Arabidopsis RGLG2, functioning as a RING E3 ligase, interacts with AtERF53 and negatively regulates the plant drought stress response

Transcriptional activities of plants play important roles in responses to environmental stresses. AtERF53 is a drought-induced transcription factor, which belongs to the AP2/ERF superfamily, and has a highly conserved AP2 domain. It can regulate drought-responsive gene expressions by binding to the GCC box and/or dehydration-responsive element (DRE) in the promoter of downstream genes. Overexpression of AtERF53 driven by the CaMV35S promoter resulted in an unstable drought-tolerant phenotype in T2 transgenic Arabidopsis plants. Using a yeast two-hybrid screening, we identified a RING domain ubiquitin E3 ligase, RGLG2, which interacts with AtERF53 in the nucleus. The copine domain of RGLG2 exhibited the strongest interacting activity. We also demonstrated that RGLG2 could move from the plasma membrane to the nucleus under stress treatment. Using an in vitro ubiquitination assay, RGLG2 and its closest sequelog, RGLG1, were shown to have E3 ligase activity and mediated AtERF53 ubiquitination for proteasome degradation. The rglg1rglg2 double mutant but not rglg2 or rglg1 single mutant exhibited a drought-tolerant phenotype when compared to wild-type plants. AtERF53-green fluorescent proteins expressed in the rglg1rglg2 double mutants were stable. The 35S:AtERF53-GFP / rglg1rglg2 showed enhanced AtERF53-regulated gene expression and has greater tolerance to drought stress than the rglg1rglg2 double mutant. In conclusion, RGLG2 negatively regulates the drought stress response by mediating AtERF53 transcriptional activity in Arabidopsis.


INTRODUCTION
In plant-stress responses, transcriptional regulatory networks affecting stressresponsive gene expression play central roles in conferring stress tolerance and protecting plants from adverse environmental conditions. High temperatures, drought, and high salinities are common abiotic stresses that adversely affect plant growth and crop production. Plant stress responses are regulated by multiple signaling pathways that activate gene transcription and the downstream machinery.
In the signal transduction network from the perception of stress signals to stressresponsive gene expressions, various transcription factors and cis-acting elements in stress-responsive promoters function in a plant's adaptation to environmental stresses (Shinozaki and Yamaguchi-Shinozaki, 2006).
The AP2/ERF superfamily is defined by the AP2/ERF domain, which consists of about 60~70 amino acids and is involved in DNA binding. The ERF family is the largest family that encodes transcriptional regulators and contains 122 genes in Arabidopsis (Nakano et al. 2006). The AP2/ERF family is divided into 12 groups based on a phylogenetic analysis of DNA sequences. The ERF domain was shown to specifically bind to a GCC box, which is a DNA sequence involved in the ethylene-responsive transcription of genes (Ohme-Takagi and Shinshi, 1995). It also recognizes the dehydration-responsive element (DRE) in target promoters. The DRE (5′-TACCGACAT-3′) was first identified in the promoter of the droughtresponsive RD29A gene from Arabidopsis (Yamaguchi-Shinozaki and Shinozaki, 1994). Similar cis-acting elements, named the C-repeat (CRT) and low-temperatureresponsive element (LTRE), both containing an A/GCCGAC motif that forms the core of the DRE sequence, regulate cold-inducible promoters (Thomashow et al., 6 1999). Many genes in group III (i.e., DREB1) and IV (i.e., DREB2) were shown to be involved in abiotic stress tolerance. DREB1 genes are induced by low temperatures, whereas DREB2 homologues are induced by drought and high-salt stress. In spite of the increase in endogenous levels of ABA after stress treatments, DREB2 is not induced by exogenous ABA, suggesting that its function is independent of this hormone (Liu et al., 1998). On the other hand, evidence was provided that both the cold-inducible DREB1 transcription factors (TFs) and non-cold-inducible DREB1D may be involved in activation of the CRT/DRE by abscisic acid (ABA) treatment (Knight et al., 2004).
Previous reports suggested that overexpression of the constitutively active DREB2A resulted in significant drought stress tolerance in transgenic Arabidopsis plants (Sakuma et al., 2006). DREB2A requires posttranslational modification for its activation, and was proven to be degraded by the 26S proteasome through DREB2A-interacting protein1 (DRIP1)-mediated ubiquitination under non-stressed conditions. DRIP1 is a RING domain E3 ligase isolated from yeast two-hybrid screening, and the drip1 drip2 double mutant exhibits a drought-tolerant phenotype (Qin et al., 2008). Moreover, overexpression of the full-length DREB2A revealed it to be more stable in a drip1-1 background than in the wild-type one. Qin et al. (2008) also found that overexpression of DRIP1 delayed the expression of DREB2Aregulated drought-responsive genes. Drought-induced gene expression was also significantly enhanced in drip1 drip2 double mutants under dehydration stress.
Ubiquitination is a common regulatory mechanism in all eukaryotes and selectively targets a diverse range of substrates, including hormone receptors, light regulators, transcription factors, and damaged proteins for degradation by the 26S proteasome and affects a range of cellular processes, like hormone signaling, embryogenesis, photomorphogenesis, circadian rhythms, floral development, senescence, disease resistance, and abiotic stress responses (Vierstra 2009;Yee and Goring, 2009).
Ubiquitin's attachment to its target protein for modification is conserved in all eukaryotes, and the conjugation cascade involves three consecutive enzymes, E1 (ubiquitin-activating enzyme), E2 (ubiquitin-conjugating enzyme), and E3 (ubiquitin ligase) ligases (Glickman and Ciechanover, 2002). Ubiquitination mediated by E1, E2, and E3 conjugates either single or multiple ubiquitin molecules to the target protein, thus enabling the ubiquitin-labeled protein to be recognized by the 26S proteasome and targeted for degradation. Ubiquitin can itself be a substrate for ubiquitination; it is a 76-amino acid protein and has seven Lys residues that can serve as sites of modification (Yin et al., 2007). Ubiquitination of the target protein can end in proteasomal degradation, or is associated with non-proteolytic signaling (Pickart and Eddins, 2004;Yin et al., 2007). Linkage of ubiquitin to Lys-48 of another ubiquitin moiety forms the most prominent chain type and targets substrates for recognition and degradation by the proteasome (Yin et al. 2007). In the process, the E3 is the last step in the transfer of ubiquitin but is also responsible for recruiting the target protein for ubiquitination. E3 is considered the major substrate recognition component of the pathway.  (Fig. S1), whereas the marker gene (RD29A for ABA treatment and EIN2 for ACC and MeJA treatments) showed specific induced expression.
Histochemical analysis of GUS expression in transgenic plants showed that there was no GUS expression under normal growth conditions. Seven-day-old seedlings, and 2-and 5-week-old T2 transgenic plants were analyzed after 1 h of drought treatment and 30 min of rehydration (Fig. 1B). In 7-day-old seedlings, transgenic plants showed weak GUS activity in the shoot apical meristem, and vascular tissues in leaves and roots. In 2-week-old transgenic plants, GUS activity was observed in leaves and roots, especially in vascular bundles (Fig. 1B). The GUS activity did not occur in guard cells of leaves (data not shown). In 5-week-old transgenic plants, GUS activity was observed in all tissues examined, with low activity in flowers and siliques (Fig. 1B). Expression was detected in sepals of very young closed buds. When the sepals and petals withered, expression was detected at the bottom of the silique, in the abscission zone, and in the pedicel region below that.
Strong GUS expression was observed in the vasculature of most tissues.
For subcellular localization of the protein, AtERF53 cDNA was fused in frame to the C-terminal side of the GFP marker gene, and the resulting construct was introduced into protoplasts isolated from Arabidopsis together with the construct carrying 35S:NLS-RFP by PEG-mediated transformation. The 35S:NLS-RFP construct served as a positive nuclear targeting control. Green and red fluorescence was detected in nuclei (Fig. 1C). Transgenic Arabidopsis plants harboring an AtERF53-GFP construct driven by the 35S promoter were generated in an attempt to determine AtERF53 protein stability in cells. As shown in Figure 2A, AtERF53-GFP transgene was highly expressed in the transgenic plants under normal growth condition, but the GFP fusion protein was not detected in Western blot analysis (Fig. 2B). The AtERF53-GFP protein was not detected in either roots or leaves under normal growth conditions. However, when these plants were treated with the 26S proteasome inhibitor, MG132, clear GFP fluorescence was observed microscopically in both roots and cotyledons (Fig. 2C). GFP-fusion proteins were also detected in the Western blot analysis (Fig. 2D). Although the AtERF53 transgene was expressed under normal growth condition, the corresponding protein is apparently unstable and may be degraded by the 26S proteasome.

Identification of a RING E3 Ligase Interacting with AtERF53
A yeast two-hybrid screen was performed to identify proteins that interact with  Approximately 1.2 x 10 6 yeast transformants were screened on a synthetic defined (SD) medium lacking Leu, Trp, and His (SD/-T-L-H) plus 1.5 mM 3-amino-1,2,4triazole. Forty-five positive clones were obtained, and the β -galactosidase activity was individually assessed for all of these clones (see Supplemental Table S1).
Sequence determination from the clone exhibiting the strongest activity identified a C3HC4 RING domain-containing protein, RGLG2 (for RING domain ligase 2, At5g14420).

Localization of RGLG2-interacting Domains with AtERF53
According to a previous study (Yin et al., 2007), the RGLG2 structure is shown in figure 3A. RGLG2 proteins contain a so-called copine (or von Willebrand factor type A) domain. Using protein analysis programs such as pfam, no transmembrane domain was indicated (Yin et al., 2007). We therefore designed three different fragments of RGLG2 according to the structure and fused them to the yeast GAL4-AD vector, and each construct was co-transformed into yeast cells with the

RGLG2 Translocalizes into the Nucleus under Salt Stress and Interacts with AtERF53 In Vivo
To further confirm the interaction between RGLG2 and AtERF53 in vivo, we first In vitro pull-down assays were performed to test the interaction between the fulllength RGLG2 protein and AtERF53. RGLG2 was fused to the C terminal of GST 13 tag in the pGEX-6p-1 vector and AtERF53 was fused to the Trx tag in the pET32a vector. As a result, the full-length Trx-AtERF53 could be pulled down by GST-RGLG2 (see supplemental Fig. S2). A BiFC system was also conducted to study the interaction between RGLG2/1 and AtERF53 in plant cells (Lee et al., 2007).
RGLG2 was fused to the C-terminal region of the CFP, and AtERF53 cDNA was fused to the N-terminal region of Venus. The two constructs were co-transfected into Arabidopsis protoplasts. The empty vectors in combination with each fusion construct were also co-transfected into Arabidopsis protoplasts as negative controls.
After about 16 h of incubation, YFP fluorescence was observed in the nuclei of protoplasts in samples co-transfected with cCFP-RGLG2 and nVenus-AtERF53, whereas samples co-transfected with the empty vector and either cCFP-RGLG2 or nVenus-AtERF53 did not yield any YFP signal (Fig. 4C). The BiFC results of RGLG1 interacting with AtERF53 were also shown in the supplemental data ( Fig.   S3). YFP signals were also observed in the nuclei. These results suggest that both RGLG2 and RGLG1 co-localize and interact with AtERF53 in nuclei of Arabidopsis protoplasts under stress condition.

RGLG2 Mediates AtERF53 Ubiquitination
We carried out in vitro ubiquitination experiments with the RGLG2 protein to confirm its function as an E3 ligase. Full-length RGLG2 was fused to GST in the pGEX vector. Purified RGLG2-GST fusion protein was mixed with His-tagged ubiquitin (ubi-His), rabbit E1, and human UBCH5c (E2). An immunoblot analysis with anti-ubi showed that ubiquitinated proteins were detected in the presence of all these components (Fig. 5A). Furthermore, an immunoblot analysis with anti-GST showed that RGLG2-GST in the presence of ubiquitin, E1, and E2, was attached to one ubiquitin monomers. As a result, one additional higher-molecular-weight protein band appeared (Fig. 5B). These results suggest that the RGLG2 protein can be autoubiquitinated in the presence of the E1 and E2 enzymes. As for RGLG1, the ubiquitination assay was also performed to demonstrate its E3 ligase activity using ubiquitin antibody (supplemental data Fig. S3).
In order to determine whether RGLG2 and RGLG1 (supplemental data) can mediate AtERF53 protein ubiquitination, we also conducted an in vitro ubiquitination assay using AtERF53 protein as a substrate. Full-length AtERF53 was fused to thioredoxin (Trx) and a His tag in the pET32a vector, and the recombinant fusion protein was purified by the His tag affinity to a nickelnitrilotriacetic acid agarose matrix. An immunoblot analysis with anti-Trx showed that a higher-molecular-weight shifted band was observed in the presence of ubiquitin, E1, E2, and RGLG2 (Fig. 5C), or ubiquitin, E1, E2, and RGLG1 (see supplemental Fig. S3). The shifted band should correspond to one additional ubiquitin monomer attached to the substrate according to the molecular weight.

The Double Mutant rglg1rglg2 Exhibits a Dehydration-tolerant Phenotype
RGLG2 mRNAs are rather abundant and ubiquitously expressed but with tissuespecific variations (Kraft et al., 2005). According to the microarray data of the 15 hypothesized that RGLG1 also participates in the interaction with AtERF53. The homozygous rglg1rglg2 double mutant showed a bush-like phenotype, completely lacking apical dominance. It also showed significant plant growth retardation and a late-flowering phenotype compared to those of the Col-0 wild-type and the two single mutants (Fig. 6A). To test the drought-stress tolerance of rglg1rglg2, water was withheld from 2-week-old double-mutant and wild-type plants for about 14～ 16 days, and then they were rewatered for 3~5 days. The survival rate of the rglg1rglg2 double mutant was 58% which was significantly higher than that of wild-type plants (2%) (Fig. 6B). No significant difference was observed between wild-type plants and single mutants (rglg1 single mutant data not shown).

AtERF53-GFP Protein Accumulation and Enhanced Drought Tolerance
To confirm RGLG2's and RGLG1's functions in the stability of the AtERF53 protein, we transformed a 35S: AtERF53-GFP construct into the rglg1rglg2 double mutant and obtained two independent transgenic lines. A real-time RT-PCR (reverse transcriptase-polymerase chain reaction) revealed that the transgene expression in both double-mutant and wild-type backgrounds were relatively high, whereas AtERF53 was expressed at very low levels under normal non-stressed conditions in wild-type plants (Fig. 7A). Interestingly, green fluorescent signals were observed in nuclei of 35S:AtERF53-GFP/rglg1rglg2, but not of 35S:AtERF53-GFP/Col-0, plants (Fig. 7C). We performed an immunoblot analysis to detect the GFP or GFP fusion protein in transgenic plants using anti-GFP. A protein band of ~67.5 kDa, which is the predicted AtERF53-GFP molecular mass, was detected in two independent 35S:AtERF53-GFP/rglg1rglg2 transgenic lines (Fig. 7B). However, no GFP fusion protein was detected in two 35S:AtERF53-GFP/Col-0 transgenic lines.
The expression of COR15B and P5CS1 was compared in two different lines from After withholding water for about 21 days, 35S:AtERF53-GFP/rglg1rglg2 transgenic plants (67%) showed much greater drought tolerance than rglg1rglg2 double mutants (21%) (Fig. 7E). We also compared the drought tolerance between WT, rglg1rglg2 double mutants, and RGLG2 overexpressor, which is supposed to be like aterf53 knockout mutants. After withholding water for about 14 days, both WT and RGLG2 overexpressor showed non-drought-tolerant phenotype (supplemental data Fig. S5), while rglg1rglg2 double mutants showed greater drought tolerance than RGLG2 overexpressor. We have also examined the proline contents in WT, double mutants, and 35S:AtERF53-GFP/rglg1rglg (supplemental data). As shown in Figure S6, we observed higher proline levels in both double mutants and overexpression lines than in the wild-type plants. Moreover, overexpression lines contained more proline than double mutants. In conclusion, protein accumulation and drought tolerance experiments provide indirect evidence that RGLG1 plays redundant role of RGLG2 in regulating the function of AtERF53.

DISCUSSION
In this study, we showed that AtERF53 was clearly induced at the transcription level by drought, salt, and osmotic stress, but not by ABA, JA, or ACC. According evidence further clarifies that the regulation of AtERF53 by RGLG2/RGLG1 is possible because of their spatial patterns. The spatial expression patterns observed in promoter:GUS transgenic plants were similar to many transcription factors such as AtMYB44, MYBC1, ANAC012, and HIPP26, which are also highly induced by many abiotic stresses (Ko et al., 2007;Jung et al., 2008;Barth et al., 2009;Zhai et al., 2010). The root system is responsible for water absorbance and transportation, and is the first organ to detect a water deficiency in the soil. High expression levels of GUS activity observed in roots and vascular bundles suggest that AtERF53 may function in the rapid drought stress response and long-distance signal transduction to mount a systemic defense in Arabidopsis plants. Interestingly, the expression patterns that were identified in leaves were not observed in guard cells, suggesting that AtERF53 does not function in stomatal closure for water conservation.
RGLG2 is categorized as being in the RING-HCa subgroup, as is COP1, which represses photomorphogenesis by targeting activators of light-responsive genes for degradation (Hardtke et al., 2000). As far as we know, only limited reports support RING domain-containing proteins functioning as E3 ligases. RGLG2 and RGLG1 both function in proteasomal degradation (this study), and are also associated with non-proteolytic signaling (Yin et al., 2007). This property of E3 ligases probably is not uncommon because AtCHIP (a CHIP-like protein in Arabidopsis) is another E3 ligase that functions in proteasomal degradation (Shen et al., 2007), and is also associated with non-proteolytic signaling (Luo et al., 2006). In addition, RGLG 20 while others are induced in response to stress. It was generalized that these E3 ligases may function as regulators during various abiotic stress responses (Cho et al., 2008).
In vitro ubiquitination assays in the presence of ubiquitin, rabbit E1, and human UBCH5c (E2) showed that RGLG2 displays E3 ligase activity and mediates ubiquitination of AtERF53. RGLG2 was reported to regulate apical dominance in Arabidopsis by forming an ubiquitin Lys-63 chain, whereas formation of Lys-48linked chains targets substrates that are recognized and degraded by the proteasome (Yin et al., 2007). RGLG2 is specific for ubiquitin Lys-63 linkages in the presence of UBC35 and MMZ2 (respective Arabidopsis homologs for Ubc13 and Mms2, which form regulatory-type ubiquitin Lys-63 chains in animals and yeast).
Predominant ubiquitin signals for protein degradation appear to be the Lys48-linked ubiquitin chain; however ubiquitin chains of other linkage types, were also reported (Johnson et al., 1995). Future studies are necessary to confirm the ability of RGLG2 to form Lys-48 chains and in vivo ubiquitination of AtERF53.
The finding of RGLG2/RGLG1, the specific E3 ligases, targets the AtERF53 protein for degradation, explaining why we could not obtain AtERF53 overexpressor at the beginning. The rglg1rglg2 double mutants displayed significant drought tolerance when compared to wild-type plants. We proved that the AtERF53 protein is more stable and is the target transcription factor conferring drought tolerance in double mutants. P5CS1 (delta1-pyrroline-5-carboxylate synthase 1) is a key enzyme involved in proline synthesis pathway. Accumulation of proline in plant cells can protect the cells from osmotic stress (Hong et al., 2000).
The expression level of P5CS1 in 35S:AtERF53-GFP/rglg1rglg2 were greater than that in wild-type plants, rglg1rglg2 double mutants, and in 35S:AtERF53-GFP/Col-0, might suggest that AtERF53 confer drought tolerance by the way of second  This result further proved that AtERF53 is a positive regulator of drought tolerance.
In conclusion, we found that RGLG2, functioning as a RING domain E3 ligase, interacts with a drought-inducible transcription factor, AtERF53. AtERF53 induced during dehydration indeed plays a positive role in plant drought signaling, and is recognized and ubiquitinated for 26S proteasome proteolysis under non-stressed conditions. RGLG2 mediates the ubiquitination and thus negatively regulates drought-stress signaling. The degradation process might be inhibited temporarily once stress signaling occurs, and consequently plant cells may acquire efficient AtERF53 to activate stress-responsive gene expression (Fig. 8).

Fluorescence Observation
The GUS expression patterns of tissues from various organs either from soil-grown Observation was conducted with a light microscope (MZ16F; Leica), and an RS Photometrics CoolSNAP camera (DFC490; Leica) was used to take the digital images, with the corresponding IM50 software. GFP fluorescence was observed with a laser scanning confocal microscope (TCS SP5 AOBS; Leica).

Yeast Two-hybrid Screening and Interaction Assay
An Arabidopsis complementary (c)DNA library was prepared in the pGADT7 SalI-BamHI and HindIII-EcoRI sites, respectively, after the 35S promoter sequence.
Transient expression in Arabidopsis protoplasts was analyzed following Jen Shen's lab protocol (Yoo et al., 2007). YFP fluorescence was observed with a laser scanning confocal microscope (Leica TCS SP5).

Fusion Protein Preparation
Recombinant GST fusion proteins were prepared as described in the GE protocol handbook (www.gelifesciences.com/protein-purification according to the manufacturer's procedures and then stored at -30 °C.

In Vitro Ubiquitination Assays and Protein Gel Blot Analysis
The ubiquitination assays were generally performed as described by Xie et al.  purified and 200 ng of the purified protein was incubated together with the ubiquitination mixture for 2 h. The mixture was then subjected to 6% SDS-PAGE and an immunoblot analysis. For the anti-GFP immunoblot analysis, the nuclear fraction was prepared according to the methods of Busk and Pages (1997) with minor modifications. Eighteen micrograms of protein was loaded and blotted for immunodetection with a monoclonal anti-GFP primary antibody (Santa Cruz Biotechnology, http://www.scbt.com/).

Stress-tolerance Tests
For the drought-tolerance test, plants were initially grown on soil under a normal watering regime for 3 weeks. Watering was then halted and observations were made after a further 14~16 d without water. When wild-type (WT) plants exhibited lethal effects of dehydration, watering was resumed, and the plants were allowed to grow for a subsequent 5 d. The survival rate was scored.

ACKOWNLEDGMENTS
We thank Prof. Andreas Bachmair for providing rglg1rglg2, rglg1, and rglg2 mutants and helpful comments on the manuscript. This work was supported by the National Science Council, Taiwan