CI group 1, it currently remains unclear whether rhizobial nifV is required.
The non-photosynthetic Bradyrhizobium strains SUTN9-2 and DOA9 have been isolated from rice fields using A. americana (CI group 1) as a trap plant (17,23). These strains have a wide host range and nodulate on not only A. americana, but also on several legume species in other genera (17). Genome sequences are available (17,19,24) and both strains possess nifV. In the present study, we generated nifV-disrupted (ΔnifV) mutants of both SUTN9-2 and DOA9 to investigate whether nifV is required for nitrogenase activity in both the free-living and symbiotic states. As host plants, we selected A. americana and Stylosanthes hamata (Dalbergioid clade, the ancestral clade of the Phaseoloid clade), as well as Indigofera tinctoria and Desmodium tortuosum (Phaseoloid clade, the ancestral clade of the Robinioids clade to which L. japonicus belongs). In the present study, we show that rhizobial nifV was differently required for symbiotic nitrogen fixation in a manner that depended on rhizobium-host plant pairing and the age of the nodule.

Phylogenetic tree of NifV proteins and map of nifV genes
The amino acid sequences of all rhizobial NifV proteins were collected from MicroScope (24) (microbe genome database, http:// www.genoscope.cns.fr/agc/microscope) using homocitrate synthase as the keyword. To collect rhizobial NifV proteins, BLASTP was also performed with NifV of SUTN9-2 as a query sequence. A neighbor-joining phylogenetic tree was generated from a ClustalW2 alignment, with Azotobacter vinelandii NifV as the outgroup. Genetic mapping of the region that includes nifV was also performed in MicroScope for both strains.

Nitrogenase activity under free-living conditions
To estimate the nitrogenase activity of Bradyrhizobium SUTN9-2, DOA9, and their ΔnifV mutants under free-living conditions, Bradyrhizobium strains grown in yeast-mannitol liquid medium were washed with BNM-B medium and resuspended in BNM-B soft agar (0.8% [w/v]) medium at OD 600 =0.15. These suspensions (2 mL) were transferred into 10-mL test tubes (BD Vacutainer, Franklin Lakes, NJ, USA). The tubes were capped, and 0.8 mL of the air in the test tube was replaced with acetylene. After an incubation (at 28°C for 7 d), ethylene, as the product of the acetylene reduction activity (ARA) of nitrogenase, was measured by gas chromatography. In complementation studies, homocitrate was added to the BNM-B soft agar medium at a final concentration of 1 mM.

Plant growth and symbiotic phenotype analyses
The symbiotic phenotypes of Bradyrhizobium SUTN9-2, DOA9, and their ΔnifV mutants were analyzed when paired with four leguminous species, A. americana, S. hamata, I. tinctoria, and D. tortuosum. The seeds of A. americana were surface-sterilized in conc. sulfuric acid for 30 min and washed with sterilized water. Seeds of S. hamata were surface-sterilized in 3% sodium hypochlorite and 0.1% Tween 20 for 5 min and then washed with sterilized water. Seeds of I. tinctoria and D. tortuosum were surface-sterilized in conc. sulfuric acid for 10 min and 0.2% sodium hypochlorite and 0.1% Tween 20 for 40 min and then washed with sterilized water. After surface sterilization, the seeds were transferred onto 0.8% agar plates and germinated at 28°C. Two-day-old seedlings were transferred to the top of a test tube containing vermiculite and BNM (5) liquid medium (BNM liquid medium without vermiculite was used for S. hamata) and grown at 28°C with a 12-h light/dark cycle. After 1 week, each seedling was inoculated with 1 mL of a rhizobial suspension adjusted to an OD 600 =1.0. Plant fresh weight, nodule number, and ARA were measured on day 10 or 21 after the inoculation.

Distribution of the nifV gene in rhizobia
To assess what percentage of rhizobial strains possess the nifV gene, we searched all rhizobia in the MicroScope database using homocitrate synthase as a keyword. We also performed BLASTP with NifV of SUTN9-2 as a query sequence. The keyword search and BLASTP analysis provided the same results. Putative nifV genes were identified in 81 out of 148 Bradyrhizobium strains (55%), in 3 out of 7 Mesorhizobium strains (43%), and in 1 out of 85 Sinorhizobium strains (1%) (Table S1). No putative nifV genes were identified in any of the 30 Rhizobium strains. Overall, 31% of rhizobia in the MicroScope database possess a nifV gene.

nifV and its homologues in SUTN9-2 and DOA9
A sequence analysis of genomic DNA revealed that SUTN9-2 and DOA9 each possess a putative nifV gene; the encoded proteins show 78 and 81% sequence identities, respectively, with the NifV protein of ORS285. We generated a phylogenetic tree of rhizobial NifV protein sequences collected from MicroScope. The predicted NifV proteins of the SUTN9-2 and DOA9 strains, which are non-photosynthetic, were found to be distinct from those of the symbionts of NF-independent Aeschynomene species (clade III) (Fig. 1A). The NifV proteins  supplementary Table S1.
The nifV genes of SUTN9-2 and DOA9 were both found to be clustered with other nif genes (Fig. 1B), as in ORS285. We constructed nifV-disruption (ΔnifV) mutants derived from the SUTN9-2 and DOA9 strains by in-frame replacing the nifV ORF with a cefotaxime-resistant gene. ORS285 contains another gene whose product shares 35% amino acid sequence identity with its own NifV protein. The encoded protein is annotated as 2-isopropylmalate synthase (EC 2.3.3.13) and belongs to the same family as homocitrate synthase (EC 2.3.3.14). SUTN9-2 and DOA9 both carry another gene, annotated as a 2-isopropylmalate synthase (leuA); the encoded proteins share 33 and 34% amino acid sequence identities with the respective NifV proteins. In both strains, the leuA genes were located outside of the nif gene cluster (Fig. 1B).

Acetylene reduction activity under free-living conditions
To verify the function of nifV in SUTN9-2 and DOA9, we measured the ARA to estimate the nitrogenase activities of both strains and their ΔnifV mutants under free-living conditions (Fig. 2). Both strains showed ARA under free-living conditions, although the ARA of SUTN9-2 was markedly lower than that of DOA9. ARA was significantly lower in the ΔnifV strains, but was still detectable in both. Supplementation with exogenous homocitrate restored the ARA of both ΔnifV mutants to the levels of their parent strains (Fig. 2).

Symbiotic phenotypes of ΔnifV mutants with four leguminous species
To investigate whether the deletion of nifV affects symbiosis, four leguminous species were inoculated with SUTN9-2, DOA9, or their ΔnifV mutants.
In comparisons with A. americana plants inoculated with wild-type SUTN9-2, those inoculated with SUTN9-2ΔnifV showed significantly decreased plant growth ( Fig. 3A and B), and ARA also decreased by 36% (Fig. 3C). These plants also had 13% more nodules (Fig. 3D). Comparisons of the nodules induced by SUTN9-2 and SUTN9-2ΔnifV on A. americana showed no marked differences, and the appearance of nodule sections did not significantly differ (Fig. 3E). A. americana plants inoculated with DOA9ΔnifV showed a normal symbiotic phenotype (Fig. 3). No significant differences were detected in plant fresh weights (Fig. 3B).
S. hamata plants inoculated with SUTN9-2ΔnifV showed a normal symbiotic phenotype (Fig. 4). An inoculation with DOA9 induced small nodules (Fig. 4E) with very low nitrogenase activity (Fig. 4C). However, the growth of the inoculated plants was 14% better than that of plants without the inoculation (Fig. 4B). No significant differences were observed between plants inoculated with DOA9 and DOA9ΔnifV (Fig. 4).
The growth of D. tortuosum plants inoculated with SUTN9-2ΔnifV was significantly decreased (Fig. 6A and B), although ARA per plant was significantly higher than in those inoculated with SUTN9-2 (Fig. 6C). The nodules of SUTN9-2 and SUTN9-2ΔnifV showed the same ARA on a per-nodule weight (Fig. 6D). The number of nodules induced by SUTN9-2ΔnifV was significantly higher than that induced by SUTN9-2 (Fig. 6E). The nodules induced by SUTN9-2ΔnifV were pink, the same as the nodules induced by SUTN9-2 (Fig. 6F). Plants of D. tortuosum inoculated with DOA9 showed lower ARA and poorer plant growth than those inoculated with SUTN9-2 (Fig. 6A, B, C, and D). No significant differences were observed between plants inoculated with DOA9 and DOA9ΔnifV (Fig. 6).
Symbiotic phenotype of D. tortuosum inoculated with SUTN9-2ΔnifV on day 10 after the inoculation Since the plants of D. tortuosum inoculated with SUTN9-2ΔnifV showed significantly decreased plant fresh weight, but   Nodule number (plant -1 ) NI 9-2WT 9-2ΔnifV DOA9WT DOA9ΔnifV 9-2WT 9-2ΔnifV DOA9WT DOA9ΔnifV    unaffected ARA per nodule weight 3 weeks after the inoculation (Fig. 6B and D), we hypothesized that the deletion of nifV in SUTN9-2 decreased ARA in younger nodules (10 d after the inoculation) on D. tortuosum, resulting in decreased plant growth 3 weeks after the inoculation. To test this hypothesis, we analyzed the symbiotic phenotypes of D. tortuosum inoculated with SUTN9-2 and SUTN9-2ΔnifV on day 10 after the inoculation. No significant differences were observed in plant fresh weight between plants inoculated with SUTN9-2 and those inoculated with SUTN9-2ΔnifV; however, nodule numbers increased by 15% (Fig. 7A and B). The ARA of SUTN9-2ΔnifV nodules was significantly lower than that of SUTN9-2 nodules (Fig. 7C). Furthermore, the nodules induced by SUTN9-2ΔnifV were paler than those induced by SUTN9-2 (Fig. 7D).
Discussion nifV, one of the genes related to nitrogenase maturation, encodes homocitrate synthase, which is essential for arranging the FeMo-cofactor in the catalytic center of nitrogenase (11). Database searches for nifV revealed that most rhizobia carrying nifV are Bradyrhizobium strains (Fig. 1A and Table S1). We generated nifV-disrupted (ΔnifV) mutants of two wide-host-range Bradyrhizobium strains, SUTN9-2 and DOA9, to investigate how nifV functions in symbiotic nitrogen fixation with four leguminous plants.
We measured the ARA of both strains and their ΔnifV mutants under free-living conditions. The ΔnifV mutants of both strains showed significantly lower, but still detectable ARA (Fig. 2). Exogenous homocitrate restored the ARA of the ΔnifV mutants of both SUTN9-2 and DOA9 (Fig. 2). These results were identical to those reported for a nifV deletion mutant of ORS285 (18), suggesting that nifV is involved in efficient nitrogenase activity under free-living conditions in both SUTN9-2 and DOA9. We have two hypotheses for why the ARA of these ΔnifV mutants did not completely disappear under free-living conditions. The nitrogenase of these ΔnifV mutant strains may retain the ability to reduce acetylene to ethylene by using citrate instead of homocitrate at its catalytic center. Although the diazotrophs Klebsiella pneumoniae and Azotobacter vinelandii use molybdenum nitrogenase, and nifV mutants retain ARA at approximately 80 and 10% of the respective wild-type strains under free-living conditions, the nitrogenases of these mutants reduce dinitrogen poorly; 7% of wild-type K. pneumoniae and 2% of wild-type A. vinelandii (13,15). A crystallographic analysis of the nitrogenase MoFe protein from the nifV mutant of K. pneumoniae showed that citrate is a ligand of FeMo-co in the catalytic center (14). Another possibility is that these strains may synthesize a small amount of homocitrate via other proteins. SUTN9-2, DOA9, and ORS285 possess putative genes for 2-isopropylmaleate synthases that show approximately 33, 34, and 35% amino acid sequence identities, respectively, with their own NifV proteins. These genes may encode enzymes that produce homocitrate in place of the NifV protein. Under free-living conditions, the ARA of SUTN9-2 was markedly lower than that of DOA9 (approximately 1/300 to 1/1,000) (Fig. 2) or of other rhizobia that possess a nifV gene, such as Bradyrhizobium ORS285, ORS278, and A. caulinodans ORS571 (25) even though the ARA of SUTN9-2 nodules was as high as that of DOA9 nodules. This suggests that the regulation of nif genes in SUTN9-2 under free-living conditions differs from that of other rhizobial strains.
We found that the nifV of DOA9 was involved in ARA under free-living conditions, but not in symbiosis with A. americana ( Fig. 2 and 3). These results suggest that the nifV in DOA9 functions for free-living conditions rather than for symbiosis. DOA9 possesses two copies of nifA (master regulator of nif genes), nifD and nifK, which are located on the chromosome and on the plasmid, respectively (19,(24)(25)(26). These two copies of nifADK are functionally redundant for symbiosis with A. americana, while the chromosomal nifADK is a major contributor for nitrogenase activity under free-living conditions (25,26). Hence, we infer that the nifV of DOA9, which is located on the chromosome, cooperates with chromosomal nifADK for nitrogenase activity under free-living conditions. SUTN9-2, not harboring any plasmid, possesses a single copy of nifA, nifD, and nifK on its chromosome (24). We found that the nifV of SUTN9-2 was involved in ARA under free-living conditions (Fig. 2). However, SUTN9-2 showed markedly lower ARA (approximately 1/300 to 1/1,000) under free-living conditions than that of DOA9 (Fig. 2). We also noted that the nifV of SUTN9-2 contributed to symbiosis with three legumes: A. americana, I. tinctoria, and D. turtuosum (Fig. 3, 4, 5, 6, and 7). These results suggest that the nifV of SUTN9-2 functions for symbiosis rather than for free-living conditions in contrast to nifV of DOA9. Plant fresh weight and ARA were lower and nodule numbers were higher in A. americana, I. tinctoria, and D. tortuosum inoculated with SUTN9-2ΔnifV than in those inoculated with wild-type SUTN9-2 ( Fig. 3, 5, 6, and 7). The low ARA of SUTN9-2ΔnifV may affect not only plant growth, but also the number of nodules on host plants. On the other hand, DOA9 showed poorer ARA than SUTN9-2 when associated with S. hamata, I. tinctoria, and D. tortuosum (Fig. 4, 5, and 6). The symbiotic phenotypes of DOA9ΔnifV indicate that the deletion of nifV did not affect symbiosis with all species in the same manner (Fig. 3, 4, 5, and 6), suggesting that DOA9 is naturally incompatible with these three host plant species.
In the case of A. americana (CI group 1), the inoculation with SUTN9-2ΔnifV decreased plant growth, whereas that with DOA9ΔnifV did not ( Fig. 3A and B). These results suggest two possibilities. In DOA9, the homocitrate needed for symbiotic nitrogen fixation with A. americana may be synthesized by 2-isopropylmalate synthase instead of by the homocitrate synthase encoded by nifV. It is also possible that A. americana does not supply sufficient homocitrate for SUTN9-2, but supplies enough for DOA9. In order to distinguish between these two possibilities, it will be necessary to investigate the symbiotic phenotype of DOA9 with the double mutation of nifV and 2-isopropylmalate synthase and to analyze the gene expression of the homocitrate synthase (FEN1 homologue) (10) of A. americana. The species of Aeschynomene CI group 1 will possess the FEN1 homologue because these species may establish symbiosis with some nifV-lacking strains, such as B. japonicum USDA110 (3). However, the FEN1 homologue of A. americana has not yet been identified.
The inoculation with SUTN9-2ΔnifV markedly decreased ARA in D. tortuosum by day 10 after the inoculation (Fig.  7C). Three weeks after the inoculation, plants inoculated with SUTN9-2ΔnifV showed less growth, but higher ARA ( Fig.  6B and C). At this time point, the nodule number of plants inoculated with SUTN9-2ΔnifV was significantly higher than that of those inoculated with SUTN9-2 (Fig. 6D). This higher nodule number may be attributed to the low ARA on day 10 after the inoculation (Fig. 7C), i.e. the low ARA induced the host plant to form more nodules. After the host plant began to supply homocitrate to the newly formed nodules, SUTN9-2ΔnifV nodules may have begun to show an ARA at the same level as SUTN9-2 nodules; this may explain the similar ARA levels observed 3 weeks after the inoculation (Fig. 6D). These results suggest that in young nodules (10 d after the inoculation) formed by SUTN9-2 on D. tortuosum, nifV is required for symbiotic nitrogen fixation. We infer that the low nitrogenase activity of SUTN9-2ΔnifV nodules caused the higher nodule number and lower plant growth observed 3 weeks after the inoculation.
A combined summary of the present results and previous findings is shown in Table 2. We found that the inoculation with SUTN9-2ΔnifV caused poor plant growth in three out of the four host plants. However, plants inoculated with SUTN9-2ΔnifV showed better growth than the non-inoculated control plants of the same species (indicated with + in the light gray cells in Table 2). This pattern is similar to that observed in NF-dependent Aeschynomene (CI group 2)/ORS285 symbiosis, but different from NF-independent Aeschynomene (CI group 3)/ORS285 symbiosis. In some NF-independent Aeschynomene species, an inoculation with ORS285ΔnifV resulted in poor plant growth, similar to non-inoculated control plants of the same species (indicated as ++ in the dark gray cell in Table 2) (18). SUTN9-2 has the canonical nod-genes to synthesize NF and may establish symbiosis with host plants in a NF-dependent manner. Thus, the present results suggest that the requirement of rhizobial nifV for symbiosis is higher in NF-independent symbiosis than in NF-dependent symbiosis, which is consistent with previous findings (18). We showed that rhizobial nifV is required for efficient symbiosis not only with Aeschynomene species of CI groups 2 and 3, but also with A. americana of CI group 1 and two legumes in the Phaseoloid clade. The symbiotic phenotypes of A. americana and D. tortuosum suggest that the requirement for rhizobial nifV depends on the symbiont-host combination, and also on the age of the nodules. These results contribute to our understanding of the mechanisms contributing to and the evolution of symbiotic nitrogen fixation.