Two novel RING-type ubiquitin ligases, RGLG3 and RGLG4, are essential for jasmonate-mediated responses in Arabidopsis.

Jasmonates (JAs) regulate various stress responses and development processes in plants, and the JA pathway is tightly controlled. In this study, we report the functional characterization of two novel RING-type ubiquitin ligases, RING DOMAIN LIGASE3 (RGLG3) and RGLG4, in modulating JA signaling. Both RGLG3 and RGLG4 possessed ubiquitin ligase activities and were widely distributed in Arabidopsis (Arabidopsis thaliana) tissues. Altered expression of RGLG3 and RGLG4 affected methyl JA-inhibited root growth and JA-inductive gene expression, which could be suppressed by the coronatine insensitive1 (coi1) mutant. rglg3 rglg4 also attenuated the inhibitory effect of JA-isoleucine-mimicking coronatine on root elongation, and consistently, rglg3 rglg4 was resistant to the coronatine-secreting pathogen Pseudomonas syringae pv tomato DC3000, suggesting that RGLG3 and RGLG4 acted in response to the coronatine and promoted JA-mediated pathogen susceptibility. In addition, rglg3 rglg4 repressed wound-stunted plant growth, wound-stimulated expression of JA-responsive genes, and wound-induced JA biosynthesis, indicating their roles in JA-dependent wound response. Furthermore, both RGLG3 and RGLG4 responded to methyl JA, P. syringae pv tomato DC3000, and wounding in a COI1-dependent manner. Taken together, these results indicate that the ubiquitin ligases RGLG3 and RGLG4 are essential upstream modulators of JA signaling in response to various stimuli.

In the absence of a stimulus, the JA signaling pathway is generally repressed by a family of jasmonate ZIM-domain (JAZ) proteins (Chini et al., 2007;Thines et al., 2007), which recruit the corepressor TOPLESS through the linker NOVEL INTERACTOR OF JAZ (Pauwels et al., 2010) and associate with numerous transcription factors to inhibit their effect on downstream JA signaling (Kazan and Manners, 2012). Once activated by environmental or developmental signals, jasmonic acid is rapidly synthesized and can be further converted into numerous conjugates, including the highly bioactive (+)-7-iso-jasmonoyl-L-isoleucine (JA-Ile; Staswick and Tiryaki, 2004;Thines et al., 2007;Fonseca et al., 2009) that is perceived by a receptor complex consisting of CORONATINE INSENSITIVE1 (COI1), JAZs, and inositol pentakisphosphate (Sheard et al., 2010). COI1 is an F-box protein that associates with ASK1/ASK2, AtCUL1, and AtRBX1 to form a large SKP/CUL/F-box complex (Devoto et al., 2002;Xu et al., 2002), a type of ubiquitin ligase that targets JAZs to be degraded by 26S proteasome upon JA perception, thus releasing their inhibitory effect on JA signaling (Chini et al., 2007;Thines et al., 2007). Downstream JA responses are regulated by an important JAZ-suppressed transcription factor, JASMONATE-INSENSITIVE1 (JIN1)/MYC2, which controls transcriptional reprogramming of a large group of JA-responsive genes (Boter et al., 2004;Lorenzo et al., 2004;Dombrecht et al., 2007). Recent reports have identified other JAZ-interacting transcription factors that cooperate to determine the specificity of JA-mediated responses (Fernández-Calvo et al., 2011;Qi et al., 2011;Song et al., 2011).
Accumulated data indicate that the JA pathway does not operate alone but rather is integrated into a complex signaling network depending on the response. Protein phosphorylation (Takahashi et al., 2007) and signal transmitters such as calcium (León et al., 1998) and nitric oxide (Huang et al., 2004) are currently known to have important roles in mediating signals in the JA pathway. Moreover, JA signaling is subjected to modification by many other phytohormones in different responses. For instance, salicylic acid antagonizes JA signaling in immune responses according to the type of invading pathogen, whereas ethylene can strengthen JA-mediated defense (Pieterse et al., 2009). Abscisic acid can also regulate defense responses and metabolic reprogramming by modulating JA signaling (Adie et al., 2007;Lackman et al., 2011). GA affects JA signaling in the defense response (Navarro et al., 2008) and stamen development (Cheng et al., 2009). DELLA proteins, essential negative regulators of GA signaling (Fleet and Sun, 2005), and JAZs interfere mutually in regulating growth and the defense response (Hou et al., 2010;Yang et al., 2012).
More research is needed to decipher the underlying molecular mechanisms coordinating JA pathwaymediated responses and examine the regulatory role of protein ubiquitination. Ubiquitination is a fundamental posttranslational modification in plants (Stone and Callis, 2007;Trujillo and Shirasu, 2010). Covalent ubiquitin attachment is completed by sequential actions of ubiquitin-activating enzyme (E1), ubiquitinconjugating enzyme (E2), and ubiquitin ligase (E3; Vierstra, 2009). Ubiquitin chains linked through Lys-48 of ubiquitin target substrates for degradation by the 26S proteasome, whereas polyubiquitin chains with alternative linkages (such as Lys-63) perform signaling functions (Bhoj and Chen, 2009). Much research has focused on E3 because of its key role in determining substrate specificity. Plants have a large number of E3s, but only a small fraction of them have been functionally defined (Santner and Estelle, 2010). In the JA pathway, ubiquitin-mediated regulation is well exemplified by COI-mediated JA perception (Chini et al., 2007;Thines et al., 2007). However, for about 10 years after COI1 was demonstrated to be an E3 (Devoto et al., 2002;Xu et al., 2002), no other ubiquitin ligase has been implicated for a substantive role in JA signaling, in contrast to the growing number of E3s described in other hormone pathways (Santner and Estelle, 2010). Therefore, identification of novel E3-mediated protein turnover or signaling events in the JA pathway should further our understanding of JA signaling in plants.
Arabidopsis (Arabidopsis thaliana) has five RING DOMAIN LIGASEs (RGLGs; Yin et al., 2007) that are in a protein family structurally characterized with a large von Willebrand A (VWA) domain at the N terminus and a RING domain at the C terminus (Stone et al., 2005). RGLG1 and RGLG2 have ubiquitin ligase activities and participate in auxin-regulated apical dominance and drought stress response (Yin et al., 2007;Cheng et al., 2012), but the functions of other RGLG members remain unknown. To address this issue, here we characterized the essential roles of RGLG3 and RGLG4 in controlling multiple biological processes by modulating the JA signaling pathway.

RGLG3 and RGLG4 Possess Ubiquitin Ligase Activities
Sequence alignment showed that the five Arabidopsis RGLGs have highly similar protein sequences, with noteworthy differences in the N-terminal peptide upstream of the VWA domain (Supplemental Fig. S1). Three members have Gly as the second amino acid residue, which is a myristoylation site (Farazi et al., 2001) that is important for protein subcellular localization (Yin et al., 2007). In contrast, RGLG4 has a very short N-terminal peptide and a Thr instead of the Gly at residue 2, and RGLG3 lacks the N-terminal peptide. Phylogenetic analysis also indicated that RGLG3 is evolutionarily closer to RGLG4 (Fig. 1A) than to other RGLGs. Hence, RGLG3 and RGLG4 were further studied to elucidate their biological functions in Arabidopsis.
Not every RING domain-containing protein functions as a ubiquitin ligase (Deshaies and Joazeiro, 2009). Hence, whether RGLG3 and RGLG4 act as E3s was determined. First, conventional E3 activities of RGLG3 and RGLG4 were characterized in the self-ubiquitination system. Ubiquitinated protein bands became intensified over time for both RGLG3 and RGLG4 (Fig. 1, B and C). These bands were not detected in the absence of any of the essential reaction components or if the His (critical for RING domain integrity) was mutated into Tyr (Fig. 1,B and D), indicating the specificity of their ubiquitin ligase activities. These results demonstrated that both RGLG3 and RGLG4 possess ubiquitin ligase activities. To date, one plant ubiquitin ligase (RGLG2) has been identified that catalyzes formation of the ubiquitin chain linked through Lys-63 in cooperation with the cognate E2, UBC35 (Yin et al., 2007). This study determined if RGLG3 and RGLG4 possessed similar activities. As RGLG2 was reported to interact with UBC35 by its C-terminal peptide containing the RING domain (Yin et al., 2007), we tested if RGLG3, RGLG3R, RGLG4, or RGLG4R (Fig. 1B) also interacted with UBC35 or its homolog UBC36 (Wen et al., 2006) using the yeast two-hybrid system. Results showed that no interaction actually existed in yeast (Supplemental Fig. S2), suggesting that RGLG3 and RGLG4 have different enzyme activities than RGLG2.

RGLG3 and RGLG4 Are Widely Distributed in Arabidopsis Tissues
Next, the gene expression of RGLG3 and RGLG4 was characterized. Reverse transcription (RT)-PCR analysis detected the expression of RGLG3 and RGLG4 mRNAs in various tissues, with relatively higher abundance in seedling roots and adult leaves (Supplemental Fig. S3). Visual confirmation of their promoter activities using 1.9-kb (RGLG3) or 1.8-kb (RGLG4) promoters to drive GUS reporter gene expression in Arabidopsis showed relatively higher intensity GUS staining in roots and leaves (Fig. 2), thus confirming the RT-PCR results. GUS staining also revealed apparent differences: RGLG3 and RGLG4 had similar GUS expression in the root tips but different expression in more mature root regions (Fig. 2, A-D). GUS staining was similar in cotyledons until 7 d of growth (Fig. 2, A-D) and was more prominent in rosette leaves, where RGLG3 was mainly expressed in the leaf veins, whereas RGLG4 had diffuse expression throughout leaves (Fig. 2, E and F). RGLG3 but not RGLG4 displayed promoter activity in the stigma, receptacle, and siliques (Fig. 2, G-J). These assays indicated that RGLG3 and RGLG4 promoter activities were widely distributed in Arabidopsis, with both similarities and distinctions.

RGLG3 and RGLG4 Modulate JA-Inhibited Root Elongation
The responses of RGLG3 and RGLG4 to different stimuli were examined to determine their functions in Arabidopsis. RGLG3 was induced by exogenously applied methyl jasmonate (MeJA), whereas RGLG4 was suppressed at 4 h after MeJA treatment (Fig. 3A). Indeed, JA-up-regulated expression of RGLG3 has been described in a gene chip data set (Pauwels et al., 2010).
To determine if RGLG3 and RGLG4 were potentially involved in the JA response, two homologous T-DNA insertion mutants, SALK_098983 for rglg3 and SALK_ 096022C for rglg4, were isolated and then crossed to obtain the double mutant rglg3 rglg4 (Supplemental Fig. S4, A and B). RGLG3 and RGLG4 mRNAs were not detected by RT-PCR, and quantitative real-time PCR confirmed a greater than 20-fold reduction in their expression (Supplemental Fig. S4, C and D), The reaction included ATP, Flagtagged ubiquitin, yeast ubiquitinactivating enzyme (E1), GST-tagged UbcH5b (ubiquitin-conjugating enzyme; E2), and recombinant GSTtagged RGLG3 or RGLG4 (ubiquitin ligase; E3). The ubiquitinated bands were detected by western blotting with anti-Flag. D, Self-ubiquitination of RGLG3 and RGLG4 in the absence of ATP, Flag-tagged ubiquitin, E1, or E2 or in the presence of fulllength, truncated, or mutated E3 (GST tagged). The reaction components were similar to those in B. After 120 min, the bands were visualized by western blotting with anti-Flag.
indicating a loss of RGLG3 or RGLG4 function in rglg3 rglg4 and the respective mutants. Cauliflower mosaic virus 35S promoter-driven transgenic lines constitutively overexpressing RGLG3 (RGLG3ox1 and RGLG3ox7) or RGLG4 (RGLG4ox18 and RGLG4ox22) showed significantly increased RGLG3 or RGLG4 expression, as confirmed by real-time PCR (Supplemental Fig. S5A). These lines were used for further functional studies.
The important effect of JA on restricting root growth was analyzed. In the absence of MeJA, RGLG3 and RGLG4 had insignificant effects on root growth, with no differences in root length between ecotype Columbia-0 (Col-0), rglg3, rglg4, and rglg3 rglg4 (Fig. 3B). Application of 50 mM MeJA restricted root elongation to a similar extent in Col-0, rglg3, and rglg4, but the inhibition was attenuated in rglg3 rglg4, although to a lesser extent than in the JA-insensitive coi1-2 ( Fig. 3B), strongly suggesting a positive and overlapping role for RGLG3 and RGLG4 in mediating the inhibitory effect of MeJA. Moreover, the root length of seedlings grown on Murashige and Skoog (MS) medium supplied with 0, 10, 25, or 50 mM MeJA showed that rglg3 rglg4 antagonized JA-induced root growth restriction in a dose-dependent manner (Fig. 3C), further demonstrating their roles in this JA effect. Additionally, 23 JA-responsive genes were found to be similarly induced by MeJA in rglg3, rglg4, and Col-0, whereas rglg3 rglg4 and coi1-2 had similarly repressed induction (Fig. 3D), indicating that loss of function of both RGLG3 and RGLG4 greatly impacted the JA signaling pathway.
Positive roles for RGLG3 and RGLG4 were further supported by the enhanced expression of RGLG3 or RGLG4 in the transgenic plants that promoted the JA inhibitory effect ( Fig. 3E; Supplemental Fig. S5A). Moreover, when RGLG3 or RGLG4 was overexpressed in the rglg3 rglg4 background (Supplemental Fig. S5B), the suppression effect of rglg3 rglg4 on JA inhibition did not occur in the presence of 25 mM MeJA, as their root lengths were shorter than rglg3 rglg4 and comparable with the wild type (Fig. 3E). Taken together, these results demonstrated that JA-inhibited root growth required the positive roles of both RGLG3 and RGLG4 despite their different responses to MeJA.

RGLG3-and RGLG4-Regulated Root Growth Requires an Intact JA Pathway
To further determine the genetic relationship of RGLG3 and RGLG4 to the key JA signaling components, rglg3 rglg4, RGLG3ox1, and RGLG4ox18 were crossed with coi1-2 (Xu et al., 2002) and myc2-2 (Boter et al., 2004), and MeJA-inhibited root growth was investigated in the generated genotypes (rglg3 rglg4 coi1-2, rglg3 rglg4 myc2-2 [Supplemental Fig. S6], RGLG3ox1 coi1-2, RGLG4ox18 coi1-2, RGLG3ox1 myc2-2, and RGLG4ox18 myc2-2). When grown on MS medium containing 50 mM MeJA, roots of rglg3 rglg4 coi1-2 and rglg3 rglg4 myc2-2 were longer than those of rglg3 rglg4 but similar to their respective coi1-2 and myc2-2 backgrounds, indicating that the JA-insensitive phenotype of rglg3 rglg4 was overcome by the more severe JA insensitivities of coi1-2 and myc2-2 (Fig. 4A). In addition, when integrated into the coi1-2 or myc2-2 background, hypersensitivities of RGLG3ox and RGLG4ox were fully suppressed, as the root lengths were comparable to coi1-2 and myc2-2 ( Fig. 4B), indicating that the RGLG3 and RGLG4 functions in JA-inhibitory root growth depended on both COI1 and MYC2. Expression of PLANT DEFENSIN1.2 (PDF1.2) and VEGETATIVE  . Function of RGLG3 and RGLG4 in JA-inhibited root growth. A, Expression patterns of RGLG3 and RGLG4 after JA treatment. Four-week-old wild-type Arabidopsis plants were treated by spraying 100 mM MeJA onto the leaves and sampled at the indicated time points (h). JR1 expression was examined to indicate the effectiveness of JA treatment. Gene expression was determined by real-time PCR. UBQ10 was used as an internal control, and expression level at 0 h was considered as 1.0. Data represent means 6 SD of four technical replicates. The experiments were repeated three times and yielded similar results. B, JA-inhibited root growth phenotypes of Col-0, rglg3, rglg4, rglg3 rglg4, and coi1-2. Nine-day-old seedlings grown on MS medium supplied with or without 50 mM MeJA were photographed. The experiments were repeated four times with similar results. Bars = 5 mm. C, Root length of 1-week-old Col-0, rglg3, rglg4, rglg3 rglg4, and coi1-2 seedlings grown on MS medium containing different concentrations of MeJA. Data represent means 6 SD (n = 12). Asterisks indicate significant differences compared with the wild type (Student's t test; **P , 0.01). This experiment was repeated three times and had similar results. D, RT-PCR analysis of JA-responsive genes in the plant materials shown in B. UBQ10 was amplified as an internal control. E, Root phenotypes (top panel) and lengths (bottom panel) of wild-type and rglg3 rglg4 seedlings overexpressing either RGLG3 or RGLG4 compared with Col-0 and coi1-2. Plants were grown on MS medium with or without 25 mM MeJA for 9 d before the roots were photographed and root lengths were measured. Bars represent SD (n = 20), and asterisks indicate significant differences from the wild type or double mutant rglg3 rglg4 (Student's t test; *P , 0.05, **P , 0.01). The experiments were repeated three times with similar results. Bar = 1 cm. F, Seedling root phenotypes (top panel) and lengths (bottom panel) of transgenic plants overexpressing either RGLG3 (RGLG3MR) or RGLG4 (RGLG4MR) in the Col-0 and rglg3 rglg4 genetic backgrounds. Two lines of each transgene were used in comparison with those of Col-0 and coi1-2. Plants were grown on MS medium with or without 25 mM MeJA for 9 d before the roots were photographed and root lengths were measured. Data represent means 6 SD (n = 20), and asterisks indicate significant differences from the wild type (Student's t test; *P , 0.05). The experiments were repeated twice with similar results. Bar = 1 cm. STORAGE PROTEIN2 (VSP2), which represent two groups of JA-inducible genes (Lorenzo et al., 2004), was also analyzed in these genetic backgrounds. In rglg3 rglg4 and coi1-2, JA-induced expression of PDF1.2 and VSP2 was suppressed (Fig. 4C), confirming the RT-PCR results (Fig. 3D). In myc2-2, JA induction of PDF1.2 was enhanced but that of VSP2 was inhibited (Fig. 4C), because MYC2 differentially regulates these two genes (Boter et al., 2004;Lorenzo et al., 2004). For PDF1.2, JA induction was suppressed in rglg3 rglg4 coi1-2 similar to rglg3 rglg4 and coi1-2 but was moderately enhanced in rglg3 rglg4 myc2 similar to myc2-2 (Fig.  4C). For VSP2, JA induction in rglg3 rglg4 coi1-2 and rglg3 rglg4 myc2 was similarly repressed compared with coi1-2 and myc2-2 (Fig. 4C). Moreover, in the transgenic overexpressors, expression levels of both PDF1.2 and VSP2 were elevated by MeJA compared with Col-0, but crosses with coi1-2 and myc2-2 negated the positive  Figure 2B. Data represent means 6 SD (n . 10). Asterisks indicate significant differences compared with the wild type (Student's t test; **P , 0.01). The experiments were repeated three times with similar results. Bars = 5 mm. C, Real-time PCR analyses of JA-responsive expression of PDF1.2 and VSP2 in the genotypes used in B. Total RNA was isolated from 9-d-old seedlings grown on MS medium with or without 50 mM MeJA. UBQ10 was used as an internal control, and expression levels were normalized to MS treatment in Col-0. Data represent means 6 SD of four technical replicates. The experiments were repeated three times with similar results. [See online article for color version of this figure.] effects on PDF1.2 and VSP2 expression (Fig. 4C), indicating that enhanced expression of RGLG3 and RGLG4 influenced downstream JA signaling in a COI1-and MYC2-dependent manner. These results suggested that RGLG3 and RGLG4 regulation of root growth and JAresponsive gene expression required an intact JA signaling pathway.
RGLG3 and RGLG4 Also Regulate Response to Coronatine and Susceptibility to Pseudomonas syringae pv tomato DC3000 Next, we used a JA-Ile mimic, CORONATINE (COR), to test if RGLG3 and RGLG4 modulated the CORstimulated response. The inhibitory effect of COR on root elongation was attenuated in rglg3 rglg4 but promoted in RGLG3ox and RGLG4ox, all in a COI-dependent manner (Fig. 5A) and just like that in MeJA-inhibited root growth assays (Fig. 4, A and B). As the hemibiotrophic pathogen Pseudomonas syringae pv tomato (Pst) DC3000 secretes COR to hijack the JA pathway to promote its virulence (Kloek et al., 2001;Fonseca et al., 2009), we further determined how RGLG3 and RGLG4 regulated Arabidopsis susceptibility to Pst DC3000. Both disease symptoms on leaves (Supplemental Fig. S7) and pathogen growth curves (Fig. 5B) showed that rglg3 rglg4 was resistant to Pst DC3000 to a lesser extent than coi1-2 but more than myc2-2, and overexpressed RGLG3 or RGLG4 promoted disease development, suggesting advantageous roles for RGLG3 and RGLG4 in Pst DC3000 virulence. Moreover, rglg3 rglg4 coi1-2 exhibited resistance to Pst DC3000 comparable to that of coi1-2, and enhanced susceptibility of RGLG3ox and RGLG4ox was eliminated by coi1-2 in RGLG3ox1 coi1-2 and RGLG4ox18 coi1-2, highlighting their dependence on coi1-2 in mediating pathogen invasion. Introduction of myc2-2 into rglg3 rglg4 weakly affected the resistance of rglg3 rglg4, Figure 5. Functional analyses of RGLG3 and RGLG4 in response to COR and the hemibiotrophic bacterial pathogen Pst DC3000. A, COR-inhibited root growth in Col-0, rglg3 rglg4, coi1-2, myc2-2, rglg3 rglg4 coi1-2, rglg3 rglg4 myc2-2, RGLG3ox1, RGLG4ox18, RGLG3ox1 coi1-2, RGLG4ox18 coi1-2, RGLG3ox1 myc2-2, and RGLG4ox18 myc2-2. Plants were grown on MS medium with or without 0.5 mM COR for 9 d before the roots were photographed and root lengths were measured. Data represent means 6 SD (n = 15), and asterisks indicate significant differences from the wild type (Student's t test; **P , 0.01). The experiments were repeated twice with similar results. Bar = 1 cm. B, Bacterial growth in the leaves of the materials used in A after infiltration with Pst DC3000 (OD 600 = 0.0001). Bacterial counts are expressed as log (colony-forming units whereas that of RGLG3ox1 and RGLG4ox18 was clearly attenuated by myc2-2 in RGLG3ox1 myc2-2 and RGLG4ox18 myc2-2, suggesting that MYC2 was also involved in the process of RGLG3-and RGLG4-regulated Pst DC3000 infection. Taken together, these results suggested that RGLG3 and RGLG4 took important roles in CORactivated JA signaling.

RGLG3 and RGLG4 Mediate JA-Dependent Wound Response
As no influence on JA-related growth and development, such as stamen development and fertility (Feys et al., 1994;Xie et al., 1998), was observed in mutants or transgenic plants of RGLG3 and RGLG4, they may act primarily in JA-mediated stress responses. Their roles in wound signaling, a representative JAmediated process (Titarenko et al., 1997;Li et al., 2002), were examined. Mechanical wounding stunts plant growth, and an impaired JA pathway inhibits this effect (Yan et al., 2007;Zhang and Turner, 2008). Consistent with these results, our wounding treatment resulted in a loss of 59.4% shoot fresh weight in wildtype Arabidopsis, and this loss was reduced to 29.1% in myc2-2. The 30.7% growth restriction in rglg3 rglg4 was comparable to that in myc2-2 and strongly indicated that loss of function of both RGLG3 and RGLG4 also suppressed wound-retarded plant growth (Fig. 6, A and B).
The wound response is characterized by the rapid induction of many JA-responsive genes. In this study, the expression of the primary response genes JAZ7 and OXOPHYTODIENOATE-REDUCTASE3 (OPR3; Koo et al., 2009) and the secondary response gene JAS-MONIC ACID RESPONSIVE2 (JR2; Rojo et al., 1999) was examined in different genetic backgrounds. rglg3 rglg4 suppressed the induction of JAZ7, OPR3, and JR2, although to a lesser extent than in coi1-2 and myc2-2, probably by limiting either the rate or level of induction (Fig. 6C), reflecting the influence of RGLG3 and RGLG4 on the entire JA signaling pathway during the wound response. In addition, the triple mutant rglg3 rglg4 coi1-2 restricted induction of these genes to a similar extent as coi1-2, and rglg3 rglg4 myc2-2 displayed some additive effects by suppressing the induction to a lower level than rglg3 rglg4 or myc2-2 (Fig.  6C). In the overexpression transgenic plants, enhanced expression of RGLG3 and RGLG4 promoted the induction of the three genes, whereas induction was significantly reduced in the coi1-2 and myc2-2 backgrounds (Fig. 6C), suggesting that functions of RGLG3 and RGLG4 in the wound response also depended on the JA signaling pathway. Suppression of the woundinduced expression of the JA biosynthesis gene OPR3 in rglg3 rglg4 suggested that wound-activated JA biosynthesis was affected, and JA quantification in Col-0 and rglg3 rglg4 showed that wound-promoted JA production was obviously suppressed (Fig. 6D). These results supported important roles of RGLG3 and RGLG4 in the JA-dependent wound response.

Responses of RGLG3 and RGLG4 Are Dependent on COI1
To determine how COI1 and MYC2 affected RGLG3 and RGLG4 expression, transcriptional dynamics of both genes upon MeJA treatment were examined in the coi1-2 and myc2-2 backgrounds in comparison with Col-0. The coi1 mutation disrupted both the induction of RGLG3 and the suppression of RGLG4 by MeJA, but in myc2-2 they were only slightly affected (Fig. 7A), suggesting that RGLG3 and RGLG4 responded to JA in a COI1-dependent manner. proRGLG3:GUS and proRGLG4:GUS constructs were introduced into Col-0, coi1-1 (Xie et al., 1998) or coi1-2, and myc2-2. GUS staining results showed that pathogen inoculation induced RGLG3 but suppressed RGLG4 in Col-0 and myc2-2 but not in coi1 (Fig. 7B) compared with the mock treatment. This indicated that both RGLG3 and RGLG4 responded to Pst DC3000 invasion, which required JA perception, similar to their responses to exogenous MeJA. However, real-time PCR analysis showed that both RGLG3 and RGLG4 were induced to a similar level upon mechanical wounding (Fig. 7C), with strong induction of OPR3 demonstrating the effectiveness of the treatment (Koo et al., 2009). The induction was suppressed by the coi1-2 mutation, indicating that wound-induced RGLG3 and RGLG4 acted mainly through a JA-dependent pathway. The response of RGLG3 to the wound signal in myc2-2 remained unaffected, but that of RGLG4 was apparently disrupted, suggesting that they were regulated by different factors in the wounding response (Fig. 7C). These results revealed different characteristics of RGLG4 in response to wounding compared with the responses to MeJA and Pst DC3000, and importantly, JA perception was required for the expression of both RGLGs.

DISCUSSION
Arabidopsis has five RGLGs, but only RGLG1 and RGLG2 have been functionally studied (Stone et al., 2005;Yin et al., 2007;Cheng et al., 2012). The work described here uncovers essential roles of RGLG3 and RGLG4 in various JA-mediated biological responses, including root growth inhibition (Figs. 3 and 4), pathogen colonization (Fig. 5), and wound response (Fig. 6). Although none of these responses was obviously influenced by a single mutation of rglg3 or rglg4, they were suppressed in the double mutant rglg3 rglg4 and uniformly promoted in the overexpression plants. In addition, the overproduction of RGLG3 or RGLG4 in rglg3 rglg4 could compensate for its functional defects (Fig. 3E). Therefore, the data demonstrate that RGLG3 and RGLG4 coordinately and positively regulate these diverse biological processes.

RGLG3 and RGLG4 Act Differently from RGLG1 and RGLG2
Our data (Fig. 2) together with a previous report (Yin et al., 2007) reveal ubiquitous expression of RGLGs in Arabidopsis tissues. Nevertheless, different characteristics exist for RGLGs. RGLG2 possesses ubiquitin ligase activities, forming both Lys-63-and Lys-48-linked ubiquitin chains that mediate signaling and protein degradation, respectively (Yin et al., 2007;Bhoj and Chen, 2009;Cheng et al., 2012), but RGLG3 and RGLG4 probably do not catalyze Lys-63 linkage-type ubiquitination, as no interaction exists between Lys-63-specific E2s and them ( Fig. 1; Supplemental Fig. S2). Moreover, it seems that RGLG functions are tightly associated with different phytohormone pathways, with RGLG1 and RGLG2 in the auxin response (Yin et al., 2007) whereas RGLG3 and RGLG4 in JA signaling. The function of RGLG5 is a mystery at the point, and if these five members act in different combinations to regulate particular biological processes remains an interesting question; Figure 6. Roles of RGLG3 and RGLG4 in wounding response. A, Stunted growth phenotypes of wounded Col-0, rglg3 rglg4, and myc2-2. Twenty-day-old plants were wounded three times total (once every 3 d) and photographed 3 d after the third wounding. Twelve plants of each genotype were wounded, and representative photographs are shown. The experiments were repeated four times with similar results. U, Unwounded; W, wounded. Bars = 2 cm. B, Shoot fresh weight of plants in the same experiment as A. The numbers above the bars indicate the percentage fresh weight reduction caused by wounding compared with the unwounded plants in each genotype. Data represent means 6 SD (n = 12). C, Real-time PCR analysis for the expression of JAZ7, JR2, and OPR3 in Col-0, rglg3 rglg4, coi1-2, myc2-2, rglg3 rglg4 coi1-2, rglg3 rglg4 myc2-2, RGLG3ox1, RGLG4ox18, RGLG3ox1 coi1-2, RGLG4ox18 coi1-2, RGLG3ox1 myc2-2, and RGLG4ox18 myc2-2 after wounding. UBQ10 was used as an internal control, and expression level at 0 min was considered as 1.0. Data indicate means 6 SD of four technical replicates. further studies on these and their target(s) will provide insight into the mode of their action.

RGLG3 and RGLG4 Are Two Novel Upstream Modulators of JA Signaling
The core JA signaling pathway involves coordinated roles of COI1 and JAZs in JA perception and downstream transcriptional reprogramming by factors like MYC2 (Browse, 2009). In the coi1 mutant, many JAmediated responses such as root growth inhibition and defense responses are severely compromised (Xie et al., 1998). In the responses we checked, rglg3 rglg4 affected them in a similar but less extensive manner than coi1 (Figs. 4-6). More importantly, genetic analysis showed that coi1 totally masked phenotypes of rglg3 rglg4 and also completely suppressed the positive effects of RGLG3ox and RGLG4ox (Figs. 4-6), indicating that RGLG3 and RGLG4 act upstream of COI1 and that JA perception is indispensable for RGLG3 and RGLG4 function. This view can also be corroborated by the genetic interaction analysis of RGLG3 and RGLG4 with MYC2. No differences were apparent for rglg3 rglg4, myc2-2, and rglg3 rglg4 myc2-2 in JA-and CORinhibited root growth, susceptibility to Pst DC3000, and the wound response (Figs. 4-6), suggesting that they similarly affect the JA pathway. Notably, enhanced JA responses upon overexpression of RGLG3 and RGLG4 were suppressed in the myc2-2 background (Figs. 4-6), although not completely in all cases, indicating that MYC2 acts downstream of RGLG3 and RGLG4. Recently identified MYC3 and MYC4, which UBQ10 was used as an internal control, and expression level at 0 h was considered as 1.0. Data represent means 6 SD of four technical replicates. The experiments were repeated twice and yielded similar results. B, GUS staining results showing responses of RGLG3 and RGLG4 to pathogen invasion. Four-week-old rosette leaves from transgenic Arabidopsis expressing proRGLG3:GUS and proRGLG4:GUS in Col-0, coi1-1 or coi1-2, and myc2-2 genetic backgrounds were inoculated with Pst DC3000 (OD 600 = 0.0001), and 3 d later the GUS activities were examined. Bars = 3 mm. C, Expression patterns of RGLG3 and RGLG4 in Col-0, coi1-2, and myc2-2 after wounding. Four-week-old Arabidopsis leaves were wounded using a hemostat, and gene expression was examined using quantitative real-time PCR. OPR3 expression indicated the effectiveness of the wounding. UBQ10 was used as the internal control, and expression levels were normalized to that measured at time zero. Data represent means 6 SD of four technical replicates. These experiments were repeated three times and had similar results.
function redundantly with MYC2 in several JA-mediated responses (Fernández-Calvo et al., 2011), may account for the incomplete repression by myc2.
Therefore, RGLG3 and RGLG4 regulate JA-related responses upstream of the core JA signaling pathway. Consistent with this view, RGLG3 and RGLG4 could influence the entire JA signaling cascade in the responses studied, including not only downstream JAresponsive defense-and growth-related genes but also core signaling members such as JAZ7, MYC2, and ETHYLENE RESPONSE FACTOR1 (Fig. 3D). Moreover, interrupted JA signaling in coi1-2 and myc2-2 impacted the effect of altered RGLG3 and RGLG4 on the induction of tested marker genes (Figs. 4C and 6C), further supporting an upstream location of RGLG3 and RGLG4 in the core JA pathway.
JA measurement indicated that the endogenous JA level was not obviously different between untreated rglg3 rglg4 and Col-0 (Fig. 6D), suggesting that RGLG3 and RGLG4 may not be components of the JA biosynthesis pathway. However, wound-induced JA production could be suppressed in rglg3 rglg4, which is consistent with the results that induction of genes important for JA production, such as OPC-8:0 COA LIGASE1 and OPR3 (Wasternack, 2007;Browse, 2009), was attenuated after MeJA and wound treatment (Figs. 3D and 6C). As we propose that RGLG3 and RGLG4 act upstream of COI1 and MYC2, and that coi1-2 and myc2-2 impaired RGLG3ox and RGLG4ox promotion of OPR3 induction in the wound response (Fig. 6C), the influence of RGLG3 and RGLG4 on JA production may be exerted through feedback regulation.

RGLG3 and RGLG4 Are Subjected to Feedback Regulation
At the transcription level, MeJA and Pst DC3000 (Figs. 3A and 7) similarly stimulated RGLG3 but suppressed RGLG4, presumably because these stresses can activate a JA signal or a JA-mimicking signal (Laurie-Berry et al., 2006). For unknown reasons, RGLG3 and RGLG4 were oppositely modulated in these responses, despite their similar functions and complementarity. The total abundance of these proteins could be delicately controlled to ensure the appropriate intensity of response. This seems to be monitored according to the nature of the response, as expression of RGLG4 was elevated upon mechanical damage (Fig. 7C). Regardless, these differences may explain why the single mutants rglg3 and rglg4 do not display obvious phenotypes and were not identified in previous genetic screens for JA-insensitive mutants.
The expression of most JA signaling components is tightly regulated by JA in a COI1-dependent manner (Devoto et al., 2005). It is true with RGLG3 and RGLG4, as demonstrated by our expression analyses (Fig. 7). However, myc2 loss of function was unable to completely disrupt the responses of RGLG3 and RGLG4 to those stimuli (Fig. 7, A and B). As mentioned above, it needs to be checked in the triple mutant myc2 myc3 myc4 background to fully address this issue (Fernández-Calvo et al., 2011). Despite this, the myc2 mutation could affect RGLG4 wound-responsive expression (Fig. 7C); therefore, MYC2 may act differently with respect to the transcriptional control of RGLG3 and RGLG4, suggesting that the JA pathway is complex and precisely regulated. The critical roles of downstream COI1 and MYC2 in the expression of RGLG3 and RGLG4 suggest that they are subjected to both positive and negative feedback regulation.
Mechanical Implications of RGLG3 and RGLG4 in JA-Mediated Responses RGLG3 and RGLG4 could modulate both MeJAand COR-induced responses (Figs. 3-5). The exogenously applied MeJA is likely first demethylated to yield JA, which is then modified into bioactive JA-Ile, although MeJA itself is also active (Seo et al., 2001;Figure 8. Possible roles for RGLG3 and RGLG4 in modulating JAmediated responses. Mechanical damage activates JA biosynthesis, which is further modified into JA-Ile (black), whereas the bacterial pathogen Pst DC3000 secretes the toxin COR (blue) to structurally and functionally mimic JA-Ile. Both JA-Ile and COR can be recognized by the core receptor component COI1, which mediates degradation of the JAZ repressors. The stress-triggered JA signals are then transmitted by MYC2 and other functionally redundant transcription factors, such as MYC3 and MYC4, and finally the different responses are activated in Arabidopsis according to the nature of the stress, inhibiting root and whole plant growth. The RING-type ubiquitin ligases RGLG3 and RGLG4 (red) probably take essential roles during the perception process by controlling the accessibility of JA-Ile (COR) to the coreceptor COI1. The gray arrows indicate possible direct signaling from JA and stresses to RGLG3 and RGLG4 that cannot be excluded by this study. The dashed arrows indicate possible feedback regulation. [See online article for color version of this figure.] Staswick and Tiryaki, 2004;Tamogami et al., 2008;Wu et al., 2008); COR can structurally and functionally mimic bioactive JA-Ile (Fonseca et al., 2009); RGLG3 and RGLG4 act upstream of COI1 and independent of the JA biosynthesis pathway (as discussed above). Taken together, this evidence suggests that RGLG3 and RGLG4 are likely involved in the signal transduction from JA-Ile to COI1.
At present, our knowledge of the regulatory details between JA and its receptor is limited. JA production is completed in the peroxisome and further modified in the cytoplasm, but perception may occur in the nucleus (Browse, 2009), as COI1 could be a nucleuslocalized protein (similar to TRANSPORT INHIBITOR RESPONSE1; Dharmasiri et al., 2005;Santner and Estelle, 2009). Therefore, a mechanism controlling the accessibility of JA-Ile to the coreceptor COI1 may be necessary, and probably RGLG3 and RGLG4 are involved in this process. Interestingly, their close homologs, RGLG1 and RGLG2, have been associated with auxin transport (Yin et al., 2007). The ubiquitin ligase activities of RGLG3 and RGLG4 definitely contribute to their roles in JA-mediated root inhibition (Fig. 3F), suggesting that an unidentified target exists and negatively regulates JA signaling. Characterization of RGLG3 and RGLG4 strongly implies that more experiments are needed to identify the target and clarify the underlying mechanism.
Ubiquitination participates in nearly every aspect of plant life, and many E3s are involved in phytohormone signaling (Santner and Estelle, 2010). The JA coreceptor COI1 forms an SCF-type E3 with other components and mediates the degradation of JAZs when perceiving JA (Xu et al., 2002;Chini et al., 2007;Thines et al., 2007). To date, no other E3 has been implicated in an essential function in JA signaling. The significance of RGLG3 and RGLG4 is self-evident with their roles in extensive JAmediated responses (Figs. 3,5,and 6). Considering that many plant E3s remain functionally uncharacterized, there are likely more players acting to coordinate JA signaling.

CONCLUSION
In summary, our data reveal that RGLG3 and RGLG4 are essential upstream modulators of JA signaling (Fig.  8) and imply a novel level of ubiquitination-mediated regulation during the perception of bioactive JA. Further identifying their targets will help us understand the complex regulatory network that coordinates JAmediated signals.
Hongwei Guo. Arabidopsis Col-0 was used as the wild type. For axenic seedlings, seeds were surface sterilized and chilled at 4°C for 72 h, then plated on solid MS medium. Seven-day-old seedlings were used for chemical treatment or transplanted into a mixture of soil:vermiculite (1:1). To treat adult plants, 4-week-old plants were sprayed with 100 mM MeJA (Sigma) on the leaves. All plants were kept in a greenhouse at 22°C under a 16-h-light/8-hdark cycle for long-day growth or a 12-h-light/12-h-dark cycle for short-day growth.

Generation of Transgenic Plants
For functional analysis of RGLG3 and RGLG4, their intact coding sequences or the RING domain-mutated forms (Supplemental Fig. S5) were cloned into pJim19, resulting in a fusion with 33 Flag and 33 hemagglutinin tags at the N terminus. For promoter activity analysis, 1.9-kb (RGLG3) or 1.8-kb (RGLG4) promoter fragments were amplified from genomic DNA by PCR and inserted separately into pBI121-GUS to drive GUS expression. All the constructs were introduced into Agrobacterium tumefaciens strain EHA105 by electroporation, and the wild-type Arabidopsis or mutant plants were transformed using the floral dip method (Clough and Bent, 1998). After selecting the transformants, segregation analysis identified single insertion lines in the T2 generation. Enhanced gene expression levels were quantified by real-time PCR (Supplemental Fig. S5) using the primers listed in Supplemental Table S1.

Root Growth Inhibition Assay
Seedlings were grown vertically on plates containing MS medium in the presence or absence of different concentrations of MeJA or COR (Sigma) for 9 d and then photographed. Primary root length was measured on scanned images using ImageJ software (http://rsbweb.nih.gov/ij).

Wounding Treatments
Wounding was performed as described (Yan et al., 2007;Koo et al., 2009). For wound-induced gene expression profiling, fully expanded leaves of 4-weekold plants were bruised across the midrib with forceps. The wounded leaves were then sampled at the indicated time points for RNA isolation. For woundinhibited growth analysis, the distal region of mature leaves of 20-d-old plants was crushed with forceps. The wounding was repeated three times every 3 d on the same leaves. Twelve plants were wounded, and four to six leaves of each plant were treated each time. Three days after wounding the third time, plants were photographed and their fresh shoot weight was recorded.

JA Quantification
For JA content measurement, 4-week-old plant leaves were wounded and collected at the indicated time points. Then, 200 mg of plant tissue was used for JA quantification as described (Cui et al., 2010).

Bacterial Growth Assay
Pseudomonas syringae pv tomato (Pst) DC3000 was cultured at 28°C on King's B medium containing rifampicin (50 mg mL 21 ). For disease testing, at least six 4-week-old plants were infiltrated with 10 mM MgCl 2 (mock treatment) or a bacterial suspension of Pst DC3000 (optical density at 600 nm [OD 600 ] = 0.0001 in 10 mM MgCl 2 ). After 3 d, leaves were harvested, homogenized in 10 mM MgCl 2 , and then serially diluted and spread on King's B medium containing rifampicin (50 mg mL 21 ). Plates were incubated at 28°C for 2 d before the colony number was determined. Data were analyzed using Sigma Plot version 10.0 software and were considered significantly different at the 0.05 level.

Sequence Alignment and Phylogenetic Analysis
Protein sequences were aligned with the ClustalW2 program (http://www. ebi.ac.uk/Tools/msa/clustalw2; Supplemental Fig. S1). The phylogenetic tree was constructed with MEGA5 software (Tamura et al., 2007) using the neighbor-joining method.

RT-PCR and Quantitative Real-Time PCR
Total RNA was extracted from Arabidopsis tissues using Trizol reagent (Invitrogen) and digested with RNase-free DNase I (Takara) for RT. Firststrand cDNA was synthesized from 2 mg of total RNA using the RevertAid First Strand cDNA Synthesis kit (Fermentas) in a 20-mL reaction. The cDNA was then diluted 10-fold with water, and 2 mL was used for RT-PCR. For realtime PCR, the cDNA was diluted 20-fold before use. Real-time PCR analysis was carried out using SYBR Green Real-Time PCR Master Mix (Toyobo) in the DNA Engine Opticon2 Continuous Fluorescence Detector (MJ Research). Data were analyzed using Opticon Monitor 2 software (MJ Research). The relative expression level for each target gene was calculated by the cycle threshold method (Livak and Schmittgen, 2001) using UBQ10 as the internal control gene. Primers used for gene expression analysis are listed in Supplemental  Table S1.

Purification of Recombinant Proteins and in Vitro Ubiquitination Assay
The full-length, truncated, and mutated coding sequences of RGLG3 and RGLG4 were amplified by PCR. The PCR products were then purified and digested by appropriate restriction enzymes before being inserted into the prokaryotic expression vector pGEX-4T-1 (New England Biolabs). All the plasmids were introduced into Escherichia coli strain BL21 (DE3) Plys for expressing recombinant proteins. The glutathione S-transferase (GST)-tagged proteins were purified using the MagneGST protein purification system (Promega).
In vitro ubiquitination assays were performed as described . A total of 500 ng of purified GST-RGLG3, GST-RGLG3DR, GST-RGLG3MR, GST-RGLG4, GST-RGLG4DR, or GST-RGLG4MR was incubated with 50 ng of E1 (Sigma), 100 ng of E2 UbcH5b (Sigma), and 3 mg of Flag-tagged ubiquitin (Sigma) in the ubiquitination buffer (50 mM Tris-HCl, pH 7.5, 2 mM ATP, 5 mM MgCl 2 , and 0.5 mM dithiothreitol) at 30°C for 2 h or for other periods as indicated. Proteins in the reactions were separated by SDS-PAGE (6% [w/v] acrylamide), and western blotting was carried out using anti-Flag (Sigma) as the primary antibody and horseradish peroxidase-conjugated anti-mouse IgG (Promega) as the second antibody.

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Amino acid sequence alignment of RGLGs.
Supplemental Figure S2. Interaction analysis of RGLG3 and RGLG4 with either UBC35 or UBC36 in yeast.
Supplemental Figure S3. Transcript levels of RGLG3 and RGLG4 in Arabidopsis tissues.
Supplemental Figure S5. Transcript levels of RGLG3 and RGLG4 in the transgenic plants used in this study.
Supplemental Figure S6. Growth phenotypes of mutant plants used in this study.
Supplemental Figure S7. Visual disease symptoms after Pst DC3000 infection.
Supplemental Table S1. List of primers used in this study.