Comparison of Symbiotic Abilities between M. loti Strain MAFF303099 and TONO.

apn1 is a unique symbiotic mutant that shows a rhizobial strain-specific Fix⁻ phenotype (32). On apn1 roots, M. loti strain MAFF303099 forms pink mature nodules with nitrogen fixation activity, whereas strain TONO forms small dark brown or black nodules with low nitrogen fixation activity (Fig. 1A). Both strains form normal nodules on wild-type L. japonicus roots (32). Detailed observation of apn1 Fix⁻ nodules revealed that the degree of browning varies in individual nodules irrespective of size and developing stage of nodule (SI Appendix, Fig. S1 A–D), and the brown pigments did not always overlap the entire infected region (SI Appendix, Fig. S1 E–G). Even in the fully developed nodules, dark brown pigments were generally observed in the limited area of central infected region and, in some cases, in the nodule periphery layer (SI Appendix, Fig. S1 E–G and N). This implies that brown pigment is induced depending on a certain physiological status of host plant cells.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Comparison of symbiotic properties between M. loti strain MAFF303099 and TONO. (A) Shoot growth (Upper) and nodule phenotype (Lower) of apn1 mutant inoculated with M. loti strain MAFF303099 (MAFF, Left) and TONO (Right). Pictures were taken at 4 wk after M. loti inoculation. (Scale bars: Upper, 2 cm; Lower, 1 mm.) (B–E) Infection thread (B and C) and infection inside nodule (D and E) of apn1 mutant visualized by GFP-expressing MAFF303099 (B and D) and TONO (C and E). Pictures are shown as a merged image of bright and fluorescence fields of GFP. (Scale bars: B and C, 100 μm; D and E, 500 μm.) (F) Frequency of infection events between MAFF303099 and TONO on wild-type (WT; MG-20 Miyakojima) and apn1 mutant at 2 wk after M. loti inoculation. Bars represent the average number of microcolonies (black bars) and infection threads (gray bars) ± SEM (n > 12). No significant differences in the frequency of infection events were detected among the combinations of plant and rhizobia. (G) Nodule number formed on WT and apn1 mutant inoculated with MAFF303099 and TONO. Number of mature pink nodules (pink bars) and Fix⁻ nodules (white bars) were counted at 4 wk after inoculation. Bars represent the average number of each type of nodule ± SEM (n > 8). (Insets) Typical pink mature and Fix⁻ nodules. Asterisk indicates statistical difference (**P < 0.01, Student’s t test) in total and white nodule number between MAFF303099 and TONO. (H) Competition assays of MAFF30399 and TONO. Nodules were counted at 2 wk after inoculation with GFP- or GUS-labeled MAFF303099 or TONO. Numbers above the graph indicate the total number of nodules analyzed. (I–K) Nodule phenotype of the apn1 mutant inoculated simultaneously with GFP-labeled MAFF30399 and GUS-labeled TONO. Bright-field (I and K) and fluorescence (J) images were captured. (Scale bars: 1 mm.)

To compare the symbiotic abilities of the two strains in more detail, we analyzed infection and nodulation phenotypes in the wild type and the apn1 mutant. Inoculation of GFP-labeled MAFF303099 and TONO with the apn1 mutant showed that formation of infection threads and infection inside nodules were similar in both strains (Fig. 1 B–E). The lack of significant differences in the frequency of infection events (Fig. 1F) suggests that apn1 is a mutant that shows symbiotic defects only in nitrogen fixation. At 4 wk after rhizobial inoculation, the number of nodules formed on the apn1 roots was higher with inoculation of TONO than with MAFF303099 (Fig. 1G). Such an increase in the number of nodules has also been reported in other Fix⁻ mutants (12, 17, 36), and this can be caused by deviation from the autoregulation of nodulation by the low nitrogen fixation activity of nodules. The competition assay with GFP- and GUS-labeled M. loti showed that, compared to strain TONO, MAFF303099 had a slightly higher competitiveness both in the wild type and the apn1 mutant (Fig. 1H). Coinoculation of MAFF303099 and TONO to the apn1 mutant did not complement the Fix⁻ nodule phenotype of TONO (Fig. 1 I–K), indicating that symbiotic phenotypes of the apn1 mutant are determined after rhizobial entry into the host plant cells.

Identification of M. loti Genes that Determine the Fix⁻ Phenotype of apn1.

The strain-specific Fix⁻ phenotype of the apn1 mutant implies that some molecular interactions between the host plant and rhizobial factors are involved in the compatibility of symbiotic nitrogen fixation. To identify rhizobial factors that determine the strain-specific Fix⁻ phenotype of apn1, we first sequenced the whole genome of strain TONO and compared its gene composition with those of strain MAFF303099. The genome of TONO was ∼8.4 Mb in size, and a total of 8,149 protein-coding genes were predicted (34). Comparison of predicted genes between MAFF303099 and TONO revealed that gene clusters involved in nodulation (nod genes), nitrogen fixation (nif and fix genes), the type III protein secretion system (T3SS), and exopolysaccharide production were highly conserved in their genome sequences and the order of genes (SI Appendix, Fig. S2), indicating that these genes were unlikely to be determinants of the Fix⁻ phenotype of apn1. This conclusion is also reasonable from the fact that both MAFF303099 and TONO can form nodules normally on wild-type L. japonicus (32). A genome-wide ortholog search between MAFF303099 and TONO identified genes present only in the TONO genome (34), but we could not specify any candidate genes because numerous genes were specific to each strain. Moreover, considering that some DNA sequence polymorphisms may affect the activity or function of genes, identification of candidate genes from the genome sequences alone is likely impossible.

As a second strategy, we constructed a transposon mutant library in strain TONO and explored mutants that could suppress the Fix⁻ phenotype of apn1. TONO mutants were created using the Tn5 transposon by following the same procedures described by Shimoda et al. (37), and a total of 16,992 independent mutants were collected. When the mixtures of TONO mutants were inoculated with apn1, pink effective nodules were occasionally formed, along with many Fix⁻ nodules, which resulted in promoted shoot growth (SI Appendix, Fig. S3). From the screening of 12,672 TONO mutants, we obtained a total of 242 pink effective nodules. To eliminate false positives, TONO mutants isolated from effective nodules were inoculated individually with apn1 to confirm the formation of Fix⁺ nodules, and transposon-disrupted genes were identified. From the series of mutant screening shown in SI Appendix, Table S1, we successfully identified one candidate TONO gene that determines the Fix⁻ phenotype of the apn1 mutant.

Characteristics of the Candidate Gene Product.

The candidate gene identified from the mutant screening was MLTONO_0844 (34), which encoded a protein similar to a bacterial outer membrane autotransporter, also known as the type V protein secretion system. We named the identified TONO autotransporter gene DCA1 (Determinant of nitrogen fixation Compatibility of APN1). Domain search analyses revealed that a signal peptide (1 to 50 aa) and an autotransporter β-domain (IPR005546) were present at the N- and C-terminal ends of DCA1, respectively (Fig. 2A). Divergent glycine-rich motifs consisting of a unit of 44 to 53 aa were found 43 times between the signal peptide and the autotransporter β-domain (Fig. 2 A and B and SI Appendix, Fig. S4). Each glycine-rich motif was predicted to encode a β-sheet and a disordered coil structure (Fig. 2B). The glycine-rich repeat region that occupied a large part of DCA1 (>250 kDa) was predicted to form a β-helix structure, a typical trait of most autotransporter proteins (38, 39), and no known functional domains were predicted in the region. A BLAST search analysis found that the autotransporter proteins with similar long glycine-rich repeat regions existed not only in other rhizobia but also in other Gram-negative bacteria and cyanobacteria (Fig. 2C), but none of these proteins have had their biological functions proven so far. The BLAST search also revealed that several Mesorhizobium species, including strains MAFF303099 and R7A (40), possessed autotransporter proteins with high similarity to DCA1 of TONO (Fig. 2C).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Structure of M. loti autotransporter DCA1. (A) Domain structure of DCA1. Gray and pale green boxes at the N- and C-terminal ends indicate the signal peptide and autotransporter β-domain (IPR005546), respectively. Black lines indicate a unit of glycine-rich motif. Arrowheads above the image indicate the location of transposon insertion sites identified from mutant screening. (B) Amino acid composition probabilities in glycine-rich motifs of DCA1 homologs predicted by using the MEME tool (73). Arrows and line below the image indicate the secondary structure of β-strand and coiled region, as predicted by the PSI-PRED program (74). (C) Amino acid alignment of glycine-rich motifs found in representative DCA1 homologs of Gram-negative bacteria (α, β, γ, and δ proteobacteria) and cyanobacteria. (D) Shoot growth and nodulation phenotypes of apn1 mutant inoculated with wild-type TONO, MAFF303099, and MAFF303099 harboring empty plasmid (MAFF_empty) or DCA1 expressing plasmid (MAFF_DCA1). MAFF_DCA1 formed small and white Fix⁻ nodules, as did wild-type TONO. (E) Complementation test of dca1 transposon mutant (dca1-5). dca1-5 harboring DCA1-expressing plasmid (dca1-5_DCA1) complemented mutant phenotype and forms Fix⁻ nodules, as did wild-type TONO. (Scale bars: Upper, 2 cm; Lower, 2 mm.)

From the large-scale mutant screening, we identified nine mutant alleles of DCA1 (Fig. 2A, arrowheads, and SI Appendix, Fig. S5A). All mutant alleles were confirmed to contain transposon insertion inside DCA1 (SI Appendix, Fig. S5 B and C) and formed pink mature nodules on apn1 roots (SI Appendix, Fig. S5D), indicating that DCA1 was involved in the regulation of the nitrogen fixation activity of the apn1 mutant. To further confirm whether DCA1 was responsible for the strain-specific Fix⁻ phenotype of apn1, DCA1 was introduced into strain MAFF303099, which formed nitrogen-fixing nodules on the apn1 roots, and its symbiotic phenotype was analyzed. MAFF303099 harboring DCA1 (MAFF_DCA1) formed small white Fix⁻ nodules on apn1 roots as in wild-type TONO, resulting in attenuated shoot growth (Fig. 2D). In contrast, the MAFF303099 control in which the empty vector was introduced (MAFF_empty) formed pink mature nodules as in wild-type MAFF303099 (Fig. 2D). In addition, DCA1 successfully complemented its transposon mutant (dca1-5), resulting in the formation of Fix⁻ nodules on apn1 roots (Fig. 2E). These results clearly show that the identified autotransporter gene of TONO, DCA1, is a determinant of the strain-specific nitrogen fixation activity of the apn1 mutant and that DCA1 alone can completely alter the nitrogen fixation phenotype of MAFF303099.

We also analyzed other phenotypes of DCA1 transposon mutants, such as growth in culture media, motility, exopolysaccharide production, and nodulation ability on wild-type plants. However, no clear differences were found between wild-type TONO and the transposon mutants in all of the phonotypes tested (SI Appendix, Figs. S6 and S7), indicating that DCA1 was not involved in these biological processes.

DCA1 Has Activity to Secrete Protein to Extracellular Spaces.

To reveal the structural features of DCA1, we conducted homology modeling. Since no appropriate PDB templates for the long glycine-rich repeat region of DCA1 (Fig. 2A) were available, we took advantage of the structure of the autotransporter β-domain (IPR005546) because some of its crystal structures had already been resolved. As a template for homology modeling, we used the crystal structure of Esterase A from Pseudomonas aeruginosa (EstA; PDB ID 3KVN) (41), which showed the highest BLAST score against the β-domain region of DCA1 (3330 to 3697 aa). The homology model by I-TASSER (42) revealed that the region of 3402 to 3697 aa formed a typical membrane barrel structure in which nearly polar and hydrophobic amino acid residues were located inside and outside of the barrel, respectively (Fig. 3A and SI Appendix, Fig. S8). The structural model also indicated that the region 3330 to 3402 aa formed an α-helix structure that passed through the barrel, and the N-terminal part from 3370 aa was located outside of the barrel.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Structure and protein secretion activity of DCA1. (A) Structure of the β-barrel region of DCA1, as predicted by homology modeling with the P. aeruginosa EstA protein as a modeling template. Structural models are colored by a gradient from blue at the N terminus to red at the C terminus. Localization of extracellular space, outer membrane, and periplasmic space is indicated according to the structure of EstA (41). Numbers on the model indicate the positions of amino acids. (B) Schematic illustrations of full-length and truncated DCA1 constructs used for the experiments in C–E. A black arrowhead indicates a cleavage site suggested in SI Appendix, Fig. S9. (C and D) Analysis of self-cleavage and secretion activity of full-length (C) and truncated DCA1 (D). CBB and WB indicate the images of Coomassie Brilliant Blue staining and Western blotting with anti-myc antibody, respectively. A black arrowhead in C is the signal of myc full-length DCA1 in cultured cell pellet. Black and white arrowheads in D indicate the signal of full-length SP_myc_3156-3697 in cultured cell pellet and secreted fragment in culture supernatant, respectively. (E) Determination of N-terminal secreted region of DCA1. Black arrowheads in pellet samples indicate the signals corresponding to the size of full-length SP_myc_51-300_3000-3697 and SP_myc_51-856_3000-3697. White arrowheads indicate the secreted fragments in culture supernatant. Western blotting was carried out against total proteins of cultured cell pellet and supernatant proteins precipitated by trichloroacetic acid.

Autotransporters have been reported to cleave a part of their own protein (the passenger domain) and to transport it to outside of the cell through the outer membrane-embedded barrel (43). To clarify whether DCA1 had the ability to transport protein to extracellular spaces, we first tried to identify the cleavage site of the DCA1 passenger domain. Because we expected that a sufficient amount of secreted proteins could not be obtained from rhizobia, we employed an IPTG-inducible protein expression system in Escherichia coli. Based on the homology model of DCA1, the protein region that was predicted to be located outside the barrel (N-terminal part from 3370 aa; Fig. 3A) was fused to the OmpA signal peptide of E. coli (MKKTAIAIAVALAGFATVAQA) (44), which allows periplasmic targeting, and three repeats of the myc epitope tag were inserted after OmpA (OmpA_myc_3252-3697; SI Appendix, Fig. S9A). The fusion fragment was cloned into the cold-inducible vector pColdI, and the protein was expressed in E. coli cells. After purification with myc tag (SI Appendix, Fig. S9B), a secreted protein fragment that could be seen with Coomassie Brilliant Blue staining was analyzed by mass spectrometry to determine the cleavage sites of DCA1. From the analysis, we successfully identified the cleavage sites of the DCA1 passenger domain at 3262 to 3265 aa (SI Appendix, Fig. S9 C and D).

To confirm the secretion of DCA1 passenger domain in M. loti cells, we next cloned the myc epitope tag-fused DCA1 into the pDG77 vector (45). As with E. coli, the myc epitope tag was placed after the signal peptide of DCA1 (1 to 50 aa) (Fig. 3B), and the resultant constructs were introduced into strain MAFF303099. We first tried to confirm the expression and secretion using full-length DCA1, but no Western blot signals other than nonspecific ones were observed in culture supernatant proteins, and only a pellet sample gave faint signals (Fig. 3C). This likely occurred because the massive size of full-length DCA1 caused a low expression level. We then created an N-terminal truncated DCA1 construct (SP_myc_3156-3697; Fig. 3B), which contained a longer N-terminal region than that used in E. coli to allow easier detection via Western blotting. Western blotting with an anti-myc tag antibody showed a signal for a cultured cell pellet at around 63 kDa, which corresponded to the size of the introduced construct (Fig. 3D, black arrowhead). In the culture supernatant proteins purified by the anti-myc beads, a Western blotting signal was detected at ∼10 kDa, and it was not seen in the empty vector control (Fig. 3D, white arrowhead). The detected signal was somewhat smaller than expected from the cleavage site determined in E. coli, suggesting that additional cleavages might have occurred in M. loti. Our results indicate that DCA1 can transport a part of its own protein to the extracellular space.

To examine where the N-terminal end of DCA1 passenger domain was secreted, we next created other truncated DCA1 constructs in which the entire or the central glycine-rich repeat region was removed. SP_myc_51-300_3000-3697 had no glycine-rich repeats, whereas SP_myc_51-856_3000-3697 contained 10 of them (Fig. 3B). Western blotting against the total protein from the cultured cell pellet and supernatant detected the signals in both constructs (Fig. 3E), indicating that up to the entire N-terminal region without the signal peptide was secreted into the extracellular space.

We also tested the functionality of several truncated DCA1 proteins. When MAFF303099 harboring several truncated DCA1 constructs was inoculated with the apn1 mutant, Fix⁻ nodules were formed only when full-length DCA1 was introduced (SI Appendix, Fig. S10). This result indicates that the glycine-rich repeats, as well as its length, are important for the functionality of DCA1. This idea is also supported by the fact that the glycine-rich repeats are conserved as a long region in DCA1 homologs of other bacteria (Fig. 2C).

Peptidase Activity of Host Plant APN1 Is Required for Its Function in Root Nodule Symbiosis.

Sequence analysis predicted that APN1 encodes an aspartic peptidase belonging to the subfamily A1B, which includes Arabidopsis CDR1 and Nodulin 41 of Phaseolus vulgaris (PvNod41) (32, 46, 47). To investigate whether APN1 had properties of aspartic peptidases, we created a recombinant APN1 protein fused to the maltose binding protein (MBP) tag in E. coli. Aspartic peptidases may exhibit proteolytic activity under acidic conditions in which they self-cleave their own protein to exert maximum activity as a mature form (48). When purified MBP-APN1 was incubated in solutions of different pH (3.0 to 6.5), self-cleavage was observed under acidic conditions (Fig. 4A). This self-cleavage was clearly inhibited by pepstatin, an inhibitor of aspartic peptidases (Fig. 4B), as well as by mutations in two catalytically active sites of APN1 (D114N and D320N; Fig. 4C). These results indicate that APN1 has properties characteristic of aspartic peptidases.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Characterization of APN1 as an aspartic peptidase and requirement of peptidase activity for APN1 function in nodule symbiosis. (A–C) Self-cleavage activity of recombinant APN1 under different pH conditions (pH 3.0 to 6.5) (A). MBP-APN1 shows self-cleavage activity under acidic pH, and the activity is inhibited by the addition of 1 μM pepstatin (B) and mutations in two catalytic active sites of APN1 (D114N and D320N; C). Black and white arrowheads indicate MBP-APN1 and its cleaved products, respectively. (D) Complementation test of apn1 mutant with wild-type (WT) APN1 and APN1 with the catalytically active site mutated (D114, D320N, and D114N_D320N). Images are representative nodules formed on transformed hairy root expressing GFP as a transformation marker. (Scale bars: 1 mm.) The accompanying graph shows the complementation activity of introduced APN1, analyzed by the percentage of mature pink nodules in total nodules formed on transgenic roots inoculated with TONO. n indicates the total number of nodules analyzed. (E) In vitro digestion of N-terminal DCA1 proteins by APN1. Black and white arrowheads indicate MBP-APN1 and two truncated DCA1 proteins (51 to 300 aa and 51 to 856 aa), respectively. Gray arrowhead indicates the GST. Asterisks indicate the cleaved products of DCA1 by APN1.

The in vitro peptidase activity of APN1 prompted us to examine whether the peptidase activity of APN1 was required for its function in root nodule symbiosis. We performed a complementation assay of the apn1 mutant with wild type and APN1 with mutated catalytically active sites (D114N and D320N). When inoculated with M. loti TONO, pink effective nodules were observed on the roots complemented with wild-type APN1, whereas nodules remained small and white with mutated APN1 (D320N and D114N_D320N; Fig. 4D). In the case of APN1(D114N), more pink nodules were formed compared to APN1(D320N) and APN1(D114N_D320N). These results indicate that the protease activity is required for the symbiotic function of APN1, and the D320 active site residue contributes significantly to the in planta activity of APN1.

Our data clearly show the one-for-one relationship between host plant APN1 and DCA1 of symbiotic rhizobia, which function as a peptidase and as protein secretion machinery, respectively. To clarify whether APN1 functions to degrade DCA1 in vitro, we created two N-terminal parts of DCA1 proteins without the signal peptide (51 to 300 aa and 51 to 856 aa), which were shown to be secreted into the extracellular space (Fig. 3E). These DCA1 proteins were expressed as fusion proteins with the GST tag. When MBP-APN1 was incubated with these substrates (GST-DCA1 51 to 300 aa and 51 to 856 aa, respectively) at pH 5.0, both substrates were degraded, whereas GST alone was not (Fig. 4E). Degradation was clearly inhibited by the addition of pepstatin (Fig. 4E) and by mutations in the active site of APN1 (D114N_D320N; SI Appendix, Fig. S11). These results indicate that DCA1 is a substrate degraded by APN1.

Promoter Activity of the M. loti Autotransporter Gene in Root Nodules.

The presence of orthologous autotransporter genes in both MAFF303099 and TONO raised the question of how these genes caused opposite nitrogen fixation phenotypes of the apn1 mutant. We first hypothesized that some specific protein regions determine the activity of respective M. loti autotransporters. To test this, we carried out a domain swapping experiment between DCA1 of TONO and its orthologous gene of MAFF303099 (gene ID mll0950). Unfortunately, we could not determine the regions responsible for the strain-specific phenotype because all swapping constructs showed the same phenotypes as intact DCA1. However, during the course of the domain swapping experiments, we found that mll0950, like DCA1, could completely convert the Fix⁺ phenotype of MAFF303099 to Fix⁻ when it was expressed by the trp-promoter of the pDG77 vector (45) (SI Appendix, Fig. S12). This result implies that the difference in the expression level of the respective autotransporter gene determines the nitrogen fixation phenotypes of apn1. Therefore, we analyzed the promoter activities of mll0950 and DCA1 in nodules. The promoter-GUS assay showed stronger activity of the DCA1 promoter compared to that of mll0950 in both wild-type and apn1 nodules (Fig. 5A).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Promoter activity of M. loti autotransporter and induction of cysteine protease genes by DCA1. (A) Promoter-GUS assay of DCA1 and mll0950 in nodules of wild-type (WT; MG-20 Miyakojima) and apn1 mutant at 14 d after inoculation. (Scale bars: 1 mm.) (B and C) Expression of two groups of cysteine protease genes, chr3.LjB03G07.10.r2.a (B) and chr3.CM0649 (C), in apn1 nodules at 7 and 10 d after M. loti inoculation (dai). Bars represent the mean values ± SEM of four independent experiments and are shown as relative expression to the level of MAFF nodules that is set to 1. All values were normalized against the expression of a reference gene, elongation factor 1 alfa. Asterisks indicate the significance of differences from MAFF nodule (*P < 0.05, Student’s t test). (D–G) Representative images of nodules formed on hairy roots transformed with the empty vector (D and E) and with vector carrying construct overexpressing cysteine protease gene chr3.LjB03G07.10.r2.a (F and G). Arrowheads indicate the nodules formed on the transformed hairy roots. Pictures were taken under bright (D and F) and fluorescence field of GFP (E and G). (Scale bars: 2 mm.) (H) Percentage of mature pink nodules in total nodules formed on hairy roots transformed with cysteine protease genes. Bars represent the mean values ± SEM of three independent experiments. n indicates the total number of nodules analyzed. Asterisks indicate the significance of differences from empty vector (**P < 0.01, Student’s t test).

Transcriptome Analysis of the apn1 Nodules.

To identify the host plant genes that were differentially regulated depending on M. loti strains, an RNA-seq analysis was carried out in the nodules of the wild type (MG-20 Miyakojima) and apn1 mutant formed by the inoculation with MAFF303099 or TONO. Differentially expressed genes specific to apn1 Fix⁻ nodules are listed in SI Appendix, Fig. S13 and Table S2. At 5 and 7 d after M. loti inoculation (dai), when no visible phenotypic differences of nodules were observed between MAFF303099 and TONO (SI Appendix, Fig. S13A), fewer genes were differentially expressed in apn1 Fix⁻ nodules than were common between Fix⁺ and Fix⁻ apn1 nodules (SI Appendix, Fig. S14 and Table S3), indicating that only minor changes of gene expression might provoke nitrogen fixation deficiency in the apn1 mutant. Among the specifically regulated genes in apn1 Fix⁻ nodules, the gene encoding cysteine protease (chr3.LjB03G07.10.r2.a) belonging to the senescence-associated gene 12 (SAG12) family (49) was significantly up-regulated at 7 dai (SI Appendix, Fig. S13C and Table S2). Orthologous cysteine protease genes of M. truncatula and soybean were reported to be induced in old or stress-treated nodules (50, 51). Real-time RT-PCR verified expression of the cysteine protease gene in the nodules formed by inoculation with MAFF303099 overexpressing DCA1 (MAFF_DCA1) or with two DCA1 transposon mutants (dca1-1 and -5). At 7 dai, chr3.LjB03G07.10.r2.a was clearly induced in nodules formed by inoculation with TONO, which was much higher in those inoculated with MAFF_DCA1 as compared to those with wild-type MAFF303099 and two transposon mutants (Fig. 5B). Induction of the cysteine protease gene was more apparent in 10-dai nodules of TONO and MAFF_DCA1, and the latter was several tens of times higher than the former (Fig. 5B). L. japonicus has another cysteine protease gene cluster of the SAG12 family (chr3.CM0649.-270.r2.d, -280.r2.d, -300.r2.d, and -310.r2.d; SI Appendix, Fig. S15), and, in the RNA-seq analysis, these cysteine protease genes were not differentially expressed in apn1 Fix⁻ nodules. Consistent with the RNA-seq data, RT-PCR analysis showed that chr3.CM0649 genes were not induced in apn1 nodules when analyzed with PCR primers that simultaneously amplify the chr3.CM0649 genes (Fig. 5C). However, these cysteine protease genes were significantly up-regulated in 10-dai Fix⁻ nodules induced by MAFF_DCA1 (Fig. 5C). Thus, our expression analyses show that, in the absence of functional APN1, DCA1 induces cysteine protease genes depending on its expression and activity.

M. loti strain TONO formed dark brown nodules on the root of apn1 mutant (Fig. 1A and SI Appendix, Fig. S1) (32). In M. truncatula, such accumulations of brown pigment in nodules were associated with the induction of defense responses (17, 36, 52, 53). To clarify whether the defense responses were induced in apn1 Fix⁻ nodules, we examined the expression of some defense-related Medicago gene homologs in L. japonicus. Up to 10 dai, no significant induction of defense gene homologs was observed in TONO Fix⁻ nodules compared to Fix⁺ nodules induced by MAFF303099 (SI Appendix, Fig. S16). In the one exception, the gene encoding a vacuole processing enzyme (VPE; Lj5g3v0369970.1) showed a slight but significant increase in 10-dai Fix⁻ nodules, which was enhanced by the overexpression of DCA1 (SI Appendix, Fig. S17).

Expression of Cysteine Protease and DCA1 Negatively Affect the Symbiotic Nitrogen Fixation.

To clarify the role of the cysteine protease gene induced by DCA1 in nodule symbiosis, we examined the effects of overexpression of cysteine protease genes on nodule phenotypes. When cysteine protease genes, chr3.LjB03G07.10.r2.a and its paralogous gene (chr3.LjB03G07.50.r2.a; SI Appendix, Fig. S15), were expressed by CaMV 35S promoter or by the nodule-specific leghemoglobin (LjLb3) promoter, formation of effective nodules was significantly decreased compared to empty vector (Fig. 5 D–H). Thus, the cysteine proteases negatively affect nodule maturation, resulting in an increase of ineffective nodules. We also analyzed the effect of DCA1 overexpression on nodule symbiosis. In the wild-type plant, MAFF303099 with DCA1 overexpression (MAFF_DCA1) formed more immature nodules compared to wild-type MAFF303099 at 4 wk after inoculation (SI Appendix, Fig. S18A), leading to reduced shoot growth and nitrogen fixation activity (SI Appendix, Fig. S18 B–D). Because the expression of DCA1 had no significant effects on the growth and nitrogen assimilation of plants (SI Appendix, Fig. S19), the reduction in the shoot growth of MAFF_DCA1 inoculated wild type plants appears to be due to the reduced nitrogen fixation by DCA1 (SI Appendix, Fig. S18D). These phenotypes likely arose because APN1 could not completely suppress the effects of DCA1. In addition, wild-type MAFF303099 formed more immature nodules on apn1 than wild-type roots, perhaps because weak but residual activity of the MAFF3033099 autotransporter appeared in the absence of functional APN1 (SI Appendix, Fig. S18A).