Molecular Mechanism Underlying the Synergetic Effect of Jasmonate on Abscisic Acid Signaling during Seed Germination in Arabidopsis

Abscisic acid (ABA) is known to suppress seed germination and post-germinative growth of Arabidopsis thaliana, and jasmonate (JA) enhances ABA function. However, the molecular mechanism underlying the crosstalk between the ABA and JA signaling pathways remains largely elusive. Here, we show that exogenous coronatine (COR), a JA analog structurally similar to the active conjugate jasmonate-isoleucine (JA–Ile), significantly enhances the delayed seed germination response to ABA. Disruption of the JA receptor, CORONATINE INSENSITIVE1 (COI1), or accumulation of the JA signaling repressor, JASMONATE ZIM-DOMAIN, reduced ABA signaling, while jaz mutants enhanced ABA responses. Mechanistic investigations revealed that several JAZ repressors of JA signaling physically interact with ABSCISIC ACID INSENSITIVE3 (ABI3), a critical transcription factor that positively modulates ABA signaling, and that JAZ proteins repress the transcription of ABI3 and ABI5. Further genetic analyses showed that JA activates ABA signaling and requires functional ABI3 and ABI5. Overexpression of ABI3 and ABI5 simultaneously suppressed the ABA-insensitive phenotypes of the coi1-2 mutant and JAZ-accumulating (JAZ-ΔJas) plants. Together, our results reveal a previously uncharacterized signaling module in which JAZ repressors of the JA pathway regulate the ABA-responsive ABI3 and ABI5 transcription factors to integrate JA and ABA signals during seed germination and post-germinative growth. Plant Cell Advance Publication. Published on October 6, 2020, doi:10.1105/tpc.19.00838 ©2020 American Society of Plant Biologists. All Rights Reserved 2 INTRODUCTION The processes of seed germination and subsequent seedling establishment are precisely regulated by external and internal cues, including phytohormones. Abscisic acid (ABA) is an important phytohormone that regulates numerous physiological processes in plants, including seed germination, stomatal aperture, and seedling growth, as well as plant responses to various abiotic and biotic stresses (Mauch-Mani and Mauch, 2005; Finkelstein et al., 2008; Fujita et al., 2011; Hauser et al., 2011; Nakashima and Yamaguchi-Shinozaki, 2013). For example, ABA promotes seed dormancy and suppresses seed germination. In the presence of ABA, the ABA receptors PYRABACTIN RESISTANCE (PYR)/REGULATORY COMPONENT OF ABSCISIC ACID RECEPTOR (RCAR) recognize the ABA molecule (Ma et al., 2009; Miyazono et al., 2009; Nishimura et al., 2009; Santiago et al., 2009). Upon binding to ABA, these receptors form a stable complex with type 2C protein phosphatases (PP2Cs), leading to the release of SNF1-related kinases 2 (SnRK2s) from PP2C-SnRK2 complexes (Cutler et al., 2010). The activated SnRK2s subsequently phosphorylate downstream transcription factors such as ABSCISIC ACID-INSENSITIVE5 (ABI5), ABI4, and ABI3 to mediate ABA responses (Kobayashi et al., 2005; Furihata et al., 2006; Fujii et al., 2007; Fujii and Zhu, 2009; Nakashima et al., 2009; Zhu et al., 2020). The transcription factors ABI3 and ABI5 are highly induced by ABA and positively modulate ABA responses to suppress seed germination (Finkelstein, 1994; Lopez-Molina et al., 2001; Brocard et al., 2002; Finkelstein et al., 2005). Mechanistic investigations have revealed that ABI3 and/or ABI5 physically interact with several critical regulators to integrate multiple signaling pathways during seed germination. For instance, Park et al. (2011) demonstrated that ABI3 forms a protein complex with PHYTOCHROME INTERACTING FACTOR3-LIKE5 (PIL5) that activates the expression of SOMNUS (SOM) in imbibed seeds. Therefore, the SOM promoter integrates ABA and light signals to mediate seed germination. Previously, we showed that ABI5 physically interacts with the BRASSINOSTEROID INSENSITIVE2 (BIN2) brassinosteroid (BR)-related protein kinases, and that the BIN2-ABI5 cascade mediates the antagonism between BRs and ABA during seed germination (Hu and Yu, 2014). Lim et al. (2013) showed that ABI3, ABI5, and DELLA proteins interact to integrate ABA and gibberellin (GA) signals to modulate the expression of a subset of high temperature-inducible genes, leading to the inhibition of seed germination. The phytohormone jasmonate (JA) functions as a critical signaling molecule with roles in plant growth, development, and stress responses (Pauwels et al., 2010; Hu et al., 2013; Qi et al., 2015; Du et al., 2017). It is perceived by the F-box protein CORONATINE INSENSITIVE1 (COI1), which assembles in the SCFCOI1 3 protein complex that targets JASMONATE ZIM-DOMAIN (JAZ) proteins for degradation (Xu et al., 2002; Chini et al., 2007; Thines et al., 2007; Yan et al., 2009). The degradation of JAZ repressors alleviates the inhibition of downstream transcription factors, thus activating the transcriptional reprogramming of JA signaling (Chini et al., 2007; Thines et al., 2007; Pauwels et al., 2010). JAZ proteins have been shown to interact with multiple transcription factors to regulate diverse aspects of JA-regulated physiological processes, such as root growth, male fertility, anthocyanin accumulation, trichome development, senescence, and stress responses (Cheng et al., 2009; Niu et al., 2011; Fernández-Calvo et al., 2011; Kazan and Manners, 2013; Schweizer et al., 2013). In addition, JA signaling is involved in seed germination and interacts with ABA signaling. For example, methyl jasmonate (MeJA) has an inhibitory effect on the seed germination of several crops and Arabidopsis thaliana (Wilen et al., 1991; Krock et al., 2002; Preston et al., 2002; Norastehnia et al., 2007; Barrero et al., 2009; Dave et al., 2011; Dave et al., 2016). Several studies have provided evidence for crosstalk between ABA and JA signaling. For example, the genes encoding the ABA receptors PYRABACTIN RESISTANCE (PYR)/PYR1-LIKE (PYL)/REGULATORY COMPONENT OF ABSCISIC ACID RECEPTOR (PYL) 4 and PYL5 were found to be up-regulated by JA, and pyl4 and pyl5 knockout mutants were hypersensitive to JA (Lackman et al., 2011). Conversely, MYC2-overexpressing plants exhibited hypersensitivity to ABA (Abe et al., 2003; Lorenzo et al., 2004). The increased inhibition of germination in a line overexpressing a mutant MYC2 was associated with ABA hypersensitivity (Goossens et al., 2015). The results of those studies indicated that the ABA receptors PYL4/5 and the transcriptional factor MYC2 play vital roles in the crosstalk between JA and ABA. Further research showed that the E3 RING ligase KEG interacts with JAZ12 and regulates its stability (Pauwels et al., 2015). Another study provided evidence for the link between ABA and JA signaling through the direct interaction of the ABA receptor PYL6 with MYC2 (Aleman et al., 2016); the pyl6 mutant was more sensitive to the combination of JA and ABA than to ABA alone. Dave et al. (2011) found that 12-oxo-phytodienoic acid (OPDA, the precursor of JA) inhibited wild-type germination and acted with ABA to regulate seed germination in Arabidopsis. The abundance of ABI5 protein was increased by the inhibitory effect of OPDA and the combination of OPDA and ABA on seed germination. Ju et al. (2019) reported that JAZ proteins interact with ABI5 to modulate seed germination. Although JA signaling has been implicated in seed germination in several species, the detailed molecular mechanism of how JA regulates those crucial physiological processes remains largely unknown. In this study, we found that exogenous application of 1 μM COR or 10 μM MeJA alone did not inhibit seed germination.

The transcription factors ABI3 and ABI5 are highly induced by ABA and positively modulate ABA responses to suppress seed germination (Finkelstein, 1994;Lopez-Molina et al., 2001;Brocard et al., 2002;Finkelstein et al., 2005). Mechanistic investigations have revealed that ABI3 and/or ABI5 physically interact with several critical regulators to integrate multiple signaling pathways during seed germination. For instance, Park et al. (2011) demonstrated that ABI3 forms a protein complex with PHYTOCHROME INTERACTING FACTOR3-LIKE5 (PIL5) that activates the expression of SOMNUS (SOM) in imbibed seeds. Therefore, the SOM promoter integrates ABA and light signals to mediate seed germination.
Previously, we showed that ABI5 physically interacts with the BRASSINOSTEROID INSENSITIVE2 (BIN2) brassinosteroid (BR)-related protein kinases, and that the BIN2-ABI5 cascade mediates the antagonism between BRs and ABA during seed germination (Hu and Yu, 2014). Lim et al. (2013) showed that ABI3, ABI5, and DELLA proteins interact to integrate ABA and gibberellin (GA) signals to modulate the expression of a subset of high temperature-inducible genes, leading to the inhibition of seed germination.
The phytohormone jasmonate (JA) functions as a critical signaling molecule with roles in plant growth, development, and stress responses (Pauwels et al., 2010;Hu et al., 2013;Qi et al., 2015;Du et al., 2017). It is perceived by the F-box protein CORONATINE INSENSITIVE1 (COI1), which assembles in the SCF COI1 However, JA significantly delayed germination in the presence of ABA, showing that JA stimulates ABA signaling to delay seed germination. We discovered that the COI1/JAZ-mediated JA signaling pathway is positively involved in ABA-delayed seed germination. Mechanistic analyses revealed that several JAZ proteins physically interact with the ABA-responsive ABI3 transcription factor and repress ABI3/ABI5-mediated transcriptional activation of downstream targets. Further genetic analyses showed that JA activates ABA signaling to delay seed germination and requires functional ABI3 and ABI5.
Overexpression of ABI3 and ABI5 simultaneously suppressed the ABA-insensitive phenotypes of coi1 mutants and JAZ-ΔJas (JAZ1, JAZ5, and JAZ8 lacking the Jas domain) plants. Our results provide a mechanistic understanding of how the integration of the JA and ABA signaling pathways is mediated by the JAZ-ABI3/ABI5 module during seed germination.

Exogenous JA Enhances ABA Signaling to Delay Seed Germination
To investigate the molecular mechanisms underlying JA regulation of seed germination, we first confirmed the regulatory effect of JA by evaluating the germination of wild-type (Col-0) Arabidopsis seeds on medium supplemented with MeJA and/or ABA. To avoid the effects of sucrose and/or nitrate on seed germination, we performed germination assays using water agar medium. Consistent with a previous study (Dave et al., 2011), the Col-0 seeds displayed similar percentages of germination and expanded cotyledons on medium with or without 10 μM MeJA (Supplemental Figure 1), confirming that 10 μM MeJA alone has no inhibitory effect on germination. We then analyzed the performance of Col-0 seeds on medium containing 10 μM MeJA with or without 0.5 μM ABA. Compared with seeds treated with MeJA or ABA alone, those treated with both MeJA and ABA showed much lower percentages of germination and expanded cotyledons (Supplemental Figure 1), suggesting that MeJA enhances ABA signaling during seed germination (Staswick et al., 1992;Ellis and Turner, 2002). To verify these observations, we used coronatine (COR), a JA analog that is structurally similar to the active conjugate jasmonate-isoleucine (JA-Ile). Whereas 1 μM COR did not inhibit germination and cotyledon expansion, Col-0 plants exhibited lower percentages of germination and of expanded cotyledons on both COR medium and on ABA medium ( Figure 1A to 1C).
To corroborate the regulatory effect of COR on ABA signaling during seed germination, we used reverse transcription quantitative PCR (RT-qPCR) to determine the transcript levels of several well-characterized ABA-responsive genes, including ALCOHOL DEHYDROGENASE1 (ADH1), LATE EMBRYOGENESIS ABUNDANT6 (EM6) and EM1 (Finkelstein and Lynch, 2000;Carles et al., 2002;Lopez-Molina et al., 2002), in Col-0 germinating seeds treated with ABA and COR simultaneously and individually. The transcript levels of these genes were higher in seeds treated with both ABA and COR than in seeds treated with ABA or COR alone ( Figure 1D). Taken together, these results demonstrate that exogenous JA enhances ABA signaling to delay seed germination.

Germination
Having ascertained that JA interacts with ABA synergistically to delay seed germination, we then queried whether critical components of the JA signaling pathway participate in these ABA-regulated processes. The JA receptor F-box protein COI1 is a positive modulator of JA signaling (Xie et al., 1998;Yan et al., 2009).
To investigate whether COI1 is involved in ABA responses during seed germination, we analyzed the loss-of-function mutant coi1-2. Compared with Col-0 seeds, coi1-2 seeds exhibited much higher percentages of germination and cotyledon greening on medium containing ABA ( nitrate enhanced this process. Then, in comparison with seeds exposed to both ABA and nitrate medium, the germination of wild type was similar to coi1-2 and coi1-16 mutant, indicating that nitrate had an effect on ABA signaling during seed germination. Similarly, there were no differences in the percentages of germination and expanded cotyledon between the coi1 mutants with Col-0 on the presence of both ABA and sucrose, suggesting that sucrose participated in ABA signaling to delay seed germination (Supplemental Figure 3). These results suggested that the phenotype of coi1 mutants is conditioned by the absence of nitrate or sucrose. Therefore, we used water agar medium to avoid the effects of sucrose and/or nitrate. In parallel experiments, compared with Col-0, in our hands and experimental conditions, coi1-16 (another loss-of-function mutant of COI1) showed reduced sensitivity to ABA during seed germination and post-germinative growth ( promoter into coi1-2 and coi1-16 mutant plants (coi1-2 COI1 and coi1-16 COI1). Expression of COI1 in coi1-2 and coi1-16 fully complemented the MeJA-insensitive root elongation phenotype (Supplemental Figure 4). Moreover, coi1-2 COI1 and coi1-16 COI1 plants had similar responses to those of Col-0 to ABA during seed germination (Supplemental Figure 5). To avoid unspecific effects due to misexpression of COI1 under the control of the 35S promoter, we also introduced COI1 into the coi1-2 and coi1-16 backgrounds under the control of its native promoter (coi1-2/COI1pro:COI1 and coi1-16/COI1pro:COI1). The coi1-2/COI1pro:COI1 and coi1-16/COI1pro:COI1 seeds consistently exhibited percentages of germination and cotyledon greening similar to those of Col-0 under ABA treatment (Supplemental Figure 5). These results suggest that endogenous JA perception or signaling positively mediates ABA responses during seed germination and early seedling growth.
To further confirm the involvement of JA signaling components in ABA responses, we investigated whether genetic manipulation of JAZ proteins, which are the substrates of the SCF COI1 E3 ubiquitin ligase and repressors of JA signaling (Chini et al., 2007;Thines et al., 2007), affects ABA-repressed seed germination. Phenotypic analysis showed that the jazQ mutant (with T-DNA insertion mutations in JAZ1/3/4/9/10 genes; Campos et al., 2016) was more sensitive than Col-0 to ABA during seed germination and seedling establishment (Figure 2A to 2D). Consistently, compared with seeds of wild-type, seeds of the jaz decuple mutant (jazD; defective in JAZ1-7,-9, -10 and -13;Guo et al., 2018) were more sensitive to ABA (Supplemental Figure 2B). To further confirm this possibility, we expressed the full-length cDNA of JAZ1 under the control of the CaMV 35S promoter (JAZ1-OE). However, there were no differences in the percentages of germination and cotyledon greening between the homozygous JAZ1-OE plants and Col-0 plants with or without ABA. Similar results were obtained for JAZ5-OE and JAZ8-OE seeds (Supplemental Figure 6A and 6B).
As JAZ1-ΔJas homozygous plants are completely sterile, seeds from heterozygous plants were used in the germination assays. When incubated on medium containing 0.5 µM ABA, JAZ1-ΔJas heterozygous (F1 progeny from a cross with wild type) lines displayed much higher percentages of germination and cotyledon greening in comparison with Col-0 ( Figure 2A to 2D). Compared with wild-type seeds, those of JAZ5-ΔJas or JAZ8-ΔJas were also less insensitive to ABA (Figure 2A to 2D). Among the JAZ1-ΔJas heterozygous seeds germinated on medium containing 0.75 µM ABA for 3.5 d, most of the seedlings with green cotyledons were JAZ1-ΔJas transgenic plants (n = 80, 91.7%). The transcript levels of several ABA-responsive genes were consistently lower in coi1-2 than in Col-0 in response to ABA ( Figure 2E), but were higher in the jazQ mutant than in Col-0 ( Figure 2E).
Taken together, these findings provide further evidence that the endogenous JA pathway is positively involved in ABA signaling during seed germination.

JAZ Repressors Physically Interact with ABI3
Recent studies have revealed that JAZ repressors physically interact with numerous transcription factors to modulate different aspects of JA signaling processes (Qi et al., 2011;Song et al., 2011;Zhu et al., 2011;Hu et al., 2013;Jiang et al., 2014). To test the interactions between JAZ1 and transcription factors, the full-length coding sequence (CDS) of JAZ1 was fused to pGBK-T7 to construct a bait vector, which was transformed into the yeast strain Y2HGOLD (Clontech, Palo Alto, CA, USA). Yeast screening was performed as described previously (Hu et al., 2013), and the cDNA library was acquired from Clontech (Cat. no. 630487). After screening, more than 10 independent colonies were isolated. Among these positive clones, clones encoding ABI5, TOE1, TOE2, MYC2, and NINJA were frequently sequenced. ABI3 was sequenced seven times through this screening procedure.
To further confirm the interaction between JAZs and ABI3, we generated the full-length ABI3 in the bait  Figure 7; Supplemental Data Set 1). ABI4 is a well-known transcription factor that positively modulates ABA responses during seed germination. However, no interaction was detected between ABI4 and JAZ ( Figure 3B), suggesting that JAZ proteins involved in the ABA response interact with ABI3 but not ABI4.
These interactions between JAZ proteins and ABI3 were confirmed in plant cells using bimolecular florescence complementation (BiFC) and co-immunoprecipitation (Co-IP) assays. For the BiFC assays, each JAZ was fused to the N-terminal yellow fluorescent protein (YFP) fragment to generate JAZ-nYFP, while ABI3 and ABI4 were each fused to the C-terminal fragment of YFP (cYFP) to produce ABI3-cYFP and ABI4-cYFP, respectively. When ABI3-cYFP was transiently co-expressed with JAZ1-nYFP, JAZ5-nYFP, or JAZ8-nYFP in leaf cells of Nicotiana benthamiana, reconstituted YFP fluorescence was observed in the nucleus, as revealed by staining with 4',6-diamidino-2-phenylindole (DAPI) ( Figure 3C). As the negative control, ABI3-cYFP (or ABI4-cYFP) was co-expressed with JAZ4-nYFP and accumulated (or JAZ1-nYFP, JAZ5-nYFP and JAZ8-nYFP) in N. benthamiana leaf cells and no fluorescence was detected ( Figure  Having ascertained that JAZ proteins physically interact with ABI3, we wondered whether JAZs are involved in ABI3-mediated ABA signaling during seed germination. To test this possibility, we further analyzed the properties of JAZs in more detail. To examine JAZ expression in different tissues, we analyzed its transcript levels by RT-qPCR. Interestingly, relatively high transcript levels of JAZ1, JAZ5, and JAZ8 were detected in dry seeds (Supplemental Figure 6C to 6E), indicating that JAZ proteins play roles in seed germination.

Domain
To examine which regions of ABI3 are required to interact with JAZ proteins, we performed directed yeast two-hybrid analyses. ABI3 was divided into the N-terminal region (BD-ABI3-N) and the C-terminal region containing the B2 and B3 domains (BD-ABI3-C; Figure 4A). Deletion of the C-terminal residues of ABI3 (BD-ABI3-C) did not affect its interactions with JAZ1 and JAZ5 (Supplemental Data Set 1); however, deletion of the N-terminal fragment of ABI3 completely abolished its interactions with these proteins ( Figure 4A). Therefore, the N-terminal region of ABI3 is required for its interactions with JAZ1 and JAZ5.
We used the same approach to identify the domains of JAZ proteins that are required for their interactions with ABI3. JAZ1 and JAZ8 (Supplemental Data Set 1) were divided into different truncated versions and then independently fused with the Gal4-activation domain to generate prey vectors ( Figure 4B and 4C). The potential interactions between these derivatives and ABI3 were then assayed using the yeast two-hybrid system. Deletion of the C-terminal 133 residues of JAZ1 (AD-JAZ1 1-120) eliminated its interaction with ABI3, while deleting its N-terminal fragment (AD-JAZ1 121-253) had no effect ( Figure 4B). Further mapping revealed that the middle region of JAZ1 containing the ZIM domain was required for the JAZ1-ABI3 interaction. Likewise, the middle region of JAZ8 containing the ZIM domain was essential for the JAZ8-ABI3 interaction ( Figure 4C).

JA Activation of ABA Signaling Requires Functional ABI3 and ABI5
The ABI3 and ABI5 transcription factors are critical regulators of ABA signaling during seed germination and early seedling growth. Phenotypic analyses have indicated that the loss-of-function mutants abi3 and abi5 are insensitive to ABA (Koornneef et al., 1984;Giraudat et al., 1992;Finkelstein, 1994;Finkelstein and Lynch, 2000). Given that several JAZ repressors of the JA signaling pathway physically associate with ABI3 and ABI5 ( Figure 3; Ju et al., 2019), we hypothesized that JA stimulates ABA responses during seed germination and that this requires ABI3 and/or ABI5. To test this possibility, we determined the percentages of germination and expanded cotyledons of abi3 (abi3-1 mutant backcrossed six times with Col-0 wild type) and abi5 (abi5-1 mutant backcrossed six times with Col-0 wild type) seeds on media containing ABA and/or COR. Consistent with previous studies (Koornneef et al., 1984;Giraudat et al., 1992;Finkelstein, 1994), abi3 and abi5 seeds germinated and grew faster than Col-0 seeds on ABA-containing medium, and the germination percentages of abi3-8 and abi5-7 were higher than that of Col-0 ( Figure 5). Furthermore, compared with Ler and Ws, the abi3-1 and abi5-1 single mutants exhibited much higher percentages of germination and expanded cotyledons on media containing ABA, COR, or both ABA and COR ( Figure 5).
Compared with abi3, abi5, and Col-0, a higher percentages of the double mutant had expanded cotyledons in the presence of both ABA and COR ( Figure 5). These observations indicate that JA activates ABA signaling to delay seed germination and that this requires functional ABI3 and ABI5.
The ABI3 and ABI5 transcription factors are recognized as key regulators of seed dormancy in Arabidopsis (Bentsink and Koornneef, 2008). Because JAZ proteins physically interact with ABI3, we questioned whether COI1 and JAZ are also involved in modulating seed dormancy. To test this possibility, we examined the fresh mature siliques of coi1-2 and JAZ1-ΔJas plants after 5 d on water-saturated filter paper at 22°C without stratification. The seed dormancy levels of coi1-2 and JAZ1-ΔJas were similar to that of wild type (Supplemental Figure 10). However, seed dormancy was significantly reduced in the abi3 mutant, consistent with previous studies (Ooms et al., 1993;Finkelstein et al., 1994;Liu et al., 2013). These results suggest that the COI1/JAZ-mediated JA pathway does not regulate seed dormancy in Arabidopsis

JAZ Proteins Repress the Transcriptional Activation Roles of ABI3 and ABI5
Because JAZ proteins directly interact with ABI3 ( Figure 3) and ABI5 (Ju et al., 2019), we investigated whether JAZ proteins interfere with the transcriptional activation roles of ABI3 and/or ABI5. To this end, we used dual luciferase (LUC) reporter assays to examine the effects of JAZ1, JAZ5, or JAZ8 on the transcriptional functions of ABI3 or ABI5 in Arabidopsis mesophyll protoplasts (Yoo et al., 2007). Because EM1, EM6, and ABI5 are downstream targets of ABI5 and ABI3 (Lopez-Molina and Chua, 2000;Nakamura et al., 2001;Carles et al., 2002;Lopez-Molina et al., 2002), their promoters were fused to the LUC gene to generate reporters ( Figure 6A). The effector constructs contained ABI5, ABI3, JAZ, or GFP driven by the CaMV 35S promoter ( Figure 6A). ABI3 expression activated LUC expression driven by the EM6 or ABI5 promoter with ABA treatment, compared with the expression of the GFP control ( Figure 6B and 6C).
Moreover, compared with expression of ABI3 alone, co-expression of JAZ1 or JAZ5 with ABI3 repressed the LUC expression level ( Figure 6B and 6C). Similarly, ABI5 expression dramatically activated LUC expression driven by the EM6 or EM1 promoter in the presence of ABA in comparison with expression of the GFP control ( Figure 6D and 6E). However, compared with expression of ABI5 alone, co-expression of ABI5 with JAZ8 suppressed LUC expression ( Figure 6D and 6E). These results demonstrate that JAZ proteins repress the transcriptional activation roles of ABI3 and ABI5, thereby preventing activation of their downstream targets.
Because ABI5 and ABI3 positively modulate ABA signaling during seed germination, we simultaneously Because ABI3 interacts with JAZ, the jazQ and jazD mutants were sensitive to ABA during seed germination, and overexpression of ABI3 and ABI5 suppressed the ABA-insensitivity of coi1-2 and coi1-16.
We speculated that signaling mediated by the receptor COI is required to induce ABA responses. Therefore, we performed RNA-seq experiments to profile the transcriptomes of coi1-2, coi1-16, and Col-0 with or without ABA treatments. Total RNAs were isolated from seeds treated with or without ABA. We sequenced each sample on the Illumina HiSeq X Ten platform. Sequenced reads were trimmed to remove adaptor sequences, and low-complexity or low-quality sequences were removed using Trimmomatic (0.36) with the following parameters: LEADING:3 TRAILING:3 SLIDINGWINDOW: 4:15 MINLEN:50. Clean reads were mapped to the genome of TAIR10.1_NCBI_year_2018 using Hisat2 (2.2.1.0) with default parameters.
The read counts of each gene were obtained using htseq-count (0.9.1).
Under normal conditions (no ABA treatment), compared with wild type, coi1-2 and coi1-16 had 84 up-regulated genes and 160 down-regulated genes ( Figure 9A). We classified the genes co-regulated by COI1 under normal conditions and found that they encode proteins with various biological functions, including seed oil body biogenesis, lipid storage, and response to ABA ( Figure 9B). Under ABA treatment, 106 genes were up-regulated and 210 genes were down-regulated in both coi1-2 and coi1-16, compared with wild type ( Figure 9C).
To identify the genes responsible for the ABA-insensitive phenotypes of the coi1-2 and coi1-16 mutants, we examined the overlap between co-regulated genes in the two mutants and those regulated by ABA signaling in the wild type. Among all the co-regulated genes in coi1-2 and coi1-16, 16 were responsive to ABA ( Figure 9D, Supplemental Data Set 2). We then compared the expression levels of these co-regulated and ABA-responsive genes under normal and ABA-treatment conditions. A heat map analysis indicated that the expression levels of most co-regulated genes were lower in coi1-2 and coi1-16 than in the wild type under both normal and ABA-treatment conditions ( Figure 9E).
We used reverse transcription-quantitative PCR to confirm the transcript levels of selected genes from the set of co-regulated and ABA-responsive genes, including CRU1, CRU3, LEA and RAB18. The analyses revealed that those genes' transcript levels were lower in coi1-2 and coi1-16 than in the wild type ( Figure   9F).
Ju et al. (2019) found that ABA treatments could promote jasmonate biosynthesis. Consistent with this, qRT-PCR confirmed that several JA biosynthesis genes (ACX1, AOC3, KAT5, and OPR3) were up-regulated by ABA treatment in wild type (Supplemental Figure 13E to 13H). The transcription of ABA biosynthesis genes (ABA1, ABA2, ABA3, and NCED3) also was up-regulated by exogenous ABA (Supplemental Figure   13A to 13D). However, under ABA treatment, the ABA induction of these ABA biosynthetic genes was reduced in coi1-2 and coi1-16 compared with wild type (Supplemental Figure 13I to 13L). Similarly, the transcript levels of these JA biosynthetic genes were decreased in the coi1-2 and coi1-16 mutants (Supplemental Figure 13M to 13P). These results indicate that ABA treatment promotes JA and ABA biosynthesis, while ABA-induced JA and ABA biosynthesis is reduced in the coi1-2 and coi1-16 mutants, and increased in the jazQ mutant (Supplemental Figure 13).

Discussion
The phytohormone JA is ubiquitous in the plant kingdom and regulates multiple physiological aspects of plant development, growth, and stress responses. Several studies have highlighted the involvement of JA in seed germination, and have suggested that it may interact with ABA to mediate these processes (Wilen et al., 1991;Krock et al., 2002;Preston et al., 2002;Norastehnia et al., 2007;Barrero et al., 2009;Dave et al., 2011). However, the detailed molecular mechanisms underlying JA regulation of seed germination and its crosstalk with ABA remain elusive. In this study, we further investigated the regulatory role of JA in modulating ABA responses during seed germination and subsequent post-germinative growth. Consistent with previous studies, our results demonstrate that exogenous JA (COR) has a positive role in activating ABA responses to delay seed germination. Col-0 seeds displayed low percentages of germination and of expanded cotyledons in the presence of both ABA and JA (Figure 1; Supplemental Figure 1), while lines with blocked endogenous JA perception or signaling were less sensitive to ABA during seed germination ( Figure 2; Supplemental Figure 2). Based on these findings, we conclude that the JA signal activates ABA responses to delay seed germination in Arabidopsis.
Intriguingly, we found that mutation of COI1 leads to decreased ABA signaling during seed germination.
This finding is in disagreement with the results of two earlier studies (Ellis and Turner, 2002;Fernández-Arbaizar et al., 2012). In those studies, the authors analyzed one of the same coi1 alleles that we used, coi1-16, and found that it was hypersensitive to ABA in comparison with Col-0. To verify the role of COI1 in mediating the ABA response, we also investigated another coi1 allele, coi1-2, in the presence of ABA. Like our coi1-16 mutant, coi1-2 seeds also exhibited much higher percentages of germination and greening cotyledons than Col-0 seeds on medium containing ABA (Figure 2; Supplemental Figure 2A). Consistent with this phenotype, the induced expression levels of several well-characterized ABA-responsive genes were lower in coi1-2 than in wild type ( Figure 2E). More importantly, to further corroborate that mutation of COI1 is responsible for the ABA-insensitive phenotypes of coi1-2 and coi1-16, we expressed the full-length COI1 gene driven by its native promoter or the CaMV 35S promoter in the coi1-2 and coi1-16 mutant backgrounds. As expected, expression of full-length COI1 in the coi1-2 or coi1-16 background resulted in the mutant plants responding similarly to Col-0 under ABA treatment (Supplemental Figure 5). Moreover, on medium containing ABA, the germination and greening percentages were much higher in heterozygous JAZ-ΔJas plants, which accumulate higher levels of JAZ proteins, than in Col-0 ( Figure 2 and 8), similar to the effects seen for the coi1 mutants. In contrast, the germination rates were lower in the jazQ and jazD mutants than in wild type under ABA treatment. Therefore, our results show that jasmonate synergistically regulates ABA signaling during seed germination.
The coi1-16 phenotype and germination rates of seeds may be dependent on our experimental conditions because many external factors, including light, humidity, temperature, and availability of nitrogenous compounds are known to affect ABA responses during seed germination. For example, the light intensity under long-light conditions is one factor that differs among various laboratories. The germination media also differ among different studies. Some studies conducted germination assays on water agar medium to avoid the effects of nitrate and sucrose (this study; Dave et al., 2011), whereas different media were used in earlier studies (Ellis and Turner, 2002;Fernández-Arbaizar et al., 2012). Differences in these factors may account for the inconsistency of the germination rates and the coi1-16 phenotype in response to ABA among different studies. In our hands and experimental conditions, the ABA sensitivity of coi1 mutants was changed under conditions where exogenous nitrate or sucrose was present (Supplemental Figure 3). The data suggested that coi1 mutants may have different ABA sensitivity according to the imbibition conditions. Furthermore, our germination data were not identical on half-strength MS and water agar medium, but seeds displayed a similar percentage of expanded cotyledons in both media.
Recently, several studies have provided evidence of crosstalk between JA and ABA signaling. The results of those studies supported that JA interacts with ABA signaling to regulate physiological process (Lackman et al., 2011;Pauwels et al., 2015). In this study, we found that JAZ proteins physically associate with an important transcription factor in ABA signaling, ABI3, and further demonstrated that those JAZ proteins repress the transcriptional activation functions of ABI3 and ABI5 (Figure 3 and 6). In our analyses, these JAZ repressors specifically interacted with ABI3 and ABI5, but did not form complexes with ABI4, an APETALA2 (AP2) family transcription factor that positively modulates ABA signaling in yeast (Figure 3; Ju et al., 2019). Further analyses revealed that the ZIM domain of JAZ is required for interactions between JAZ and ABI3 (Figure 4). Deletion of the C-terminal 133 residues of JAZ that contain the ZIM domain eliminated the interaction between JAZ with ABI3, while deletion of the N-terminal region (121-253 residues) did not affect this interaction.
Interestingly, this finding differs from that of several previous studies, which showed that the Jas domain is essential for most of the interactions between JAZ repressors and downstream transcription factors (Pauwels and Goossens, 2011). Recently, Zhai et al. (2015) found that the deletion of a 53-amino acid residue N-terminal region (including the ZIM domain) of JAZ1 eliminated its interaction with TOE1, indicating that this region is required for the JAZ-TOE1 interaction. Other studies also reported that the region containing the ZIM domain is vital for interactions with other proteins (Cheng et al., 2011;Song et al., 2011;Jiang et al., 2014). Importantly, the ZIM domain of most JAZ proteins can recruit TPL or TPR proteins indirectly through the ethylene-response factor amphiphilic repression (EAR) motif-containing NOVEL INTERACTOR OF JAZ (NINJA) adaptor protein to repress JA responses (Shyu et al., 2012;Thatcher et al., 2016;Howe et al., 2018). In addition, sequence variations in the hypervariable region of the degron affect JAZ stability and JA-regulated physiological responses. JAZ8-mediated repression depends on an EAR motif at the JAZ8 N terminus, which binds the corepressor TOPLESS and represses its transcriptional activation function (Shyu et al., 2012). Therefore, different domains of JAZ proteins interact with different proteins. In future studies, the identification of JAZ-associated transcription factors and further mapping of specific JAZ domains or residues required for these interactions may enhance our understanding of JAZ-regulated targets.
The transcription factors ABI3 and ABI5, which are mainly expressed in seeds and are strongly induced by ABA, play critical roles in modulating ABA responses to suppress seed germination and early seedling growth (Finkelstein, 1994;Finkelstein and Lynch, 2000;Lopez-Molina and Chua, 2000;Lopez-Molina et al., 2001, 2002Suzuki et al., 2001;Brocard et al., 2002;Finkelstein et al., 2005;Bedi et al., 2016). Previous studies have shown that the loss-of-function mutants abi3 and abi5 are insensitive to ABA treatment (Koornneef et al., 1984;Giraudat et al., 1992;Finkelstein, 1994;Finkelstein and Lynch, 2000). Interestingly, our phenotypic analyses showed that abi3, abi5, and an abi5 abi3 double mutant also exhibited higher percentages of expanded cotyledons than Col-0 on medium containing ABA and COR ( Figure 5). These findings demonstrate that JA stimulates ABA signaling to delay seed germination and post-germinative growth, and that this requires functional ABI3 and ABI5. We found that overexpression of ABI3 and ABI5 simultaneously suppresses the ABA-insensitive phenotypes of coi1-2 and JAZ8-ΔJas mutants (Figure 7 and   8). This finding further supports the hypothesis that ABI3/ABI5 and JAZ proteins modulate ABA and JA signaling during seed germination. Together, the results of our biochemical and genetic analyses reveal a previously unknown signaling module in which JAZ repressors in the JA pathway directly modulate ABA-responsive ABI3 and ABI5 transcription factors to integrate JA and ABA signals during seed germination and post-germinative growth.
In a recent study (Pan et al., 2018), two VQ (containing the conserved FxxxVQxxTG motif) proteins, VQ18 and VQ26, were shown to physically associate with ABI5 and interfere with its transcriptional role.
As with the JAZ proteins in the present study, VQ18 and VQ26 do not mediate seed dormancy but, rather, negatively modulate ABA signaling during seed germination and early seedling establishment. However, Liu family, regulates ABI5 expression to stimulate seed dormancy but represses ABA responses during seed germination. Together, the results of these studies show that the components that regulate or interact with ABI3 and/or ABI5 may have distinct roles in regulating seed dormancy or ABA signaling. Further research is required to dissect the crucial regulators of ABI3/ABI5 and elucidate the exact mechanisms underlying ABI3/ABI5-mediated seed dormancy and regulation of the ABA signaling network.
JA is a hormone induced by pathogen attack, wounding, and some abiotic stresses (Kazan, 2015;Howe et al., 2018), and ABA also is induced by several abiotic stresses. It is easy to imagine that when seeds are exposed to soil, environmental factors induce the accumulation of JA and ABA. In plants, ABA and JA are two important hormones that regulate diverse aspects of plant developmental and stress responses. Some observations have revealed clues about the crosstalk between the ABA and JA signaling pathways. For instance, ABA induces the expression of PLIP2 and PLIP3, which may participate in JA biosynthesis (Wang et al., 2018). Another study showed that ABA does indeed promote JA biosynthesis (Ju et al., 2019).
Consistent with these studies, we found that ABA treatments enhanced JA and ABA biosynthesis. Then, the accumulated JA may cooperate with ABA to delay seed germination for avoiding a bad environment. Our results reveal some of the mechanisms underlying the crosstalk between ABA and JA: JAZ proteins physically interact with the transcription factors ABI3/ABI5 and repress their activities to delay seed germination. As shown in Figure 10, we propose the following working model: under unfavorable growth conditions, JAZ proteins are degraded, then ABI3 and ABI5 are released from JAZ-mediated repression to activate ABA signaling; simultaneously, the transcription factors ABI3 and ABI5 are activated by ABA, thereby delaying the process of seed germination.

Generation of Transgenic Arabidopsis Lines
To generate JAZ-and ABI3-OE transgenic lines, the full-length cDNA of ABI3 was cloned into the binary vector pOCA30 (Pan et al., 2018) in the sense orientation behind the CaMV 35S promoter (Hu et al., 2013).
The JAZ1, JAZ5, or JAZ8 cDNAs had the Jas domain deleted (Thines et al., 2007), and each was independently cloned into the binary vector pOCA30 in the sense orientation behind the CaMV 35S promoter. The heterozygous JAZ-ΔJas plants were the F1 progeny of JAZ-ΔJas crossed with wild type, and the seeds of this cross were used in a germination assay. The primers and restriction enzyme sites used to amplify sequences and generate vectors are listed in Supplemental Table 1. Because the abi5-1 mutant was in the Ws background, we backcrossed it with the Col-0 wild type six times before use. Similarly, the abi3-1 mutant (in the Ler background) was backcrossed with the Col-0 wild type six times before use. The abi5 abi3 double mutant was generated by crossing the backcrossed abi5 mutant with the backcrossed abi3 mutant.

Seed Germination and Dormancy Assays
Plants of each genotype were grown side by side under the same conditions. Seeds of each genotype were harvested from independent plants and stored at the same time, and pooled before germination assays. We scored germination as radicle emergence from the seed coat and endosperm, and we also recorded cotyledon opening and cotyledon expansion (Dave et al., 2011;Piskurewicz and Lopez-Molina, 2016). For the stratification treatment, seeds were stratified at 4°C in the dark for 3 d.
To examine the phenotypes of germinated materials, every batch of seeds was grown on water agar medium supplemented with or without ABA and/or COR for the indicated periods in long-day conditions (14-h light and 10-h dark). Every batch of dry mature seeds for each genotype was pooled from at least 15 independent plants (each figure represents data from seeds that matured and were harvested at the same time). Average values were calculated from three biological replicates and compared using statistical tests to determine the significance of differences. For every biological replicate, we tested the seeds from the same batch at least three times as technical replicates.
For seed dormancy analysis, for each genotype, we examined fresh mature siliques on plants grown under the same conditions and maturing at the same time after 5 d of growth on water-saturated filter paper at 22°C without stratification (Liu et al., 2013).

Expression Analyses
Seeds were placed on media containing hormones as indicated, and total RNA was extracted from seeds after the indicated periods. Total RNA was extracted using TRIzol reagent (Invitrogen) for real-time RT-PCR analysis. Superscript II was used according to the manufacturer's instructions (Invitrogen, Carlsbad, CA, USA), and 1 µg DNase-treated RNA was reverse transcribed in a 20 µl reaction volume. Each qRT-PCR was conducted using 1 µl cDNA and the SYBR Premix Ex Taq kit (Takara) on a Roche LightCycler 480 real-time PCR instrument. At least three independent biological samples for each replicate were analyzed.
AT1G13320 was used as an internal control. The independent biological data were subjected to ANOVA (Supplemental Data Set 3). The 2 -ΔΔCt method was used for relative quantification of gene transcript levels (Livak and Schmittgen, 2001). The RT-qPCR primers used in these analyses are listed in Supplemental Table   2.

RNA Sequencing and Data Analysis
The Col-0, coi1-2, and coi1-16 seeds were grown on media with or without 1 µM ABA for 24 h, and then the materials were collected. Three biological replicates were prepared for each sample, and RNA was extracted from a mixed sample of seeds. RNA was extracted using TRIzol reagent (Invitrogen) using the ethanol precipitation protocol, and then sequenced (OEBIOTECH, Shanghai, China). Clean reads were mapped to the Arabidopsis genome (TAIR10; www.arabidopsis.org) after screening and trimming. Cufflinks software was used to determine expression values (Tarazona et al., 2011). Genes with estimated absolute fold changes ≥1 were identified as reliable differentially expressed genes (DEGs). DEGs were analyzed using DESeq (Anders and Huber, 2012). Multiple testing was corrected via false discovery rate estimation and q-values below 0.05 were considered to indicate differential expression. Subsequently, gene ontology (GO) enrichment analysis was performed using TopGO (Alexa and Rahnenfuhrer, 2010). The RT-qPCR primers used in the analysis shown in Figure 9F are listed in Supplemental Table 2.

Y2H Screening and Confirmation
The full-length CDS of ABI3 was fused to pGBKT7 to construct the bait vector, which was transformed into the yeast strain Y2HGOLD (Clontech). Yeast screening was performed as described previously (Hu et al., 2013). Each of the 12 JAZs (JAZ1, JAZ2, JAZ3, JAZ4, JAZ5, JAZ6, JAZ7, JAZ8, JAZ9, JAZ10, JAZ11, and JAZ12) was introduced into the prey vector pGADT7. These plasmids were co-transfected into yeast strain AH109. The transfected yeast cells were plated on SD/-Leu/-Trp medium and SD/-Ade/-His/-Leu/-Trp medium and cultured at 28°C for 4 days. The primers and restriction enzyme sites used to amplify sequences and generate vectors are listed in Supplemental Table 3.

BiFC Assays
The full-length CDSs of JAZ1, JAZ4, JAZ5, and JAZ8 were fused in-frame to the YFP C-terminus of the C-YFP to form JAZ1-cYFP, JAZ4-cYFP, JAZ5-cYFP and JAZ8-cYFP, respectively. The ABI3 and ABI4 full-length CDSs were cloned in-frame to the YFP N-terminus, respectively. Each of the cloning plasmids was introduced into Agrobacterium tumefaciens strain EHA105, and then infiltrated into N. benthamiana leaves as described previously (Wang et al., 2016). At 48 h after infiltration, infected tissues were analyzed under a confocal laser-scanning microscope (Leica, Wetzlar, Germany). The primers and restriction enzyme sites are listed in Supplemental Table 3.
Combinations of plasmids were introduced into mesophilic protoplasts from Arabidopsis according to the protocol of Sheen (2001). Transfected cells were cultured for 10 to 16 h with or without 5 µM ABA, and then relative LUC activity was determined using the Dual-Luciferase Reporter Assay Protocol machine (Promega, Madison, WI, USA), which measures the activities of firefly LUC and the internal control Renilla luciferase (REN). The primers and restriction enzyme sites are listed in Supplemental Table 4.

Statistical analysis
Analysis of variance (ANOVA) was performed using SPSS software. A value of P < 0.05 was considered to be statistically significant. The results of statistical analyses are shown in Supplement Data Set 3.

Accession Numbers
Arabidopsis Genome Initiative numbers for the genes discussed in this article are as follows: ABA1,    COR, respectively, then those seeds grown on these medium 4 h in long-day conditions. The average and significance were calculated over three biological replicates. Total RNA was extracted from at least three batches of seeds as biological replicates. Each batch of seeds for the wild type was pooled from 50 independent plants. For each biological replicate, more than 250 seeds of the same batch were used for RNA extraction. AT1G13320 gene was used as control. Data from three biological replicates were analyzed by ANOVA. Values with different letters are significantly different from each other (P < 0.05).
Data from three biological replicates were analyzed by ANOVA. Values with different letters are significantly different from each other (p < 0.05). Error bars show SD from three biological replicates. were calculated over three biological replicates by analyzing seeds of different batches. Each batch of seeds for each genotype was pooled from at least 18 independent plants. For every biological replicate, we examined the seeds from the same batch at least three times as technical replicates. The value of each biological replicate was the average calculated over three technical replicates by analyzing more than 150 seeds. Data from three biological replicates were analyzed by ANOVA. Values with different letters are significantly different from each other (p < 0.05). Error bars show SD from three biological replicates. were germinated on medium containing 0.5 µM ABA treated with 4 h. The average and significance were calculated over three biological replicates. Total RNA was extracted from at least three batches of seeds as biological replicates. Each batch of seeds for each genotype was pooled from at least 20 independent plants. For each biological replicate, more than 250 seeds of the same batch were used for RNA extraction. AT1G13320 gene was used as control. Three biological replicates data were analyzed by ANOVA. Values with different letters are significant different from each other (P < 0.05).