Arabidopsis Ubiquitin Conjugase UBC32 Is an ERAD Component That Functions in Brassinosteroid-Mediated Salt Stress Tolerance

Plants modify their growth and development to protect themselves from detrimental conditions by triggering a variety of signaling pathways, including the activation of the ubiquitin-mediated protein degradation pathway. Endoplasmic reticulum (ER)-associated protein degradation (ERAD) is an important aspect of the ubiquitin-proteasome system, but only a few of the active ERAD components have been reported in plants. Here, we report that the Arabidopsis thaliana ubiquitin-conjugating enzyme, UBC32, a stress-induced functional ubiquitin conjugation enzyme (E2) localized to the ER membrane, connects the ERAD process and brassinosteroid (BR)-mediated growth promotion and salt stress tolerance. In vivo data showed that UBC32 was a functional ERAD component that affected the stability of a known ERAD substrate, the barley ( Hordeum vulgare ) powdery mildew O (MLO) mutant MLO-12. UBC32 mutation caused the accumulation of bri1-5 and bri1-9, the mutant forms of the BR receptor, BRI1, and these mutant forms subsequently activated BR signal transduction. Further genetic and physiological data supported the contention that UBC32 plays a role in the BR-mediated salt stress response and that BR signaling is necessary for the plant to tolerate salt. Our data indicates a possible mechanism by which an ERAD component regulates the growth and stress response of plants.


INTRODUCTION
As sessile organisms, plants must adapt their growth and development to detrimental conditions by activating a variety of signaling pathways, which includes the interplay of stress response genes. Extreme environments often increase free radical levels, which cause protein denaturation and damage to the plant. The removal of these proteins by various quality control pathways within the ubiquitin-26S proteasome system is critical for cell survival (Smalle and Vierstra, 2004). Usually, proteins that are designated for degradation by the 26S proteasome are covalently modified by the attachment of a polyubiquitin chain. This is achieved in a multistep reaction, sequentially involving an E1 enzyme (ubiquitinactivating enzyme), an E2 enzyme (ubiquitin-conjugating enzyme [UBC]), and an E3 enzyme (ubiquitin ligase). Substrate specificity is mainly determined by both E2 and E3 enzymes. In the Arabidopsis thaliana genome, there are 37 E2s and over 1400 E3s (Vierstra, 2009). Although the functions of numerous E3s have previously been demonstrated, our knowledge regarding the role of E2s in signal transduction pathways is limited. Consistent with the fact that E2s are conserved in different eukaryotes, many E2s in Arabidopsis perform a similar role to their homologs in yeast or mammalian cells. For example, Arabidopsis UBC1 and UBC2 play a redundant role in histone 2B monoubiquitination and have a similar function to their yeast homolog, RAD6 (Zwirn et al., 1997;Cao et al., 2008;Gu et al., 2009;Xu et al., 2009). Accumulated data suggest that abiotic stresses disturb endoplasmic reticulum (ER) homeostasis and induce ER stress . Cells invoke an ER stress response known as the unfolded protein response (UPR) that involves distinct signal transduction pathways. The UPR has been well studied in yeast and mammalian cells. When cells sense ER stress, they reduce the rate of protein synthesis and translocation into the ER through RNA-dependent protein kinase-like ER eIF2a kinase pathways and subsequently upregulate the expression of ER chaperones, such as binding protein, disulfide isomerase, and calnexin (CNX), through the IRE1 and ATF6 pathways. When prolonged ER stress extensively impairs the homeostasis of the ER, apoptosis is activated by the induction of CHOP (encoding the CCAAT/ enhancer binding protein homologous protein) (Araki et al., 2003).
The accelerated degradation of misfolded proteins by the ERAD pathway is another important response to alleviate ER stress. The ERAD process consists of substrate recognition, targeting, retrotranslocation, polyubiquitination, and degradation by the 26S proteasome. In yeast, two transmembrane E3 complexes that are localized in the ER contribute to the ubiquitination events in the ERAD pathway. One is the Hrd1p-Hrd3p complex, which functions alongside the ubiquitin-conjugating enzymes, Ubc7p and Ubc1p. The other is the Doa10p complex, which cooperates with Ubc6p and Ubc7p. Ubc6p is an ER membrane-localized ubiquitin-conjugating enzyme that has its catalysis domain facing the cytosol, whereas Ubc7p is anchored to the ER membrane via Cue1p (Hirsch et al., 2009).
Compared with the level of knowledge regarding ER stress signaling in yeast and mammalian cells, research in plants is limited. Genomic analysis has shown that the ER stress response is integrated with other important cellular processes (Martínez and Chrispeels, 2003;Kamauchi et al., 2005;Irsigler et al., 2007). For example, the ER stress signal is transduced by Ire1 (Koizumi et al., 2001;Noh et al., 2002;Deng et al., 2011), the bZIP transcription factor family (Iwata and Koizumi, 2005;Liu et al., 2007), and heterotrimeric G protein signaling .
Brassinosteroids (BRs) are a group of plant polyhydroxy steroidal hormones that control a variety of significant plant growth and development processes, including cell division and elongation, photomorphogenesis, xylem differentiation, seed germination, and adaptations to abiotic and biotic environmental stresses (Clouse and Sasse, 1998). Exogenous BR treatment confers tolerance to a range of abiotic stresses, including drought, salt, and high temperature (Kagale et al., 2007). BR signaling is initiated by the plasma membrane-localized receptor, BRI1, which transduces information to the nucleus by a series of phosphorylation events (Kim and Wang, 2010). Mutant BR receptors bri1-9 and bri1-5, which are encoded by genes with a point mutation and thus have a single amino acid substitution, are reportedly retained in the ER by an overvigilant ER quality control system (Jin et al., 2007) for degradation by the ERAD mechanism (Hong et al., 2008(Hong et al., , 2009. Recently, HRD3A and HRD1 were reported to be ERAD components in plants that were responsible for the degradation of bri1-9 and bri1-5 Su et al., 2011), but no ubiquitin conjugating enzyme in the ERAD pathway has been reported in plants to date. Another article has indicated that two transcription factors, bZIP17 and bZIP28, not only activate ER chaperone genes, but can also activate BR signaling (Che et al., 2010). However, components that are directly responsible for both ERAD and BR signaling activation have not been reported.
In this study, we characterized the molecular features of the Arabidopsis ubiquitin-conjugating enzyme UBC32 using a combination of genetic and cell biology studies. Our data showed that UBC32 was involved in stress response through the regulation of BR signaling by the ERAD pathway.

RESULTS
The Stress-Induced UBC32 Encodes an ER-Localized Ubiquitin-Conjugating Enzyme To identify Arabidopsis ubiquitin-conjugating enzyme (E2) members in stress responses, an in silico gene expression set of Arabidopsis E2 genes was analyzed from several publicly available stress-related microarray studies (Seki et al., 2002). Interestingly, the UBC32 gene was highly induced by both drought and salt (NaCl) treatments. We could detect the expression of UBC32 throughout the life cycle in all tissues and at a particularly high level in germinated roots and senescent leaves (see Supplemental Figures 1A and 1B online). Quantitative RT-PCR analysis ( Figure  1A) and UBC32 promoter-b-glucuronidase (GUS) staining ( Figure  1B) confirmed the microarray data and revealed that the transcript level of UBC32 was upregulated by stressors such as NaCl, mannitol, and drought and increased according to the concentration of NaCl and mannitol. A whole-genome BLAST search (http:// blast.ncbi.nlm.nih.gov/) indicated that there are two other genes that are homologous to UBC32 in Arabidopsis: UBC33 and UBC34. All three genes are homologous to the yeast Ubc6p (Bachmair et al., 2001) and mammalian UBE2J1 and UBE2J2 (Lenk et al., 2002), which all participate in the ERAD process. Similar to the homolog in metazoa, the plant Ubc6p homolog evolved into two subfamilies, and UBC32 was found to be similar to the larger one, UBE2J1, whereas UBC33 and UBC34 belong to the UBE2J2 subfamily (Lenk et al., 2002;Oh et al., 2006;Younger et al., 2006) (see Supplemental Figure 1C online). As the homologous yeast and mammalian genes participate in the ER stress response, we assessed the expression of UBC32 after treatment with ER stressors. Consequently, our data showed that UBC32 was also induced by L-azetidine-2-carboxylic acid (AZC), tunicamycin (Tm), and DTT ( Figures 1A and 1B). These results demonstrated that UBC32 was induced by different types of abiotic stress and ER stress and suggested that UBC32 was involved in both the abiotic stress and ER stress responses in plants.
UBC32 encodes a protein of 309 amino acids with a predicted molecular mass of ;34.32 kD. The Simple Modular Architecture Research Tool program (http://smart.embl-heidelberg.de/) predicted the presence of a transmembrane domain located at the C terminus of the UBC32 protein and a ubiquitin-conjugating enzyme catalytic domain (UBCc domain) with a conserved active Cys at position 93 located at the N terminus ( Figure 1D; see Supplemental Figure 1C online). To discover the UBC32 subcellular localization, the UBC32-green fluorescent protein (GFP) fusion construct driven by the cauliflower mosaic virus 35S promoter was introduced into wild-type Arabidopsis. Confocal scanning of the transgenic seedling roots revealed that the free GFP control was located in both the cytosol and nucleus ( Figure  1Ca), whereas the UBC32-GFP protein was observed in the perinuclear region as well as peripheral structures (Figure 1Cb) but not in the nucleus. The results of plasmolysis treatment with 0.8 M mannitol indicated that the UBC32-GFP protein was localized in intracellular membranes rather than in the cell wall (Figures 1Cc to 1Ce). A colocalization experiment further showed that UBC32-GFP was mostly colocalized with an ER marker, ER-Tracker Red dye (Invitrogen), and was circumjacent to a nuclear marker, 49,6-diamidino-2-phenylindole (DAPI) (Figures 1Cf to 1Ci). To test whether UBC32 was a functional E2, an in vitro thiolester formation assay was performed. To increase protein solubility, we produced a transmembrane domain-deleted UBC32 mutant (UBC32nTM) in Escherichia coli as a fusion protein with a 63His tag and purified the His-UBC32nTM protein from the soluble fraction. In the presence of His-tagged ubiquitin, E1, and ATP, the purified His-UBC32nTM protein formed an adduct with ubiquitin that was lost in the presence of the thiol-reducing agent DTT, indicating that a thioester linkage was formed between ubiquitin and UBC32 ( Figure 1E).  (A) Quantitative RT-PCR was used to show that UBC32 is induced by NaCl, mannitol, and the ER stress elicitors Tm, DTT, and AZC. Bars (left to right) indicate water, 150, 200, 250, 300, or 400 mM NaCl, 5 mg/mL Tm, 2 mM DTT, and 5 mM AZC. Each bar represents the mean 6 SD of three independent repeats. Student's t test (P < 0.05) was used to compare all treatments with the water control. After another 9 d, ubc32 grew better than the other two lines: ;62% were green and had true leaves, ;25% turned yellow and lacked true leaves, and 13% were pale. However, the percentage of pale seedlings was highest in 35S-UBC32 lines, which had 58 and 55% positive for ovx2.5 and ovx3.7, respectively, while the wild-type plants were in between this range ( Figure 2A). We also measured the root growth inhibition of same lines on vertical growth plates with different concentrations of NaCl. Similar phenotypes were observed, with increased inhibition in the overexpression line and reduced inhibition in the mutant lines (see Supplemental Figure 2G online). A similar response to KCl was observed (see Supplemental Figures 2H and 2I online), while no notable differences were found among the different lines after treatment with mannitol ( Figure 2A). These data showed that ubc32 mutants were more tolerant to NaCl than wild-type plants, whereas 35S-UBC32 plants were less tolerant, and the response to NaCl was salt specific rather than ionic or osmotic. It is well known that abscisic acid (ABA) plays a very important role in plant salt signaling; thus, we also detected the response of these lines to ABA. As shown in Supplemental Figure 3 online, the ubc32 lines were more insensitive than wild-type plants to ABA in both the germination and postgermination stages, whereas the 35S-UBC32 line were more sensitive (see Supplemental Figure 3 online).
As the yeast and human homologs of UBC32 function in the ERAD pathway and our data showed that UBC32 was induced by ER stress elicitors ( Figures 1A and 1B), we expected that the ubc32 mutants and 35S-UBC32 plants would have an altered response to ER stress. To test this assumption, 2-d-old ubc32 mutants and wild-type and 35S-UBC32 plants were transferred to half-strength MS medium containing 0.45 mg/mL Tm. After growing for 4 d, ubc32 mutants had the highest true leaf emergence rates and the lowest proportion of dead seedlings with yellow cotyledons, whereas the growth status was the worst in 35S-UBC32 plants. No major differences were observed between seedlings grown on control plates that contained the same volume of the Tm solvent DMSO ( Figure 2B). These data showed that ubc32 mutants were less sensitive to Tm than wild-type plants but that 35S-UBC32 plants were more sensitive to Tm. This suggested that UBC32 was involved in the ER stress response.

UBC32 Is an Active Component of the Plant ERAD Compartment
The yeast homolog of UBC32, Ubc6p, participates in the Doa10 complex of the ERAD pathway and is involved in the degradation of a broad range of ERAD substrates (Kostova et al., 2007). We speculated that UBC32 is involved in the plant Doa10 complex of the ERAD system as all of the gene expression, genetic, and

of 12
The Plant Cell subcellular localization studies support this hypothesis. To confirm this, we initially detected whether UBC32 was a component of the ERAD complex, and we analyzed its interaction with the putative Arabidopsis DOA10B  in vivo by performing a well-established Agrobacterium tumefaciensbased firefly luciferase complementation imaging assay in Nicotiana benthamiana. DOA10B and UBC32 were fused with N-terminal luciferase (NLuc) and C-terminal luciferase (CLuc) regions, respectively. A. tumefaciens strains that harbored CLuc-UBC32 and DOA10B-NLuc constructs and/or the vectors were mixed and infiltrated into different positions at the same leaf of N. benthamiana, as shown in Supplemental Figure  4 online. Only the combination of DOA10B-NLuc and CLuc-UBC32 showed LUC complementation, which suggested that UBC32 interacts with DOA10B in plants.
We then tested the stability of several known ERAD substrates in plants using either overexpression or knockdown of the UBC32 protein. It was reported that a series of mlo alleles, mutant forms of barley powdery mildew resistance o (MLO), with single amino acid replacements mostly generated destabilized mutant MLO forms that were specifically degraded by ERAD in yeast and plant protoplasts (Mü ller et al., 2005). We therefore generated a MLO-12 construct with a C-terminal 63MYC tag, which was reported to be rapidly degraded in plants. This construct was coinfiltrated with constructs that expressed the UBC32-GFP protein or a point-mutated form of UBC32 (C93S)-GFP to the same leaf area of N. benthamiana by agroinfiltration analysis for the in vivo protein degradation assay . To confirm that the mutation of C93S had no secondary effect on the cellular localization of UBC32, we checked the localization of both wildtype and mutated GFP fusion proteins in plant cells. Two days after infiltration, the fluorescence was observed by confocal laser scanning. As shown in Figure 3A, the UBC32-GFP protein exhibited typical ER localization, which was confirmed by colocalization with HDEL-RFP, an ER marker. The data from both transient expression and the stable transgenic plants (Figure 1C) showed that UBC32-GFP was localized to the ER. As shown in Figure 3A, UBC32 (C93S)-GFP was localized on the ER. This indicated that C93S had no secondary effect on the cellular localization of UBC32. Then, samples were collected for the detection of both the protein and RNA levels of transfected constructs. Overexpression of UBC32-GFP protein resulted in a reduced amount of MLO-12, whereas the overexpression of UBC32 (C93S)-GFP led to the increased accumulation of MLO-12 compared with infiltration with a vector control in repeated experiments ( Figure 3B). Furthermore, infiltration with the 26S proteasome inhibitor MG132 resulted in the accumulation of MLO-12 protein in the UBC32-GFP overexpression tobacco leaves ( Figure 3C). In this experiment, HA-GFP was detected as an internal control, and the mRNA levels of both GFP and MLO-12-Myc were analyzed by RT-PCR to ensure that equal amounts of MLO-12 were expressed in parallel coinfiltrations (Figures 3B and 3C). To understand the effects of UBC32 on the ubiquitination of MLO-12, we performed an in vitro ubiquitination assay. UBC32-GFP and MLO-12-Myc were transiently expressed separately in N. benthamiana and then the UBC32-GFP and MLO-12-Myc protein extracts were purified individually by immunoprecipitation with anti-GFP and anti-Myc antibodies, (A) Subcellular localization of UBC32-GFP and UBC32 (C93S)-GFP in transiently expressed tobacco leaves. HDEL-RFP is an ER localization peptide motif fused with red florescent protein (RFP). Bar = 10 mm. (B) UBC32 facilitates the degradation of MLO-12 in tobacco leaves. Different combinations of transformed Agrobacterium were coinfiltrated into tobacco leaves. Proteins from the tobacco leaves temporarily overexpressing the indicated constructs were processed for protein gel blot analysis (top four panels). Total RNAs from the same tobacco leaves were analyzed by RT-PCR (bottom two panels). (C) The effect of MG132 on MLO-12 degradation. MG132 was infiltrated into tobacco leaves 12 h before sample collection. Protein gel blot analysis (top two panels) and RT-PCR (bottom two panels) are as described in (B). (D) UBC32 enhanced the ubiquitination of MLO-12. UBC32-GFP extracts and MLO-12-Myc extracts were immunoprecipitated with anti-GFP and anti-Myc antibodies, respectively (left panels). The immunoprecipitated products were used in an in vitro ubiquitination assay using an anti-Ub antibody (right panel). E1 (from wheat) and 6x His-tagged ubiquitin (Ub) were added to the reaction as shown above the gel images. The numbers in the middle show the molecular masses of marker proteins in kilodaltons. ERAD in Salt Stress and BR Signaling 5 of 12 respectively. The immunoprecipitated products were analyzed in an in vitro ubiquitination reaction. We found that MLO-12 expressed in plants already had Ub-like modifications ( Figure  3D), and even in the reactions without the addition of UBC32, the increased ubiquitination of MLO-12 was observed. This might have been due to the immunoprecipitated MLO-12 that was already bonded with the tobacco UBC32 homolog(s).
In the presence of E1 and His-tagged ubiquitin (His-Ub), dramatically enhanced higher molecular mass species were detected by both the anti-Myc and anti-Ub antibodies when UBC32 was added into the reaction ( Figure 3D). Interestingly, we also detected that UBC32 itself was also ubiquitinated in these plants ( Figure 3D, right last lane). These data indicated that UBC32 contributed to the ubiquitination and degradation of MLO-12, and the degradation depended on the function of 26S proteasome.
To further analyze the function of UBC32 in the plant ERAD pathway, genetic analysis was performed with the mutant forms of BRI1, which are other known ERAD substrates. BRI1 is a membrane-localized BR receptor, and the BR signal transduction pathway is interrupted by BRI1 mutations (Wang et al., 2001). The point-mutated forms of BRI1, bri1-5 and bri1-9, have been reported to be retained in the ER by the ER quality control system as ERAD substrates (Hong et al., 2008(Hong et al., , 2009Liu et al., 2011). To determine whether UBC32 was involved in the degradation of bri1-9, we crossed the ubc32 mutant with bri1-9. The bri1-9 ubc32 double mutant partially suppressed the phenotype of bri1-9 and exhibited longer embryonic stems with dark, larger rosette leaves and longer floral stems at maturity than the bri1-9 plants (Figures 4A and 4B). We speculated that bri1-9 ubc32 double mutants reduced the degradation of bri1-9 because of the loss of UBC32,and that bri1-9 was sequentially transported to the plasma membrane where it functioned as the native BRI1. To test this hypothesis, we determined the protein level of bri1-9 in the bri1-9 ubc32 double mutant by protein gel blotting with an anti-BRI1 antibody. The bri1-9 protein accumulated at a considerably higher level in the bri1-9 ubc32 double mutant compared with the wild-type and bri1-9 plants ( Figure 4C). To confirm our speculation, a BES1 dephosphorylation assay was performed, and the results showed that epibrassinolide (eBL) treatment resulted in the partial dephosphorylation of BES1 in the bri1-9 ubc32 double mutant, whereas the phosphorylated form of BES1 in bri1-9 plants was mostly unaffected by eBL treatment. This (A) Phenotypes and hypocotyl lengths of 4-d-old dark-grown seedlings. Bar = 0.5 cm (left panel). Each bar represents the mean 6 SD of three independent repeats (n $ 30). Student's t test **P < 0.01 (right panel). WT, wild type. (B) Phenotypes of 3-week-old soil-grown (top panels, bar = 1 cm) or 7-week-old mature (bottom panel, bar = 10 cm) Col wild type, bri1-9, and bri1-9 ubc32 double mutants. (C) The accumulation of bri1-9 in Col wild type, bri1-9, and bri1-9 ubc32 double mutants. Tubulin was used as a loading control in the bottom panel.
(D) BR-induced changes in the phosphorylation status of BES1. Five-week-old seedlings of Col wild type, bri1-9, and bri1-9 ubc32 double mutants were treated with 10 mM eBL for 2 h in liquid half-strength MS medium. BES1 was detected by anti-BES1 antibody. Ribulose-1,5-bisphosphate carboxylase/ oxygenase (Rubisco) stained with Ponceau S was used as a loading control.

of 12
The Plant Cell indicated that at least a partial restoration of BR signaling had occurred in bri1-9 ubc32 double mutants ( Figure 4D). We additionally crossed bri1-5 with ubc32 mutants. The same phenomenon was observed in the bri1-5 ubc32 double mutants (see Supplemental Figure 5 online), although bri1-5 is a Wassilewskija ecotype and ubc32 is a Columbia (Col) ecotype. All mature bri1-5 ubc32 double mutants were taller than the homozygous bri1-5 in the F2 progenies of $ bri1-5 3 # ubc32, which indicated that the alternative phenotypes of bri1-5 ubc32 were due to that mutation of UBC32 rather than as result of heterosis (see Supplemental Figure 5F online). Taken together, these data demonstrated that UBC32 was an active ERAD component with significant roles in the plant ERAD system. UBC32 is involved in the plant ERAD process, and it is a homolog of Ubc6p. We then hypothesized that the function of UBC32 might be conserved with Ubc6p and considered whether UBC32 could complement the yeast Ubc6p mutant. We then expressed UBC32 in the yeast mps2-1 ubc6 strain. MPS2 in yeast encodes a membrane protein, and it is known that MPS2 is targeted to the ER degradation pathway. The Mps2-1p mutant protein level is markedly reduced compared with wild-type Mps2p, so yeast cannot grow at a nonpermissive temperature (378C). With mutations in ubc6p, the defect in Ubc6p restores the Mps2-1p level to nearly wild-type levels, which allows the yeast mps2-1 ubc6 strain to grow at 378C (McBratney and Winey, 2002; see Supplemental Figure 6 online). Yeast Ubc6p and UBC32 were cloned into the same vector by the same strategy and then transformed into the mps2-1 ubc6 strain. We observed the phenotype complementation by Ubc6p expression but not by UBC32 (see Supplemental Figure 6 online). The same phenomenon was also observed from the failure of complementation by two human ubc6 homologs in yeast (Lenk et al., 2002). The existence of two families in many metazoa and plants indicates that the function of the single Ubc6p found in yeast may be subsumed by two distinct membrane-bound ubiquitin-conjugating enzymes in higher eukaryotic cells (see Supplemental Figure 1C online; Lenk et al., 2002).

The Participation of UBC32 in ERAD Is Involved in BR-Mediated Salt Stress Response
BR and its derivatives have been used to protect economically important plants against environmental hazards. BR treatment has been reported to increase tolerance to a number of stresses, including salt stress, in Arabidopsis and Brassica napus (Kagale et al., 2007). We speculated that the enhancement of the endogenous BR signaling would also increase the salt tolerance of plant cells, whereas a reduction of BR signaling would produce the opposite phenotypes. As the UBC32 mutation positively regulated BR signaling through the accumulation of the bri1-9 protein (Figure 4), we assessed the salt tolerance capacity of bri1-9 and bri1-9 ubc32 double mutant plants. Our data showed that bri1-9 plants were more sensitive to salt in comparison to wild-type seedlings. Furthermore, bri1-9 ubc32 double mutants partially rescued the salt-sensitive phenotype of bri1-9 not only at the germination stage, but the effects endured throughout the life cycle (Figures 5A to 5D; see Supplemental Figure 7 online). The evidence from the genetic studies indicated that BR signal-ing is necessary for plants to become tolerant to salt. The above results suggested that the salt tolerance capacity of ubc32 mutants might be conferred by enhanced BR signaling.
To further analyze whether UBC32 affected BR signaling, we performed dose-response studies in wild-type, ubc32, and 35S-UBC32 plants with eBL. eBL is a type of commercial BR that can promote hypocotyl elongation in plants, and BL treatment reflects sensitivity of plants to BR. Mutants of UBC32 seedlings were more sensitive to exogenous eBL and had relatively longer hypocotyls, whereas 35S-UBC32 seedlings were less sensitive to exogenous eBL than wild-type plants, particularly when treated with a high concentration of eBL ( Figures 5E and 5F). These data indicated that UBC32 is involved in BR signaling, and the salt tolerance phenotype of ubc32 mutants is probably due to the enhanced BR signaling through the accumulation of BRI1 protein. In conclusion, our results showed that UBC32 affected the accumulation of certain misfolded proteins, such as MLO-12, bri1-9, bri1-5, and probably the natural protein BRI1, through its participation in the ERAD pathway; UBC32 was involved in the BR-mediated salt stress response through affecting the accumulation of BRI1 and appears to be the key factor that connects ERAD with plant growth and the stress response.

DISCUSSION
In this study, we demonstrated that the Arabidopsis ubiquitinconjugating enzyme UBC32, a stress-induced functional E2 localized on the ER membrane, connected the ERAD process and BR-mediated growth promotion and salt stress tolerance.
BRs are well known for their function in development, such as cell division and elongation (Clouse and Sasse, 1998). In this study, we found that BR signaling contributes to plant salt tolerance. The ubc32 mutants were more sensitive to eBL in the hypocotyl elongation experiment, which showed that UBC32 negatively regulated the BR signaling pathway. It is reported that BRI1 localizes to both the plasma membrane and early endosomal compartments and increasing the endosomal localization of BRI1 enhances BR signaling (Geldner et al., 2007). In mammalian cells, ;60 to 75% of the wild-type cystic fibrosis transmembrane conductance regulator was degraded during protein folding in the ER (Ward and Kopito, 1994). Thus, we speculated that the folding status of BRI1 may be similar to that of cystic fibrosis transmembrane conductance regulator. Our results suggested a possibility that the protein quality control of BRI1 in the ER was interrupted by the mutation of UBC32, leading to the increased accumulation of wild-type or structurally imperfect but biochemically functional BRI1 in the ER. This could then be transported to the plasma membrane by endosomal trafficking, which thus enhances BR signaling.
In our previous study concerning the Arabidopsis ERAD component HRD3A of the HRD complex, we found that the ERAD contributes to plant salt tolerance . In this study, we found that UBC32 is a functional ERAD component, and the ubc32 mutants showed salt-tolerant and Tm-tolerant phenotypes. We thought that the different responses to salt stress of ubc32 and hrd3a were due to the specificity to different substrates. Although UBC32 shared the common substrates of (A) Growth phenotypes of Col wild type (WT), bri1-9, and bri1-9 ubc32 double mutant seedlings germinated on half-strength MS medium with (top panels) or without (bottom panels) 125 mM NaCl. Photographs of the seedlings were taken 4 d after germination. Bar = 0.1 cm. (B) Percentages of Col wild type, bri1-9, and bri1-9 ubc32 double mutant seedlings germinated (radical emergence) on 125 mM NaCl were normalized in each line by the germination rate on half-strength MS control plates. Each bar represents the means 6 SD of three independent experiments (n = 50). Student's t test **P < 0.01. (C) Growth phenotype of Col wild type, bri1-9, and bri1-9 ubc32 double mutant seedlings grown on half-strength MS medium with (top panels) or without (bottom panels) 150 mM NaCl for 2 weeks. Bar = 0.5 cm.
(D) The percentages of different phenotypes in (C). One representative image of three independent experiments is shown (n = 50). (E) and (F) Hypocotyl elongation in response to eBL. (E) Phenotypes of seedlings grown on vertical plates with 10 mM eBL or the same volume of 80% ethanol (CK) for 4 d at 228C under long-day conditions (16 h light/8 h darkness, 30 mmol m À2 ·s À1 white light). Bar = 0.5 cm.
(F) Hypocotyl lengths of seedlings grown on plates with different eBL concentrations. Results are the means 6 SD of three independent repeats (n $ 30). Student's t test *P < 0.05 and **P < 0.01.

of 12
The Plant Cell bri1-9 and bri1-5 with HRD3A, the substrate spectrum of UBC32 may have its own features. For example, UBC32 contributed to the degradation of MLO-12, which has not been checked in hrd3a mutants. Furthermore, UBC32 contributed to the degradation of BRI1, and the ubc32 mutants and 35S-UBC32 plants showed an altered phenotype to eBL. However, this eBL response was not checked in the hrd3a plants. Considering that the yeast homolog genes of UBC32 and HRD3A belong to different ERAD complexes (Doa10 and Hrd1 complexes), and the Hrd1 complex mainly recognize ERAD-L substrates, while Doa10 mainly contributes to the ERAD-M and ERAD-C substrates (Kostova et al., 2007), the diversification of the substrate ranges of UBC32 and HRD3A is reasonable. From another aspect, there are already some examples that ERAD may lead to favorable effects for organisms under specific situations. For example, mutations in yeast ubc6 suppress the temperature-sensitive phenotype of the sec61 mutant, which leads to the survival of the sec61 mutant at the normally restrictive temperature (Sommer and Jentsch, 1993). Based on this phenomenon, we speculated that it might be important for the cell to maintain the homeostasis of ER function. Tm treatment should dramatically inhibit the glycosylation of many important glycoproteins that control the growth and development of plants and induce severe ER stress. Thus, ERAD defects such as the mutation of UBC32 could compromise ER stress by preventing the degradation of certain structurally imperfect but functional proteins and allowing them to be transported to their destinations. This is the case of bri1-5 and bri1-9 and results in the tolerance of ubc32 mutants to Tm.
Moreover, the activation of UPR enhances the protein folding capacity in the ER, which, combined with the protein degradation functions of ERAD, conditions plant responses to different stresses. On the other hand, the overexpression of UBC32 might lead to the overzealous ERAD degradation of functional proteins and explain the observation that 35S-UBC32 plants showed growth retardation with shorter roots and smaller aerial parts under standard growth conditions and that the 35S-UBC32 plants showed a sensitive phenotype upon Tm treatment. A similar mechanism may also exist in the salt stress response. We also observed that the lower expression of 35S-UBC32 only contributed a small component to the phenotype, and this may indicate that a certain amount of UBC32 in plants is necessary to optimize the function of the ERAD pathway. Furthermore, our results suggested that enhanced BR signaling was responsible for the development of salt tolerance in ubc32 mutants. Our studies on UBC32 provide a better understanding of the mechanism used by plants to maintain homeostasis under different stress conditions.
The duplication of UBC genes, from one copy, Ubc6p, in yeast to multiple copies in metazoa and plants, and the failure of complementation by both human homologs (Lenk et al., 2002) and UBC32 in yeast suggested that the function of single yeast Ubc6p in higher eukaryotic cells is split between two subfamilies, UBC32/ UBE2J1 and UBC33/UBC34/UBE2J2. Thus, the study of the double or triple mutants, such as ubc32 ubc33, ubc32 ubc34, and ubc32 ubc33 ubc34, will enable further understanding of the role of this membrane-spanning domain group of UBC proteins in Arabidopsis. The coexpression of UBC32 with UBC33 or UBC34 in the yeast ubc6p mutant can also explain the evolutionary relationship between yeast and plant Ubc6p homologs.
It is known that ER stress signals will activate ER-localized membrane-associated bZIP transcription factors, such as bZIP17, bZIP28, and bZIP60 in Arabidopsis. bZIP17 and bZIP28 transduce stress signals from the ER to the nucleus during UPR by transporting those bZIPs to the Golgi apparatus so they can be processed by Golgi-localized proteases. Additionally, the translocation of the Arabidopsis bZIP60 to the nucleus has been found to be regulated by the nonconventional splicing of Arabidopsis bZIP60 mRNA by IRE-1 (Deng et al., 2011). The processed forms of bZIPs are imported into the nucleus to activate stress response genes, such as chaperones (Liu and Howell, 2010). The increase in the expression levels of the ER chaperones may play a role in the development of increased salt tolerance. It has been reported that BR signaling might be influenced by the capacity of the Golgi bodies to perform regulated intramembrane proteolysis, suggesting a link between ER stress and BR signaling (Che et al., 2010). However, our study provides evidence that BR signaling is also directly regulated by the ERAD mechanism, which opens the question of whether ERAD and regulated intramembrane proteolysis function together or independently. This would be accordingly addressed by an analysis of the genetic interactions between ERAD and UPR components.
BRs may also enhance abiotic stress tolerance through the interaction with other plant hormones, such as ABA (Divi et al., 2010). For example, it has been reported that the crosstalk between BR and ABA occurs after BR perception but at or before BIN2, so a large portion of BR responsive genes are also regulated by ABA (Zhang et al., 2009a). The reported data show that ABA plays very important role in the establishment of seed dormancy capacity during embryonic maturation and the inhibition of seed germination, while BRs promote seed germination, perhaps through the enhancement of the growth ability of embryos to antagonize the effects of ABA (Zhang et al., 2009b). Our data showed that the bri1-9 ubc32 double mutants are more insensitive to NaCl-induced germination inhibition compared with bri1-9 and indicated another possible mechanism by which BR and ABA interact through the modulation of the amount of BRI1 by the ERAD mechanism. This possibility was further supported by the alternative ABA sensitivity of both germination and postgermination growth of the ubc32 and UBC32 overexpressors. Even with our current data, we still could not exclude the possibility of crosstalk between ABA and BR in the ERAD-regulated plant salt response. This hypothesis can be addressed in the future by investigations into the genetic interactions between BR, ABA, and ERAD mutants. Further studies on other ERAD components in Arabidopsis will also facilitate a better understanding of the correlation of the ERAD mechanism with different physiological processes in plants.

Plant Materials and Growth Conditions
Arabidopsis thaliana ecotype Col-0 was the parental line for mutants and transgenic plants with the following exception: bri1-5 (Wassilewskija [Ws-2]) was used for genetic analysis. The T-DNA insertion mutants of ubc32-1 (SALK_055050) and ubc32-2 (SALK_082711) were obtained ERAD in Salt Stress and BR Signaling 9 of 12 from the ABRC (The Ohio State University, Columbus, OH). Methods for seed sterilization and conditions for plant growth were described previously  with the slight modification that half-strength MS medium was used instead of MS medium. The half-strength MS medium was supplemented with 1.5% Suc and with NaCl, mannitol, KCl, Tm, DTT, and AZC as needed.

Subcellular Localization
For the subcellular localization, the UBC32-GFP fusion construct was put under the control of the 35S promoter and the NOS terminator. Roots of 1-week-old seedlings were stained with DAPI (10 mg/mL; Sigma-Aldrich) or the ER marker ER-tracker Red dye (1 mM; Invitrogen) and visualized using a laser confocal microscope (Leica). To minimize the crosstalk between the partially overlapping emission spectra when DAPI or ER marker staining was combined with GFP, the sequential scanning mode was used. The plasmolysis of the root cell was induced by 0.8 M mannitol treatment for 5 min in order to separate the cell wall and plasma membrane.

E2 Activity and in Vitro Ubiquitination Assays
The 63His UBC32DTM fusions were expressed in Escherichia coli strain BL21 (DE3) and purified by nickel-nitrilotriacetic acid agarose (Qiagen). Thioester assays were performed in a total reaction volume of 30 mL, consisting of 50 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , 10 mM ATP, 100 ng wheat (Triticum aestivum) E1 (Gene ID: 136632), 2 mg of recombinant UBC32, and 10 mg His-tagged Arabidopsis ubiquitin (UBQ14). Reactions were split after incubation for 5 min at 378C and terminated by SDS sample buffer with DTT or 8 M urea sample buffer without DTT and boiled for 10 min. Finally, the reaction products were separated by SDS-PAGE and detected by nickel-horseradish peroxidase. The in vitro ubiquitination assay was performed as described by . The immunoprecipitation product of UBC32-GFP and MLO-12-Myc were used for an in vitro ubiquitination assay. Reactions were performed at 308C with agitation in an Eppendorf Thermomixer for 1.5 h. For the immunoblots, the anti-Myc, anti-GFP, and antiubiquitin antibodies described below were used to detect the corresponding proteins.

Tobacco Infiltration Assay
Tobacco (Nicotiana benthamiana) infiltration assay was performed as described previously . Leaves were harvested days after infiltration and ground into powder with liquid nitrogen for protein gel blot assay or RT-PCR assay. The UBC32 (C93S) and MLO-12 mutants were prepared using the QuickChange site-directed mutagenesis kit (Stratagene) according to the protocol provided by the manufacturer.

Transformation Vectors and Construction of Transgenic Plants
To produce 35S-UBC32 plants, a 963-bp fragment containing the UBC32 cDNA was cloned into the vector pCambia1300-221-HA. The 35S-UBC32-GFP construct was created by cloning the entire coding region of UBC32 with GFP into binary vector pCambia1300-221-HA. The 35S-UBC32 (C93S)-GFP construct was created by site-directed mutagenesis (primer sequences in Supplemental Table 1 online). For the UBC32 promoter and GUS fusion construct, a 59 flanking sequence (a 2025-bp promoter region upstream of the ATG start codon of UBC32) was amplified from genomic DNA and verified by sequencing, and then the fragment was cloned into the binary vector pCambia1300-221 to obtain a transcriptional fusion of the UBC32 promoter and the GUS coding sequence. Transformation of Arabidopsis was performed by the vacuum infiltration method (Bechtold and Pelletier, 1998) using the Agrobacterium tumefaciens strain EHA105. For the phenotypic analysis, T3 or T4 homozygous lines were used. Two independent lines of homozygous T4 plants containing a single insertion of each construct were used for the detailed analysis.

RT-PCR and Quantitative PCR Amplification
To examine the expression of UBC32 by RT-PCR, DNase I-treated total RNA (2 mg) was denatured and subjected to reverse transcription using Moloney murine leukemia virus reverse transcriptase (200 units per reaction; Promega) at 428C for 60 min followed by heat inactivation of the reverse transcriptase at 708C for 15 min. Quantitative PCR was performed using the CFX96 real-time system (Bio-Rad). Three biological replicates were performed. Gene expression was quantified at the logarithmic phase using the expression of the housekeeping ACT7 gene as an internal control. Primer sequences are given in Supplemental Table 1 online.

GUS Bioassays
Young seedlings after germination, and different parts from mature transgenic plants, were collected and used for histochemical detection of GUS expression. Materials were stained at 378C overnight in GUS staining solution. To test the induction of GUS expression by salt and drought, 12-d-old transgenic seedlings were transferred from agar plates to half-strength MS liquid medium containing different concentrations of NaCl, mannitol for salt stress and osmotic stress treatment, and AZC for ER stress treatment or to a filter exposed in the air with 70% RH for drought treatment. The treated and control transgenic seedlings were stained in 1 mg/mL X-Gluc, 0.03% Triton X-100, and 20 mM HEPES buffer, pH 7.0, for 2 h for histochemical detection.

Luciferase Complementation Imaging Assay
Luciferase complementation imaging assays were performed as described by Chen et al. (2008). DOA10B and UBC32 were amplified by PCR. Primer sequences are given in Supplemental Table 1 online. The PCR products of DOA10B and UBC32 were digested with KpnI or SalI and inserted into pCambia-NLuc and pCambia-CLuc separately. All the constructs were transformed into the A. tumefaciens strain, EHA105. An equal volume of A. tumefaciens harboring pCambia-NLuc and pCambia-CLuc (or their derivative constructs) was mixed to a final concentration of OD 600 = 1.5. Four different combinations of A. tumefaciens were infiltrated into four different positions at the same leaves of N. benthamiana. Plants were grown at 238C and allowed to recover for 3 d. A low-light cooled charge-coupled device imaging apparatus (NightOWL II LB983 with indiGO software) was used to capture the LUC image.

Yeast Complementation
The yeast strains mps2-1 and mps2-1 ubc6 were used in this work. The coding sequences of Ubc6p and UBC32 were amplified by PCR and inserted into pYES2C individually. The pYES2C was a modified form of plasmid pYES2 (Invitrogen), with the GAL1 promoter replaced by the CYC1 promoter. The empty plasmid and constructs were transformed into the corresponding yeast strains. The preparation of yeast media and protocol of transformed yeast cells were performed as described previously (Xie et al., 2002).

Protein Extraction and Protein Gel Blot Analysis
Two-week-old seedlings grown on half-strength MS media were ground in liquid nitrogen and extracted with 23 SDS buffer (Yan et al., 2009). Crude extracts were separated by SDS-PAGE and transferred to nitrocellulose membranes. The membranes were stained with 0.2% Ponceau S. Antibodies to GFP and c-Myc were purchased from Santa Cruz Biotechnology. Anti-ubiquitin was produced by our laboratory .

of 12
The Plant Cell

Quantification of Hypocotyl Length
Hypocotyl length of dark-grown seedlings and eBL-treated seedlings were quantified by ImageJ software (http://rsbweb.nih.gov/ij; National Institutes of Health).

Supplemental Data
The following materials are available in the online version of this article. Supplemental Table 1. Oligonucleotide Primers Used in This Study.