ABSTRACT
The purB and purH mutants of Mesorhizobium loti exhibited purine auxotrophy and nodulation deficiency on Lotus japonicus. In the presence of adenine, only the purH mutant induced nodule formation and the purB mutant produced few infection threads, suggesting that 5-aminoimidazole-4-carboxamide ribonucleotide biosynthesis catalyzed by PurB is required for the establishment of symbiosis.
Nitrogen-fixing nodules are formed as a consequence of a series of signal exchanges between rhizobia and their leguminous plant hosts. Flavonoids, which are plant signal molecules, induce the expression of nodulation genes in the rhizobia, resulting in the production of host-specific lipooligosaccharide signal molecules called Nod factors (7). The Nod factors in turn induce root hair curling and cortical cell division in the host roots and the formation of infection threads (12, 20). The rhizobia then enter and travel down the root hairs, are released into the cortical cells, and differentiate into bacteroids that fix nitrogen.
Most purine auxotrophs of Rhizobium species are defective in symbiosis due to the abortion of the infection process or the formation of pseudonodules that lack bacteria. For example, a purF mutant of Rhizobium etli CFN42 bv. phaseoli, CE106, elicits pseudonodules on Phaseolus vulgaris (17). RL106 and RL110, purF and purY mutants of Rhizobium leguminosarum bv. viciae 128C56, induce small bumps lacking bacteria on Pisum sativum (16). purQ and purL mutants of Sinorhizobium (basonym, Rhizobium) fredii HH103 induce small bumps or pseudonodules on Glycine max (4, 16). ANU2861 (purM::Tn5) and ANU2866 (purY:: Tn5), mutants of the broad-host-range Rhizobium strain NGR234, are defective in nodule formation on a wide variety of legumes. These mutants elicit root hair curling and nodule meristem initiation on Macroptilium atropurpureum cv. Siratro but fail to induce full nodule formation (9, 10, 24). A possible exception has been observed in the symbiosis between Sinorhizobium meliloti and Medicago sativa: pur mutants of S. meliloti form nodules with infection threads and infected cells, although they do not fix nitrogen (Fix−) (8, 22).
The nodulation deficiency of RL106, which is a purF mutant of R. leguminosarum bv. viciae, was partially reversed by the addition of inosine or adenine to the root environment, although the nodules contained few or no bacteria (16). In many other cases, however, despite the recovery of growth, the defect in symbiosis of purine auxotrophs was not restored by the addition of purine (17, 18). By contrast, it has been reported that 5-aminoimidazole-4-carboxamide (AICA) riboside, which is a nonphosphorylated derivative of the purine biosynthetic intermediate AICA ribonucleotide (AICAR), promoted the symbiosis of purine mutants. Similar findings have been reported in several other cases (10, 15, 16, 17).
At the outset of the current study, we noted that a deletion mutant of Mesorhizobium loti exhibited purine auxotrophy and a nonnodulating (Nod−) phenotype on the model legume Lotus japonicus. However, unlike other purine auxotrophs, the nodulation deficiency of the mutant was not restored by the application of AICA riboside. This suggested that AICA riboside does not participate in promoting nodulation in the association between M. loti and L. japonicus. However, we could not exclude an alternative explanation—namely, that M. loti is incapable of utilizing AICA riboside due to defective uptake or lack of conversion to AICAR. In the present study, we constructed purB and purH mutants, which were blocked prior to and after AICAR production, respectively. We examined their capacities for infection and nodule development to explore the involvement of AICAR in nodulation.
Genetic and nutritional characterization of deletion mutant DM002S.
We carried out systematic gene disruption experiments on M. loti MAFF303099 using an ordered cosmid library, in order to identify the genes involved in symbiosis (13). One of the mutants, DM002S, which lacked a 21.9-kbp chromosomal region encompassing the RhizoBase (http://bacteria.kazusa.or.jp/rhizobase/) genome coordinates 53204 and 75103, exhibited purine auxotrophy and a Nod− phenotype on L. japonicus B-129. Genetic and nutritional analysis revealed that an open reading frame with locus tag mll0079, which was similar to the purB gene of Escherichia coli and other bacteria, is responsible for purine auxotrophy and the Nod− phenotype of DM002S. purB encodes adenylosuccinate lyase (EC 4.3.2.2), which catalyzes two well-separated steps in the purine biosynthesis pathway: the conversion of 5-phosphoribosyl-5-aminoimidazole-4-N-succinocarboxamide to AICAR and the conversion of adenylosuccinate (sAMP) to AMP (Fig. 1). However, unlike other purine auxotrophs, the nodulation deficiency of the mutant was not restored by the application of AICA riboside.
Purine biosynthetic pathway in M. loti. Gene annotations and reaction steps are based on information from the Kyoto Encyclopedia of Genes and Genomes (KEGG; http://www.genome.ad.jp/kegg/). Genes are indicated in italics. PRPP, 5-phosphoribosyl pyrophosphate; AICAr, AICA riboside; FAICAR, 5-phosphoribosyl-5-formaminoimidazole-4-carboxamide; apt, adenine phosphoribosyltransferase gene; hpt, hypoxanthine phosphoribosyltransferase gene; xpt, xanthine phosphoribosyltransferase gene; ade, adenine deaminase gene; deoD, purine nucleoside phosphorylase gene; add, adenosine deaminase gene.
Construction of purB and purH mutants.
In order to directly address the involvement of AICAR in nodulation, we constructed purB and purH mutants which were blocked prior to and after AICAR production, respectively. A purB mutant (ML0079) was constructed by a modification of the one-step method described by Baba et al. and Datsenko and Wanner (1, 6). First, the levan sucrase-spectinomycin adenylyltransferase gene cassette (sacB-aadA) was amplified from pKS808 (13) with primers Pm4-FRT-sacB and Pm1-TRF-Adaa (Table 1) and then cloned into the pCR2.1-TOPO vector yielding template plasmid pKST001. The sacB-aadA gene cassette was further amplified from pKST001 using the primers mll0079U and mll0079L (Table 1). The resultant linear PCR products, containing short sequences homologous to mll0079 flanking the sacB-aadA gene cassette, were introduced by electroporation into E. coli DH10B harboring both cosmid 2 and plasmid pKD46 expressing lambda red recombinase; in this system, the products were recombined into the cosmid by the plasmid-encoded red recombinase. The cosmid derivative with deletion-insertion loci was confirmed by restriction digestion and PCR and transferred to M. loti MAFF303099 by conjugation. The deletion mutant ML0079 was obtained as a result of homologous recombination. The purH mutant ML4101 (Δmlr4101) was constructed similarly using cosmid 165 and the primers mlr4101U and mlr4101L (Table 1). Both mutants were confirmed by PCR and Southern blot hybridization.
Bacterial strains, plasmids, and oligonucleotides used in this study
Both ML0079 and ML4101 failed to grow in B− minimal medium (23), and growth was restored by 0.1 mM adenine or adenosine (Table 2). Inosine restored the growth of ML4101 but not of ML0079 or DM002S, possibly due to the defective conversion of sAMP to AMP in ML0079, which is mediated by PurB (Fig. 1). Plant assays were carried out using agar slants made with B&D nitrogen-free medium (3) or a plant box (CUL-JAR300; Iwaki, Tokyo, Japan) containing sterile vermiculite watered with B&D nitrogen-free medium. Seed sterilization, germination, and plant cultivation were carried out as described previously (13). Evaluation of the numbers of bacteria colonized on the L. japonicus roots revealed poor colonization of both ML0079 and ML4101 without adenine supplementation: colonization of the root surface was 1,000-fold lower than that of wild-type MAFF303099 (Fig. 2A). With 0.1 mM or 0.5 mM adenine supplementation, colonization of the mutants on the associated roots was restored to the same level as that of the wild-type strain (Fig. 2A).
Colonization, infection, and nodulation properties of wild-type and purine mutants of M. loti MAFF303099 in the presence and absence of adenine. Plants inoculated with the wild-type M. loti strain MAFF303099 are indicated by solid bars, those inoculated with ML0079 are indicated by open bars, and those inoculated with ML4101 are indicated by hatched bars. (A) Root colonization of L. japonicus roots by M. loti strains 14 days after inoculation. L. japonicus plants were inoculated with essentially the same population of each M. loti strain using inocula of approximately 3 to 5 × 106 CFU·ml−1. The roots were excised and suspended in 1 ml of 0.1% Tween 20. After agitation of the roots on a vortex mixer for 2 min at top speed, the suspension was diluted and plated on tryptone-yeast medium containing 100 μg·ml−1 phosphomycin in order to count the number of CFU per mg fresh root. (B) Numbers of infection threads observed 14 days after inoculation. (C) Numbers of nodules formed 28 days after inoculation. Mean values with standard deviations were obtained for at least eight plants.
Growth of wild-type and purine mutants of Mesorhizobium loti MAFF303099 in B− minimal medium supplemented with purine derivatives
Infection and nodulation properties of the purB and purH mutants.
In the absence of adenine supplementation, neither ML0079 nor ML4101 induced nodule formation in L. japonicus 28 days after inoculation (Fig. 2C). With 0.1 mM or 0.5 mM adenine supplementation, plants inoculated with ML4101 formed nodules that were unpigmented (white) and smaller than those formed with the wild-type strain (Fig. 3D and E). The number of nodules on the roots inoculated with ML4101 in the presence of 0.5 mM adenine was greater than that on the roots inoculated with the wild-type strain. Inoculation with ML0079 did not produce discernible nodules, even upon supplementation with 0.5 mM adenine (Fig. 3G); however, nodule meristems were observed upon adenine supplementation (Fig. 3H).
L. japonicus nodules induced by the wild-type M. loti strain MAFF303099 and purine mutants in the presence of 0.5 mM adenine. Roots and nodules were photographed 28 days after inoculation with the wild-type strain MAFF303099 (A, B, and C), ML4101 (D, E, and F), or ML0079 (G, H, and I). M. loti strains were marked by the gusA expression plasmid pFAJPcycA. Nodules were sectioned and stained (blue) with GUS staining solution (C, F, and I). Bars = 1 mm (A, D, and G), 0.5 mm (B, E, and H), and 250 μm (C, F, and I).
In order to examine the infection and the nodule-like structures induced by the M. loti strains, we constructed a constitutive gusA-expressing plasmid, pFAJPcycA. The 410-bp promoter region of the cytochrome c2 gene (cycA) in Rhodobacter capsulatus was excised from plasmid pCHB500 (2) and inserted in the XbaI and BamHI site of the plasmid pFAJ1701 (11). The plasmid pFAJPcycA was transferred to M. loti strains by conjugation. L. japonicus roots inoculated with each pFAJPcycA-harboring strain were immersed in a β-glucuronidase (GUS) staining solution [100 mM phosphate buffer (pH 7.0), 0.50 μM K3(Fe(CN)6), 0.50 μM K4(Fe(CN)6), 0.3% Triton X-100, and 200 mg·ml−1 of 5-bromo-4-chloro-3-indolyl-β-d-glucuronic acid (X-Gluc)]. The roots were subjected to vacuum treatment for 15 min and incubated at 28°C overnight, and the number of infection threads was counted with a stereoscopic microscope. Without adenine supplementation, MAFF303099(pFAJPcycA) induced approximately 50 infection threads per plant 14 days after inoculation, whereas neither ML0079(pFAJPcycA) nor ML4101(pFAJPcycA) induced recognizable infection threads (Fig. 2B). Adenine clearly enhanced infection thread formation and nodulation by ML4101(pFAJPcycA) (Fig. 2B and C), while it barely affected infection thread formation by ML0079(pFAJPcycA) (Fig. 2B).
We carried out a cytological comparison of the mature nodules induced by MAFF303099(pFAJPcycA) and the nodule-like structures induced by ML4101(pFAJPcycA). Fresh nodules were embedded in Jung tissue-freezing medium (Leica Instruments, Nussloch, Germany) and sliced into 20-μm-thick sections using a cryomicrotome (CM1510; Leica Instruments GmbH, Germany). Sections were stained with the GUS staining solution and viewed through a bright-field microscope. Bright-field microscopy of transversely sectioned nodules induced by the wild-type strain revealed a structure typical of determinate nodules with a large circular central region of infected cells (Fig. 3C). The nodules induced by ML4101(pFAJPcycA) contained more vacuoles and fewer infected cells (Fig. 3F), and many contained no infected cells discernible by GUS staining. This is probably attributable to the fact that the amount of adenine delivered to the bacteria within the nodule limits the proliferation of the purine mutant, which leads to the formation of immature nodules. Alternatively, AICAR accumulating in ML4101 might cause changes in gene expression that lead to the formation of Fix− nodules, as reported for R. etli, where AICA riboside downregulates the expression of fixNOPQ (21).
While ML0079 induced root hair curling and nodule meristem formation when adenine was added (Fig. 3H), it seems to produce Nod factors under these conditions. This agrees with the fact that purine mutants of R. etli and the Rhizobium strain NGR234 also induce nodule meristems on inoculated roots (9, 17). However, infection thread formation was barely observed with ML0079(pFAJPcycA), whether or not adenine was added (Fig. 2B). In addition, no blue coloration was detected in the nodule meristems after inoculation of ML0079(pFAJPcycA) (Fig. 3I). This suggests that M. loti purB is required after meristem initiation for infection thread formation and nodule development.
We also examined the nodulation phenotype of ML0079 in the presence of both AICA riboside (0.1 or 0.5 mM) and adenine (0.5 mM). Under these conditions, however, we could not detect significant effects of AICA riboside on nodulation (data not shown). This result suggested the incapability of M. loti to utilize AICA riboside. Therefore, to explore the ability of M. loti to take up and utilize AICA riboside, we examined the effect of AICA riboside on the growth of pur mutants which were blocked prior to AICAR production. The purE mutant STM-6T01c04 (mll3809::Ω), the purD mutant STM-4T02h01 (mll7447::Ω), the purN mutant STM-18T01a05 (mll7961::Ω), and the purM mutant STM-3T01d01 (mll7962::Ω) were screened from a signature-tagged mutant library (Y. Shimoda, S. Sato, and S. Tabata, unpublished data). None of the purine mutants could grow in B− minimal medium, but growth could be restored by the application of adenine (Table 2). Although the growth level was much lower than in the case of adenine, AICA riboside could also restore the growth of pur mutants, suggesting that M. loti possesses the capabilities to take up and metabolize AICA riboside. It has been reported that R. leguminosarum 128C56 and R. fredii HH303 were less able to take up AICA riboside than R. etli CE3 and that R. fredii might be defective in metabolizing AICA riboside (16). Interestingly, AICA riboside restored infection and nodulation by the R. leguminosarum and R. fredii purine auxotrophs, even though the compound did not restore their growth effectively. It can be speculated that the uptake and/or metabolism of AICA riboside in M. loti is even less efficient or that M. loti requires a higher intracellular level of AICAR for infection than R. leguminosarum and R. fredii.
AICAR is involved in many of the physiological functions of bacteria. It is reported to affect lipopolysaccharide (LPS) production (9), and this compound affects the symbiosis between R. leguminosarum and pea plants (18). However M. loti LPS has little effect on nodule development in Lotus tenuis plants (5). Worland et al. (24) reported that purine mutants exhibited many changes in protein expression, which could be responsible for the effects on LPS biosynthesis, exopolysaccharide synthesis, and symbiotic performance. Therefore, the involvement of purine pathway intermediates in nodulation has not conclusively been established, because purine deficiency affects many biochemical pathways. Transcriptional and proteomic analysis using ML0079 and ML4101 could be useful in identifying cellular components regulated by AICAR.
In conclusion, we have presented evidence that the purB gene of M. loti is required for infection thread formation and nodule development in L. japonicus, probably because of AICAR production rather than purine nucleotide synthesis. This is the first genetic evidence that AICAR is an important factor modulating infection and nodule formation in the model legume L. japonicus.
ACKNOWLEDGMENTS
We thank the Frontier Science Research Center of the University of Miyazaki, Japan, for supplying L. japonicus seeds. We also thank J. Michiels (Katholieke Universitet Leuven) for providing plasmid pFAJ1701 and Y. Shimoda, S. Sato, and S. Tabata (Kazusa DNA Research Institute, Kisarazu, Japan) for providing interposon mutants prior to publication.
This work was supported, in part, by a grant-in-aid for scientific research (KAKENHI) on the “Comparative Genomics” priority area from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (to K.S.) and by a postdoctoral fellowship for research abroad from the Japan Society for the Promotion of Science (to S.O.).
FOOTNOTES
- Received 21 May 2007.
- Accepted 29 August 2007.
↵▿ Published ahead of print on 7 September 2007.
- American Society for Microbiology
