Deubiquitination of BES1 by UBP12/UBP13 promotes brassinosteroid signaling and plant growth

As a key transcription factor in the brassinosteroid (BR) signaling pathway, the activity and expression of BES1 (BRI1-EMS-SUPPRESSOR 1) are stringently regulated. BES1 degradation is mediated by ubiquitin-related 26S proteasomal and autophagy pathways, which attenuate and terminate BR signaling; however, the opposing deubiquitinases (DUBs) are still unknown. Here, we showed that the ubp12-2w/13-3 double mutant phenocopies the BR-deficient dwarf mutant, suggesting that the two DUBs UBP12/UBP13 antagonize ubiquitin-mediated degradation to stabilize BES1. These two DUBs can trim tetraubiquitin with K46 and K63 linkages in vitro. UBP12/BES1 and UBP13/BES1 complexes are localized in both cytosol and nuclei. UBP12/13 can deubiquitinate polyubiquitinated BES1 in vitro and in planta, and UBP12 interacts with and deubiquitinates both inactive, phosphorylated BES1 and active, dephosphorylated BES1 in vivo. UBP12 overexpression in BES1OE plants significantly enhances cell elongation in hypocotyls and petioles and increases the ratio of leaf length to width compared with BES1OE or UBP12OE plants. Hypocotyl elongation and etiolation result from elevated BES1 levels because BES1 degradation is retarded by UBP12 in darkness or in light with BR. Protein degradation inhibitor experiments show that the majority of BES1 can be degraded by either the proteasomal or the autophagy pathway, but a minor BES1 fraction remains pathway specific. In conclusion, UBP12/UBP13 deubiquitinate BES1 to stabilize the latter as a positive regulator for BR responses.


INTRODUCTION
Brassinosteroids (BRs) play essential roles in diverse plant developmental processes, including hypocotyl and petiole elongation and leaf expansion (Nolan et al., 2020;Oh et al., 2020). Major components of the BR signaling pathway have been identified Lv and Li, 2020). The signaling cascade is initiated by the perception of BRs by the receptor-like kinase BRASSINOSTEROID-INSENSITIVE 1 (BRI1) on cell membranes, which phosphorylates and activates the co-receptor BRI1-ASSOCIATED KINASE 1 (BAK1). The signals are fine-tuned by inhibition of the glycogen synthase kinase 3-like kinase BRASSI-NOSTEROID-INSENSITIVE 2 (BIN2) and activation of the transcription factors BRI1-EMS-SUPPRESSOR 1 (BES1) and BRAS-SINAZOLE RESISTANT1 (BZR1). Eventually, BES1 and BZR1 directly regulate the transcription of thousands of downstream genes in response to various environmental conditions, such as light, drought, temperature, nutrients, and immune response toward pathogens (Sun et al., 2010;Yu et al., 2011;Li et al., 2018;Lv and Li, 2020;Nolan et al., 2020;Oh et al., 2020).
As key transcription factors for BR signaling, it is not surprising that levels of BES1/BZR1 proteins are finely regulated. Depending on external stimuli and during plant growth transition, various ubiquitin E3 ligases such as COP1 (Kim et al., 2014), MAX2 (Wang et al., 2013b), SINATs , and PUB40 (Kim et al., 2019) are known to polyubiquitinate and destabilize these factors by preparing them for 26S proteasomal degradation (Callis, 2014). Polyubiquitinated BES1 can also be degraded via selective autophagy Wang et al., 2021a). In general, polyubiquitinated proteins can be rescued from destruction by deubiquitinating enzymes (deubiquitinases [DUBs]), which remove mono-or polyubiquitin from target proteins, thereby increasing the stability of the latter. Such dynamic ubiquitination and deubiquitination reactions are important for fine-tuning signaling pathways. However, deubiquitination enzymes specific for BES1/ BZR1 have not yet been identified.
Twenty-seven out of 64 Arabidopsis DUBs have been categorized as ubiquitin-specific proteases (UBPs), and all of them contain a conserved cysteine residue in their catalytic domain (Yan et al., 2000;March and Farrona, 2018). As members of a small UBP subfamily, UBP12 and UBP13 show 91% amino acid similarity and are functionally redundant. In addition to the catalytic domain, these two UBPs contain a conserved MATH (meprin and tumor necrosis factor receptor-associated factor homology) domain at their N terminus, which mediates protein-protein interactions (Cui et al., 2013;Eletr and Wilkinson, 2014). UBP12 and UBP13 have been implicated in diverse biological functions, including plant immune response (Ewan et al., 2011), onset of flowering (Cui et al., 2013), shade avoidance responses (Zhou et al., 2021), silencing of polycomb group protein genes (Derkacheva et al., 2016), jasmonic acid (JA) response (Jeong et al., 2017), root growth , leaf senescence under nitrogen deficiency , circadian clock , and leaf size (Vanhaeren et al., 2020). Our previous study showed that ubp12-2w/ubp13-3 double mutant plants (hereafter called ubp12-2w/13-3) are dwarf with small curved leaves and short petioles . Because these phenotypes are typical of BR signaling and biosynthesis-deficient mutants, we hypothesized that UBP12/UBP13 might promote BR signaling. Here, we show that UBP12/UBP13 can indeed deubiquitinate BES1 to increase the stability of this transcription factor during plant cell elongation triggered by BR.

RESULTS
Mutant plants deficient in UBP12/UBP13 exhibit BR hyposensitivity UBP12 and UBP13 contain the MATH domains and the catalytic UBP domain. The null double mutant ubp12/ubp13 shows stunted growth and a pleiotropic phenotype. It is either infertile or rarely fertile, setting only a few seeds (Cui et al., 2013). Using Western blots, we analyzed protein levels of endogenous  and in UBP12-and UBP13-overexpressing plants (UB-P12 OE and UBP13 OE ) using an anti-UBP12 antibody. We also compared relative gene expression levels of UBP12 and UBP13 in these plants (Supplemental Figure 1). We confirmed that ubp12-2w/13-3 produced the truncated protein UBP12DC, probably corresponding to a C-terminal truncation of UBP12, whereas UBP12 OE and UBP13 OE plants showed increased production of UBP12 and UBP13.
Preliminary observations showed that ubp12-2w/13-3 plants phenocopied mutants deficient in BR biosynthesis or BR signaling. Wang et al. (2013a) previously used liquid chromatographymass spectrometry and yeast-two hybrid experiments to show that UBP12 interacts with BZR1, which is functionally redundant with BES1. To explore a possible connection between UBP12/ UBP13 and BR signaling, we analyzed cell elongation in ubp12-2w/13-3 as well as plants overexpressing UBP12 and UBP13. Compared with the wild type (WT), hypocotyl length, petiole length, and the ratio of leaf length to width were significantly decreased in ubp12-2w/13-3 but slightly increased in UBP12 OE and UBP13 OE (Supplemental Figure 2A-2E). Expression levels of BR-responsive genes paralleled the observed phenotypes (Supplemental Figure 2F). In ubp12-2w/13-3 plants, CONSTITUTIVE PHOTOMORPHOGENIC DWARF (CPD) and C-22 hydrolase (DWF4), whose expression is repressed by activated BR signaling, were expressed at higher levels compared with WT levels under normal conditions. By contrast, two BR-inducible genes, SMALL AUXIN UP RNA 1 FROM ARABIDOPSIS THALI ANA ECOTYPE COLUMBIA (SAUR-AC1) and ACC SYNTHASE 5 (ACS5), were expressed at a lower level under BR treatment. These results indicate that BR signaling is attenuated in ubp12-2w/13-3 plants.
Because ubp12-2w/13-3 showed reduced BR sensitivity, we compared its morphological and molecular phenotypes with those of plants with reduced BES1 expression (BES1-RNAi; Yin et al., 2005). qRT-PCR analysis showed that BES1 transcript levels in the BES1-RNAi plants were about 40% of WT values (Supplemental Figure 3). Figure 1A shows that 12-day-old seedlings of ubp12-2w/13-3 exhibited a phenotype similar to that of BES1-RNAi seedlings. Hypocotyl elongation of ubp12-2w/13-3 seedlings was hyposensitive to 24-epibrassinolide (eBL), an active form of BR, but hypersensitive to brassinazole (Brz), a BR biosynthesis inhibitor ( Figure 1B and 1C). WT seedlings showed a 94% increase in hypocotyl length upon eBL treatment, whereas the comparable number in ubp12-2w/13-3 was only 9%, very similar to that obtained with BES1-RNAi. With Brz treatment, hypocotyl lengths of WT seedlings were decreased by 61% compared with the non-treated group, but ubp12-2w/13-3 seedlings showed a 73% inhibition of hypocotyl growth like BES1-RNAi seedlings. Time-course analysis of BRinducible genes showed that SAUR-AC1 and ACS5 transcript levels were increased in WT but not in ubp12-2w/13-3 under the same conditions. On the other hand, expression of the BRrepressed genes CPD and DWF4 was 1.5-2 times higher in ubp12-2w/13-3 compared with WT, but the expression levels declined rapidly in both genotypes during BR treatment. Control experiments showed no change in BES1 and UBP12 expression levels in response to BR treatment ( Figure 2D). These results showed that the BR insensitivity could be attributed to UBP12 and UBP13 deficiency, suggesting a positive role for these two DUBs in BR responses.
UBP12/UBP13 bind to BES1 in vitro, and UBP12 interacts with both phosphorylated and dephosphorylated BES1 in plants To investigate possible targets of UBP12/UBP13, we performed in vitro pull-down assays using purified proteins. Maltose-binding protein (MBP)-BES1-myc was pulled down by GST and GSTtagged full-length UBP12/UBP13 (Full) and by derivatives carrying the N-and/or C-terminal region (N and/or C). Figure 2A shows that UBP12 and UBP13 could directly interact with BES1. Further analysis of UBP12 derivatives showed that BES1 was specifically bound to the N-terminal region, which contains the MATH domain and the cysteine/histidine-containing catalytic domain. By dividing the BES1 protein into nine peptide fragments, we identified the BES1 fragment containing amino acids 99-164 as the region that interacts with UBP12 and UBP13 (Supplemental Figure 4).
We first examined the UBP12-BES1 interaction in plants by semiin vivo pull-down assays using 35S:BES1-myc plants treated with or without eBL. pBES1-myc (phosphorylated BES1-myc) and BES1-myc (dephosphorylated BES1-myc) in protein extracts were pulled down with GST resin-bound UBP12. Both forms of BES1 showed strong interaction with GST-UBP12 but not with GST ( Figure 2B). Under normal conditions, UBP12 interacted with both pBES1-myc and BES1-myc. The addition of BR facilitated BES1 dephosphorylation and accumulation of BES1-myc. Next, we performed co-immunoprecipitation (coIP) assays using extracts derived from 35S:BES1-myc/UBQ10:UBP12-HA seed- (A) ubp12-2w/13-3 plants show growth retardation like BES1-RNAi plants. WT,, and BES1-RNAi plants were grown on MS medium for 12 days under normal growth conditions. Scale bars correspond to 1 cm. (B and C) Hypocotyl length (mm) of WT, ubp12-2w/ 13-3, and BES1-RNAi grown on MS medium containing 0.5 mM epi-brassinolide (eBL) under normal light (100 mmol m À2 s À1 ) (B) and 0.8 mM brassinazole (Brz) under dim light (25 mmol m À2 s À1 ) (C) for 7 days. Fifteen independent biological samples (n = 15) were analyzed. *P < 0.05, **P < 0.01 (two-tailed t-test). (D) RNA expression dynamics of the BRresponsive genes SAUR-AC1, ACS5, CPD, and DWF4 as well as BES1 and UBP12. All samples were collected from 7-day-old seedlings of WT and ubp12-2w/13-3 treated with 1 mM of eBL for 1, 2, 4, and 6 h. Expression levels of WT were set to 1. All qRT-PCR experiments were performed in triplicate. Average values of three independent biological samples (n = 3) are presented with standard deviations. ACT2 expression levels were used as a control for normalization. lings to examine the UBP12-BES1 interaction in vivo. pBES1-myc/BES1-myc and UBP12-HA proteins were immunoprecipitated with a-HA and a-myc antibodies, respectively, and the immunoprecipitates were analyzed by immunoblotting ( Figure 2C). Figure 2C shows that UBP12-HA immunoprecipitates contained predominantly pBES1-myc and BES1-myc in the absence and presence of BR, respectively. Furthermore, BES1-myc immunoprecipitates also contained UBP12-HA. Bimolecular fluorescence complementation (BiFC) assays in tobacco leaves showed that BES1 interacted with full-length UBP12/ UBP13 and their N-terminal derivatives, and the complex was detected in both the cytosol and nuclei under normal conditions ( Figure 2D). BZR1, functionally redundant with BES1, also interacted with UBP12 and UBP13 in vitro, and these associations were confirmed in vivo by BiFC analysis (Supplemental Figure 5). Collectively, our results showed a direct association of UBP12/UBP13 with BES1 and its homolog BZR1. Moreover, UBP12/UBP13 strongly bind to active BES1 in plants treated with BR.

UBP12/UBP13 can deubiquitinate poly-ubiquitinated BES1 in vitro
The UBP12-BES1 interaction suggested that BES1 may be a substrate of UBP12. To examine this hypothesis, we explored whether UBP12 could function as a BES1 DUB in vitro. Because SINAT2 has been shown to polyubiquitinate BES1 , we used SINAT2 as an E3 ligase to produce polyubiquitinated BES1. GST-UBP12 and GST-UBP13 were treated with poly-ubiquitinated MBP-BES1-myc substrate for 0, 2, and 4 h. Figure 3A shows that UBP12 and UBP13 could indeed deubiquitinate polyubiquitinated BES1 in vitro. UBP12/ UBP13 also interacted with SINAT2 in vitro and in vivo, but these two DUBs could not remove ubiquitin from polyubiquitinated SINAT2, indicating that this E3 ligase is not a target (Supplemental Figure 6).
In another experiment, full-length UBP12 and three different derivatives were co-incubated with MBP-BES1-myc substrate, enzymes E1, E2, and E3, and ubiquitin ( Figure 3B). The three UBP12 derivatives were a partially inactive form of UBP12 in which cysteine 208 was substituted with serine (GST-UBP12_C208S) (Jeong et al., 2017), the N-terminal region of UBP12 (amino acids 1-541) (GST-UBP12-N), and the C-terminal region of UBP12 (amino acids 542-1116) (GST-UBP12-C) (Figure 2A). GST protein was used as a negative control (GST). The deubiquitinating enzymes and derivatives were added in increasing amounts. Figure 3B shows that UBP12 deubiquitination activity was significantly reduced by an active site mutation (C208S) in the enzyme. Interestingly, the N-terminal region of UBP12, which still contains the active site, did not show any enzyme activity, suggesting that the (A) UBP12/UBP13 and BES1 interaction in vitro. MBP-BES1-myc was pulled down by GST and GST-tagged full-length UBP12/UBP13 (Full) and derivatives carrying the N-and/or C-terminal regions (N and/or C). The upper panel shows a schematic diagram of UBP12, and numbers indicate the amino acid positions. MATH, meprin and tumor necrosis factor receptor-associated factor homology domain; catalytic domain (called UBP in Supplemental Figure 1), ubiquitin-specific protease domain containing Cys and His motifs; Ubl, ubiquitin-like motifs. Bands were detected by immunoblots using anti-myc. Membrane (mb) was stained with Coomassie brilliant blue. About 2 mg of each protein was used in all experiments. The asterisks mark resinbound bait proteins. MBP, maltose-binding protein; GST, glutathione-S-transferase; kDa, kilodalton. The positive control (PC) sample contained 50 ng MBP-BES1-myc. (B) Semi-in vivo UBP12 and BES1-myc interaction. Seven-day-old BES1-myc-overexpressing plants (35S:BES1-myc) were treated with 1 mM epibrassinolide (eBL) for 6 h. BES1-myc from the total protein extracts was pulled down by resin-bound GST or GST-UBP12. The bands were detected by western blotting with anti-myc. GST, GST-UBP12, and total input proteins stained with Coomassie brilliant blue were used to show equivalent loading (mb). (C) In vivo interaction of UBP12/UBP13 and phosphorylated/dephosphorylated BES1 (pBES1/BES1) in transgenic Arabidopsis thaliana (Col-0) plants expressing 35S:BES1-myc/UBPQ10:UBP12-HA. Seven-day-old seedlings were treated with 50 mM MG132 in the presence or absence of 1 mM epibrassinolide (eBL) for 6 h. The pBES1-myc/BES1-myc and UBP12-HA proteins were immunoprecipitated with a-HA and a-myc antibodies, respectively. -Ab indicates the negative control without any antibodies. pBES1-myc/BES1-myc and UBP12-HA proteins in the immunoprecipitates were analyzed by immunoblots using a-myc and a-HA, respectively. Input proteins are shown. Levels of b-actin (Actin) and RbcL (the large subunit of ribulose-1,5-bisphosphate carboxylase) were used as loading controls. (D) Interaction between UBP12/UBP13 and BES1 in tobacco (Nicotiana benthamiana) leaves assayed by bimolecular fluorescence complementation (BiFC). nYFP-BES1 was transiently co-expressed with UBP12-cYFP or UBP13-cYFP in leaves of N. benthamiana infiltrated with Agrobacterium. N termini of UBP12 and UBP13 were used as positive controls, whereas the C termini of the same proteins were used as negative controls. Localization is shown by merging YFP, mCherry-NLS (nuclear localization sequence), and DIC images (Merged). Confocal microscopy images were taken 3 days after infiltration. DIC, differential interference contrast. Scale bars correspond to 50 mm. Western blots and BiFC assays were analyzed in three independent experiments, and images of a representative set are shown. C-terminal region, which comprises five ubiquitin-like (UBl) motifs ( Figure 2A), may modulate the activity of DUB (Kim et al., 2016) through an allosteric effect.
Ubiquitin K63 linkages have been implicated in receptor cycling via endocytosis as well as in autophagy (Erpapazoglou et al., 2014;Romero-Barrios and Vert, 2018). A recent study reported that UBP13 can cleave the K63-linked polyubiquitin chain from BRI1 (Luo et al., 2022). Therefore, we examined the substrate specificity of UBP12/UBP13 using K48 and K63 tetra-ubiquitin as substrates. Figure 3C shows that UBP12/UBP13 actively cleave K48-as well as K63-linked ubiquitin and that the cysteine mutation at the active site blocked these activities.
We used a cell-free degradation assay to compare the fate of the recombinant protein MBP-BES1-myc in extracts prepared from plants of various genotypes: ubp12-2w, ubp12-2w/13-3, and UBP12 OE . MBP-BES1-myc was most rapidly degraded in ubp12-2w/13-3 extracts compared with ubp12-2w and WT extracts, but MBP-BES1-myc levels were unaffected in extracts of UBP12 OE plants ( Figure 3D). Taken together, our results show that the half-life of BES1 can be prolonged by UBP12 and UBP13, which remove ubiquitin from BES1.

UBP12 deubiquitinates both phosphorylated and dephosphorylated BES1 in plants
So far, our results showed that UBP12 bound to both pBES1 and BES1 in coIP assays and that UBP12/UBP13 could deubiquitinate polyubiquitinated BES1 in vitro. To see whether the phosphorylation status of BES1 would influence UBP12-mediated deubiquitination in plants, we performed in vivo ubiquitination assays ( Figure 4). These assays were performed using WT, BES1 OE , BES1 OE /UBP12 OE , and BES1 OE /ubp12-2w/13-3 treated with or without eBL. Ubiquitinated BES1-myc proteins were immunoprecipitated with TUBE (Tandem Ubiquitin Binding Entities) magnetic beads, and Ub n -BES1-myc proteins in the immunoprecipitates were analyzed by immunoblots. Figure 4 shows that eBL treatment converted pBES1-myc into BES-myc. Both pBES1myc and BES1-myc protein levels in BES1 OE , BES1 OE /UBP12 OE , and BES1 OE /ubp12-2w/13-3 plants increased with increasing UBP12 levels. Polyubiquitinated-BES1-myc levels were lower in BES1 OE /UBP12 OE but higher in BES1 OE /ubp12-2w/13-3 compared with those in BES1 OE . Together, these results support the notion that UBP12 is active toward both pBES1 and BES1, suggesting that the phosphorylation status of BES1 does not significantly affect the deubiquitination activity of UBP12. Because pBES1 is initially localized in the cytosol, our results also suggest that UBP12 is active in this cellular compartment as well.

UBP12 promotes plant growth by maintaining BES1 levels in BR signaling
We analyzed morphological and molecular phenotypes in WT, and UBP12 OE /BES1-RNAi plants to assess whether UBP12/ UBP13 could stabilize BES1 in vivo. We generated different plant genotypes by Agrobacterium-mediated transformation ( Figure 5). BES1-RNAi/ubp12-2w/13-3 was obtained by genetic crosses of BES1-RNAi and ubp12-2w/13-3 plants. We used UBP12 as a representative of UBP12/UBP13 because plants overexpressing UBP12 alone exhibited a significant phenotype (Supplemental Figure 2). Figure 5 shows that hypocotyl and petiole lengths, the ratio of leaf length to width, and leaf area were markedly decreased in BES1-RNAi, ubp12-2w/13-3, and BES1-RNAi/ ubp12-2w/13-3 compared with the WT.  Figure 8). In both det2-1 and BES1-RNAi plants, the overexpression effect of UBP12 OE was mild, whereas in the WT background (UBP12 OE ), overexpression of UBP12 was able to promote growth resulting from BR signaling mediated by BES1 and its homologs (Supplemental Figure 2). These results indicate that the growthpromoting properties of UBP12 require BR and its major transcription factors BES1 and homologs.
The dominant mutant bes1-D harbors a point mutation (P233L) in the BES1 PEST domain. This mutant displays BR signal enhancement, including longer hypocotyls and petioles and curved leaf shapes. We found that bes1-D also interacted with UBP12/ UBP13 in vitro (Supplemental Figure 7). Plants overexpressing 35S:bes1-D-YFP in ubp12-2w/13-3 exhibited only partial restoration of hypocotyl growth and reversal of dwarfism (Supplemental Figure 9), suggesting that the bes1-D protein is more unstable in the double mutant than in the WT. Moreover, UBP12 overexpression promoted the growth of bes1-D plants, resulting in longer hypocotyls and petiole lengths and larger leaf sizes. UB-P12 OE /bes1-D seedlings also showed hyponasty due to greater growth of the lower petiole surface, resulting in a narrower angle between two leaves compared with bes1-D seedlings (Supplemental Figure 10). These results suggest that UBP12 can stabilize bes1-D protein as well and that UBP12/UBP13 regulate BES1 in the same BR signaling process.
Plants of various genotypes were grown in an eBL-containing medium, and hypocotyl length was measured to investigate the functions of UBP12 and UBP13 in plant growth under the influence of BR hormone ( Figure 5E). Compared with untreated controls, BR treatment increased WT hypocotyl length by 102%; the corresponding values for BES1 OE and UBP12 OE plants were 114%-115% and 135%-138%, respectively. The effect of BES1 and UBP12 double overexpression was almost additive, producing an increase of 151%-155% in hypocotyl length. By contrast, under the same conditions, hypocotyl lengths of BES1-RNAi, ubp12-2w/13-3, BES1-RNAi/ubp12-2w/ 13-3, and BES1 OE /ubp12-2w/13-3 seedlings increased by only 32%-37%. These results are consistent with the phenotypes and BR-responsive gene expression observed in Figure 5A-5D. Our collective evidence indicates that UBP12 and UBP13 promote plant growth by maintaining BES1 levels in BR signaling.

UBP12 stabilizes BES1 protein under BR treatment and/ or dark conditions
To investigate the in vivo relationship between BES1 protein levels and UBP12, we analyzed pBES1 and BES1 levels in mutants and transgenic plants with varying UBP12 levels. BES1 OE , BES1 OE /UBP12 OE , and BES1 OE /ubp12-2w/13-3 plants were treated with or without eBL and with or without MG132, which blocks protein degradation ( Figure 5F). Control experiments showed that BES1 transcript levels were comparable among Ubiquitination assays were performed in vivo with 7-day-old seedlings of WT (Col-0), BES1 OE , BES1 OE /UBP12 OE , and BES1 OE /ubp12-2w/13-3, which were treated with 1 mM epi-brassinolide (eBL) for 6 h. Ubiquitinated BES1-myc proteins were immunoprecipitated with TUBE (Tandem Ubiquitin Binding Entities) magnetic beads. Ub n -BES1-myc proteins in the immunoprecipitates were analyzed by immunoblotting using a-myc antibody. Western blotting with a-Ubi antibody was used as a positive control for the amount of immunoprecipitate in each reaction. Input pBES1-myc and BES1-myc and UBP12 proteins were detected by immunoblotting using a-myc and a-UBP12 antibodies, respectively. Levels of RbcL (the large subunit of ribulose-1,5-bisphosphate carboxylase) were used as a loading control. All assays were analyzed in three different experiments, and images of a representative set are shown.
(D) Transcript levels of BR-responsive genes were analyzed in 7-day-old seedlings of various genotypes treated with 1 mM eBL for 6 h using qRT-PCR. Expression levels of mock-treated WT plants were set to 1.
(E) Hypocotyl lengths of seedlings treated with BR. Seedlings were grown on MS medium containing 0.5 mM eBL for 7 days under normal conditions. Fifteen independent biological samples (n = 15) were obtained, and images from one representative experiment are shown. *P < 0.05, **P < 0.01 (twotailed t-test).
(F) BES1 and UBP12 protein levels under BR treatment. MG132 (20 mM) was added to 7-day-old seedlings of BES1 OE /WT, BES1 OE /UBP12 OE , and BES1 OE /ubp12-2w/13-3 (+MG132), and an equivalent amount of DMSO was added to control seedlings (ÀMG132). The two groups of seedlings were immediately treated with and without 1 mM eBL and then incubated for 6 h. WT was used as a negative control. Levels of the phosphorylated form (legend continued on next page) these three genotypes (Supplemental Figure 11). Relative levels of pBES1-myc and BES1-myc were normalized using RbcL as a loading control, and the gel bands corresponding to these forms are marked on the figure. Under  Previous studies have shown that BES1 is regulated by different E3 ligases depending on growth conditions: SINATs operate in the light , whereas COP1 is active in darkness (Kim et al., 2014). These findings led us to explore the possible regulation of BES1 by UBP12 under various growth conditions. In the light, BES1 OE showed a major band, pBES1myc, and a minor band, BES1-myc, at the beginning of the experiment (0-h time point). In the presence of cycloheximide, which blocks new protein synthesis, levels of both BES1 forms decreased with time, indicating instability. Comparative experiments showed that protein decay was decreased by UBP12 overexpression (BES1 OE /UBP12 OE ) but increased in the double mutant background (BES1 OE /ubp12/2w/13-3). eBL treatment of BES1 OE seedlings in the light caused dephosphorylation of pBES1-myc and the corresponding accumulation of active BES1-myc. BES1-myc was also the predominant species in etiolated BES1 OE seedlings ( Figure 6A). No significant differences were detected for BES1-myc levels among these three genotypes at the beginning point. On the other hand, in the light with eBL and in darkness, BES1-myc levels decreased much faster in BES1 OE /ubp12/2w/13-3 but were more stable in BES1 OE /UB-P12 OE compared with BES1 OE ( Figure 6A). Hypocotyl lengths of 7-day-old seedlings of BES1 OE , BES1 OE /UBP12 OE , and BE-S1 OE /ubp12-2w/13-3 grown under continuous dark conditions were consistent with BES1 protein stability ( Figure 6B). These results show that UBP12 functions to stabilize both pBES1 and BES1 under normal light conditions, but it also prevents BES1 from being degraded when BR signaling is stimulated endogenously in darkness or by exogenous hormone application.
Because BES1 can also be degraded via the selective autophagy pathway Wang et al., 2021a) we tested the effect of E64d, an autophagosome-specific inhibitor, on pBES1 and BES1 levels. Figure 6C shows that, similar to the results of MG132 treatment, E64d also increased pBES1/BES1 levels in mock-treated samples but not to the extent obtained with MG132. If the two degradative pathways process different populations of pBES1/BES1 we would expect an additive effect of the two inhibitors. We found only a moderate increase in pBES1/ BES1 levels when the two inhibitors were added together, sug-gesting that the majority of pBES1/BES1 can be processed by either pathway and that only a minor fraction of pBES1/BES1 is degraded by specific pathways. Notwithstanding differences in the initial levels of pBES1/BES1 in BES1 OE /UBP12 OE and BE-S1 OE /ubp12-2w/13-3, similar responses to MG132 and E64d were seen in mock-treated seedlings of these samples. We also checked the regulatory function of UBP12 with respect to pBES1/BES1 proteins when BR signaling was activated by eBL and when BR biosynthesis was blocked by Brz. Irrespective of genotype, eBL addition converted inactive pBES1 to active BES1 compared with the mock-treated samples, whereas the opposite result was seen with the Brz inhibitor, although the effects were moderate. In these samples, both MG132 and E64d inhibited pBES1 and BES1 protein degradation, and pBES1/BES1 levels were obtained when the two inhibitors were added together. Collectively, these results suggest that UBP12 can stabilize pBES1 and BES1 by rescuing them from degradation by the 26S proteasomal as well as the autophagic pathway in the presence or absence of BR.

DISCUSSION
During the last decade, at least four different ubiquitin E3 ligases have been shown to downregulate BES1/BZR1 by polyubiquitination Lv and Li, 2020). In the dark, COP1 facilitates the degradation of phosphorylated BZR1, and skotomorphogenesis is induced with an increase in dephosphorylated BZR1 levels (Kim et al., 2014). In the light, on the other hand, SINAT E3 ligases destabilize BES1 to regulate BR-induced plant growth . The F-box protein MAX2 marks BES1 for destruction to control the number of shoot branches (Wang et al., 2013b), and in root tissues, the instability of phosphorylated BZR1 is promoted by PUB40 (Kim et al., 2019). In all these cases, BES1/BZR1 polyubiquitinated by E3 ligases are destroyed primarily via 26S proteasomes whose activity is blocked by MG132. In addition, particularly under stress conditions, polyubiquitinated BES1/BZR1 can also be removed by selective autophagy, which is sensitive to E64d Wang et al., 2021a). Further complexity may arise through crosstalk between the two pathways (Marshall et al., 2015;Zientara-Rytter and Subramani, 2019). We found that BES1 levels can be elevated by MG132 and also by E64d, although to a lesser extent ( Figure 6C). The moderate synergistic effect of the two inhibitors suggests that the major fraction of polyubiquitinated BES1 can be redundantly degraded by either pathway, although a minor fraction remains pathway specific.
Because ubiquitination is a reversible process, it is not surprising to expect the participation of DUBs in BR signaling. However, nothing is known about the DUBs that counteract the action of these ubiquitin E3 ligases. Here, we show that UBP12/UBP13 can remove polyubiquitin chains from BES1 and rescue the latter from destruction. The following lines of evidence support our notion that UBP12/UBP13 are positive regulators of BR signaling: (1) UBP12/UBP13 can directly interact with (pBES1-myc) and dephosphorylated form (BES1-myc) were analyzed using an anti-myc antibody. The transgene-derived UBP12-HA, the endogenous UBP12, and the truncated-UBP12 proteins were detected with anti-UBP12 antibodies (produced by rabbits immunized with an N-terminal UBP12 fragment). To determine the relative protein levels (indicated by numbers) of pBES1-myc and BES1-myc, the related protein level of each sample treated with MG132 was set to 1. RbcL levels were used as a loading control and for normalization. All assays were performed with three independent experiments, and images from a representative set are shown.  (A) Time course of BES1 protein levels after cycloheximide (CHX) treatment. Seven-day-old seedlings of BES1 OE /WT, BES1 OE /UBP12 OE , and BES1 OE / ubp12-2w/13-3 grown under normal conditions (16 h light/8 h dark) were used. After 12 h of light, the seedlings were divided into two groups. For experiments in the dark, seedlings grown in continuous dark conditions were treated with 20 mM MG132 for 16 h in darkness. After rinsing with water, seedlings were treated with 200 mM CHX, and samples were taken hourly for 4 h. New protein synthesis in the treated seedlings was blocked by CHX, and levels of the phosphorylated form (pBES1-myc) and dephosphorylated form (BES1-myc) were analyzed by immunoblotting using an anti-myc antibody. The transgene-derived UBP12-HA, the endogenous UBP12, and the truncated-UBP12 (UBP12DC) proteins were detected with anti-UBP12 antibodies (produced by rabbits immunized with an N-terminal UBP12 fragment). To determine the relative protein levels (indicated by numbers) of pBES1-myc and BES1-myc, the related protein level at 0 h was set to 1. RbcL levels were used as a loading control and for normalization. All assays were analyzed in three different experiments, and images of a representative set are shown. Our findings on the relationship between UBP12 and pBES1/ BES1 presented here can be summarized by the simple working model shown in Figure 7. Upon BR signaling, pBES1 is imported into the nucleus, where it is dephosphorylated by PP2A to give the active BES1 (Wang et al., 2021b). Both forms of BES1 are marked for proteasomal degradation by a number of E3 ligases (Wang et al., 2013b;Kim et al., 2014Kim et al., , 2019Yang et al., 2017). UBP12/UBP13 can oppose this process and increase BES1 stability by deubiquitination. The elevated BES1 levels activate or depress downstream gene expression. Luo et al. (2022) recently reported that UBP12/UBP13 could remove K63-linked polyubiquitin chains from the BRI1 protein, thereby limiting endocytosis-mediated degradation of BRI1. Here, we show that UBP12 can stabilize pBES1 and BES1 targeted for degradation via 26S proteasomes as well as by autophagy in both the presence and absence of BR. Under Brz treatment, the BR biosynthesis inhibitor UBP12 can also rescue BES1 proteins from destruction ( Figure 6C). These results pertaining to the regulation of BES1 by UBP12/UBP13 in the nucleus and cytosol are related to but clearly independent of BRI1 regulation at the plasma membrane. Therefore, UBP12/UBP13 can regulate BR signaling not only at the signal perception level (BRI1) but also at the transcription factor (BES1/BZR1) level. The ability to finetune both at the top and the bottom of the signaling pathway allows for more flexible and rapid responses to changing BR levels.
Our in vitro experiments showed that UBP12 and UBP13 could associate with the amino acid 99-164 region of BES1, which is enriched in serine/threonine residues. The region is close to the phosphorylation domain that binds the 14-3-3 protein (Gampala et al., 2007;Ryu et al., 2007) (Supplemental Figure 4). The phosphorylation domain of BZR1 is required for its nuclear export triggered by the BIN2 kinase (Li et al., 2001;Li and Nam, 2002;House et al., 2003). Our observation that UBP12 binds to active BES1 under BR treatment suggests that the UBP12/ BES1 complex may shield the bound BES1 from phosphorylation by BIN2 kinase. In this regard, UBP12 and UBP13 are also expected to oppose the action of 14-3-3 protein and BIN2, both of which are negative regulators of BR signaling.
Although plant DUBs have not been as extensively investigated as ubiquitin E3 ligases, nonetheless, results to date show that these enzymes play crucial roles in plant growth and development (Majumdar and Nath, 2020). The functionally redundant UBP12 and UBP13 are expressed at a reasonably high level throughout the plant life cycle. The null double mutant ubp12/ ubp13 is lethal (Cui et al., 2013;An et al., 2018), indicating that the two encoded DUBs are essential for plant survival. In the ubp12-2w/13-3 double mutant, the UBP12 coding sequence is interrupted at its 3 0 end by a T-DNA insertion. The resulting UBP12DC form, which is expressed from the truncated gene and maintained at a low level, can confer partial function (Supplemental Figure 1), allowing the survival of ubp12-2w/13-3. Plants of this double mutant are dwarf and partially fertile, producing only a limited number of seeds (Cui et al., 2013) and displaying a BR-deficient phenotype. UBP12 and UBP13 have been shown to bind to the latent peptidases DA1, DAR1, and DAR2 in vivo and to inactivate protease function by removing ubiquitin (Vanhaeren et al., 2020). DA1, DAR1, and DAR2 act as negative regulators of leaf growth and are activated via multiple mono-ubiquitination by the E3 ligases BIG BROTHER and DA2. These latent peptidases inhibit plant growth by cleaving and degrading growth regulators such as UBP15, TCP14, TCP15, and TCP22 (Dong et al., 2017). As such, UBP12 and UBP13 can BR signaling triggers import of pBES1 (the inactive form) into the nucleus where it becomes dephosphorylated by nuclear-localized PP2A (Wang et al., 2021b). Both forms of BES1 are polyubiquitinated by various E3 ligases (SINATs, COP1, etc.), targeting them for degradation by 26S proteasomes. However, polyubiquitinated BES1 and pBES1 can be rescued from destruction by the deubiquitinases UBP12/UBP13, which stabilize and increase cytosolic and nuclear BES1 and pBES1 levels. The active BES1 transcription factor activates downstream genes SAUR-AC1 and ASC5, resulting in cell elongation. Concurrently, BES1 represses the expression of the BR biosynthetic genes CPD and DWF4. Recently, Luo et al. (2022)  induce plant growth by stabilizing BRI1 and BES1 as well as by eliminating the activity of DA1, DAR1, and DAR2 proteases, which restrain growth. These findings support a pivotal role for UBP12 and UBP13 in the promotion of plant growth and development.
Given the reported roles of UBP12/UBP13 in several signaling pathways, we can surmise that the dwarf phenotype of the double mutant cannot be attributed solely to a deficiency in BR signaling. Our finding that the BR-deficient phenotype of ubp12-2w/13-3 is only partially restored by overexpression of bes1-D supports this view (Supplemental Figure 9). In addition to BRs, gibberellins (GAs) and auxins are important phytohormones with broad functions in hypocotyl elongation and leaf and root growth. Mutants defective in GA or auxin signaling pathways all display a dwarf stature with reduced hypocotyl, rosette leaf, and root growth (Koornneef and van der Veen, 1980;Griffiths et al., 2006;Kasahara, 2016). In addition, cross-talk of BR with GA and auxin has been extensively reported (Ohri et al., 2019). Therefore, it is not surprising that UBP12/UBP13 should target key components of GA and auxin signaling pathways as well.
The Arabidopsis genome encodes more than 1400 ubiquitin E3 ligases (Vierstra, 2009;Wang and Deng, 2011) but only 64 potential DUBs (Isono and Nagel, 2014), suggesting that one DUB may oppose the action of several E3 ligases. Our observations that UBP12/UBP13 can counteract the action of PUB10 E3 ligase and NLA E3 ligase in JA signaling and nitrogen starvation-induced leaf senescence, respectively, support this notion (Jeong et al., 2017;Park et al., 2019). Because BR is also involved in accelerating leaf senescence during normal growth, it is likely that BES1 may also function along with UBP12/UBP13 in N-induced leaf senescence. We previously observed that UBP12/UBP13 transcript levels increased threefold in response to N starvation . However, we found here that UBP12 transcript and protein levels did not change for 6 h under eBL treatment ( Figures 1D and 5F); moreover, no change in protein levels was observed even under light and dark conditions ( Figure 6). Whether UBP12/ UBP13 activities can be regulated by post-translational events remains to be investigated.
In summary, during BR signaling, BES1 is polyubiquitinated by various E3 ligases and rapidly degraded via 26S proteasomes and/or autophagy, but polyubiquitinated BES1 can be rescued from destruction by polyubiquitin chain removal mediated by UBP12/UBP13. The delicate balance between BES1 degradation and stabilization controls the expression of downstream genes, such as SAUR-AC1, ASC5, CPD, and DWF4, thereby regulating cell growth (Figure 7). Such dynamic fine-tuning of BES1 protein levels provides plants with the flexibility to rapidly regulate plant tissue elongation in response to changes in diverse developmental cues. The positive regulation of plant growth mediated by UBP12/UBP13 can be exploited to improve crop plant yield.

Plasmid DNA construction
Full-length BES1 and BZR1 coding sequences were amplified from cDNA of WT plants using RT-PCR. 35S:BES1-myc was used to generate overexpressing plants, and 35S:nYFP-BES1 was produced for BiFC assays in tobacco. Derivatives of BES1 (DC1, DC2, DC3, DN1, DN2, DN3, M, M1, and M2) for in vitro pull-down assays were produced using the indicated primer pairs. Vectors to produce BES1 and BZR1 were constructed by fusing the appropriate cDNA sequences with DNA sequences encoding N-terminal MBP-tag and C-terminal 6x myc-tag. All constructs were generated by the Gateway cloning system with BP Clonase II Enzyme (Invitrogen, cat. no. 11789100) and LR Clonase II Enzyme (Invitrogen, cat. no. 11791100). Primers used for vector construction are listed in Supplemental Table 1.

Phenotype analysis
For hypocotyl length analysis, plants were grown under 16 h dim light (25 mmol m À2 s À1 ) and 8 h dark at 22 C for the indicated period. Petiole lengths and leaf length/width ratio were measured using 3-week-old plants grown at 22 C under long-day conditions. Sixteen 10-day-old seedlings of each type were transferred onto a large Petri dish (d = 120 mm) containing new MS medium. There was sufficient space between seedlings to avoid interference with plant growth. Plants were then incubated under long-day conditions (16 h light/8 h dark; light intensity 100 mmol m À2 s À1 ) for an additional 11 days, as shown in Figure 5 and The average value of three independent biological replicates (n = 3) is presented with standard deviation (±SD). The qRT-PCR data were statistically compared with the control using a two-tailed t-test (*P < 0.05, **P < 0.01). Primers used in this study are listed in Supplemental Table 1.

In vitro pull-down assays
In vitro pull-down assays were performed as described previously (Park et -myc and MBP-BZR1-myc were individually incubated with GST  resin-bound proteins: GST, GST-UBP12, GST-UBP13, GST-UBP12-N, GST-UBP12-C, GST-UBP13-N, and GST-UBP13-C. After washing with a wash buffer, the beads were boiled with a protein loading buffer. Pulled down proteins separated by SDS-PAGE were immunoblotted with antimyc (Proteintech, cat. no. 16286-1-AP) or anti-MBP (Proteintech, cat. no. 15089-1-AP) antibody using the ECL Prime Western Blotting System (GE Healthcare, cat. no. RPN2232). Two micrograms of recombinant protein were used in all experiments. Pull-down assays for the truncated forms of MBP-BES1 with GST resin-bound protein GST-UBP12 and GST-UBP13 were performed using the same procedure. All assays were analyzed in three independent experiments.
In vivo coIP assay

Semi-in vivo pull-down assays
Seven-day-old seedlings overexpressing BES1-myc were grown under normal conditions, and seedlings were treated with or without 1 mM eBL (Sigma, cat. no. E1641) for 6 h. Total protein extracts were prepared from ground powders by resuspension in a protein extraction buffer containing 100 mM Tris-HCl (pH 7.5), 300 mM NaCl, 2 mM EDTA (pH 8.0), 1% Triton X-100, 10% glycerol, and protease inhibitor cocktail (Roche, cat. no. 04693159001) . Supernatants were pre-incubated with GST resin at 4 C for 1 h and then incubated with GST resin-bound GST and GST-UBP12 at 4 C for 2 h. After the beads were washed 5 times with a protein extraction buffer, the precipitated proteins were separated by SDS-PAGE and analyzed by immunoblotting with anti-myc antibody (Santa Cruz, cat. no. SC-9996). All assays were repeated in three independent experiments.

Cell-free protein degradation assay
Seven-day-old seedlings grown under normal conditions were frozen with liquid nitrogen and then ground to a powder. After resuspension in a degradation buffer containing 25 mM Tris-HCl (pH 7.4), 10 mM NaCl, 10 mM MgCl 2 , 5 mM DTT, 10 mM ATP, and protease inhibitor cocktail (Roche, cat. no. 04693159001), the extracts were centrifuged at 12 000 g for 10 min at 4 C. Protein concentrations in the supernatants were determined by Bradford assay (Bio-Rad, cat. no. 500-0006), and the supernatants were adjusted to equal protein concentrations. For individual assays, approximately 100 mg of total protein was incubated with 400 ng of the recombinant protein MBP-BES1 in a 100-ml reaction mixture. The reaction mixtures were incubated at the indicated times and temperatures. The collected protein mixtures were separated by SDS-PAGE and analyzed by immunoblotting with anti-MBP antibody (Proteintech, cat. no. 15089-1-AP). Band intensities were measured with ImageJ software. All assays were repeated in three independent experiments.