Genome Editing Reveals Both the Crucial Role of OsCOI2 in Jasmonate Signaling and the Functional Diversity of COI1 Homologs in Rice

Abstract

Jasmonic acid (JA) regulates plant growth, development and stress responses. Coronatine insensitive 1 (COI1) and jasmonate zinc-finger inflorescence meristem-domain (JAZ) proteins form a receptor complex for jasmonoyl-l-isoleucine, a biologically active form of JA. Three COIs (OsCOI1a, OsCOI1b and OsCOI2) are encoded in the rice genome. In the present study, we generated mutants for each rice COI gene using genome editing to reveal the physiological functions of the three rice COIs. The oscoi2 mutants, but not the oscoi1a and oscoi1b mutants, exhibited severely low fertility, indicating the crucial role of OsCOI2 in rice fertility. Transcriptomic analysis revealed that the transcriptional changes after methyl jasmonate (MeJA) treatment were moderate in the leaves of oscoi2 mutants compared to those in the wild type or oscoi1a and oscoi1b mutants. MeJA-induced chlorophyll degradation and accumulation of antimicrobial secondary metabolites were suppressed in oscoi2 mutants. These results indicate that OsCOI2 plays a central role in JA response in rice leaves. In contrast, the assessment of growth inhibition upon exogenous application of JA to seedlings of each mutant revealed that rice COIs are redundantly involved in shoot growth, whereas OsCOI2 plays a primary role in root growth. In addition, a co-immunoprecipitation assay showed that OsJAZ2 and OsJAZ5 containing divergent Jas motifs physically interacted only with OsCOI2, whereas OsJAZ4 with a canonical Jas motif interacts with all three rice COIs. The present study demonstrated the functional diversity of rice COIs, thereby providing clues to the mechanisms regulating the various physiological functions of JA.

Introduction

Jasmonic acid (JA) is a plant hormone that regulates plant growth, development and stress responses (Wasternack and Hause 2013). In rice, JA-deficient mutants, such as hebiba, cpm2, pre1 and osjar1, exhibit elongation of coleoptile and young leaves, indicating that JA is involved in growth inhibition (Riemann et al. 2008, 2013, Hibara et al. 2016). JA-deficient mutants also exhibit low fertilization rates and abnormal flower formation, indicating that JA plays a crucial role in flower development and fertility in rice (Riemann et al. 2008, 2013, Li et al. 2009, Cai et al. 2014, Xiao et al. 2014, Hibara et al. 2016). Moreover, as a signaling molecule JA plays a pivotal role in the defense response to biotic and abiotic stresses (Wasternack and Hause 2013). JA-related rice mutants are more susceptible to pathogens and herbivores (Ye et al. 2012, Riemann et al. 2013, Shimizu et al. 2013).

Jasmonoyl-l-isoleucine (JA-Ile) is a biologically active form of JA. The receptor for JA-Ile is coronatine insensitive 1 (COI1), an F-box protein of the E3 ubiquitin ligase complex components (Katsir et al. 2008, Yan et al. 2009, Sheard et al. 2010). COI1 recognizes JA-Ile by forming a receptor complex with the jasmonate zinc-finger inflorescence meristem-domain (JAZ) protein, which serves as a repressor of JA responses. COI1 is diffusely localized to the nucleus, while JAZ repressors are located in nucleus bodies by transcription factors when the JA level is low (Withers et al. 2012, Zhang et al. 2015, Yan et al. 2018). The diffusely localized COI1 serves as the JA receptor to initially bind bioactive JA molecules directly (Yan et al. 2009, 2018, Chen et al. 2021). The formation of the COI1–JA-Ile complex provides a platform to subsequently recruit JAZ repressors that are always trapped in nucleus bodies by transcription factors for ubiquitination and degradation, thus activating JA signal transduction (Chini et al. 2007, Thines et al. 2007, Yan et al. 2007, Hu et al. 2022).

The Arabidopsis thaliana genome contains a single COI1 gene, AtCOI1, which regulates JA response (Xie et al. 1998, Katsir et al. 2008). In contrast, there are three COI1 homologs in the rice genome: OsCOI1a (Os01g0853400 in the Rice Annotation Project Database, https://rapdb.dna.affrc.go.jp/), OsCOI1b (Os05g0449500) and OsCOI2/OsCOI1c (Os03g0265500). Lee et al. (2013) reported no tissue specificity in the gene expression levels of the three rice COIs. In addition, when referring to data from the public microarray database RiceXPro, no clear tissue-specific expression was observed (Sato et al. 2012; https://ricexpro.dna.affrc.go.jp/publication.html). Complementation analyses have revealed that OsCOI1a, OsCOI1b and point-mutated OsCOI2 can restore JA signal transduction in the Arabidopsis coi1-1 mutant, suggesting that the three rice COIs potentially function as JA-Ile receptors (Lee et al. 2013). Loss-of-function studies have revealed the physiological functions of OsCOI1a and OsCOI1b in rice. Suppression of rice OsCOI1a and OsCOI1b expression using RNA interference (RNAi) caused a reduction in JA-induced growth inhibition and elongation of the internode and grain, suggesting that OsCOI1a and/or OsCOI1b are involved in growth regulation (Yang et al. 2012). Silencing of OsCOI1a by RNAi caused a reduction in defense responses and an increase in susceptibility to herbivorous insects, indicating that OsCOI1a plays a role in defense responses against herbivores (Ye et al. 2012, 2013, 2017). The transfer DNA (T-DNA) insertion mutant of OsCOI1b exhibits delayed leaf senescence and low fertility, indicating that OsCOI1b is involved in senescence and fertility (Lee et al. 2015). OsCOI1a- and OsCOI1b-knockdown rice generated using RNAi and OsCOI1a-knockout rice generated using genome editing were found to be more susceptible to the rice stripe virus than the wild type (WT) (Yang et al. 2020). However, the physiological function of OsCOI2 in rice has not been analyzed. Moreover, there has been no research using mutant lines for all three rice COIs or even a study comparing them.

The present study revealed the physiological functions of three rice COIs. First, we generated a mutant of each rice COI gene using clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9)–mediated genome editing. The oscoi2 mutants exhibited extremely low fertility, indicating a pivotal role of OsCOI2 in rice fertilization. Transcriptomic analysis revealed that most transcriptional changes in methyl jasmonate (MeJA)–treated rice leaves were OsCOI2-dependent. We also found that OsCOI2 mainly contributed to the JA response in rice leaves, including antimicrobial secondary metabolite production and chlorophyll degradation. Our results clearly show a crucial role of OsCOI2 in fertility, defense responses and senescence in rice. In contrast, exogenous JA inhibited shoot growth in the oscoi1a, oscoi1b and oscoi2 mutants. Moreover, the second leaf sheaths of the oscoi1a oscoi1b double mutants showed prolonged shoot growth compared to those of the WT and parental oscoi1a and oscoi1b mutants, in either the absence or presence of JA. These results suggest that rice COIs have redundant functions in the regulation of shoot growth, unlike in fertility and senescence. However, JA-mediated growth inhibition was reduced in the roots of the oscoi2 mutants compared with that in the WT and other rice COI mutants. These results show that OsCOI1a and OsCOI1b function redundantly in shoots and OsCOI2 mainly functions in the roots of rice seedlings. In addition, we showed that OsJAZ2 and OsJAZ5 containing divergent Jas motifs (Tian et al. 2019) interact specifically with OsCOI2, whereas OsJAZ4 with a canonical Jas motif interacts with all three rice COIs. The present study clearly demonstrated the functional diversity of rice COIs.

Results

Knockout of OsCOI1a, OsCOI1b and OsCOI2 genes using CRISPR/Cas9-mediated genome editing

To determine the physiological functions of the three rice COIs (OsCOI1a, OsCOI1b and OsCOI2), we generated a mutant of each COI gene using CRISPR/Cas9-mediated genome editing. Ge nome editing was performed as described by Mikami et al. (2015), using the pZH_gYSA_MMCas9 vector. First, we designed guide RNAs (gRNAs) for the coding region of each COI gene (Fig. 1A) using CRISPR-P (Lei et al. 2014; http://crispr.hzau.edu.cn/CRISPR/) based on the criteria of sequence specificity in the rice genome. We then introduced a Ti plasmid vector, which harbors a gRNA sequence, Cas9 gene cassette and hygromycin phosphotransferase gene, into rice calli derived from the scutellum of Nipponbare by Agrobacterium-mediated transformation. After selection by hygromycin, DNA fragments, including the targeted region of genome editing, were amplified and sequenced to screen for rice plants harboring a mutation in the targeted genes in regenerated plants (T₀). Next, we analyzed targeted DNA editing and transgene segregation in T₁ and T₂ plants of the selected lines. Finally, we selected homozygous lines that did not harbor transgenes. As a result, we obtained two lines of oscoi1a mutants (#1 and #2), which possessed one nucleotide deletion and one nucleotide insertion in the targeted region, respectively (Fig. 1B). We also acquired three independent oscoi1b mutant lines (#1, #2 and #3). One line (#1) had nucleotide sequence substitutions and deletions around and away from the target site, resulting in a stop codon (Fig. 1C, Supplementary Fig. S1). The other two lines (#2 and #3) possessed a 4- and 16-bp deletion in the targeted region, respectively (Fig. 1C). Moreover, we obtained two lines of oscoi2 mutants (#1 and #2), which had one nucleotide deletion and one nucleotide insertion in the targeted region, respectively (Fig. 1D). As shown in Supplementary Figs. S2–S4, the COI proteins were expected to be truncated in the mutants before the alanine residue was involved in JA-Ile recognition (Monte et al. 2018). Next, we generated an oscoi double mutant. A double mutant of oscoi1a and oscoi1b (oscoi1a oscoi1b) was obtained by crossing oscoi1a (#2) and oscoi1b (#3). However, we were unable to obtain oscoi1a oscoi2 and oscoi1b oscoi2 double mutants.

The oscoi2 mutant showed severely reduced fertility

First, we investigated the vegetative growth of the oscoi1a, oscoi1b, oscoi2 and oscoi1a oscoi1b mutants. These mutants did not show morphological abnormalities in the vegetative stage under greenhouse conditions, although rice COI mutants tended to elongate compared to the WT (Supplementary Fig. S5). Next, we examined mutants during the reproductive stage. The fertility rate of WT plants was 97–99% under our growth conditions. While the oscoi1a and oscoi1b mutants showed a slightly reduced fertility rate compared to the WT, the fertility rate of the two oscoi2 mutant lines was approximately 4% (Fig. 2A). The fertility rate of the oscoi1a oscoi1b double mutant was slightly lower than that of the WT, similar to that of the parental oscoi1a and oscoi1b mutants. (Fig. 2B). These results indicated a crucial role of OsCOI2 in rice fertility. However, the spikelets of the oscoi2 mutants did not show morphological abnormalities (Supplementary Fig. S6), unlike those of the previously reported JA-deficient mutants (Riemann et al. 2008, 2013, Cai et al. 2014, Xiao et al. 2014, Hibara et al. 2016). We also analyzed the internode lengths of the oscoi2 mutants. As a result, the first and fifth internodes of the oscoi2 mutants were more prolonged than those of the WT (Supplementary Fig. S7).

Next, we observed the flowers of the oscoi2 mutants using scanning electron microscopy (SEM) to investigate why the oscoi2 mutants showed severely reduced fertility. We observed anther dehiscence and pollen release outside the anthers in the WT flowers. In contrast, anther dehiscence was not observed in the oscoi2 mutant flowers (Fig. 2C). However, pollen was found inside the anthers of oscoi2 mutants after cutting with a microknife (Fig. 2D–E). These observations suggest that OsCOI2 is involved in the regulation of anther dehiscence.

ZmCOI2a and ZmCOI2b, orthologs of OsCOI2 in Zea mays, are involved in gametophytic male sterility, resulting in segregation distortion in the zmcoi2a and zmcoi2b mutants (Qi et al. 2022). We analyzed the segregation ratio to assess whether a similar distortion was observed in the oscoi2 mutant. We backcrossed the oscoi2 mutant (#1) with the WT. The resultant heterozygotes (F₁) were self-pollinated to obtain F₂ seeds. We then analyzed the segregation ratio of the OsCOI2 genotype in F₂ plants. Mutations in oscoi2 created a polymorphism between DNA digested with restriction enzymes from the WT and oscoi2 mutants (Supplementary Fig. S8). Analysis of the 72 F₂ plants revealed that five were homozygous for the oscoi2 mutation, 26 were heterozygous and 41 were homozygous for the WT. This segregation ratio deviated from the expected ratio of 1 : 2 : 1 (oscoi2:heterozygous:WT) in the tested F₂ plants (chi-square test, P < 0.001). This deviating segregation ratio may have made it difficult to generate oscoi1a oscoi2 or oscoi1b oscoi2 double mutants by crossing single mutants.

Yang et al. (2012) reported that the OsCOI1a and OsCOI1b RNAi lines exhibited a prolonged grain length. However, the grain lengths of oscoi mutants, including the oscoi1a oscoi1b double mutant, did not significantly change compared to those of the WT, unlike OsCOI-RNAi rice (Yang et al. 2012), although oscoi1a #2 exhibited a significantly prolonged grain length (Supplementary Fig. S9).

Transcriptomic analysis after MeJA treatment

We performed transcriptomic analysis of MeJA-treated WT and rice COI single mutants using a genome-wide DNA microarray to reveal the contribution of each COI to the JA response. We used WT, oscoi1a mutants (#1 and #2), oscoi1b mutants (#2 and #3) and oscoi2 mutants (#1 and #2) for transcriptomic analysis. We analyzed the gene expression in leaf disks (6 mm diameter) punched from 4-week-old seedlings at 24 h with or without 500 μM MeJA treatment. Three biological replicates were used for each sample.

After quality control (QC) of the microarray data to remove noise and non-expressing probes, 37,039 probes were subjected to statistical analyses (Supplementary Table S1). We searched for probes whose expression was altered upon MeJA treatment in the WT and each mutant according to the following criteria: the fold change between the MeJA-treated samples and controls was >2 or <0.5, and the false discovery rate (FDR) was <0.01. The number of selected probes is summarized in Table 1, and lists of selected probes are shown in Supplementary Datasets S1 and S2.

We performed hierarchical clustering for the upregulated and downregulated probes following MeJA treatment in the WT. As shown in Fig. 3A, 3,599 upregulated probes in the WT exhibited similar expression patterns in oscoi1a and oscoi1b mutants. We found that two clades (I and II) showed characteristic expression patterns. Probes belonging to clade I or II are shown in Supplementary Dataset S1. The 758 probes were classified as clade I with strongly repressed MeJA-induced expression in the oscoi2 mutants. The expression of 285 probes belonging to clade II was inversely repressed by MeJA treatment in the oscoi2 mutants. Interestingly, the MeJA inducibility of clade II also appeared to be reduced in the oscoi1a mutants. The MeJA inducibility of the remaining probes was also reduced in the oscoi2 mutants, although they still had MeJA inducibility (Fig. 3A). Among the 4,624 downregulated probes, the oscoi1a and oscoi1b mutants showed transcriptional changes similar to those in the WT. In contrast, the suppression of these 4,624 probes was mitigated in the oscoi2 mutants. However, these downregulated probes were still responsive to the MeJA treatment. These results suggested that OsCOI2 plays a primary role in the transcriptional changes caused by MeJA treatment in rice leaves. However, the oscoi2 mutants remained responsive to MeJA in the expression levels of these genes possibly because of the involvement of OsCOI1a and OsCOI1b.

We compared the differentially expressed genes (DEGs) by MeJA treatment in the WT and rice COI mutants. There are cases in which a single gene corresponds to multiple probes in the Agilent Oligo Microarray for Rice. The number of genes after removing the duplicated probes is shown in Table 1. The 3,599 and 4,624 probes, which were upregulated or downregulated in the WT, contain 2,642 and 3,337 genes, respectively (Table 1). A comparison of DEGs in the WT and oscoi1a mutants showed that 488 genes, 18% of upregulated genes in the WT, were not induced in either oscoi1a #1 or #2 (Supplementary Fig. S10A). The 274 genes induced in the WT, but not in either oscoi1b #2 or #3, corresponded to 10% of the 2,642 upregulated genes in the WT (Supplementary Fig. S10B). However, 83% of the genes upregulated in the WT (2,183 genes) were not induced in either of the two lines of the oscoi2 mutants (Supplementary Fig. S10C). The 407 genes, 12% of the 3,337 genes downregulated in the WT strain, were not repressed by either oscoi1a #1 or #2 (Supplementary Fig. S10D). The 425 genes induced in the WT, but not in either oscoi1b #2 or #3, were 13% of the downregulated genes in the WT (Supplementary Fig. S10E). However, 81% of the genes suppressed in the WT, 2,712 genes, were not downregulated in either of the two lines of the oscoi2 mutants (Supplementary Fig. S10F). Taken together, >80% of DEGs in the WT did not show MeJA responsiveness in the oscoi2 mutants above the criteria, suggesting a primary role of OsCOI2 in the transcriptional changes caused by MeJA treatment in rice leaves, similar to hierarchical clustering. However, 10–20% of DEGs in the WT did not exhibit MeJA responsiveness in the oscoi1a and oscoi1b mutants, implying the partial involvement of OsCOI1a and OsCOI1b in the transcriptional changes induced by MeJA.

We further performed gene ontology (GO) analysis of the genes that showed MeJA responsiveness in WT, but not in either oscoi2 #1 or #2. The 2,183 or 2,712 genes were upregulated or downregulated by MeJA treatment in WT, but not in the oscoi2 mutants (Supplementary Tables S2 and S3). As shown in Table 2, 11 GO terms for biological processes were enriched in 2,183 upregulated genes compared to the reference gene list with a criterion (FDR < 0.05). GO terms related to the various metabolic processes of trehalose, cell wall, fatty acid, nucleotide sugar and respiration were enriched in 2,183 genes (Table 2). The GO terms regarding environmental stress responses, such as ‘regulation of response to stress’ and ‘response to chemical’, were also enriched (Table 2), suggesting the involvement of OsCOI2 in the JA-induced defense response to stress. Nine GO terms for the biological process were significantly enriched for 2,712 downregulated genes (Table 3). GO terms related to plastid function and photosynthesis (‘regulation of photosynthesis’, ‘chlorophyll metabolic process’, ‘chloroplast organization’ and ‘photosynthesis’) were included (Table 3), implying the involvement of OsCOI2 in JA-induced leaf senescence accompanied by chlorophyll degradation and a decrease in photosynthesis. We subsequently evaluated whether OsCOI2 was involved in JA-induced defense responses and leaf senescence.

OsCOI2 is required for JA-induced phytoalexin production in rice leaves

Phytoalexin accumulation in rice is a typical JA-induced defense response (Shimizu et al. 2013, Yamane 2013). We measured phytoalexin accumulation in the leaves of the WT, oscoi1a, oscoi1b, oscoi2 and oscoi1a oscoi1b mutants upon MeJA treatment to elucidate the role of OsCOI2 in the JA-induced defense response. Leaf disks of the WT and mutant plants were treated with 0 or 500 μM MeJA. We measured the accumulation of phytoalexins in each leaf disk using liquid chromatography with electrospray ionization tandem mass spectrometry (LC–ESI-MS/MS). The levels of flavonoid-type phytoalexin sakuranetin in the oscoi1a, oscoi1b and oscoi1a oscoi1b mutants were almost equal to or greater than those in the WT after MeJA treatment (Fig. 4A, C). In contrast, sakuranetin levels in the oscoi2 mutants after MeJA treatment were significantly lower than those in the WT (Fig. 4B). The accumulation levels of momilactone and phytocassane (major diterpenoid phytoalexins in rice) in the oscoi1a, oscoi1b and oscoi1a oscoi1b mutants were almost equal to or higher than those in the WT upon MeJA treatment (Fig. 4D, F, G, I). However, momilactone and phytocassane levels in the oscoi2 mutants were significantly lower than those in the WT plants after MeJA treatment (Fig. 4E, H).

Microarray data showed that the inducible expression of genes relevant to phytoalexin biosynthesis (OsCPS2, OsKSL7, OsCPS4, OsKSL4, OsNOMT and OsMYC2) was suppressed in oscoi2 mutants after MeJA treatment (Supplementary Fig. S11). Therefore, we further analyzed the expression levels of these genes by quantitative reverse transcription polymerase chain reaction (qRT-PCR) using biologically independent samples from microarray analysis. OsCPS2 (Os02g0571100) and OsKSL7 (Os02g0570400) are phytocassane biosynthetic genes, whereas OsCPS4 (Os04g0178300) and OsKSL4 (Os04g0179700) are momilactone biosynthetic genes (Yamane 2013). OsNOMT (Os12g0240900) encodes sakuranetin synthase (Shimizu et al. 2012), and OsMYC2 (Os10g0575000) regulates JA-induced expression of OsNOMT (Ogawa et al. 2017). The messenger RNA (mRNA) levels of these genes were induced by treatment with 500 µM MeJA in WT plants (Supplementary Fig. S12). The MeJA responsiveness of these genes was severely reduced in oscoi2 mutants (Supplementary Fig. S12), similar to the microarray analysis (Supplementary Fig. S11).

In addition, we analyzed the mRNA levels of these genes at a low concentration of 50 µM MeJA. OsCPS2, OsCPS4 and OsNOMT were induced by 50 μM MeJA treatment in the WT, although the induction levels were lower than those in the 500 µM MeJA treatment (Supplementary Figs. S12, S13). In the oscoi2 mutant, OsCPS2, OsCPS4 and OsNOMT expression levels decreased compared to those in the WT after 50 μM MeJA treatment (Supplementary Fig. S13A, B, E). In contrast, OsKSL4 and OsKSL7 were not induced by 50 μM MeJA treatment in WT plants (Supplementary Fig. S13C, D). Therefore, we could not evaluate the inducibility of these genes in the oscoi2 mutants after treatment with 50 μM MeJA. However, these genes were induced by 50 μM MeJA treatment only in oscoi1b #3 plants, although the reason for this is unclear (Supplementary Fig. S13C, D). The inducibility of OsMYC2 after 50 μM MeJA treatment was comparable to that of the 500 μM treatment in the WT (Supplementary Figs. S12F, S13F). The oscoi2 mutants treated with 50 μM MeJA did not show the inductive expression of OsMYC2, similar to that observed after 500 µM MeJA treatment (Supplementary Fig. S13F). These results show that OsCOI2 is essential for the JA-induced accumulation of phytoalexins, such as sakuranetin, phytocassanes and momilactones.

OsCOI2 plays a crucial role in JA-induced senescence of rice leaves

GO analysis revealed that the GO terms related to plastid functions were enriched in genes whose suppression by MeJA was OsCOI2-dependent (Table 3). JA treatment promotes leaf senescence through the degradation of chlorophyll in plants (Wasternack and Hause 2013, Zhu et al. 2017). We investigated the involvement of each COI gene in JA-induced leaf senescence under light conditions. Leaf disks of the WT and oscoi1a, oscoi1b, oscoi2 and oscoi1a oscoi1b mutants were treated with 0, 200 or 500 μM MeJA. We then measured the chlorophyll content of each leaf disk using a spectrophotometer. As shown in Fig. 5, the chlorophyll content in the leaf disks of the WT plants decreased upon treatment with 200 and 500 µM MeJA. The chlorophyll content in leaf disks from the MeJA-treated oscoi1a, oscoi1b and oscoi1a oscoi1b mutants was comparable to that in leaf disks from the WT after treatment with 200 or 500 µM MeJA, although oscoi1a (#1) showed a significantly higher chlorophyll content without MeJA treatment (Fig. 5A, B, D). In contrast, the chlorophyll content in the oscoi2 mutants was significantly higher than that in the WT after MeJA treatment (Fig. 5C). These results indicated that OsCOI2 plays a significant role in JA-induced chlorophyll degradation during leaf senescence.

JA-mediated growth inhibition in the rice COI mutants

JA treatment inhibits seedling growth in plants (Wasternack and Hause 2013). Yang et al. (2012) reported that RNAi lines of OsCOI1a and OsCOI1b are resistant to JA treatment, suggesting that OsCOI1a and/or OsCOI1b are involved in growth regulation. In the present study, we attempted to clarify the contribution of each rice COI gene to the generated mutants.

Seeds of the WT, oscoi1a, oscoi1b, oscoi2 and oscoi1a oscoi1b mutants were germinated on agar media containing 0, 5, 10 or 20 μM JA, following which the length of the second leaf sheath was measured 10 d after germination. As shown in Fig. 6, the second leaf sheath length of the WT decreased in a dose-dependent manner. In the absence of JA, the second leaf sheaths of the single mutants were longer than those of the WT. Moreover, the second leaf sheaths of the oscoi1b mutants were significantly longer than those of the WT plants after treatment with JA (Fig. 6B). The second leaf sheaths of the oscoi2 mutants were also longer than those of the WT in the presence of 10 or 20 μM JA (Fig. 6C). However, the oscoi1a, oscoi1b and oscoi2 mutants were still susceptible to JA, suggesting that rice COIs have redundant functions in regulating the growth inhibition of the second leaf sheath. The second leaf sheaths of the oscoi1a oscoi1b double mutants were more extended than those of the WT and parental oscoi1a and oscoi1b mutants, in either the absence or presence of JA (Fig. 6D). These results suggested that OsCOI1a and OsCOI1b function redundantly to inhibit the growth of the second leaf sheath. However, the growth of the second leaf sheaths of the oscoi1a oscoi1b double mutants was still inhibited by JA, implying the partial involvement of OsCOI2.

We also measured the root length of each plant 10 d after germination on agar media containing 0, 5, 10 or 20 μM JA. As shown in Fig. 7C, the root lengths of oscoi2 mutants were greater than those of WT plants in the presence of JA, especially at a concentration of 20 μM. In contrast, the lengths of the oscoi1a mutant roots were equal to or shorter than those of the WT roots in the absence or presence of JA (Fig. 7A). The root length of oscoi1b (#2) was slightly longer than that of the WT in the presence of 10 or 20 μM JA, whereas oscoi1b (#3) showed the same tendency as the WT (Fig. 7B). These results suggested that OsCOI2 is primarily involved in root growth inhibition. Consistent with this, oscoi1a oscoi1b double mutant roots were sensitive to JA, although the root lengths of the double mutants were significantly longer than those of the WT and parental oscoi1a and oscoi1b mutants (Fig. 7D).

Physical interaction between OsJAZ2/OsJAZ5 and OsCOI2

COI1 forms a receptor complex with JAZs in the presence of JA-Ile. Therefore, we investigated the protein–protein interactions between rice COIs and JAZs to determine why OsCOI2 plays a different role than OsCOI1a and OsCOI1b. Among the 15 rice JAZs, 13 JAZs have 27 conserved amino acid sequences called the Jas motif, which contains the canonical JAZ degron (X₂PXARR/KX), a core sequence (SLX₂FX₂KRX₂R) and a C-terminal motif (X₅PY). In contrast, OsJAZ2 (Os07g0153000) and OsJAZ5 (Os04g0395800) have sequences known as divergent Jas motifs with low JAZ degron conservation (Fig. 8A) (Tian et al. 2019). The Jas motif is involved in COI-JA-Ile-JAZ interactions (Chung et al. 2010, Takaoka et al. 2019, Tian et al. 2019). Therefore, we focused on OsJAZ2 and OsJAZ5 to investigate the interaction specificity of OsCOI1a, OsCOI1b and OsCOI2.

We performed an in vitro co-immunoprecipitation assay using recombinant COI proteins and Jas motif peptides (Takaoka et al. 2019, 2020, Saito et al. 2021). The JAZ peptide was more stable than the full-length JAZ protein. This in vitro system enabled us to use the bioactive form (+)-7-iso-JA-Ile, which is easily converted to (−)-trans-JA-Ile upon pH change or temperature elevation (Fonseca et al. 2009). We prepared glutathione-S-transferase-fused OsCOI (GST-OsCOI) and fluorescein (Fl)-tagged rice JAZ peptides. The prepared COI1 and JAZ peptides were mixed and immunoprecipitated using an anti-Fl antibody. Immunoprecipitated proteins were analyzed by protein gel blot analysis using an anti-GST antibody. GST-OsCOI2 was detected by immunoprecipitation using Fl-tagged OsJAZ2 and OsJAZ5 peptides in the presence of (+)-7-iso-JA-Ile. However, GST-OsCOI1a and GST-OsCOI1b were not detected by immunoprecipitation in the presence or absence of (+)-7-iso-JA-Ile. An Fl-tagged OsJAZ4 (Os09g0401300) peptide with a canonical Jas motif was co-immunoprecipitated with the GST-OsCOI1a, GST-OsCOI1b and GST-OsCOI2 proteins in the presence of (+)-7-iso-JA-Ile (Fig. 8B). These results revealed that OsJAZ2 and OsJAZ5, which contain divergent Jas motifs, interact specifically with OsCOI2, whereas OsJAZ4, which harbors a canonical Jas motif, interacts with all three rice COIs.

Discussion

Rice COI mutants showed elongated phenotypes in the vegetative and reproductive stages

In this study, we investigated the physiological functions of three rice COIs using a mutant for each COI generated using genome editing. At the vegetative stage, rice COI mutants tended to elongate compared to WT (Supplementary Fig. S5), similar to JA-deficient mutants pre1 (Hibara et al. 2016), suggesting that the three rice COI proteins are redundantly involved in the regulation of vegetative growth under JA signaling. Moreover, oscoi2 mutants showed longer internode lengths than the WT (Supplementary Fig. S7). OsCOI1a and OsCOI1b-knockdown rice show an elongated phenotype similar to that of the internode (Yang et al. 2012). In the present study, we suggest that OsCOI2 is also involved in regulating the internode length. Notably, suppression of OsCOI1a and OsCOI1b causes elongation of all internodes (Yang et al. 2012), whereas oscoi2 mutants showed elongation of the first and fifth internodes (Supplementary Fig. S7). The site where OsCOI2 functions might differ from those of OsCOI1a and OsCOI1b, although there is no information on the expression pattern of rice COIs in the internodes.

OsCOI2 is essential for rice fertility, but not for flower morphogenesis

The oscoi2 mutants exhibited severely reduced fertility (Fig. 2A), similar to that observed in JA-deficient rice mutants (Li et al. 2009, Riemann et al. 2013, Cai et al. 2014, Xiao et al. 2014, Hibara et al. 2016). A similar phenotype between the JA-deficient mutants and the oscoi2 mutants suggests a crucial role of OsCOI2 in JA signaling in rice fertility. Electron microscopy revealed that anther dehiscence did not occur in the oscoi2 mutants (Fig. 2C), which could explain the decreased fertility. In addition, the segregation ratio deviated from the expected ratio in the oscoi2 heterozygous plants (Supplementary Fig. S8). Qi et al. (2022) reported similar segregation distortions in zmcoi2a and zmcoi2b mutants due to gametophytic male sterility caused by the loss of ZmCOI2a and ZmCOI2b. Because OsCOI2 is an ortholog of ZmCOI2a and ZmCOI2b (Qi et al. 2022), the loss of OsCOI2 may cause gametophytic male sterility.

JA-deficient mutants exhibit reduced fertility and abnormal flower morphology (Riemann et al. 2008, 2013, Li et al. 2009, Cai et al. 2014, Hibara et al. 2016). In contrast, the present study showed normal flower morphology in oscoi2 mutants (Supplementary Fig. S6), despite severely reduced fertility. Cai et al. (2014) reported that eg2-1, which has a point mutation in the Jas motif of OsJAZ1, exhibits abnormal flower morphogenesis. The mutated OsJAZ1 protein was not degraded by JA treatment, resulting in a dominant eg2-1 phenotype. OsJAZ1 represses the activity of OsMYC2, resulting in transcriptional repression of E-class genes such as OsMADS1. In addition, OsJAZ1 interacted with OsCOI1b in a yeast two-hybrid assay in the presence of the JA-Ile analog coronatine, suggesting the involvement of OsCOI1b in rice flower morphogenesis (Cai et al. 2014). Furthermore, the overexpression of rice JAZs causes similar abnormal flower morphogenesis (Hori et al. 2014, Mehra et al. 2022). These reports indicate that the accumulation of rice JAZ proteins affects the regulation of flower morphogenesis via MADS-box transcription factors. However, we did not observe abnormal flower or seed formation in the rice COI mutants (Supplementary Figs. S6, S9), indicating that OsCOI1a, OsCOI1b and OsCOI2 have redundant functions and that rice JAZs involved in flower morphogenesis are degraded in the presence of at least one rice COI. Future studies using double and triple mutants, which the present study could not generate, will reveal the detailed functions of each rice COI gene in flower morphogenesis.

OsCOI2 plays a primary role in the JA signaling in rice leaves

We performed transcriptomic analysis using a genome-wide DNA microarray to analyze transcriptional changes after MeJA treatment in the WT and COI single mutants. A comparison of DEGs of microarray analysis and hierarchical clustering of JA-responsive genes showed that mutation of oscoi2 diminished the transcriptional changes in MeJA-treated leaves, although the partial contribution of OsCOI1a and OsCOI1b was also suggested (Fig. 3, Supplementary Fig. S10).

The GO terms related to various biological processes were enriched in genes whose MeJA-responsive genes were OsCOI2-dependent (Tables 2 and 3). The GO terms ‘trehalose metabolism in response to stress’ and ‘trehalose biosynthetic process’ were the most strongly enriched. Trehalose 6-phosphate plays a crucial role in the regulation of sugar metabolism, plant growth and stress responses (Ponnu et al. 2011, Joshi et al. 2019). JA and OsCOI2 may be involved in trehalose 6-phosphate signaling. Other enriched GO terms related to the metabolic processes of the cell wall, nucleotide sugar and respiration suggest the possible involvement of OsCOI2 in these metabolic processes. JA regulates the trade-off between plant growth and defense responses (Savchenko et al. 2019). OsCOI2 may function under JA signaling to regulate trade-offs. We also found that the GO terms ‘regulation of response to stress’ and ‘response to chemical’ were enriched, suggesting that OsCOI2 is involved in the JA-induced defense response. Most GO terms enriched in the genes downregulated by MeJA were related to plastid function and photosynthesis (Table 3). These GO terms were consistent with the physiological functions of JA, which induces leaf senescence and chlorophyll degradation.

MeJA-induced phytoalexin production and chlorophyll degradation were severely repressed in the oscoi2 mutant leaves (Figs. 4, 5). These results strongly suggest that OsCOI2 plays a primary role in JA signaling in rice leaves. Sakuranetin production depends on JA-Ile accumulation (Riemann et al. 2013, Shimizu et al. 2013, Miyamoto et al. 2016). Furthermore, OsMYC2-knockdown lines generated using RNAi show reduced sakuranetin production and OsNOMT expression, indicating that OsMYC2 plays a crucial role in sakuranetin production (Ogawa et al. 2017). Combined with our results of reduced sakuranetin production in oscoi2 mutants, OsCOI2 may activate OsMYC2 by degrading OsJAZs in MeJA-treated leaves. Indeed, our microarray and qRT-PCR analyses revealed that JA-induced OsMYC2 expression was disrupted in the oscoi2 mutants (Supplementary Figs. S11–S13). Diterpenoid phytoalexins, momilactones and phytocassanes also accumulate after JA treatment (Shimizu et al. 2013). It is unclear whether diterpenoid phytoalexin production is OsMYC2-dependent. Future studies are needed to elucidate the effect of the possible involvement of OsMYC2 in diterpenoid phytoalexin production on the downstream signaling of OsCOI2.

Leaf senescence, with chlorophyll degradation, is a representative physiological function of JA. JA-deficient or JA-insensitive mutants show delayed senescence with chlorophyll degradation in various plants (Zhang and Zhou 2013, Zhu et al. 2017). The present study showed that MeJA-induced chlorophyll degradation was inhibited in oscoi2 mutants, suggesting a central role of OsCOI2 in JA-induced leaf senescence under light conditions. In contrast, Lee et al. (2015) reported that a mutation in OsCOI1b delays leaf senescence. The oscoi1b mutant harboring a T-DNA insertion in the 3ʹ-untranslated region showed delayed senescence after the heading stage. In addition, detached leaves of oscoi1b mutants exhibited decreased chlorophyll degradation after incubation in the dark (Lee et al. 2015). However, the oscoi1b mutants generated by genome editing in this study exhibited no significant changes in chlorophyll degradation after MeJA treatment. One possible explanation for these conflicting results is that senescence-inducing signaling differs between the MeJA and dark treatments. In addition to JA, other endogenous phytohormones are involved in leaf senescence (Zhang and Zhou 2013, Zhu et al. 2017), and complex signal transduction may be present in dark-induced senescence. Overexpression of OsMYC2 promotes JA-induced chlorophyll degradation in rice leaves, indicating the involvement of OsMYC2 in JA-induced leaf senescence (Uji et al. 2016). OsMYC2 may function downstream of OsCOI2 to promote chlorophyll degradation and senescence in the JA-treated leaves.

Differential signaling for growth inhibition by JA in the shoot and root of rice seedlings

Upon exogenous application of JA, shoot growth inhibition occurred in the oscoi1a, oscoi1b and oscoi2 seedlings, whereas the oscoi1b and oscoi2 mutants were slightly resistant to JA compared to the WT (Fig. 6A–C). These results suggest that rice COI is redundantly involved in growth inhibition by JA in the shoot, unlike its involvement in fertility and senescence, in which OsCOI2 plays a crucial role. We also generated oscoi1a oscoi1b double mutants and analyzed their shoot growth in the absence or presence of JA, revealing that the second leaf sheaths of the double mutants were longer than those of the WT and parental oscoi1a and oscoi1b mutants (Fig. 6D). Consistent with this, OsCOI1a- and OsCOI1b-knockdown rice generated using RNAi has been shown to exhibit a reduction in JA-induced growth inhibition (Yang et al. 2012). These results suggested that OsCOI1a and OsCOI1b function redundantly to inhibit the growth of the second leaf sheath. However, shoot growth of the oscoi1a oscoi1b double mutants was still inhibited by JA (Fig. 6D), implying the partial involvement of OsCOI2.

In contrast, the roots of the oscoi2 mutants were significantly resistant to JA, whereas the oscoi1a and oscoi1b mutants showed root growth inhibition after JA treatment (Fig. 7). The root growth of the oscoi1a oscoi1b double mutants was also inhibited by JA, although the root lengths of the double mutants were significantly longer than those of the WT and parental oscoi1a and oscoi1b mutants in the absence and presence of JA (Fig. 7D). These results suggest that OsCOI2 plays a primary role in the growth inhibition of roots upon JA treatment and that different signaling transductions for JA-induced growth inhibition exist between rice shoots and roots. Lee et al. (2013) reported no organ-specific transcriptional changes in OsCOI1a, OsCOI1b or OsCOI2 mRNA levels in rice seedlings. There is a need for future analyses of the protein levels of rice COIs and the expression of downstream signaling components, such as JAZ and transcription factors, to investigate the different signaling mechanisms of JA-induced growth inhibition in shoots and roots.

OsJAZ2 and OsJAZ5 specifically interact with OsCOI2

COI1 forms a receptor complex with JAZ in the presence of JA-Ile. After JA-Ile perception, ubiquitination and degradation of JAZ occur, leading to the activation of JA responses. JAZ forms a multigene family. The rice genome encodes 15 JAZs (Tian et al. 2019). In the present study, we focused on OsJAZ2 and OsJAZ5, which have divergent Jas motifs, and found that OsJAZ2 and OsJAZ5 physically interact with OsCOI2, but not with OsCOI1a and OsCOI1b (Fig. 8). OsJAZ2 and OsJAZ5 may act in the downstream signaling of OsCOI2 to regulate fertility and JA response in rice leaves. There is a need for a comprehensive interaction analysis between the three rice COIs and 15 JAZs and a functional analysis of the specific interaction of JAZs with OsCOI2 to clarify the molecular mechanisms of JA perception and signal transduction in rice. Yan et al. (2016) reported that OsCOI2 interacts with OsJAZ1 (Os10g0392400), which is called OsJAZ12 by Tian et al. (2019), in the presence of five coronafacic acid (CFA)-amino acid conjugates (CFA-AAs) that is mimic of JA-amino acid conjugates. In contrast, OsCOI1a and OsCOI1b showed narrower selectivity for CFA-AAs. This report suggests that OsCOI2 differs from OsCOI1a and OsCOI1b in forming the COI–JA-Ile–JAZ complex and may be consistent with our results that OsCOI2 has different JAZ selectivity and plays a crucial role in rice JA response.

The interaction between OsCOI2 and OsJAZ2 needed 0.1 μM of (+)-7-iso-JA-Ile, whereas other interactions (OsCOI2–OsJAZ5 and OsCOIs–OsJAZ4) were detected at 0.01 μM. This result suggests that the combination of OsCOI2 and OsJAZ2 has a lower affinity for (+)-7-iso-JA-Ile than other combinations. The difference in the affinity of COI–JAZ for JA-Ile might be involved in regulating different downstream signals in response to different intracellular (+)-7-iso-JA-Ile concentrations. However, our expression analysis showed that OsCOI2 was mainly involved in the inductive expression of phytoalexin biosynthesis–related genes by both 50 and 500 μM MeJA treatments, although OsKSL4 and OsKSL7 were not induced by 50 μM MeJA (Supplementary Figs. S12, S13). Quantitative analyses of the intracellular abundance of COIs, JAZs and (+)-7-iso-JA-Ile and elucidation of the downstream signal of each COI–JAZ are needed to clarify the biological significance of the difference in affinity of the COI–JAZ combination for JA-Ile.

Conclusion

The present study showed the crucial role of OsCOI2 in fertility and the response to JA in rice leaves using a genome editing approach. However, a redundant function of rice COIs in JA-induced shoot growth inhibition has also been suggested. These results revealed the existence of functional diversity in rice COIs despite the absence of organ-specific changes in their mRNA levels of rice COIs. In addition, we observed differences in the specificity of the interaction between OsJAZ2 and OsJAZ5 among rice COIs. These findings will contribute to the elucidation of the mechanisms regulating the various physiological functions of JA in rice.

Materials and Methods

Plant materials and growth conditions

Oryza sativa L. ‘Nipponbare’ was used as WT. Rice seeds were surface-sterilized in 2% sodium hypochlorite solution and then washed with sterilized water. Surface-sterilized seeds were grown in 0.8% agar medium under continuous white light at 25°C for 7 d. Rice plants were transferred to a composite soil, Bonsoru No. 2 (Sumitomo Chemical, Tokyo, Japan), and grown under natural daylight conditions in a glasshouse at Teikyo University (Utsunomiya, Tochigi, Japan; 36.60°N, 139.88°E). The temperature in the greenhouse was maintained at approximately 28°C during the day and at 25°C at night. To analyze vegetative growth, plant height was measured in 6-week-old rice plants. Fully ripened rice plants were used to measure the internode length.

Genome editing

We selected the target site (Fig. 1A) for genome editing using CRISPR-P (Lei et al. 2014). CRISPR/Cas9-mediated genome editing was performed as previously described by Mikami et al. (2015). Oligo DNA, including each target sequence with a BbsI site (Supplementary Table S4), was annealed and cloned into pU6gRNA. Target sequences with the OsU6 promoter were transferred to the pZH_gYSA_MMCas9 vector (Mikami et al. 2015). Rice transformation was performed as previously described (Toki et al. 2006). The pSuperAgro vector (Inplanta Innovations Inc., Kanagawa, Japan) was used to improve transformation efficiency. The target sites were amplified from regenerated T₀ plants by PCR. The amplified DNA fragments were treated with ExoSAP-IT (Thermo Fisher Scientific, Waltham, MA, USA) and directly sequenced to select mutants for each COI gene. PCR and sequencing were performed to select homozygous strains that lost the gene cassette for genome editing in the T₁ and T₂ plants. The primers used for screening are listed in Supplementary Table S4.

Artificial pollination of rice

The unpollinated spikelets were maintained at 43°C for 7 min in a water bath. The pollen of the dehiscent anthers was pollinated to emasculate spikelets using forceps. Panicles with pollinated spikelets were covered with paper bags, and mature seeds were collected after 1 month.

Measurement of fertility

Seeds were collected from each rice plant and were classified as mature or immature. The ratio of mature seeds to whole seeds was defined as the fertility rate. At least 600 seeds per plant were counted.

SEM observations

The anthers were observed immediately after flowering using a low-vacuum SEM (TM3030; Hitachi High-Tech Co., Tokyo, Japan) at an accelerating voltage of 15 kV. We observed three independent rice plants each (grown in a greenhouse) for WT and oscoi2 mutants.

Analysis of the segregation ratio

To analyze the segregation ratio of oscoi2 (#1) heterozygotes, we amplified the DNA fragments of the target site of OsCOI2 and digested them with BsrI. The digested DNA fragments were subjected to gel electrophoresis using Tris-acetate buffer on 2% agarose S gels (Nippon Gene Co., Tokyo, Japan). Fragmented patterns are shown in Supplementary Fig. S8.

MeJA treatment

MeJA (Tokyo Chemical Industry Co Ltd., Tokyo, Japan) was dissolved in dimethyl sulfoxide (DMSO) at 500 mM for the stock solution. The stock solution was diluted with distilled water to a concentration of 50 or 500 μM. An equal volume of DMSO was added to the control treatment instead of the MeJA stock solution. Leaf disks (6 mm diameter) were punched from the fully expanded youngest leaves of 4-week-old rice plants and then floated on the MeJA solution under continuous light.

Microarray analysis

Total RNA was extracted from 10 leaf disks treated with or without 500 μM MeJA for 24 h using the RNeasy^® Plant Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. Cy-3-labeled complementary RNA synthesis, hybridization and scanning of a 60-mer rice oligo microarray with 44k features (Agilent Technologies, Santa Clara, CA, USA) on glass slides were performed as previously described according to the manufacturer’s instructions. Data normalization and statistical analyses were performed using the Subio Platform with Basic Plug-in (version 1.22; Subio, Aichi, Japan). All raw signal intensities were normalized by 75 percentile-shift, log-2-transformed and then centered. QC was performed based on the following criteria: we first removed the probes containing the flag ‘gIsBGNonUnifOL’ or ‘gIsFeatNonUnifOL’ from at least one sample as noise probes caused by cracks or dust. We removed the probes containing the flag ‘gIsWellAboveBG’ from all samples as non-expressing probes. After QC, 37,039 probes, which corresponded to 25,807 genes after removal of duplication, were subjected to statistical analyses (Supplementary Table S1). Genes upregulated or downregulated by MeJA treatment were selected according to the following criteria: the fold change between the MeJA-treated samples and controls was >2 or <0.5, and the q-value was <0.01. Hierarchical clustering was performed using Pearson’s correlation for upregulated or downregulated genes in the WT. A comparison of DEGs was performed, and Venn diagrams were drawn on the website of Bioinformatics & Evolutionary Genomics, Ghent University (http://bioinformatics.psb.ugent.be/webtools/Venn/). GO analysis was performed using PANTHER 17.0 (Mi et al. 2012, Thomas et al. 2022; http://pantherdb.org/). The 25,807 genes (Supplementary Table S1) passing QC were used as a reference gene list for GO analysis.

Quantification of phytoalexin levels

Leaf disks were treated with or without 500 µM MeJA for 72 h under continuous light. Extraction and quantification of phytoalexins using LC–ESI-MS/MS were performed as described previously (Miyamoto et al. 2016).

Assessment of gene expression using qRT-PCR

Leaf disks were treated with 0, 50 or 500 µM MeJA for 24 h under continuous light. Total RNA was extracted using the RNeasy^® Plant Mini Kit (Qiagen, Hilden, Germany) and subjected to complementary DNA (cDNA) synthesis using a PrimeScript RT reagent Kit with gDNA Eraser (Takara Bio Inc., Shiga, Japan). qRT-PCR was performed using a Fast SYBR Green Master Mix (Applied Biosystems, Waltham, MA, USA) on an ABI PRISM 7500 Fast Real-Time PCR System (Applied Biosystems). Primer sequences are provided in Supplementary Table S4. The expression level of each gene was calculated using the ΔΔCt method. The relative expression of each gene was normalized to that of the ubiquitin domain-containing protein gene (Os10g0542200).

Quantification of chlorophyll contents

The leaf disks were treated with 0, 200 and 500 µM MeJA for 72 h under continuous light and then homogenized in 500 µl of N,N-dimethylformamide. The mixture was then centrifuged at 17,800×g for 10 min, and the supernatants were used for chlorophyll quantification. Chlorophyll content was measured using a spectrophotometer (NanoDrop 2000c, Thermo Fisher Scientific) as described previously (Porra et al. 1989).

Growth inhibition assay

Surface-sterilized seeds were germinated on 0.8% agar medium containing 0, 5, 10 or 20 µM JA in a magenta box (GA-7; Merck, Darmstadt, Germany) and incubated under continuous light at 25°C. After 10 d, the second leaf sheath and root lengths were measured.

Expression of rice COI proteins

Plasmids harboring GST-fused AtCOI1 or AtASK1 (pFB-GTE-COI1 and pFB-HTB-ASK1, respectively) were purchased from Addgene (https://www.addgene.org/). The cDNA clones of OsCOI1a, OsCOI1b and OsCOI2 were obtained from the NARO Genebank (https://www.gene.affrc.go.jp/index_en.php). The coding sequences of OsCOI1a, OsCOI1b and OsCOI2 were cloned into pFB-GTE-COI1 to obtain expression vectors for the GST-fused OsCOI proteins. GST-fused OsCOI1a, OsCOI1b and OsCOI2 were expressed in insect cells with AtASK1 and purified using Glutathione Sepharose 4B (GE Healthcare, Chicago, IL, USA) as previously described (Sheard et al. 2010, Takaoka et al. 2020, Saito et al. 2021).

Synthesis of OsJAZ peptides

The amino acid sequences of the OsJAZ2, 4 and 5 peptides are shown in Fig. 8A. The N-terminal end of each peptide was labeled with fluorescein (Fl). Fl-tagged OsJAZ peptides were synthesized and purified as previously described (Takaoka et al. 2020, Saito et al. 2021). High-performance liquid chromatography (HPLC) and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry were used to characterize each peptide (Supplementary Fig. S14).

Preparation of (+)-7-iso-JA-Ile

The diastereomeric mixture of JA-Ile was synthesized from racemic JA and l-isoleucine as previously described (Ogawa and Kobayashi 2008). A mixture of (−)-trans-JA-Ile and (+)-7-iso-JA-Ile was prepared using an HPLC system (LC-20AD, Shimadzu, Kyoto, Japan) equipped with a Senshu Pak octadecyl silyl column (250 mm long, 10 mm internal diameter; Senshu Scientific Co., Tokyo, Japan) as previously reported (Jikumaru et al. 2004). We then purified (+)-7-iso-JA-Ile as previously described, with slight modifications (Fonseca et al. 2009, Takaoka et al. 2019). The HPLC system (LC-20AD) was equipped with a NUCLEOSIL 50-5 column (250 mm long, 10 mm internal diameter; Chemco Plus Scientific Co., Osaka, Japan), with hexane-ethanol-acetic acid (96 : 4 : 0.15, v/v/v) as the eluent, at a flow rate of 3.0 ml min^–1. The retention time of (+)-7-iso-JA-Ile was 32 min.

Co-immunoprecipitation analysis

Purified GST-OsCOI (5 nM), Fl-tagged OsJAZ peptide (10 nM) and (+)-7-iso-JA-Ile (0, 0.01, 0.1, 1 and 10 µM) were mixed with 350 µl of incubation buffer [50 mM Tris pH 7.8, 100 mM NaCl, 10% (v/v) glycerol, 20 mM 2-mercaptoethanol, 0.1% (v/v) Tween 20 and 100 nM inositol-1,2,4,5,6-pentakisphosphate] in the presence of a complete EDTA-free protease inhibitor cocktail (Merck). After incubation at 4°C for 1 h, 0.2 µl of anti-Fl antibody (GeneTex, Irvine, CA, USA) was added to each sample. Samples were incubated for 1 h at 4°C with gentle mixing. SureBeads Protein G Magnetic Beads (Bio-Rad, Hercules, CA, USA) were added to each sample. Finally, after incubation for 1 h at 4°C with gentle mixing, the beads were washed three times with 350 µl of phosphate-buffered saline containing 0.1% (v/v) Tween 20. The washed beads were resuspended in 35 µl of double-diluted sample buffer solution (2ME+) (×2) (Fujifilm Co., Tokyo, Japan) containing 100 mM dithiothreitol. The beads were boiled at 60°C for 10 min and subjected to protein gel blot analysis.

Protein gel blot analysis

The boiled samples were subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis on 8% (w/v) polyacrylamide gels and transferred to a nitrocellulose membrane (Bio-Rad). GST-OsCOI proteins were detected using an anti-GST horseradish peroxidase (HRP) conjugate [dilution 1 : 5,000 (v/v)] (Cytiva, Tokyo, Japan) and an iBind™ Flex Western Device (Thermo Fisher Scientific). Chemiluminescent detection was performed using the Immobilon Western Chemiluminescent HRP Substrate (Merck) and ChemiDoc™ imaging system (Bio-Rad).

Statistical analysis

Dunnett’s test was performed using Excel (Microsoft Co., Redmond, WA, USA) and Statcel—the Useful Addin Forms on Excel, 3rd edn.—software (OMS Publishing, Inc., Saitama, Japan).

Supplementary Data

Supplementary data are available at PCP online.

Data Availability

Transcriptome data were deposited in the National Center for Biotechnology Information Gene Expression Omnibus database (ID: GSE205750). Data supporting the findings of this study are available from the corresponding author upon reasonable request.

Funding

Japan Society for the Promotion of Science (JSPS) KAKENHI (JP18K14399 to K.M.; JP20H02922 to K.O.; JP17H06407, JP18KK0162, and JP20H00402 to M.U.; JP18H02101, JP19H05283, JP21H00270 to Y.T.); Advanced Comprehensive Research Organization (ACRO) Research Grants of Teikyo University (TeTe20-01 to K.M. and M.A.); the Ministry of Education, Culture, Sports, Science, and Technology-Supported Program for the Strategic Research Foundation at Private Universities (S131052A01 to H.Y.).

Acknowledgements

We thank Masaki Endo and Masashi Mikami (the National Agriculture and Food Research Organization) for providing vectors for genome editing. We also thank to Yoshiaki Nagamura (the National Agriculture and Food Research Organization) for his advice and assistance in microarray analysis.

Author Contributions

H.Ina. performed all experiments. K.Hay., Y.T. and M.U. prepared the COI and JAZ peptides and performed the co-immunoprecipitation analysis. H.It., Y.F., A.Y.-N., X.C., M.S.-N., E.H., K.Hat., Y.H., M.I. and T.S. performed the transformation and screening of rice COI mutants. E.Y. carried out the LC–ESI-MS/MS analysis. M.A. performed electron microscopy observations. K.U. synthesized JA-Ile. H.Ina. and K.M. performed computational analyses. K.O., H.Y. and K.M. designed the study. Y.T., M.A., K.O., H.Y., M.U. and K.M. supervised the experiments. H.Ina. and K.M. wrote the manuscript. K.Hay., M.A., K.O. and H.Y. revised the manuscript. All authors have read and approved the final manuscript.

Disclosures

The authors have no conflicts of interest to declare.

References