Regulation of growth–defense balance by the JASMONATE ZIM‐DOMAIN (JAZ)‐MYC transcriptional module

1Plants continuously integrate information from the environment to tailor their growth, development and defensive capabilities in ways that optimize fitness. Much of this phenotypic plasticity is orchestrated by the concerted action of a small number of plant hormones (Pieterse et al., 2009; Santner et al., 2009). Among the hormones whose biosynthesis and action is exquisitely tuned by changing environmental conditions is the oxylipin jasmonate (JA) (Browse, 2009; Bhosale et al., 2013). JA controls a multitude of transcriptional programs affecting plant growth, development, and responses to biotic and abiotic stress (Howe & Jander, 2008; Baldwin & Wu, 2010; Wasternack & Hause, 2013). In the past decade, tremendous progress has been made in understanding how JA regulates gene expression and also how JA responses are integrated with other signaling pathways (Pauwels & Goossens, 2011; Kazan & Manners, 2012; Campos et al., 2014; Huot et al., 2014; Chini et al., 2016).

2When endogenous JA levels are below a threshold concentration, the expression of JA‐responsive genes is switched off through active repression of bHLH‐type MYC transcription factors (TFs) (Chini et al., 2016). This repression is mediated by JASMONATE ZIM‐DOMAIN (JAZ) proteins, which bind directly to MYCs to impede transcription by two distinct mechanisms (Chini et al., 2007; Thines et al., 2007; Yan et al., 2007). First, MYC‐bound JAZs recruit the corepressor TOPLESS (TPL) either directly (Shyu et al., 2012) or indirectly through the NOVEL INTERACTOR OF JAZ (NINJA) adaptor protein (Pauwels et al., 2010). Second, binding of the Jas motif of JAZ to the N terminus of MYC restricts access of MYC to the MED25 coactivator subunit of the mediator complex (Çevik et al., 2012; Chen et al., 2012; Zhang et al., 2015). A subset of JAZ repressors, including alternative splice variants of JAZ10, contain a cryptic MYC‐interaction domain (CMID) that tightly binds MYC and represses target gene expression (Chung & Howe, 2009; Chung et al., 2010; Moreno et al., 2013; Goossens et al., 2015; Zhang et al., 2017). Transcription of JA‐responsive genes is activated upon accumulation of jasmonoyl‐l‐isoleucine (JA‐Ile), the production of which is tightly controlled by environmental and developmental cues (Staswick & Tiryaki, 2004; Suza & Staswick, 2008; Koo & Howe, 2009). JA‐Ile promotes the formation of a nuclear co‐receptor complex consisting of the CORONATINE INSENSITIVE1 (COI1) F‐box protein and JAZ (Xie et al., 1998; Thines et al., 2007; Katsir et al., 2008; Melotto et al., 2008; Fonseca et al., 2009; Koo et al., 2009; Yan et al., 2009; Sheard et al., 2010). JA‐Ile‐dependent recruitment of JAZs to the E3 ubiquitin ligase Skp1‐Cullin‐F‐box protein (SCF)COI1 results in proteolytic destruction of JAZ repressors by the ubiquitin‐proteasome system, resulting in relief of transcriptional repression on MYC TFs (Chini et al., 2007; Thines et al., 2007).

3A major gap in our understanding of JA signaling is how receptor activation maintains spatial and temporal control over diverse transcriptional outputs. Several lines of evidence suggest that the size and complexity of the JAZ–TF interactome may account, at least in part, for the diversity of JA responses. First, higher plants produce a large repertoire of JAZ proteins encoded by a family of JAZ genes (e.g. 13 in Arabidopsis), many of which are alternatively spliced to produce multiple JAZ isoforms (Vanholme et al., 2007; Yan et al., 2007; Chung & Howe, 2009; Chung et al., 2010; Bai et al., 2011; Thireault et al., 2015). Second, JAZs repress the transcriptional activity of not only MYCs (bHLH superfamily clade IIIe) but also several other bHLH TFs, as well as transcriptional regulators belonging to other TF families (Wager & Browse, 2012; Chini et al., 2016; Goossens et al., 2016). The combinatorial complexity resulting from interaction of multiple JAZs with multiple TFs could, in theory, explain much of the specificity and diversity of JA responses. Most JAZ–TF interactions, however, have been studied with in vitro approaches that alone are insufficient to delineate the biological consequences of specific JAZ–TF interactions. Among the factors that are likely to influence the biological outputs of JAZ–TF interactions are cell‐ and tissue type‐specific expression pattern, binding affinity of JAZ for TF targets, posttranslational modification, and recruitment of additional regulatory proteins to JAZ–TF complexes. It is therefore necessary to develop experimental approaches that provide insight into the in vivo function of specific JAZ–TF modules. Recent progress in this direction has come from studies showing that JAZ2 regulates a specific MYC‐dependent transcriptional cascade to modulate stomatal dynamics during pathogen infection (Gimenez‐Ibanez et al., 2017).

4Here, we employed a genetic approach to define epistatic interactions between components of the core JA pathway and also to assess the contribution of the JAZ‐MYC signaling module to JA responses in Arabidopsis. Our experimental approach leveraged a jaz quintuple mutant (jazQ) that exhibits both enhanced responsiveness to exogenous JA and constitutive growth–defense antagonism as a consequence of mutations in JAZ1/3/4/9/10 (Campos et al., 2016). We show that phenotypes of jazQ are largely dependent on intact pathways for JA biosynthesis and perception. Detailed phenotypic comparison between jazQ and a jazQ myc2/3/4 octuple mutant demonstrated a key role for MYC TFs in restricting leaf growth concomitant with activation of leaf defense pathways, and also revealed JAZ‐repressible processes that do not require MYC TFs. Our collective data provide a genetic framework to understand how specific JAZ–TF transcriptional modules control discrete branches of the JA response.

5Arabidopsis thaliana (L.) Heynh. ecotype Columbia‐0 (Col‐0) was the wild‐type (WT) genetic background for all experiments. Construction of the quintuple jaz mutant (jazQ) and the jazQ suppressor screen has been described previously (Campos et al., 2016). suppressor of jaz quintuple10 (sjq10) and sjq66 suppressor mutants were identified by visual screening for larger rosette size from soil‐grown M2 plants. Phenotype heritability was confirmed in the M3 generation. To generate jazQ allene oxide synthase (aos) and jazQ coi1, jazQ was crossed to aos (Park et al., 2002) or to coi1‐1, respectively. glabrous1 mutations were removed from aos and coi1‐1 lines by backcrossing to Col‐0 (Yoshida et al., 2009). The myc2 myc3 myc4 triple mutant (mycT) was generated by combining the previously described mutants jin1‐7/myc2‐1 (SALK_040500) (Boter et al., 2004), myc3‐1 (GK‐445B11) (Fernandez‐Calvo et al., 2011), and myc4‐1 (GK‐491E10) (Fernandez‐Calvo et al., 2011). The myc5‐1 mutant (SALK_060048) (Figueroa & Browse, 2015; Qi et al., 2015a) was used for construction of the myc2345 quadruple mutant. Pedigrees describing the details of construction of jazQ mycT octuple and myc2345 quadruple mutants are provided in Supporting Information Fig. S1. We note that this breeding scheme was not designed specifically for the construction of jazQ mycT and myc2345. Rather, the pedigrees shown were part of a broad strategy to combine mutations affecting all signaling components of the core JA pathway and to identify, in subsequent segregating populations, specific mutant combinations. PCR‐based genotyping of mutants was performed using primer sets flanking T‐DNA insertion sites, with a third primer specific for the T‐DNA border (Table S1) (Campos et al., 2016). Seeds were stratified for 3–4 d at 4°C before germination. Plants were grown in environmentally controlled chambers with cool‐white fluorescent light for all experiments. Unless stated otherwise, conditions were 21–20°C with a 16 h: 8 h, day (100 μE m−2 s−1): night photoperiod.

6Root growth inhibition assays (Shyu et al., 2012) were performed with seedlings grown on Petri plates (Thermo Fisher Scientific, Waltham, MA, USA) containing LS medium (0.5× Linsmaier and Skoog (Caisson Labs, Smithfield, UT, USA), 0.7% w/v phytoblend agar (Caisson Labs) and 0.8% w/v sucrose) supplemented with the concentration of methyl‐JA (MeJA; Sigma‐Aldrich) indicated in the legends to Figs 1 and 2. Primary root length of WT and mutant lines (grown on the same plate) was determined in 8‐ to 11‐d‐old seedlings using imagej software (http://imagej.nih.gov/ij/). Growth parameters, including leaf dry weight, leaf area, petiole length, rosette diameter, and flowering time, were determined as described previously (Campos et al., 2016).

7JA‐induced chlorophyll degradation assays were performed as previously described (Qi et al., 2015b). Briefly, third and fourth rosette leaves were gently removed from 3‐wk‐old plants and floated on distilled water or 100 μM MeJA, and were kept in the dark at 21°C for 4 d. For chlorophyll extraction, leaves were incubated overnight in methanol in the dark. Absorbance was measured at 652, 665 and 750 nm, and the chlorophyll content was calculated as previously described (Porra et al., 1989). Total chlorophyll was normalized to either leaf area or leaf fresh weight.

8Insect feeding assays were performed as described previously (Herde et al., 2013; Campos et al., 2016). Plants were grown on soil at 20°C with an 8 h: 16 h, day (120 μE m−2 s−1): night photoperiod. To each of 12 plants (6 wk old) per genotype, four neonate Trichoplusia ni larvae (Benzon Research, Carlisle, PA, USA) were reared for 10 d, after which larval weights were measured. Pseudomonas syringae (Pst) pv. DC3000 infection assays were performed as described previously (Katagiri et al., 2002). Plants were grown on soil at 22°C with a 12 h: 12 h, light (120 μE m−2 s−1): dark photoperiod. Five‐week‐old Arabidopsis plants were dip‐inoculated with a Pst DC3000 suspension (1 × 108 colony forming units (CFUs) ml−1) containing 0.025% Silwet L77 (Lehle Seeds, Round Rock, TX, USA). Bacterial population was determined by serial dilution and plating 4 d after inoculation.

9Global gene expression profiling in 8‐d‐old whole seedlings (Col‐0 WT, jazQ, mycT and jazQ mycT) was assessed by mRNA sequencing (RNA‐seq), as described previously (Campos et al., 2016). mycT and jazQ mycT seedlings were grown and processed, and RNA‐seq analysis was performed in parallel with our previous analysis (Campos et al., 2016) of Col‐0, jazQ, phytochrome B (phyB), and jazQ phyB to facilitate cross‐comparisons; data for Col‐0 and jazQ are from Campos et al. (2016), while data for mycT and jazQ mycT are new here. Seedlings were grown in continuous light on solid medium supplemented with sucrose, and each sample was pooled from c. 200 seedlings, with three independent RNA samples (biological replicates) sequenced per genotype. Single‐end (50‐bp) sequencing was performed on the Illumina (San Diego, CA, USA) HiSeq 2000 platform at the Michigan State University Research Technologies Service Facility (https://rtsf.natsci.msu.edu). Filtered reads (Illumina quality control tools and Fastx toolkit; http://hannonlab.cshl.edu/fastx_toolkit/) were mapped to the arabidopsis information resource, genome release 10 (TAIR10) gene models with Rsem (v.1.2.11; default parameters; Li & Dewey, 2011). deseq (v.1.18.0; Anders & Huber, 2010) was used to normalize expected counts from Rsem and to assess differential gene expression relative to WT. The average transcripts per million (TPM) ± error and P‐values for all Arabidopsis genes are provided in Table S2. Gene onthology (GO) analysis of enriched functional categories was performed using David (v.6.7; Huang et al., 2009). The hypergeometric test with Benjamini & Hochberg's false discovery rate (FDR) correction was used to calculate over‐ and underrepresented GO categories among differentially expressed genes, using a P‐value < 0.05. RNA‐seq data are deposited at the National Center for Biotechnology Information Gene Expression Omnibus (GEO) as series record GSE98389.

10Plants grown in plastic containers (‘Cone‐tainers’; Steuwe & Sons, Tangent, OR, USA) with an 8 h 19°C: 16 h 16°C, light (120 μE m−2 s−1): dark period were used for gas exchange measurements (Campos et al., 2016). CO2 and light response curves were obtained from single mature rosette leaves (attached) of 8‐ to 10‐wk‐old plants on an LI‐6400XT system (Li‐Cor Biosciences, Lincoln, NE, USA) outfitted with a standard leaf chamber (chamber area = 6 cm2). Assimilation rates were normalized to projected leaf area as measured by image analysis with the Gimp software (www.gimp.org).

11Allocation of metabolic resources to the production of plant defense compounds is often associated with reduced growth and biomass accumulation. These apparent growth–defense tradeoffs are of considerable interest for understanding plant form and function in both natural and agricultural ecosystems (Havko et al., 2016; Karasov et al., 2017; Züst & Agrawal, 2017). In our previous studies, we employed a JAZ‐depleted quintuple mutant (jazQ) to better understand how JA signaling balances growth and defense, and also to identify suppressor mutations that mitigate the antagonistic relationship between leaf growth and anti‐insect defense (Campos et al., 2016). This work showed that phyB mutations resulting in loss of the photoreceptor (phyB) rescue the reduced growth of jazQ rosette leaves but do not compromise the enhanced leaf defense traits of jazQ. Thus, the apparent tradeoff in jazQ between leaf biomass and defense can be effectively uncoupled through loss of phyB activity. The jazQ suppressor screen also identified a distinct class of mutants in which recovery of leaf growth was associated with visible loss of anthocyanin pigment in leaves. Two such sjq lines, sjq10 and sjq66, exhibited features of male sterility that are associated with impaired JA biosynthesis or signaling (Fig. S2) (Thines et al., 2013). DNA sequencing of candidate genes in the JA pathway identified a C→T non‐sense mutation in codon 56 of the AOS gene in sjq10 and a C→T missense mutation in codon 86 of the COI1 gene in sjq66 (Fig. S2). These results suggested that constitutive growth–defense antagonism (i.e. reduced growth and enhanced defense of shoots) resulting from the loss of JAZ1/3/4/9/10 in jazQ depends on intact pathways for JA biosynthesis and signaling.

12To further investigate the relationship between jazQ and other components of the JA pathway (Fig. 1a), we reconstructed sextuple mutant lines in which jazQ was combined with mutant alleles of aos or coi1. Root growth inhibition assays showed that jazQ coi1 plants resemble coi1 single mutants in being strongly insensitive to JA treatment (Fig. 1b). We also found that coi1 is epistatic to jazQ with respect to male sterility. Whereas jazQ flowers developed normally, jazQ coi1 flowers had short anther filaments and lacked viable seed production (Fig. 1c,d). jazQ coi1 flowers were also indehiscent at the time when stigmatic papillae are receptive to pollen for fertilization. Reconstructed jazQ aos lines maintained the JA hypersensitivity of jazQ plants and were also male sterile (Fig. 1e,f). These findings demonstrate that coi1 and aos are epistatic to jazQ with respect to leaf growth and fertility traits, and that coi1 abolishes the responsiveness of jazQ roots to exogenous JA.

13jazQ roots are not only hypersensitive to exogenous JA but also are shorter than WT roots in the absence of JA (Fig. 1b; Campos et al., 2016). To determine whether this constitutive short‐root phenotype results from increased sensitivity to endogenous JA, we compared the root length of jazQ seedlings grown in the absence of applied JA to that of jazQ coi1 and jazQ aos plants. In contrast to our expectation, neither coi1 nor aos rescued the short root of jazQ (Fig. 1b,e). This result indicated that constitutive root shortening in jazQ does not depend on JA signaling, and raised the possibility that one or more of the JAZs (JAZ1/3/4/9/10) affected by jazQ positively regulate root growth when JA concentrations are low. Comparison of jaz single mutants grown side by side on medium lacking JA showed that jaz3 roots were significantly shorter than those of other mutants (Fig. 1g), suggesting that JAZ3 promotes root growth under these conditions. Only jaz10‐1 roots were shorter than WT roots on medium containing JA, as shown in previous studies (Demianski et al., 2012; Moreno et al., 2013).

14Among the diverse TFs that physically interact with JAZs are MYC2 and the closely related proteins MYC3 and MYC4, which perform prominent roles in JA signaling (Dombrecht et al., 2007; Fernandez‐Calvo et al., 2011; Schweizer et al., 2013). MYC5, which is also a member of the group IIIe subfamily of bHLH TFs to which MYC2/3/4 belong, plays a role in JA‐mediated stamen development (Figueroa & Browse, 2015; Qi et al., 2015a). We assessed whether MYC5 might contribute to JA signaling in nonfloral tissues by comparing responses of vegetative organs of a myc2 myc3 myc4 triple mutant (referred to hereafter as mycT) to those of a myc2 myc3 myc4 myc5 quadruple mutant (myc2345). No differences in root growth, leaf growth, or leaf anthocyanin concentrations were observed between mycT and myc2345 (Fig. S3). Subsequent studies were thus focused on investigating the genetic interaction between jazQ and mycT.

15To investigate the collective role of MYC2/3/4 in jazQ phenotypes, we compared JA responses in jazQ to those in mycT and the corresponding jazQ mycT octuple mutant. Root growth assays performed on JA‐free medium showed that mycT and WT roots were similar in length, whereas jazQ mycT plants displayed the constitutive short‐root trait of jazQ (Fig. 2a). Assays performed on JA‐supplemented media showed that mycT roots were partially insensitive to JA (Fig. 2a), consistent with previous studies (Lorenzo et al., 2004; Fernandez‐Calvo et al., 2011; Gasperini et al., 2015). Moreover, the response of jazQ mycT roots to JA was indistinguishable from that of mycT plants. These data indicate that the hypersensitivity of jazQ roots to JA is mediated by MYC2/3/4 but that the constitutive short‐root phenotype of jazQ does not require these TFs.

16We used a dark‐ and JA‐induced chlorophyll degradation assay to investigate how jazQ and mycT interact to mediate shoot responses to exogenous JA. Control experiments showed comparable levels of chlorophyll in WT, jazQ, mycT and jazQ mycT plants, in both intact rosette leaves and in detached leaves incubated in the dark in the absence of exogenous JA (Fig. S4; Table 1). Exogenous JA promoted chlorophyll degradation in detached WT leaves incubated in the dark (Fig. 2b,c), as previously reported (Zhu et al., 2015). JA‐induced leaf degreening was modestly but reproducibly exacerbated in jazQ leaves relative to Col‐0, and this effect was abolished in both mycT and jazQ mycT lines. These results show that the responsiveness of jazQ leaves to exogenous JA is dependent on MYC2/3/4.

17To determine whether MYC2/3/4 activity plays a role in JA‐mediated restriction of shoot growth, we compared leaf growth traits of 21‐d‐old plants grown in soil under our standard conditions. Consistent with previous studies (Campos et al., 2016), the shoot biomass, projected leaf area, petiole length, and leaf number were all decreased in jazQ relative to WT (Fig. 3). Loss of MYC2/3/4 in mycT plants had the opposite effect on shoot growth, such that leaf area, biomass and petiole length of mycT plants were greater than those of WT. Significantly, loss of MYCs in the jazQ genetic background completely recovered the restricted growth of jazQ leaves, and the area and biomass of jazQ mycT leaves also exceeded those of WT (Fig. 3b–e). These data demonstrate that JAZ1/3/4/9/10 and MYC2/3/4 act as positive and negative regulators, respectively, of leaf growth and biomass accumulation.

18We next determined whether MYC TFs are required for enhanced resistance of jazQ plants to insect herbivory (Campos et al., 2016). Feeding assays performed with Trichoplusia ni (cabbage looper) showed that, as expected, larval growth on jazQ plants was reduced relative to growth of WT‐reared insects. By contrast, the mass of larvae reared on either mycT or jazQ mycT plants exceeded that on WT (Fig. 4a,b). These differences in caterpillar performance were generally reflected by the amount of leaf tissue consumed during the course of the bioassay (Fig. 4c). We conclude that enhanced defense of jazQ to T. ni feeding is dependent on MYC2/3/4 TFs. The finding that the weight gain of caterpillars grown on mycT slightly exceeded that of larvae reared on jazQ mycT plants (Fig. 4b) suggests that other regulatory factors may contribute to jazQ‐mediated anti‐insect defense. We also tested the effect of mycT on leaf anthocyanin concentrations, which are elevated in jazQ (Campos et al., 2016). Anthocyanin concentrations in mycT leaves were modestly but consistently lower than those in WT. Leaf anthocyanin concentrations in jazQ mycT were intermediate between those of jazQ and WT (Table 1). These results indicate that, although MYC2/3/4 positively regulate anthocyanin production, these TFs are not sufficient to account for all anthocyanin accumulation in jazQ.

19The bacterial pathogen P. syringae pv. tomato DC3000 (Pst DC3000) exploits the host JA signaling pathway to promote virulence. We therefore sought to determine whether the constitutive activation of JA signaling by jazQ is sufficient to alter host susceptibility to this pathogen, which uses the JA‐Ile mimic coronatine as part of its virulence strategy to suppress host defenses (Melotto et al., 2006). Indeed, bacterial infection assays showed that Pst DC3000 multiplies to a higher level on jazQ than on WT leaves (Fig. 5a,b). The enhanced susceptibility of jazQ was particularly evident from strong disease symptoms in young emerging leaves, which in WT plants typically show few disease symptoms after Pst DC3000 infection (Fig. 5c). Consistent with previous reports (Laurie‐Berry et al., 2006; Fernandez‐Calvo et al., 2011), mycT plants were more resistant than WT to Pst DC3000 infection. Moreover, we found that jazQ mycT plants resisted Pst DC3000 infection to a similar level to mycT (Fig. 5a). These results show that MYCs TFs are required for the increased susceptibility of jazQ to Pst DC3000 infection, and are consistent with the idea that this pathogen activates the JA pathway as part of a virulence strategy to suppress salicylate‐based immunity (Zhao et al., 2003; Katsir et al., 2008; Fernandez‐Calvo et al., 2011; Demianski et al., 2012; Zheng et al., 2012).

20Several studies have reported that JA signaling through the COI1‐JAZ pathway delays the onset of flowering in Arabidopsis (Robson et al., 2010; Yang et al., 2012; Song et al., 2013; Zhai et al., 2015). Consistent with these observations, we found that jazQ plants are developmentally delayed in flowering but have the same number of leaves as WT at the time of bolting (Fig. 6; Campos et al., 2016). Under our standard long‐day growth conditions, mycT alone did not have an obvious effect on flowering time. Unexpectedly, however, the combination of jazQ and mycT retarded flowering time even later than jazQ, and also increased the number of leaves at the time of bolting (Fig. 6). These data indicate that MYC2/3/4 do not mediate the delayed flowering of jazQ.

21To better understand the contribution of MYC TFs to changes in gene expression resulting from loss of JAZ1/3/4/9/10, we used mRNA sequencing (RNA‐seq) to compare transcript profiles of WT, jazQ, mycT, and jazQ mycT seedlings grown in the absence of exogenous JA (Table S2). We used stringent statistical criteria to define a set 99 JAZ1/3/4/9/10‐repressible, MYC2/3/4‐inducible transcripts whose abundance relative to Col‐0 is higher in jazQ but not in jazQ mycT (Figs 7a, S5). Likewise, 159 MYC‐independent genes were identified as being up‐regulated in both jazQ and jazQ mycT (Figs 7a, S5). Based on this analysis, we estimate that MYC2/3/4 activity is required for the increased expression of c. 38% of all genes that are up‐regulated in jazQ seedlings. Among the 258 genes that were up‐regulated in jazQ relative to Col‐0, the MYC‐dependent gene set was associated with gene ontologies for JA biosynthesis and glucosinolate (GLS) metabolism, and also included known wound‐response genes such as vegetative storage protein 2 (VSP2) and tyrosine aminotransferase 1 (TAT1) (Fig. 7b,c). That several of these MYC‐dependent genes, most notably genes involved in GLS biosynthesis, were strongly repressed by mycT relative to Col‐0 (Figs 7c, S6) is consistent with a role for MYC TFs in maintaining basal expression of JA‐responsive genes.

22Among the genes that were up‐regulated most robustly in jazQ (relative to WT) and independently of MYC2/3/4 were those associated with triterpenoid biosynthesis and GLS catabolism (Fig. 7b). Marneral and thalianol triterpenoids are synthesized in the Arabidopsis root epidermis by enzymes encoded within metabolic gene clusters (Field & Osbourn, 2008). Strikingly, transcript levels for all genes within these clusters were similarly elevated in both jazQ and jazQ mycT, and mycT alone did not affect the basal expression level observed in WT (Fig. 7d). We also found that the expression of genes immediately flanking the marneral and thalianol gene clusters was not altered by jazQ or mycT. Thus, the effect of jazQ in increasing the expression of triterpenoid biosynthetic genes is not only independent of MYC2/3/4, but is also spatially restricted to genes within the clusters.

23In intact plant cells, GLSs are stored as inert glycosides physically separated from GLS catabolic enzymes (Halkier & Gershenzon, 2006). Tissue disruption by herbivores allows myrosinases to deglycosylate GLSs, whereas ancillary specifier proteins catalyze the formation of cyanate and nitrile toxins from the corresponding aglycones (Fig. S7a). We found that genes encoding enzymes involved in GLS breakdown were significantly up‐regulated in both jazQ and jazQ mycT (Fig. S7b,c). This expression pattern was highly distinct from that of MYC‐dependent GLS biosynthetic genes (Figs S6, S7), suggesting that specific JAZ–TF modules control different aspects of GLS metabolism. Further analysis of the broader set of 159 genes that are up‐regulated in jazQ independently of MYC2/3/4 (Fig. 7a) revealed a substantial overlap with genes that are up‐regulated in roots of the ninja‐1 mutant (Fig. S5c) (Gasperini et al., 2015). This overlap was significantly greater than the portion of MYC‐dependent genes that overlapped with ninja‐1 (P = 0.0013; Fisher's exact test), suggesting that NINJA cooperates with JAZ1/3/4/9/10 to repress the expression MYC‐independent genes.

24Among the 88 genes expressed to higher levels in mycT than Col‐0, there was little overlap with genes misregulated by either jazQ or jazQ mycT (Fig. S5a). We found that the majority of genes (37/72 genes; 51.4%) uniquely up‐regulated in mycT are associated with chloroplast processes and, in particular, photosynthesis (Fig. S8). Gas exchange experiments showed that mycT leaves had higher CO2 assimilation rate per unit leaf area than Col‐0 leaves under our plant growth conditions (Fig. S9). Photosynthetic rates measured in jazQ and jazQ mycT leaves were comparable to those in WT (Fig. S9). These findings suggest a potential role for the MYC2/3/4 TFs as negative regulators of photosynthesis, and further highlight the complex relationship between JA signaling and photosynthetic capacity (Campos et al., 2016).

25Genetic epistasis provides a powerful approach to dissect complex interactions between core components of the JA pathway, as exemplified by recent studies of root responses to JA (Acosta et al., 2013; Gasperini et al., 2015). We previously employed a genetic suppressor screen to identify mutations that uncouple growth–defense antagonism in leaves of the jazQ mutant (Campos et al., 2016). Here, we report on a new class of jazQ suppressor mutants in which defects in either JA‐Ile biosynthesis (aos) or perception (coi1) suppress both the slow growth and enhanced defense traits of jazQ leaves. Given that JA‐Ile biosynthesis and perception are both required for turnover of JAZ repressors, the most likely mechanistic explanation for the observed jazQ aos and jazQ coi1 phenotypes is that one or more of the remaining JAZs in jazQ are stabilized by the loss of the JAZ degradation machinery. This model implies that the remaining complement of JAZs in jazQ can strongly repress JA responses in the absence of JAZ1/3/4/9/10, thus providing evidence for functional redundancy among JAZs. The genetic interactions defined in our study are generally consistent with current models of JA‐mediated signal transduction in which JA‐Ile and COI1 work together to degrade JAZs, thereby relieving repression on target TFs (Fig. 1a).

26A unique attribute of jazQ in comparison to jaz single mutants is its enhanced sensitivity to exogenous JA. One explanation for this phenotype is that genetic depletion of JAZs increases the capacity of SCFCOI1 to ubiquitylate the remaining pool of JAZ substrates in jazQ, which is consistent with evidence that COI1 dosage modulates sensitivity to JA (Feng et al., 2003). Alternatively, the enhanced responsiveness of jazQ to JA treatment may reflect the loss of JAZ proteins that desensitize JA responses once the signaling pathway is initiated (Campos et al., 2014). JAZ1 and JAZ10 probably contribute to this role. First, JAZ10 alternative splice variants are resistant to JA‐induced degradation, and loss of these isoforms is associated with increased sensitivity to JA (Chung & Howe, 2009; Moreno et al., 2013). Second, the N termini of both JAZ10 and JAZ1 contain a cryptic MYC‐interaction domain (CMID) that may enhance the capacity of these proteins to repress MYCs in JA‐elicited cells (Moreno et al., 2013; Goossens et al., 2015; Zhang et al., 2017). The up‐regulation of JA biosynthetic genes in jazQ raises the additional possibility that changes in endogenous JA/JA‐Ile concentrations contribute to the phenotypes observed in this line. Regardless of the mechanisms that confer enhanced sensitivity of jazQ to JA, it is evident that jazQ mutant phenotypes are relatively subtle in comparison to WT plants subject to chronic JA exposure. This observation supports the notion of genetic redundancy among JAZ genes and provides a strong rationale for further analysis of higher order jaz mutants.

27In addition to JA hypersensitivity, jazQ roots are c. 25% shorter than those of WT in the absence of exogenous JA (Campos et al., 2016; this work). The unexpected finding that constitutive root shortening also occurs in jazQ aos, jazQ coi1, and jazQ mycT mutants indicates that this phenotype is not dependent on core components of the JA pathway. Analysis of jaz single mutants further suggested that JAZ3 may serve a role in promoting root growth in the absence of JA, but additional studies are needed to test this hypothesis. We are not aware of other studies suggesting JA‐independent roles for JAZ proteins. However, it was recently reported that overexpression of Arabidopsis TIFY8, a ZIM domain‐containing protein belonging to the TIFY family to which JAZs belong, stunts primary root growth in the absence of exogenous JA (Perez et al., 2014). TIFY8 also represses transcriptional activity through recruitment of the NINJA−TPL corepressor complex, but does not affect the expression of JA‐responsive genes. Acosta et al. (2013) found that null mutations in NINJA led to JA‐independent reduction in root cell elongation to generate a short‐root phenotype similar to that of jazQ. Given that JAZs and NINJA physically interact and negatively regulate JA signaling in roots (Pauwels et al., 2010; Acosta et al., 2013), it is possible that JAZ−NINJA complexes have a role in promoting root growth under specific conditions or in specific cell types. Unlike jazQ, ninja null mutations do not confer hypersensitivity to exogenous JA (Acosta et al., 2013), presumably because loss of NINJA does not impede JA‐Ile‐dependent JAZ degradation via SCFCOI1 and the 26S proteasome. In future studies it will be informative to determine whether jazQ and ninja mutations mediate constitutive root shortening through the same or parallel pathways.

28A current challenge in JA research is to understand how the hormone controls diverse aspects of plant growth, development, and responses to the environment. As summarized in Fig. 8, our genetic data suggest that JAZ‐mediated control over MYC2/3/4 TFs plays a major role in executing the majority of JA‐related phenotypes in jazQ. We found that MYCs are largely responsible for root hypersensitivity to JA, reduced leaf biomass, leaf defense against insect herbivory, and JA‐induced chlorophyll degradation. The latter observation supports previous studies showing that exogenous JA and leaf damage reduce the abundance of photosynthetic proteins in Arabidopsis leaves, perhaps as a strategy to mobilize resources for defense (Gfeller et al., 2011; Shan et al., 2011; Zhu et al., 2015). Consistent with previous work showing that MYCs are critical for the effects of coronatine on virulence of Pst DC3000 (Fernandez‐Calvo et al., 2011; Zheng et al., 2012; Schweizer et al., 2013; Gimenez‐Ibanez et al., 2017), we also found that jazQ‐mediated enhanced susceptibility to this pathogen is eliminated by mycT. These collective data support a dominant role for MYC TFs in mediating JA responses in Arabidopsis and further demonstrate that MYC activity can be enhanced through loss of a specific subset of MYC‐interacting JAZs in vivo.

29Our findings support an increasing number of studies showing that JAZs functionally interact with TFs other than MYCs (Fig. 8) (Wager & Browse, 2012; Chini et al., 2016; Goossens et al., 2016). One example is jazQ‐mediated anthocyanin accumulation, which was reduced but not eliminated by mycT. The role of MYC TFs as positive regulators of anthocyanin biosynthesis is well established (Niu et al., 2011; Nakata et al., 2013). Other JAZ‐interacting TFs, however, including YABBY, MYB, and transparent testa 8/glabrous 3, also contribute to JA‐inducible anthocyanin accumulation (Shan et al., 2009; Qi et al., 2011; Boter et al., 2015). It thus seems likely that these non‐MYC TFs contribute to the anthocyanin accumulation observed in jazQ mycT plants (Fig. 8).

30We also demonstrate that MYC2/3/4 activity is not required for delayed flowering of jazQ. This finding agrees with recent models in which a subset of JAZ proteins (i.e. JAZ1/3/4/9) repress TARGET OF EAT1 (TOE1) and TOE2 TFs, which delay flowering time by repressing the expression of FLOWERING LOCUS T (FT) (Zhai et al., 2015). Although additional work is needed to understand how mycT further delays the flowering time of jazQ, inspection of RNA‐seq data (Campos et al., 2016; this work) showed a strong correlation between FT transcript levels and flowering time in various mutant lines. For example, FT mRNA levels were high in the early‐flowering phyB and jazQ phyB lines, whereas FT transcript levels were strongly reduced (relative to Col‐0) in jazQ and even lower in jazQ mycT (Fig. S10).

31Although JA has long been recognized as a potent elicitor of specialized metabolism (Gundlach et al., 1992), the underlying transcriptional mechanisms that control these metabolic pathways in specific cell types and morphological structures are only beginning to be elucidated (De Geyter et al., 2012). Consistent with previous studies (Schweizer et al., 2013; Campos et al., 2016), our results establish a direct link between JAZ1/3/4/9/10 and their interacting MYC TFs in controlling GLS production. Interestingly, however, we also found that genes encoding myrosinases and associated specifier proteins involved in GLS breakdown are up‐regulated in jazQ independently of MYC2/3/4, suggesting that additional JAZ‐interacting TFs govern distinct aspects GLS metabolism (Fig. 8). This interpretation is consistent with recent work showing that GLS accumulation is dependent on a complex transcriptional network in which many TFs optimize metabolite accumulation across multiple tissue types and environmental conditions, thus allowing for tight yet flexible control of the ‘mustard oil bomb’ (Li et al., 2014).

32Our results also provide new insight into processes underlying the expression of metabolic gene clusters for triterpenoid biosynthesis. We found that genes within the thalianol and marnerol clusters are coordinately up‐regulated in jazQ seedlings, which is consistent with the ability of JA to elicit triterpenoid production (Hayashi et al., 2003). MYC2‐like TFs are implicated in the control of metabolic gene clusters in several plant species (Shang et al., 2014; Cardenas et al., 2016). However, our finding that mycT does not suppress the elevated expression of the thalianol and marnerol clusters in jazQ argues against a role for MYC2/3/4 in controlling these genes. The observation that genes within these clusters are strongly up‐regulated in roots of ninja mutants (Gasperini et al., 2015) further supports the notion that recruitment of co‐repressors by JAZ‐NINJA complexes (Pauwels et al., 2010) may negatively regulate triterpenoid production. Indeed, recent studies show that chromatin modifications within the thalianol and marnerol clusters are associated with silencing and activation of individual cluster genes (Nützmann et al., 2016; Yu et al., 2016). Mutants affected in NINJA and JAZ function may provide useful tools to delineate the contribution of core JA signaling components to chromatin signatures and TF modules that coordinate the expression of metabolic gene clusters.

33JA is a potent inhibitor of leaf growth and biomass accumulation in Arabidopsis (Yan et al., 2007; Zhang & Turner, 2008; Noir et al., 2013; Attaran et al., 2014). Similar effects are observed in monocots (Yang et al., 2012; Hibara et al., 2016), suggesting that the underlying pathways for JA‐mediated restriction of leaf growth are conserved. JA acts primarily to reduce leaf cell number through perturbation of the cell cycle but effects on leaf cell expansion have also been documented (Zhang & Turner, 2008; Noir et al., 2013; Havko et al., 2016). A role for the JAZ‐MYC module in mediating these effects is supported by the observation that myc2 mutation, as well as overexpression of the JAZ10.3 alternative splice variant, partially inhibits JA‐ and wound‐induced growth stunting (Yan et al., 2007; Zhang & Turner, 2008). Our data showing that mycT completely rescues the reduced size and biomass of jazQ are consistent with this view, as is the finding that mycT leaves are larger than Col‐0 leaves under our growth conditions. These collective observations highlight a key role for MYC2/3/4 TFs as negative regulators of leaf growth and biomass accumulation.

34The mechanism by which MYC TFs restrict leaf growth remains unknown but several hypotheses can be considered. First, the ability of MYCs to repress the expression of photosynthesis‐associated genes (Yadav et al., 2005; this work) suggests that JA‐induced reduction in photosynthetic efficiency may be linked to reduced leaf growth. The photosynthetic robustness of JA‐elicited Arabidopsis leaves, however, does not support this hypothesis (Attaran et al., 2014; Campos et al., 2016). Second, MYCs may directly repress the activity of positive regulators of leaf cell division or expansion (Pauwels et al., 2008; Zhang & Turner, 2008; Noir et al., 2013). Such a mechanism would be analogous to the role of MYC2 in repressing the activity of PLETHORA TFs, which promote auxin‐dependent control of cell proliferation specifically in the root stem cell niche (Chen et al., 2011). A third possibility is that JA‐induced activation of MYC activity increases the production of GLSs and other defensive compounds whose biosynthesis limits resource allocation to growth (Paul‐Victor et al., 2010). Recent studies, however, show that GLS‐based leaf defenses can be expressed in the absence of a growth penalty, indicating that growth–defense antagonism in this and perhaps other genotypes cannot simply be explained by allocation costs (Campos et al., 2016; Kliebenstein, 2016; Züst & Agrawal, 2017). In this context, uncoupling of growth–defense tradeoffs in jazQ phyB implies that rewiring of phyB−JA crosstalk can override MYC‐mediated growth restriction while leaving MYC‐mediated defenses intact (Campos et al., 2016). Although DELLA proteins have been implicated in JA‐mediated growth inhibition of Arabidopsis roots (Hou et al., 2010) and hypocotyls (Yang et al., 2012), DELLAs are not required for wound‐ and JA‐induced growth stunting of leaves (Zhang & Turner, 2008). Nevertheless, we cannot exclude the possibility that changes in MYC stability (Chico et al., 2014), DELLA protein abundance (Leone et al., 2014), or direct interaction between DELLA and MYC TFs (Hong et al., 2012) contributes to the mechanism by which leaf growth is attenuated by MYC activity. A better understanding of how plants balance leaf growth and defense traits may benefit from research to determine how JA signaling is integrated into regulatory networks that control leaf cell division and expansion (Chitwood & Sinha, 2016; Nelissen et al., 2016; Mao et al., 2017).

35I.T.M., Y.Y. and G.A.H. designed the research plans; I.T.M., Y.Y., M.L.C., G.K., D.d.O.F., K.S., and X‐F.X. performed the experiments; I.T.M., Y.Y., G.K., K.S., X‐F.X., S.Y.H., and G.A.H. analyzed the data; I.T.M. and G.A.H. wrote the manuscript.