Auxin methylation by IAMT1, duplicated in the legume lineage, promotes root nodule development in Lotus japonicus.

Legumes attract symbiotic bacteria and create de novo root organs called nodules. Nodule development consists of bacterial infection of root epidermis and subsequent primordium formation in root cortex, steps that need to be spatiotemporally coordinated. The Lotus japonicus mutant “daphne ” has uncoupled symbiotic events in epidermis and cortex, in that it promotes excessive bacterial infection in epidermis but does not produce nodule primordia in cortex. Therefore, daphne should be useful for exploring unknown signals that coordinate these events across tissues. Here, we conducted time-course RNA sequencing using daphne after rhizobial infection. We noticed that IAA carboxyl methyltransferase 1 (IAMT1) , which encodes the enzyme that converts auxin (IAA) into its methyl ester (MeIAA), is transiently induced in wild-type roots at early stages of infection but shows different expression dynamics in daphne. IAMT1 serves an important function in shoot development of Arabidopsis, a nonsymbiotic plant, but the function of IAMT1 in roots has not been reported. Phylogenetic tree analysis suggests a gene duplication of IAMT1 in the legume lineage, and we found that one of the two IAMT1s (named IAMT1a) was induced in roots by epidermal infection. IAMT1a knockdown inhibited nodule development in cortex; however, it had no effect on epidermal infection. The amount of root MeIAA increased with rhizobial infection. Application of MeIAA, but not IAA , significantly induced expression of the symbiotic gene NIN in the absence of rhizobial infection. Our results provide evidence for the role of auxin methylation in an early stage of root nodule development.

Legumes develop de novo organs known as root nodules to accommodate symbiotic bacteria called rhizobia. Nodule formation involves two distinct processes: rhizobial infection involving host–microbe communication via signaling molecules in root epidermis, and nodule primordium development accompanied by cell division in root cortex. In epidermis, rhizobia-derived lipochitin oligosaccharide (Nod factor) binds to LysM receptor-like kinases (NFR1/5) in host root hair cells (1–3). This triggers periodic calcium spiking, which is decoded by the calcium/calmodulin-dependent protein kinase (CCaMK) (4–7). CYCLOPS, a direct phosphorylation substrate of CCaMK, acts as a transcription activator of NODULE INCEPTION (NIN) and ERF REQUIRED FOR NODULATION (ERN) (8–10), which are necessary to form microcolonies in infection chambers and infection threads (ITs; plant-derived intracellular tube-like structures) (11–15).

Cortical cell division, which occurs just below the site of rhizobial infection in epidermis, is required for primordium development. Phytohormones are important for cortical cell division. Exogenous cytokinin application induces ectopic cortical cell division (16, 17). Some cytokinin receptor genes, such as Lotus histidine kinase (LHK), are induced in dividing cortical cells upon rhizobial infection (18). Gain-of-function LHK1 causes spontaneous nodulation (19). Loss of function or knockdown (KD) of LHK1 or the homologous gene, Medicago truncatula CRE1, inhibits nodulation (16, 20). Cortical cell division also provides an indispensable scaffold for IT progression from epidermis to cortex (21), and rhizobia are released from intracellular ITs ramified in nodule primordia, leading to successful nodule organogenesis. Recently, it has been reported that symplastic communication by callose turnover at plasmodesmata is important for coordinating epidermal infection and nodule development (22). These findings suggest that spatiotemporal coordination across epidermis and cortex is essential for this symbiotic organogenesis. Epidermal expression of genes required for calcium spiking, such as CASTOR and POLLUX (DOES NOT MAKE INFECTIONS 1 [DMI1] in Medicago) (23, 24), NUP85 (25), and NUP133 (26), is sufficient for nodule formation (27), suggesting that some kinds of signals generated in epidermis trigger cortical cell division. However, little is known about the mechanism that coordinates these two events.

Among various symbiotic mutants of Lotus japonicus, daphne is an intriguing nonnodulation mutant in which epidermal infection is uncoupled from cortical cell division (28). In daphne, excess ITs are formed in the epidermis but cortical cell division is not activated. The daphne mutation is a chromosome translocation 7 kb upstream of the NIN start codon, resulting in a lack of NIN expression in cortex but not in epidermis. NIN expression is involved in both epidermal IT formation and initiation of cortical cell division (10, 11, 29), and the regulatory nucleotide sequences for NIN expression differ between root tissues (28, 30). CYC box, the CYCLOPS binding site, in the NIN promoter is sufficient for IT formation. The cytokinin response element in the NIN promoter is required for nodule formation in Medicago (30). Cortical NIN not only induces cortical cell division but also represses excessive infection in epidermis, and a lack of cortical NIN expression causes the daphne phenotype (28). Therefore, in daphne, signals derived from infection in epidermis should be overproduced, and signals after cortical NIN-derived cell division should be reduced. On the other hand, signals that are not reflected in the daphne phenotype, including those that induce cortical cell division upstream or independent of cortical NIN and are derived from epidermal infection, may be overproduced in daphne. Therefore, use of daphne to explore transcriptional profiles may allow us to uncover genes and factors that coordinate these two events, in addition to the molecular mechanisms of epidermal infection and cortical cell division.

Here, we conducted a time-course transcriptome analysis of daphne, and identified genes that showed different expression patterns in daphne and wild type (WT). Among these genes, we found IAA CARBOXYL METHYLTRANSFERASE 1 (IAMT1), which encodes the enzyme that specifically converts auxin (indole-3-acetic acid; IAA) into methyl-IAA (MeIAA) (31–33). IAMT1 is essential for shoot development and differential growth in Arabidopsis, a nonsymbiotic plant (34, 35), but, as far as we know, there have been no reports on detailed expression and function analysis of IAMT1 in roots. In this study, we found that IAMT1 is duplicated in the legume lineage, and one of the duplicates (named IAMT1a) is mainly expressed in epidermis, whereas reverse genetic analysis showed that IAMT1a is crucial for nodule development, rather than for epidermal infection. A significant MeIAA increase after rhizobial infection was detected by using daphne roots. Furthermore, expression of NIN in WT roots increased after MeIAA treatment, in contrast to IAA treatment. Based on these findings, herein we discuss how MeIAA properties differ from those of IAA and how MeIAA may be a signaling molecule that links different events in epidermis and cortex.

Results

Time-Course Transcriptome Analysis of L. japonicus MG-20 and daphne.

We performed time-course RNA sequencing (RNA-seq) on L. japonicus WT MG-20 and daphne at early time points after rhizobial inoculation. We set four time points (0 d after inoculation [DAI] [noninoculation], and 1, 2, and 3 DAI): At 1 and 2 DAI, root hair deformation and microcolony entrapment were observed. At 3 DAI, ITs were observed in root epidermis and cortical cell division occurred in WT, while no cortical cell division was observed in daphne, despite excessive IT formation.

To identify significant differentially expressed genes (DEGs) during the time course, maSigPro (36) was used. Using a false discovery rate <0.05 as a cutoff, 4,871 genes were classified as time-course DEGs (Dataset S1). Hierarchical clustering of time-course DEGs that changed >2-fold (1,076 genes) revealed four subgroups, based upon expression patterns (Fig. 1): In cluster I (473 genes), transcript levels increased at 1 DAI in WT but increased to a greater extent and more persistently in daphne (Fig. 1). In cluster II (204 genes), transcription was activated at 1 DAI in daphne whereas, in WT, transcription was unchanged or attenuated (Fig. 1). Cluster III included 222 genes that were more highly up-regulated in WT than daphne (Fig. 1). Cluster IV grouped 177 genes that displayed temporal up-regulation in WT but for which expression was not altered in daphne (Fig. 1). For example, genes associated with infection events in epidermis, such as genes involved in IT formation and/or that act from infected epidermis to cortex, may be included in clusters I and III, which show increased expression in WT. In addition, genes involved in excessive IT formation in daphne are most likely to be included in cluster I. On the contrary, genes that positively regulate nodule primordium formation and/or act repressively from cortex to epidermal infection can be included in cluster IV, where no induction of expression occurs in daphne.

Fig. 1.

Time-course RNA-seq in WT and daphne. (A) Classification of DEGs with fold change >2 (1,181 genes) into four subgroups (I to IV) by hierarchical clustering. (B) Expression modules of genes with significant differences between WT (black lines) and daphne (orange lines) during early infection. Each dot in each cluster represents an average value.

Phylogenetic Analysis and Expression of L. japonicus IAMT1.

Lj2g3v3222870 was one of the most differentially expressed DEGs in cluster I (P = 3.97 × 10−97). A phylogenetic tree showed that Lj2g3v3222870 is included in the IAMT1 clade of the SABATH family, which comprises a group of small-molecule methyltransferases (Fig. 2). IAMT1 encodes an enzyme that specifically converts IAA into its methyl ester (31–33) (Fig. 2) and, unlike other SABATHs, it has an amino acid substitution (Trp-256 to Gly-256 in AtIAMT1) that is required for recognition and binding of the IAA indole ring (32, 37). This amino acid substitution was also confirmed in Lj2g3v3222870 (). Based on this feature and its phylogenetic relationship, we conclude that Lj2g3v3222870 is an ortholog of Arabidopsis IAMT1. A BLASTP search identified another IAMT1 gene (Lj6g3v0819010) in L. japonicus; hence, the two LjIAMT1 genes (Lj2g3v3222870 and Lj6g3v0819010) were named LjIAMT1a and LjIAMT1b, respectively. An analysis of a phylogenetic tree of IAMT1 proteins from various plant species suggested that the gene duplication of IAMT1 occurred in the common ancestor of legumes (Fig. 2).

Fig. 2.

Expression patterns of two IAMT1 genes in L. japonicus and phylogenetic trees containing these genes. (A) Arabidopsis IAMT1 specifically converts IAA into MeIAA in vitro (31–33). (B) Phylogenetic tree of Arabidopsis carboxyl methyltransferases in the SABATH family, including OsIAMT1 (37), and Lj2g3v3222870. (C) Fabaceae lineage-specific duplication of IAMT1. An asterisk indicates the duplication in the Fabaceae. (D) Lj2g3v3222870 (LjIAMT1a), but not Lj6g3v0819010 (LjIAMT1b), was detected as a DEG in time-course RNA-seq analysis. mRNA abundance of WT (gray bars) and daphne (black bars) in LjIAMT1a and LjIAMT1b at 0 (noninoculation), 1, 2, and 3 DAI. Error bars indicate means ± SDs of three biological replicates.

Although both LjIAMT1a and LjIAMT1b highly share a conserved sequence containing the amino acid substitution characteristic of IAMT1 (), IAMT1a but not IAMT1b was differentially expressed in time-course RNA-seq as well as qRT-PCR (Figs. 2 and 3 and ). Despite a highly conserved similarity in the legume lineage, the messenger RNA (mRNA) abundance of IAMT1a in roots estimated from RNA-seq data was ∼100 times higher than that of IAMT1b (Fig. 2).

Fig. 3.

Genetic dependencies of IAMT1a expression in the early infection stage. Time-course qRT-PCR analysis of IAMT1a expression in WT, daphne, nin-9, ccamk-14, and ern1-6 at 0 (noninoculation), 1, 2, and 3 DAI. Data are means of three or more biological replicates and are displayed as values relative to WT at 0 DAI. Error bars indicate means ± SDs (n = 12 plants for each biological replicate). Statistical analysis was performed using ANOVA followed by Tukey’s honest significant difference (HSD) test (P < 0.05) in each genetic background. Different letters indicate significant differences. There were no significant differences (n.s) in ccamk-14 and ern1-6.

IAMT1a Expression Pattern in the Early Infection Stage.

To determine the genetic dependency of transcriptional changes in IAMT1a, we conducted time-course qRT-PCR experiments on a series of symbiotic mutants after rhizobial infection. In a nin-null mutant (nin-9), as well as in daphne, IAMT1a was induced more highly and continuously than in WT (Fig. 3). This indicates that NIN is at least unnecessary for induction of IAMT1a expression. In contrast, IAMT1a was not induced in ccamk-14 or ern1-6 (Fig. 3), indicating that IAMT1a is induced downstream of CCaMK and ERN1 in the symbiotic pathway.

To identify the expression site of IAMT1a during early rhizobial infection, we performed a histochemical analysis. β-Glucuronidase (GUS) signals driven by the 2.9-kbp IAMT1a promoter in the WT background were detected in the rhizobia-susceptible region at 2 DAI (Fig. 4 ). Expression of proIAMT1a:tripleYFP-nls was observed in root epidermis of the susceptible region at the same time (Fig. 4). However, the GUS signal was attenuated after epidermal ITs were formed (Fig. 4 ). These changes in IAMT1a promoter activity were consistent with transient increases in its mRNA levels in WT as detected by RNA-seq and qRT-PCR (Figs. 2 and 3). In contrast, in the daphne background, the susceptible window remains open (28), and GUS signals were detected in the broader root region after inoculation (Fig. 4 ). Interestingly, GUS signals were detected in the region in which epidermal IT formation was observed in daphne (Fig. 4). These patterns are consistent with persistent increases in its mRNA levels in daphne (Figs. 2 and 3).

Fig. 4.

Spatiotemporal profile of IAMT1a expression in WT and daphne roots inoculated with or without rhizobia. WT (A–D and H–J) and daphne (E–G and K) roots were transformed with proIAMT1a:GUS or proIAMT1a:tripleYFP-nls (I). GUS activity was observed at 0 DAI (noninoculation; A and E) and 2 DAI (B) and after ITs developed (C and F). The arrowhead indicates active GUS sites in WT. DsRed-labeled M. loti was infected in epidermis (D and G). Magnified images of the susceptible region of WT at 2 DAI (H and I) and the root region where epidermal ITs were observed in WT (J) and daphne (K). Images merged with DsRed fluorescence are shown. (Scale bars, 0.5 mm [A–G] and 100 µm [H–K].)

IAMT1a Knockdown Affected Cortical Events, but Not Epidermal Infection.

To examine involvement of IAMT1a in nodulation, we performed RNA interference (RNAi) KD analysis of IAMT1a in L. japonicus. We prepared three constructs for KD that targeted different sequences (5′ untranslated region or coding sequence). IAMT1a and IAMT1b expression levels were analyzed in roots with real-time RT-PCR, 3 wk after inoculation. In roots transformed with RNAi constructs, IAMT1a transcription levels were reduced to less than half of controls (10−4 < P < 10−2) (). Transcription levels of IAMT1b also tended to decrease (0.2 < P < 0.4) (). On average, IAMT1a-RNAi-2 reduced IAMT1a transcripts to 10% of control levels. IAMT1a-RNAi-2 was the most effective for decreasing IAMT1a transcripts, but IAMT1a-RNAi-2 had the weakest effect on reducing IAMT1b transcripts (). When the number of nodules was measured 3 wk after inoculation, the number of nodules decreased significantly in hairy roots transformed with IAMT1a-RNAi vectors (Fig. 5). Nodules were not observed in 21 to 33% of plants with hairy roots harboring IAMT1a-RNAi vectors, although nodules formed in all controls (Fig. 5). IAMT1a-RNAi also significantly inhibited formation of nodule primordia at 7 DAI (). Interestingly, in IAMT1a-RNAi hairy roots without nodules, ITs wandered in epidermis but did not enter cortex (). This is similar to the symbiotic phenotype of L. japonicus vag1 and daphne mutants (21, 28).

Fig. 5.

IAMT1a-RNAi inhibits nodulation. (A) Representative phenotype of hairy roots of WT harboring an empty vector (EV) as a control (Left) and the IAMT1-RNAi-2 vector (Right) 3 wk after inoculation. Roots expressing green fluorescent protein as a transformation marker were selected. Root nodules are indicated by arrowheads. (Scale bars, 1 cm.) (B) The nodule number in hairy roots harboring an EV (controls) and IAMT1-RNAi vectors 3 wk after inoculation. Each dot represents the nodule number of each plant. n = 37 (control), 40 (RNAi-1), 40 (RNAi-2), and 28 (RNAi-3). Asterisks indicate that differences are statistically significant (Welch’s t test).

IAMT1a-RNAi seemed not to affect epidermal infection of rhizobia in WT (). To further confirm this, we performed IAMT1a-RNAi using a daphne nonnodulating mutant, which has excessive ITs due to deficient negative feedback by cortical NIN. As a result, excessive ITs of daphne were kept in IAMT1a-RNAi hairy roots 2 wk after inoculation without reduction (). These results suggested that IAMT1a contributes more to cortical events than to epidermal infection.

To assess IAMT1a function in nodule development, we performed IAMT1a-RNAi in the absence of rhizobia, using spontaneous nodule formation (snf) mutants such as constitutively expressing a gain-of-function CCaMKT265D (snf1-like) (7, 38) or a gain-of-function LHK1 cytokinin receptor (snf2) (19). IAMT1a-RNAi inhibited spontaneous nodulation in snf1-like (). This indicated that IAMT1a acts downstream of CCaMK, consistent with the fact that IAMT1a expression was not induced in the ccamk mutant after rhizobial inoculation (Fig. 3). On the other hand, IAMT1a-RNAi did not affect spontaneous bump formation in the snf2 mutant (). This indicates that the function of IAMT1a is not under control of LHK1-mediated cytokinin signaling in nodule development.

Overexpression of IAMT1a Promoted Nodulation in the tml-4 Mutant.

To investigate whether IAMT1a positively regulates nodule development, we overexpressed IAMT1a. IAMT1a overexpression had no effect on nodule number in WT (Fig. 6). However, in tml-4 mutants, which produce excessive ITs and nodules due to lack of autoregulation of nodulation (39, 40), an increased number of nodules was observed in overexpressed IAMT1a (Fig. 6). In addition, we confirmed the correlation between expression levels of IAMT1a and nodule number (). These results show that IAMT1a is a positive regulator of nodule development.

Fig. 6.

IAMT1a overexpression promotes nodulation. Transgenic hairy roots harboring the proLjUBQ:IAMT1a vector and EV control were generated in WT (Left) and the tml-4 mutant (Right). Nodules were counted 3 wk after inoculation. Each dot represents the nodule number of each plant. n = 28 (control/WT), 30 (ox/WT), 30 (control/tml-4), and 22 (ox/tml-4). Asterisks indicate that differences are statistically significant (Welch’s t test).

Involvement of Auxin Methylation in Nodule Development.

To clarify the presence of endogenous MeIAA during nodulation, we tried to detect MeIAA before and after rhizobial infection. Identification of endogenous MeIAA is generally difficult, because the amount of MeIAA is much less than that of IAA (35). Therefore, we used daphne, in which rhizobial infection and accumulation of IAMT1a transcripts were enhanced (Figs. 2 and 3). The use of daphne could facilitate the capture of quantitative change of MeIAA during nodulation. First, we confirmed that overexpression of IAMT1a in hairy roots increased MeIAA levels (). Then, we detected the critical MeIAA peak especially in infected roots of daphne at 2 DAI (). We measured amounts of IAA and MeIAA at 0 DAI (noninoculation) and 2 DAI in WT and daphne. Although no significant change in the amount of IAA or MeIAA could be detected in WT before or after rhizobial infection, a significant MeIAA increase after rhizobial infection was detected in daphne (Fig. 7). Furthermore, we performed constitutive expression of MES17, which encodes the enzyme that converts MeIAA to IAA (Fig. 7) (41), to counteract the catalytic function of IAMT1a during nodulation. Constitutive expression of Lj2g3v2171910, a gene homologous to Arabidopsis MES17 (), resulted in a statistically significant decrease in MeIAA levels and nodule number compared with WT (Fig. 7 and ). These data indicate the importance of auxin methylation in nodule development. Furthermore, to gain insight into the role of auxin methylation, we tested the effect of exogenous MeIAA on NIN expression. NIN is a key transcription factor of cortical cell division for nodule development (29). Treatment with IAA did not induce NIN expression in L. japonicus roots (Fig. 7), consistent with findings of Soyano et al. (42). However, treatment with MeIAA did induce NIN expression (Fig. 7). This induction of expression was not detected in daphne (Fig. 7), suggesting that MeIAA affects cortical NIN expression. Finally, NIN expression was induced at 7 DAI in hairy roots harboring control vectors, but was poorly induced in hairy roots harboring IAMT1a-RNAi constructs (Fig. 7). These findings indicate that auxin methylation by IAMT1a is involved in nodule development by affecting NIN expression.

Fig. 7.

Auxin methylation and NIN expression. (A) Relative amounts of IAA and MeIAA at 0 DAI (noninoculation) and 2 DAI in WT and daphne. Error bars indicate means ± SDs of three biological replicates (n = 40 plants for each biological replicate). (B) MES17 demethylates MeIAA in vitro (41). (C) Nodule numbers in controls and constitutive expression of LjMES17 3 wk after inoculation. Each dot represents the nodule number of each plant. n = 34 (control) and 36 (ox). (D) Relative expression levels of NIN after treatment with dimethyl sulfoxide (DMSO) as mock, IAA (10−7 M), or MeIAA (10−7 M) for 24 h in WT. (E) Relative expression levels of NIN after treatment with DMSO as mock or MeIAA (10−7 M) for 24 h in WT and daphne. Error bars indicate means ± SDs of six biological replicates (n = 10 plants for each biological replicate) (D and E). Asterisks indicate that differences are statistically significant (Welch’s t test) (A and C–E). (F) Relative expression levels of NIN in hairy roots harboring an EV as controls and IAMT1a-RNAi-2 vectors at 0 DAI (noninoculation) or 7 DAI. Error bars indicate means ± SDs of three or five biological replicates in control or RNAi, respectively (n > 10 hairy roots for each biological replicate). Statistical analysis was performed using ANOVA followed by Tukey’s HSD test (P < 0.05) in each genetic background. Different letters indicate significant differences.

Discussion

IAMT1 has been characterized as a gene encoding carboxyl methyltransferase, which specifically converts IAA to MeIAA in vitro (31–33). In Arabidopsis, a nonsymbiotic plant, IAMT1 participates in MeIAA biosynthesis in vivo (35) and in shoot development and differential growth (34, 35). On the other hand, the function of IAMT1 in roots is unknown. This study demonstrates that L. japonicus IAMT1 functions in root nodule development. We found an IAMT1 gene duplication in the Fabaceae lineage and characterized one of two IAMT1 genes, named IAMT1a, induced in roots after rhizobial infection, as a positive regulator of nodule development. Notably, we identified the increase of MeIAA in roots after rhizobial infection using daphne (Fig. 7 and ). Because MeIAA is much less abundant than IAA (35), a quantitative change of endogenous MeIAA in biological processes has not been reported. In this study, however, the use of daphne allowed us to detect a significant increase of MeIAA levels associated with induction of IAMT1a expression mediated by rhizobial infection.

We documented induction of IAMT1a expression in nodulation using time-course RNA-seq in early symbiotic stages using daphne. IAMT1a is one of the most significant genes in a cluster of DEGs that are transiently induced in WT roots but continuously and more strongly expressed in daphne roots after rhizobial infection (Fig. 1). Consistent with changes of IAMT1a transcript levels detected using RNA-seq and qRT-PCR, upon epidermal infection, IAMT1a promoter activity is transiently observed in a local infectable region, but not throughout entire roots in WT, whereas it is persistently observed in wide regions of daphne roots (Figs. 2, 3, and 4). daphne lacks the promoter region of NIN expression in cortex (28), and the nin-null mutant shows persistent expression of IAMT1a, as well as in daphne (Fig. 3), suggesting that characteristic spatiotemporal expression patterns of IAMT1a in daphne result from a lack of cortical NIN. Cortical NIN provides negative feedback and suppresses persistent, widespread epidermal infection (21, 30, 43). Early nodulin 11 is extensively expressed in the Mtnin-1 mutant (10). Given this evidence, IAMT1a expression is probably under negative feedback control by cortical NIN.

A BLAST search of legumes and phylogenetically closely related nonlegumes showed that legumes have two IAMT1 genes, and a phylogenetic tree suggested that IAMT1a and IAMT1b genes originated from IAMT1 duplication in the common ancestor of legumes (Fig. 2). IAMT1b, which shares 90% amino acid sequence identity with IAMT1a, also has a conserved amino acid sequence for the auxin-binding pocket (). Although it is assumed that both share the same molecular function, mRNA levels of IAMT1a are about 10-fold higher than those of IAMT1b in noninoculated roots. IAMT1a, but not IAMT1b, is induced after rhizobial infection (Fig. 2 and ) and the difference in expression levels then becomes about 400-fold (Fig. 2). Recently, the existence of genomic clusters, termed symbiotic islands, has been demonstrated in legume genomes, in which symbiotic genes are colocalized (44). In the L. japonicus MG-20 ecotype, IAMT1a is located on chromosome 2 while IAMT1b is on chromosome 6. Establishment of a new gene locus by gene duplication in IAMT1 may have driven IAMT1a expression in roots, further leading to involvement of auxin methylation in nodule development. Considering that Arabidopsis IAMT1 is required for development and differential growth in shoots, and that it functions in leaf development and gravitropic reorientation in hypocotyls (34, 35), IAMT1b may have inherited the function of IAMT1 in shoots of nonlegumes.

Auxin is involved in various processes in nodule symbiosis. During early infection stages, auxin biosynthesis occurs in epidermis, and auxin signaling has been observed in infected epidermis (45). Auxin response factor (ARF) 16a participates in IT formation (46). During the postcortical cell-division stage, LjYUCCA1/MtYUC8 and LjYUCCA11/MtYUC2, encoding auxin biosynthetic enzymes, are expressed (47, 48), and these are downstream factors of SHORT INTERNODES/STYLISH (STY), required for nodule emergence (47). GmYUC2a is induced after rhizobial infection and is involved in nodule formation (49). Posttranscriptional control via miR160 regulates ARF10/16/17 in soybean (50) and Medicago (51) in nodule developmental stages.

Considering that auxin biosynthesis occurs in epidermis during early stages of infection, auxin accumulation in epidermis may provide a substrate for IAMT1a in vivo. As evidence to support this, MeIAA levels showed an increasing trend in WT and were significantly elevated in daphne due to rhizobial infection (Fig. 7), where IAMT1a is extensively expressed (Fig. 4). In the beginning, we assumed the involvement of IAMT1a in epidermal events, such as IT formation. However, unexpectedly, IAMT1a KD had no effect on IT number but inhibited nodule and primordium development involving cortical infection (). These findings indicate that IAMT1a is involved in cortical events for nodule development. Consistent with this result, spontaneous nodulation with constitutive expression of CCaMK [a gain-of-function type of CCaMK (7, 38)] was inhibited by IAMT1a KD (Fig. 6). Since the ccamk mutant shows no induction of IAMT1a expression after infection (Fig. 3), IAMT1a is positioned as a downstream factor of CCaMK. On the other hand, spontaneous nodulation in the snf2 mutant [a gain-of-function mutant of LHK1 (19)] was not significantly inhibited by IAMT1a KD (Fig. 6). Given that LHK1 expression in cortex but not epidermis is sufficient to restore bump formation in lhk1 mutants in the absence of rhizobial infection (52), although LHK1 is expressed in epidermis and cortex (18), IAMT1a can act either in parallel with or upstream of cytokinin signaling via LHK1 in cortex.

Arabidopsis MES17 has been identified as an MeIAA esterase in vitro (41). Constitutive expression of the homolog in L. japonicus decreased endogenous MeIAA levels and nodule number (Fig. 7 and ). Nodule development was inhibited by IAMT1a KD (Fig. 5) and further enhanced by its overexpression in the tml background (Fig. 7 and ). These results show that auxin methylation is an important process during nodule development. It is interesting to note that MeIAA has different properties from those of IAA. Soyano et al. (42) found that expression of NIN is not induced by exogenous IAA in L. japonicus, and the present work confirmed that finding (Fig. 7). In contrast, exogenous MeIAA does induce NIN expression (Fig. 7). Expression induction does not occur in daphne (Fig. 7), which lacks the promoter region for cortical NIN expression, and IAMT1a KD inhibited NIN expression during the nodule developmental stage (Fig. 7), suggesting that MeIAA contributes to induction of cortical NIN expression for nodule development.

Results of these experiments suggest that auxin methylation is not simply due to alteration of auxin homeostasis, and suggest the following hypotheses. First, in intracellular signaling, MeIAA is a probable signaling molecule distinct from IAA. Since the IAA receptor, TIR1, recognizes the carboxyl group of IAA (53), MeIAA is not likely to be recognized by TIR1, which supports our hypothesis. Second, in intercellular signaling, MeIAA could have different mobility characteristics, as previous studies have noted (34, 54). MeIAA is a nonpolar molecule, unlike auxin and other auxin conjugates. In general, nonpolar molecules can penetrate cell membranes, but molecules such as abscisic acid and glycerol move via transporters. MeIAA rescues a part of the phenotype of the Arabidopsis aux1 mutant (54), suggesting that MeIAA can traverse the membrane or that it moves via the AUX1-independent influx system into cells, where it may work as a donor of IAA. During nodule development, it may be that MeIAA moves from epidermis into cortex and induces cortical NIN. MeIAA could migrate from epidermis to cortex, and be hydrolyzed in cortex to produce IAA. Investigating this possibility and whether IAA can induce NIN in cortex are interesting issues for future research. Cortical cell division just below epidermal cells infected with rhizobia is essential for symbiotic nodule formation. An analysis of auxin methylation and MeIAA function should open a new avenue for understanding the linkage between infection and development in nodule symbiosis.

Materials and Methods

Plant Material and Growth Conditions.

L. japonicus Miyakojima MG-20 ecotype (55) was used as WT and the common genetic background for the following mutants: ccamk-14 (56), daphne, ern1-6, nin-9, snf2, and tml-4 (13, 28, 40, 57, 58). Three-day-old seedlings were transferred to culture vessels containing sterilized vermiculite with B&D medium (59) and grown for 3 d for adaptation. Plants were then inoculated with Mesorhizobium loti MAFF303099 [DsRed-labeled for microscopic observation (60)] suspended in B&D medium.

Time-Course RNA-Seq.

Roots of WT and daphne cultured without (0 DAI) or with inoculation (1, 2, and 3 DAI) were harvested. Total RNA was isolated using the RNeasy Plant Mini Kit (QIAGEN) and with DNA removed by treatment with DNase (QIAGEN). Three independent biological replicates (each with n = 20 roots) were included for each time point. RNA-seq libraries were prepared using the NEBNext Ultra II RNA Library Prep Kit for Illumina (NEB) and NEBNext Poly(A) mRNA Magnetic Isolation Module (NEB). Libraries were sequenced using an Illumina HiSeq 2000 and generated 66-bp single-end reads. A detailed description of bioinformatic analysis is provided in . The raw reads have been deposited in the DNA Data Bank of Japan (DDBJ) Sequence Read Archive (DRA) under accession number DRA013121.

Alignment and Phylogenetic Tree Construction.

Alignment of sequences retrieved from the National Center for Biotechnology Information Genome database and Phytozome v12 (https://phytozome.jgi.doe.gov/pz/portal.html) was performed using Clustal X. Poorly aligned regions were automatically removed by trimAl (61). Phylogenetic trees were estimated by the maximum-likelihood method and constructed using the trimmed amino acid sequence and IQ-TREE (62). ModelFinder (63) was used for model selection.

Expression Analysis.

Primers used for qPCR are listed in . Roots were harvested for RNA extraction using PureLink Plant RNA reagents (Invitrogen). After treatment with DNase I (Takara) to remove genomic DNA, complementary DNA was synthesized from <0.1 µg of RNA using ReverTra Ace qPCR RT Master Mix (Toyobo). qPCR was performed with Thunderbird SYBR qPCR Mix (Toyobo) on a LightCycler 96 System (Roche) according to the manufacturer’s protocol. Expression of LjUBQ was used as a reference.

Plasmid Construction and Transformation.

Detailed information is provided in .

Microscopy.

Bright-field and fluorescence microscopy was performed with a BX50 upright microscope (Olympus) or with an A1R confocal microscope (Nikon). Images were acquired and analyzed using DP Controller (Olympus) and NIS Elements (Nikon).

Histochemical Analysis.

Hairy roots of WT and daphne were transformed with the GUS-reporter gene fused to the IAMT1a promoter. Roots were incubated in histochemical GUS staining solution (100 mM NaPO4, pH 7.0, 0.5 mg⋅mL−1 5-bromo-4-chloro-3-indolyl-β-glucuronic acid, 2 mM K4Fe[CN]6, 2 mM K3Fe[CN]6, and 0.1% Triton X-100) for <60 min at 37 °C after a 10-min vacuum filtration.

Quantification of IAA and MeIAA.

Forty roots were harvested in each biological replicate of WT and daphne. For preparation of plant extracts, frozen plant material was dissolved in 400 μL of 80% MeOH containing 60 pmol [2H5]MeIAA in a tube, and pulverized using a multibead shocker (Yasui Kikai) for 2 min at 1,500 rpm and 4 °C. After centrifugation for 3 min at 13,000 rpm and 4 °C, 300 μL of the supernatant was transferred to another tube. Then, 300 μL of hexane was added. After vortexing and centrifugation at 15,000 rpm for 5 min, the lower layer was collected, dried using a centrifugal evaporator, and dissolved in 20 μL of 80% MeOH after centrifugal evaporation. Mass spectroscopy analysis was performed using a TripleTOF 5600 mass spectrometer (SCIEX), coupled with a microLC 200 System (SCIEX). Metabolites were separated using a HALO fused C18 column (500-μm internal diameter × 5 cm, 2.7-μm particles) with a gradient elution of mobile phase A (0.5% formic acid/H2O) and mobile phase B (methanol) (0 min: 5% B; 10 min: 95% B; 13 min: 95% B) at an eluent flow rate of 25 μL/min and room temperature (RT). The mass spectrometer was operated in positive-mode electrospray ionization with multiple-reaction monitoring (MRM). MRM transitions are m/z 190.1 to 130.106 for MeIAA, m/z 176.2 to 130.06 for IAA, and m/z 195.1 to 135.11 for [2H5]MeIAA. Source parameters are curtain gas, 25 psi; spray voltage, 5.5 kV; temperature, 550 °C; ion-source gas 1, 25 psi; ion-source gas 2, 35 psi.

Application of MeIAA and IAA.

With reference to Yang et al. (41), MeIAA and IAA dissolved in 95% ethanol were diluted 1:1,000 in medium to a final concentration of 10−7 M. Five-day-old seedlings transferred to beakers containing either MeIAA or IAA were incubated at RT in the dark, based on Murray et al. (20). After 24 h, roots were harvested to analyze gene expression.