Significance

Abstract

Results

B. elkanii Nodulates nfr1 Mutant Soybeans in a T3SS-Dependent Manner.

In B. elkanii, the rhcJ gene is in an operon encoding several T3SS components, and strain BErhcJ is a mutant defective for protein secretion by this system (12). In wild-type Enrei, the mutant BErhcJ induced 20–40% fewer nodules than wild-type B. elkanii (Fig. 1A). Nodule numbers derived by BEttsI, the mutant of transcriptional activator TtsI, also decreased to similar levels observed for BErhcJ. In En1282 (a mutant of Enrei), however, nodulation was observed in roots inoculated with USDA61, whereas few nodules were observed following inoculation with T3SS mutants or Bradyrhizobium japonicum USDA110 (Fig. 1B). In comparison with wild-type Enrei (Fig. 1A), En1282 formed significantly fewer nodules when inoculated with B. elkanii (Figs. 1B and ​and2).2). Notably, the nodulation timing was delayed in En1282; nodules became visible on the roots of Enrei 7–8 d after B. elkanii inoculation, compared with 14–16 d on the roots of En1282. Similar nodulation phenotypes were observed when cultivar Clark-rj₁ was selected as the host plant (Fig. S1). These results demonstrate that B. elkanii USDA61 is capable of nodulating nfr1 mutant soybeans of two different cultivars.

Open in a separate window

Fig. 1.

Nodulation of G. max cv. Enrei (A) and En1282 (B) inoculated with wild-type and mutant strains of B. elkanii USDA61 and B. japonicum USDA110. Nodule numbers were counted 30 d after inoculation. Nodulation tests were performed at least three times, values shown are means obtained from at least 18 plants, and error bars indicate SDs. Statistical analyses (Student t test) were performed comparing mutants with the wild-type strain. *P < 0.05; **P < 0.01.

Open in a separate window

Fig. 2.

Roots of G. max cv. Enrei and En1282 inoculated with B. elkanii USDA61 and type III mutant BErhcJ. Roots were photographed 8, 16, and 30 d postinoculation (dpi). (Scale bars, 4 mm.)

Synergistic Effects of NFs and T3SS Signaling on the Efficiency of Rhizobial Infection.

To explore further the importance of NFs in the interaction between B. elkanii and nfr1 soybean, we constructed the nodC mutant of B. elkanii, BEnodC, which is unable to synthesize NFs, and examined its nodulation phenotype in wild-type and nfr1 mutant soybeans. Contrary to expectations, BEnodC induced nodule formation on Enrei roots, although the nodule number was significantly reduced compared with that of the B. elkanii wild-type strain (Fig. 1A). The bacteria reisolated from nodules retained the same phenotype (Table S1). The nodule numbers of En1282 inoculated with BEnodC were roughly halved compared with those inoculated with the B. elkanii wild-type strain. The nodule numbers of Enrei and En1282 inoculated with BEnodC were not significantly different (Student t test P < 0.05). Almost all nodules induced by BEnodC on both Enrei and En1282 showed irregular or tumor-like morphology (Figs. 1B and 3 G and H). In En1282 inoculated with BEnodC, ~20–50% of nodules showed tumor-like structures that contained no infected cells (Fig. 3 H and L). However, when both nodC and ttsI were mutated (BEttsnod), the nodulation ability of B. elkanii was mostly abolished (Fig. 1 A and B). These results indicate that B. elkanii is capable of infecting and nodulating soybean in a T3SS-dependent manner in the absence of NFs, although the nodulation efficiency was much lower than NF-derived nodulation.

Open in a separate window

Fig. 3.

Root hairs and nodules of G. max cv. Enrei and nfr1 mutant En1282 inoculated with B. elkanii strains. (A) Root-hair curling and infection threads in Enrei inoculated with B. elkanii USDA61. (B) Root-hair curling and infection threads in Enrei inoculated with type III mutant BErhcJ. (C) Root hairs of nfr1 mutant En1282 inoculated with USDA61. (D) Root hairs of En1282 inoculated with BEnodC. (E) Nodule of Enrei inoculated with USDA61. (F) Nodule of Enrei inoculated with BErhcJ. (G) Nodule of En1282 inoculated with USDA61. (H) Nodule of En1282 inoculated with BEnodC. (I) Thin section of Enrei nodule infected by USDA61. (J) Thin section of Enrei nodule infected by BErhcJ. (K) Thin section of En1282 nodule infected by USDA61. (L) Thin section of En1282 nodule infected by BEnodC. [Scale bars, (A–D) 20 μm; (E–H) 200 μm; (I–L) 100 μm.]

Dissection of NF- and T3SS-Derived Gene Expression in Soybean by Microarray Analysis.

Plant inoculation tests suggested that nodule formation in nfr1 soybeans is dependent on B. elkanii T3SS. To investigate the molecular basis for T3SS-dependent nodulation and distinguish it from NFR-dependent nodulation, we examined gene-expression profiles in soybean roots using microarray analysis. Comparative analysis of microarray data was performed to dissect soybean genes up- or down-regulated by NF or T3SS signaling (Fig. 4). To extract soybean genes up- or down-regulated by NFR signaling, we compared gene expression in the roots of Enrei and En1282 inoculated with BErhcJ (Fig. 4A). Eight days after inoculation, roots were sampled as suitable material for target RNA preparation. To detect soybean genes up- or down-regulated by signals derived by B. elkanii T3SS, we compared the gene expression of En1282 roots inoculated with USDA61 and BErhcJ (Fig. 4A).

Open in a separate window

Fig. 4.

Experimental design and summary of microarray analysis. (A) Experimental design. To extract soybean genes up- or down-regulated by Nod-factor signaling, we compared gene expression in the roots of Enrei and En1282 inoculated with BErhcJ. To detect soybean genes up- or down-regulated by the signals derived by B. elkanii T3SS, we compared the gene expression of En1282 roots inoculated with USDA61 and BErhcJ. (B) Summary of microarray analyses of genes up-regulated by NF and T3SS signaling. Soybean genes with signal intensities more than 2-fold that of uninfected roots were considered to be up-regulated. (C) Summary of microarray analyses of genes down-regulated by NF and T3SS signaling. Soybean genes with signal intensities less than 0.5-fold that of uninfected roots were considered to be down-regulated.

The microarray analysis results are summarized in Fig. 4 B and C. Soybean genes with signal intensities more than 2-fold and less than 0.5-fold that of uninfected roots were considered up- and down-regulated, respectively. We identified 173 and 991 genes potentially up- and down-regulated by NFR signaling (Table S2 and Dataset S1, respectively), and 29 and 36 genes potentially up- and down-regulated by T3SS signaling (Tables S3 and S4, respectively). The 173 genes potentially up-regulated by NF signaling include a number of known nodulins and symbiosis-related genes (Table S2). Of these, 16 are commonly up-regulated by T3SS signaling, including symbiosis-related genes such as root-nodule extensin, ENOD40, and nodule inception (NIN) (Table 1). The fold changes of signal intensity were not comparable between NF and T3SS signaling. Nodulins 21 and 26 were up-regulated only by NF signaling (Table S2), whereas T3SS signaling triggered the up-regulation of endo-β-1,4-glucanase, cyclin, and early nodulin GmN93 (Table S3).

Table 1.

G. max genes up-regulated or down-regulated by both Nod-factor and T3SS signaling

ID no.	Similarity	NF pathway*		T3SS pathway†
		Fold change	P value	Fold change	P value
Up-regulated
BI316036.1	Root-nodule extensin	217.69	0.0021	71.55	0.0063
BI970567.1	G. max uncharacterized protein	178.16	0.0007	2.36	0.0323
BI971063.1	Subtilisin-like protease	81.41	0.0048	6.45	0.0037
BW679106.1	F-box/kelch-repeat protein	45.95	0.0027	10.67	0.0021
CO978759.1	CASP-like protein N24–like protein	34.24	0.0015	4.63	0.0126
CO985886.1	Gibberellin 3-β-dioxygenase 1–like protein	30.45	0.0030	14.11	0.0117
BW676134.1	G. max cDNA clone Gm-c1082-2286	17.90	0.0066	2.85	0.0044
BW657022.1	Nodule inception protein (nin)	15.73	0.0104	5.10	0.0135
BI425123.1	Cytochrome P450 71A1-like protein	15.10	0.0115	4.10	0.0028
BW676262.1	ENOD40-2	10.90	0.0071	3.30	0.0336
CF922378.1	Early nodulin GmN36a	9.17	0.0081	3.33	0.0230
AW830341.1	G. max hypothetical 17.8-kDa protein	6.60	0.0213	7.60	0.0032
BI969878.1	G. max uncharacterized protein	6.03	0.0068	2.27	0.0081
BE346992.1	G. max uncharacterized protein	2.50	0.0346	2.13	0.0271
BI426827.1	Indole-3-acetate O-methyltransferase 1-like–protein	2.46	0.0017	2.31	0.0031
BI967256.1	G. max uncharacterized protein	2.17	0.0409	3.60	0.0019
Down-regulated
CO981522.1	Tryptophan aminotransferase-related protein	0.05	0.000	0.15	0.043
BQ080234.1	Callose synthase 9-like, transcript variant	0.09	0.002	0.28	0.012
BM177253.1	G. max uncharacterized protein	0.11	0.004	0.13	0.016
CD405795.1	G. max mRNA induced by salicylic acid	0.13	0.014	0.27	0.011
BM892625.1	G. max uncharacterized protein	0.20	0.021	0.20	0.042
AW620254.1	17.9-kDa class II heat shock protein	0.20	0.003	0.48	0.043
AW705695.1	Vicilin-like antimicrobial peptides 2-2–like protein	0.26	0.007	0.46	0.025
CA938283.1	Chlorophyll a-b–binding protein	0.27	0.047	0.34	0.001
CD391543.1	G. max uncharacterized protein	0.31	0.027	0.41	0.041
BE346566.1	Myb-related protein 306–like transcription factor	0.33	0.046	0.38	0.048
BG839502.1	NAC domain-containing protein 8–like protein	0.37	0.016	0.49	0.045
BW683725.1	G. max uncharacterized protein	0.40	0.047	0.10	0.002

Open in a separate window

^*Microarray fold changes in the roots of Enrei inoculated with BErhcJ versus the roots of En1282 inoculated with BErhcJ (Fig. 4A).

^†Microarray fold changes in the roots of En1282 inoculated with USDA61 versus the roots of En1282 inoculated with BErhcJ (Fig. 4A).

NF signaling down-regulated 991 genes, of which 12 were commonly down-regulated by T3SS signaling. These include tryptophan aminotransferase-related protein, callose synthase, and salicylic acid-inducing genes (Table 1). By contrast, down-regulation of 24 genes was identified only in T3SS signaling, including vacuolar processing enzyme 2, chalcone synthase, and vicilin-like antimicrobial peptides (Table S4).

T3SS-Dependent Expression of Symbiosis Genes ENOD40 and GmNIN in Soybean.

Microarray analysis indicated that T3SS-derived signaling from B. elkanii activates several symbiosis-related genes, which are known to be expressed by the recognition of compatible NFs via host legume NFRs. To investigate further the importance of NFs and NFRs in the expression of these symbiosis genes, we compared their expression in the roots of Enrei and En1282 inoculated with T3SS and nodC mutants using real-time RT-PCR. Two well-characterized symbiosis genes, ENOD40 (15) and NIN (16), were selected as marker genes because their expression is induced early after NF perception (17, 18). In both Enrei and En1282 roots, relative expression levels of ENOD40 and NIN were significantly induced by the inoculation of USDA61 (Fig. 5). Deletion of the nodC gene decreased the up-regulation of these genes, but it remained significant compared with the mock treatment. However, the up-regulation of both genes was abolished by inoculation of the T3SS mutant BErhcJ on En1282. These results indicated that the expression of ENOD40 and NIN was induced by T3SS-derived signaling and that NF production enhances the expression even in roots of the nfr1 mutant En1282.

Open in a separate window

Fig. 5.

Real-time RT-PCR analysis of G. max ENOD40 and G. max NIN transcripts in roots of Enrei and En1282 inoculated with wild-type and mutants of B. elkanii USDA61. RNAs were isolated from the roots of Enrei and its nfr1 mutant En1282 4 and 8 d postinoculation. cDNA synthesis and real-time PCR of G. max ENOD40 (A) and G. max NIN (B) were performed with gene-specific primers (Table S5) as described in Materials and Methods. The values are means of triplicate measurements, and the error bars indicate SDs. Statistical analyses (Student t test) were performed comparing each inoculation with the mock treatment. *P < 0.05; **P < 0.01.

Discussion

The present study investigated the activation of soybean nodulation signaling by T3SS in B. elkanii, homologs of which have been widely observed in pathogenic bacteria, and which play crucial roles in infecting their animal and plant hosts. We demonstrated that the activation does not require the initial step of symbiotic molecular interaction involving rhizobium NF and host legume NFR via the observation that wild-type B. elkanii, but not the T3SS-deficient mutant, was able to form nitrogen-fixing nodules on soybean nfr1 mutants En1282 and Clark-rj₁. Nodulation of these soybean mutants was observed by infecting the B. elkanii NF-deficient mutant BEnodC unless T3SS genes were mutated.

Microarray and RT-PCR analysis revealed that T3SS-derived signaling from B. elkanii can induce the expression of symbiosis genes ENOD40 and NIN in the absence of NFR1 and NF (Fig. 5 and Table 1). These symbiosis genes were previously thought to be up-regulated only following the recognition of NFs via associated NFRs. Although the mechanism for their activation by T3SS remains to be elucidated, recent findings from plant mutants that form nodules in the absence of NFs or rhizobia suggest a number of possibilities. The L. japonicus mutant spontaneous nodule formation 1 (snf1) carries a mutation in CCaMK that leads to autoactivation of its kinase domain and the development of spontaneous nodules in the absence of rhizobia (19). snf1 and other CCaMK gain-of-function mutations lead to the induction of nodulation genes such as ENOD11, resulting in the development of spontaneous nodules in the absence of either NFs or rhizobia (19). Similarly, the gain-of-function mutant in the cytokinin receptor LHK1 was shown to develop spontaneous nodules, concomitant with the expression of NIN in the absence of rhizobia (20, 21). Moreover, exogenous cytokinins induce the expression of early nodulin genes including ENOD40 and NIN in several legume species (22, 23). It remains to be determined whether B. elkanii T3SS activates CCaMK or LHK, or affects cytokinin levels in the nfr1 mutant soybean, leading to nodule development.

Of particular interest are the nature and function of B. elkanii T3Es in activating soybean nodulation signaling. B. japonicum, a close relative of B. elkanii, also possesses a functional T3SS that is activated by soybean genistein, and secretes some proteins that are also secreted by B. elkanii (6, 24). However, this and other studies show that B. japonicum is unable to nodulate on either En1282 or Clark-rj₁, suggesting that B. elkanii possesses a unique T3E(s) responsible for activation of nodulation signaling. Although we have not yet detected any B. elkanii unique T3E candidate genes, B. elkanii encodes several genes for T3E candidates, including nopL, nopM, and nopP, which have been reported to play either a positive or a negative role in symbiosis in other rhizobia (9, 25). Intriguingly, CCaMK and LHK1, mentioned above, are protein kinases that play pivotal roles in nodulation and should participate in phosphorylation systems (26). As a number of T3Es in plant pathogens are known to target host kinases (27), and the rhizobial T3Es NopL and NopP are phosphorylated in host cytoplasm (9), it is conceivable that some T3Es perturb the host phosphorylation state for nodulation.

Intriguingly, in the present study, the nodC mutant induced fewer nodules than wild-type B. elkanii on En1282 or Clark-rj₁ soybean (Fig. 1 and Fig. S1), indicating that rhizobial NF signals even play a positive role in the nodule formation of the nfr1 mutant soybean. The positive effects of NFs on nfr1 soybean were also confirmed by expression analysis of ENOD40 and NIN, as the expression of both genes in the roots of En1282 was higher for USDA61 than BEnodC inoculation (Fig. 5). The NFRs belong to lysine motif (LysM) receptor-like kinases (LysM-RKs), and besides NFR1/NFR5 homologs that directly bind to NFs (28), legume plants are known to possess larger numbers of LysM-RKs than nonleguminous plants (18, 29). Together, these results suggest that rhizobial NFs are perceived by additional receptors in soybean roots, and that the derived signal and that from NFRs synergistically activate nodulation signaling.

Another intriguing observation was the lack of root-hair curling or infection-thread formation in the roots of En1282 inoculated with USDA61. This suggested that T3SS plays a role in alternative infection processes such as crack entry or intercellular infection, although further microscopic analysis will be necessary to elucidate the nature of the infection enhanced by T3SS. Crack-entry infection is well-studied in A. hypogea, one of the host plants of B. elkanii, although the nodulation to A. hypogea is likely to be NF-dependent (30). In the course of A. hypogea nodulation, neither root hairs nor cortical infection threads are formed. Instead, rhizobia invade through fissures in epidermal cells, and the intercellular rhizobia directly internalize cortical cells via an endocytosis-like process (30). Recently, photosynthetic Bradyrhizobium spp. lacking the canonical nod genes were shown to infect their host legume Aeschynomene spp. through gaps between axillary root hairs and invade the cortical cells by invagination of the host cell wall (31). Moreover, Madsen et al. reported that infection between M. loti nodC and L. japonicus nfr1-1 snf1 mutants involves intercellular infection with no infection threads in either root hairs or nodule tissues (26). Such ground-state infection occurs at a 20- to 100-fold lower frequency compared with normal root-hair invasion (26). Similarly, we observed limited but significant nodulation on Enrei and En1282 by the B. elkanii nodC mutant and no nodulation by the nodC-ttsI double mutant (Fig. 1). This suggests that B. elkanii T3SS can promote intercellular infection. In the presence of NFs and NFRs, however, this poorly efficient infection would be overridden by that accompanying root-hair infection threads. Although the details of the T3SS-related infection of B. elkanii remain to be elucidated, some insights might be gained by examining the T3SS-mediated Salmonella infection process. Salmonella spp. use T3Es to manipulate cellular processes in promoting their uptake and subsequent survival within vacuoles (32), which makes a fitting analogy with rhizobia being entrapped by root cortical cells via an endocytosis-like process that develops symbiosomes of infected cells in nodules.

Although the B. elkanii nodC mutant was shown to form nodules in the present study (Fig. 1), most of these nodules were poorly infected (Fig. 3). This suggests that concerted expression, rather than simple copossession, of T3SS and nodulation genes is required for the formation of effective nodules. Such a requirement might be why most known rhizobial T3SS genes are situated downstream of flavonoid-responsive signaling (6, 7). Host-derived flavonoid induction is found in most T3SS-harboring rhizobia, except for the β-rhizobium Cupriavidus taiwanensis, which incorporates a T3SS similar to that of the β-proteobacterium Burkholderia, and its expression is more likely to be glutamate-dependent. These findings suggest that the acquisition of T3SS from ancestor pathogenic bacteria has been adapted for the orchestrated expression of symbiotic genes during coevolution with host legume plants.

Another interesting aspect of rhizobial T3SS is that T3 effector proteins can cause restriction of nodulation. Our previous study revealed that soybean cultivar Hill, which carries the Rj4 gene and is known to restrict nodulation by B. elkanii USDA61, formed nodules when inoculated with type III knockout mutants of USDA61 (12). Although the Rj4 gene has not been identified, the soybean genes Rj2 and Rfg1 were recently identified as allelic genes encoding the Toll-interleukin receptor/nucleotide-binding site/leucine-rich repeat (TIR-NBS-LRR) class of plant resistance (R) proteins that restrict nodulation with B. japonicum USDA122 and Sinorhizobium fredii USDA257, respectively (33). Introduction of the Rfg1 allele of Williams 82 into Peking (genetically defined as rfg1/rfg1) abolished nodulation of S. fredii USDA257, whereas the T3SS knockout mutant of S. fredii USDA257 exhibited the ability to nodulate soybean carrying the Rfg1 allele (33). These findings imply that rhizobial T3Es, when recognized by cognate host R proteins, trigger a strong defense reaction (effector-triggered immunity) that halts rhizobial infection.

In conclusion, rhizobia use a pathogenic secretion system for symbiotic purposes. B. elkanii T3SS hijacks soybean symbiosis genes in the absence of canonical NFs and NFRs that are known to be dispensable for the initiation of nodulation. We propose a model of nodulation signaling and modulation by the rhizobial type III secretion system in Fig. 6. Rhizobial T3SS is likely to promote intercellular infection, which can be superimposed by more efficient root-hair infection mediated by canonical NF and NFR signaling. Our efforts are now being directed toward the identification of the B. elkanii effector protein that hijacks soybean symbiosis signaling, to acquire a better understanding of the evolutionary aspects of interactions between endosymbionts and their eukaryotic hosts.

Open in a separate window

Fig. 6.

Model of nodulation signaling and modulation by the rhizobial type III secretion system. A host plant-derived flavonoid induces the production of Nod factors in rhizobia. Recognition of NFs by NF receptors triggers a signaling cascade leading to nodulation. The flavonoid also induces rhizobial T3SS, which injects effector proteins into host cells. One effector modulates nodulation signaling toward nodulation, whereas another is recognized by the host defense system, which is capable of aborting the nodulation process.

Materials and Methods

Supplementary Material

Acknowledgments

Footnotes

References