Hydrogen peroxide positively regulates brassinosteroid signaling through oxidation of the BRASSINAZOLE-RESISTANT1 transcription factor

Hydrogen peroxide (H2O2) is an important signaling molecule in plant developmental processes and stress responses. However, whether H2O2-mediated signaling crosstalks with plant hormone signaling is largely unclear. Here, we show that H2O2 induces the oxidation of the BRASSINAZOLE-RESISTANT1 (BZR1) transcription factor, which functions as a master regulator of brassinosteroid (BR) signaling. Oxidative modification enhances BZR1 transcriptional activity by promoting its interaction with key regulators in the auxin-signaling and light-signaling pathways, including AUXIN RESPONSE FACTOR6 (ARF6) and PHYTOCHROME INTERACTING FACTOR4 (PIF4). Genome-wide analysis shows that H2O2-dependent regulation of BZR1 activity plays a major role in modifying gene expression related to several BR-mediated biological processes. Furthermore, we show that the thioredoxin TRXh5 can interact with BZR1 and catalyzes its reduction. We conclude that reversible oxidation of BZR1 connects H2O2-mediated and thioredoxin-mediated redox signaling to BR signaling to regulate plant development.

R eactive oxygen species (ROS) are initially considered to be toxic by-products that accumulate under stress conditions and cause irreversible damage to cells 1 . Recently, ROS are recognized as important signaling molecules that regulate normal plant growth and development, as well as stress responses 1,2 . ROS include superoxide anion (O 2 •− ), hydrogen peroxide (H 2 O 2 ), singlet oxygen ( 1 O 2 ), and hydroxyl radical (OH • ). Among these, H 2 O 2 is considered an important redox signaling molecule, given its specific physical and chemical properties, including a remarkable stability within cells (half time of 10 −3 s), and rapid and reversible oxidation of target proteins 3 . H 2 O 2 can cause oxidation of thiol groups of cysteines in target proteins, which either activates, inactivates, or alters their structure and function 3 . Elucidation of how H 2 O 2 is sensed and how this signal is transmitted to the cell machinery is central to our understanding of so-called "redox signaling" pathways.
H 2 O 2 interplays with diverse phytohormone signals to regulate plant developmental processes and stress responses 4 . Abscisic acid is a stress hormone that has important roles in regulating plant responses to abiotic stress as well as controlling seed germination and stomatal movement. ABA treatment induces the production of H 2 O 2 , which in turn activates the plasma membrane calcium channel and promotes the stomatal closure. ABA-insensitive loci ABI1 and ABI2, which encode protein phosphatase 2C, are the key negative components of the ABA signaling pathway, in which H 2 O 2 may directly inhibit PP activity of ABI1 and ABI2 to positively regulate the transduction of ABA 5,6 . Salicylic acid (SA) plays crucial roles in the plant defense response to biotrophic pathogens. Upon exposure to biotrophic pathogens, SA accumulates and triggers an early burst of H 2 O 2 by binding to CATALASE2 (CAT2), thereby inhibiting catalysis of H 2 O 2 decomposition to water and oxygen. An increasing level of H 2 O 2 induces the sulfenylation of the auxinbiosynthesis enzyme, tryptophan synthetase B subunit 1 and subsequently reduces the auxin levels. At the same time, SA binding to CAT2 also inhibits the stimulation of the enzymatic activity of jasmonic acid (JA) biosynthesis genes ACYL-CoA OXIDASE 2 and 3 (ACX2 and ACX3) to repress JA production. The decrease in auxin and JA accumulation reinforces plant immunity against biotrophic pathogens 7 . Following the SA-induced increase in oxidation potential, which did not last long, plant cells overcompensate with both enzymatic and nonenzymatic antioxidants to protect against oxidative stress. NONEXPRESSOR of PR GENES 1 (NPR1) senses the SAmediated cellular redox changes and undergoes a change from the oligomeric state to the monomeric form. The redox-activated NPR1 is translocated to the nucleus, where it interacts with TGA transcription factors to activate the expression of PR genes and the plant immune response 8 .
Brassinosteroids (BRs) are a group of plant steroid hormones that play fundamental roles in plant growth, development, and stress responses 9,10 . BR perception through the plasma membrane-localized receptor kinase BRASSINOSTEROID INSENSITIVE1 (BRI1) promotes the association of BRI1 with coreceptor BRI1-ASSOCIATED KINASE1 (BAK1) and induces a phosphorylation-dephosphorylation cascade that activates BRASSINAZOLE-RESISTANT1 (BZR1) and BRI1-EMS-SUPPSSOR1 (BES1), two homologous transcription factors, to regulate expression of thousands of BR-responsive genes 11,12 . In the absence of BR, BZR1 and BES1 are phosphorylated by the GSK3-like kinase BRASSINOSTEROID-INSENSITIVE2 (BIN2), they lose their DNA-binding activity, and are retained in the cytosol due to binding to the 14-3-3 proteins 12 . When BR levels are high, BZR1 is dephosphorylated by protein phosphatase 2A (PP2A), and subsequently translocated into the nucleus to bind and regulate target gene expression 13,14 .
Recent studies uncovered a role of H 2 O 2 in BR regulation of plant growth and stress responses 15,16 . Low BR levels induce a transient H 2 O 2 production and change the cellular redox status in guard cells, thus resulting in stomata opening 15 . High BR levels induce prolonged H 2 O 2 accumulation, which facilitates stress responses and stomata closure. Inhibition of H 2 O 2 accumulation by chemical agents, such as ascorbic acid or diphenylene iodonium (DPI) 17 , blocked BR-induced stomata closure, indicating an essential role for H 2 O 2 in BR-induced stomata closure. However, the interplay between H 2 O 2 and BR signaling in plant development remains unclear.
In this study, we demonstrate a critical role of H 2 O 2 in BR regulation of gene expression and seedling growth and development. We find that BR not only inhibits the kinase activity of BIN2 to promote BZR1 dephosphorylation and DNA-binding ability, but also triggers the accumulation of H 2 O 2 to induce BZR1 oxidation, which increases the transcriptional activity of BZR1 by enhancing its binding affinity to its partners such as PIF4 and ARF6. We further show that oxidized BZR1 is reduced by a thioredoxin protein, TRXh5. Mutations of oxidative sites in BZR1 and overexpression of TRXh5 both reduce BZR1 functions in cell elongation and QC cell division. Our study demonstrates that BZR1 activity is regulated by the opposing redox actions of H 2 O 2 and TRXh5.  18 showed BR induction of H 2 O 2 in wild-type and BR-deficient mutant rot3-2, but not in the BR-insensitive mutant bri1-116, and this induction was impaired in the presence of the NADPH oxidase inhibitor, DPI 17 ( Supplementary Fig. 1b-f). These results suggest that BR signaling through BRI1 triggers the production of H 2 O 2 through a NADPH-dependent pathway.

Results
To evaluate the potential roles of H 2 O 2 in BR-mediated seedling development, we analyzed BR regulation of hypocotyl elongation under the H 2 O 2 -deficient condition. Our results showed that DPI treatment, which reduces the H 2 O 2 levels in plants, significantly inhibited the effects of BR on hypocotyl elongation, with high concentrations of DPI causing seedlings insensitivity to BR (Fig. 1a). H 2 O 2 effectively restored the BR sensitivity in DPI-treated plants in a dose-dependent manner (Fig. 1b), indicating that the accumulation of H 2 O 2 is required for BR-induced hypocotyl elongation.
Because BR treatment increased H 2 O 2 level in root tips, including the root stem cell niches ( Supplementary Fig. 1b-d), we tested whether H 2 O 2 plays an important role in BR-mediated root meristem development. In agreement with previous results 19,20 , exogenous BR treatment increased the frequency of QC cell division about 6-fold, from 10 to 59%, in 5-day-old wild-type seedlings. Co-treatments with DPI and potassium iodide strongly counteracted the BR effects and reduced the frequency of QC division from 59 to 21 and 44%, respectively (Supplementary Fig. 2a), suggesting the production of H 2 O 2 is necessary for BR promotion of QC cell division.
To corroborate these pharmacological data, we analyzed the effects of H 2 O 2 on BR-induced cell elongation and QC cell division in the rbohDrbohF double mutant 21 and catalaseoverexpressing plants (CAT2-Ox and CAT3-Ox) 22 , which are defective in normal production of H 2 O 2 . The results showed that rbohDrbohF, CAT2-Ox, and CAT3-Ox all showed BR hyposensitivity in hypocotyl elongation (Fig. 1c, d) and QC cell division ( Supplementary Fig. 2b, c). Together, these results demonstrated that the BR-triggered production of H 2 O 2 contributes to the BRpromoted cell elongation and QC cell division.
H 2 O 2 contributes to BR signaling by targeting BZR1. The GSK3 kinase BIN2 is a key negative regulator in BR signal transduction, and is involved in several other signaling pathways by interacting with and/or phosphorylating additional substrates 23 . The kinase activity of BIN2 was recently reported to be inhibited by nitric oxide in a dose-dependent manner 24 . H 2 O 2 and NO sometimes share common target proteins to transduce redox signals. For example, NPR1, the master regulator in plant immunity, is induced to occur oxidative modification by both H 2 O 2 and NO and then undergoes polymerization and loss of its activity 8,25 . To determine whether H 2 O 2 regulates BR signal transduction through BIN2, we examined the effects of H 2 O 2 on the cell elongation and QC cell division phenotypes in the bin2-3bil1bil2 mutant lacking BIN2 and its two homologs. The results showed that the bin2-3bil1bil2 mutant exhibited a long hypocotyl phenotype in the presence of the BR biosynthesis inhibitor propiconazole (PPZ), but this phenotype was suppressed by DPI treatment (Supplementary Fig. 3a). Microscopic observation showed that bin2-3bil1bil2 displayed higher frequency of QC cell division than wild-type seedlings. Both DPI and KI treatments significantly reduced the frequency of QC cell division in bin2-3bil1bil2 ( Supplementary Fig. 3b). Considering BIN2 belongs to a GSK3 family containing 10 members in the Arabidopsis genome, and among which seven members act redundantly with BIN2 in the BR signaling pathway, two specific GSK3 kinase inhibitors, lithium chloride (LiCl) and bikinin 26,27 , were used to block the kinase activity of GSK3 family proteins. The results showed that both LiCl and bikinin significantly increased the cell elongation and the frequency of QC cell division, as observed upon genetic or ligand-induced activation of the BR pathway. In contrast removal of H 2 O 2 by DPI treatment decreased the hypocotyl length and the frequency of QC cell division, suggesting that involvement of H 2 O 2 in the BR signaling pathway is likely independent of BIN2 and its homologous proteins ( Supplementary Fig. 3c-f).
Further confirmation of the crosstalk between H 2 O 2 and BR downstream of BIN2 was obtained by analyzing cell elongation and QC cell division in transgenic plants expressing BZR1ΔDM, a mutant BZR1 with deletion of the 12-amino-acid BIN2-docking motif resulting in the loss of BIN2 association and phosphorylation 28 . These plants exhibited longer hypocotyl and higher frequency of QC cell division than wild-type plants, while DPI treatment also inhibited these phenotypes ( Supplementary  Fig. 3g, h). Consistent with these results, DPI treatment reduced the cell elongation in the BR-activated mutants bzr1-1D and bes1-D mutants grown on the medium containing the BR biosynthesis inhibitor PPZ, which causes inactivation of wild-type BZR1 and BES1 but not bzr1-1D and bes1-D proteins (Fig. 1e, f and Supplementary Fig. 3i), suggesting that BZR1/BES1-driven cell elongation is abrogated by DPI treatment. DPI also reduced QC cell division in bzr1-1D and bes1-D ( Supplementary Fig. 3j, k). Together, these genetic and physiological data suggest that the H 2 O 2 regulates the BR activity downstream of BIN2, and probably through the transcription factors BZR1 and BES1.  29 . It has been reported that biotinconjugated iodoacetamide (BIAM) and H 2 O 2 could selectively and competitively react with cysteine residues that exhibit a low pK a in target proteins 30 , and thus the H 2 O 2 -sensitive and H 2 O 2oxidized cysteine residues can be detected by BIAM labeling assay and biotin-switch assay, respectively (Fig. 2a, b). To test whether H 2 O 2 induces oxidation of BZR1, maltose-binding protein (MBP) and MBP-BZR1 were pretreated with or without H 2 O 2 , and then incubated with BIAM. SDS-PAGE followed by immunoblot  , which irreversibly modifies free thiols in proteins. The NEM-modified MBP-BZR1 was then precipitated, treated with DTT to reduce any oxidized cysteines pre-existing before NEM treatment. The newly exposed free thiol groups were then labeled with BIAM. The BIAM-tagged proteins in the samples were then immunoprecipitated with streptavidin beads and revealed by western blot analysis. The results show that H 2 O 2 causes oxidation of BZR1 in vitro (Fig. 2d). Five cysteine residues in BZR1 are putative target sites of oxidation, two of them, cys-63 and cys-73, are located in the BZR1 N-terminal DNA-binding domain, while cys-91, cys-174, and cys-240, are located in the BZR1 C-terminal phosphorylation domain ( Supplementary Fig. 4a, b). To investigate which cysteine sites of BZR1 are oxidized by H 2 O 2 , we analyzed tryptic fragments derived from recombinant BZR1 protein sequentially treated with H 2 O 2 , NEM, and BIAM by liquid chromatographytandem mass spectrometry, and identified cys-63 (Fig. 2e), cys-91, cys-174, and cys-240 ( Supplementary Fig. 5a-c) as oxidized residues. Then, we mutated each cysteine (C) to serine (S) to determine which cysteine residues are the major oxidized sites in BZR1. Simultaneous mutagenesis of cys-91, cys-174, and cys-240 had no significant effect on oxidation of BZR1, while simultaneous mutagenesis of cys-63 and cys-73 significantly decreased BZR1 oxidation (Fig. 2f). Further experiments showed that MBP-BZR1 C73S exhibited similar oxidation levels with wild-type MBP-BZR1, but MBP-BZR1 C63S showed a very weak oxidative modification (Fig. 2g)  were first incubated with NEM to irreversibly block the free thiols, then treated with DTT to reduce the pre-existing oxidized cysteines in MBP-BZR1. The newly exposed free thiol groups were then relabeled with BIAM. The BIAM-tagged proteins in the samples were then captured with streptavidin beads and detected by gel blot using anti-MBP antibody. e Mass spectrometry analysis of the tryptic fragments of of MBP-BZR1 protein treated as in d.  (Fig. 2h). In addition, we found that overexpression of CAT2 significantly reduced the oxidative modification of BZR1 (Fig. 2i). These results suggest that H 2 O 2 regulates BZR1 redox state by oxidation.
Cys-63 is important for the functions of BZR1. To understand whether oxidation of cys-63 modulates the function of BZR1 in plants, we generated transgenic Arabidopsis plants expressing Mock PPZ Hypocotyl length (mm)

PRE1
EXP8  . 3a-e). The QC cell division ratio of bzr1-1D 5S (10%) was significantly lower than that of bzr1-1D transgenic plants (Fig. 3f). Moreover, conversion of the cysteine-84 of bes1-D to serine also suppressed the phenotypes of bes1-D ( Supplementary Fig. 6c-f). Together, these results demonstrate that the conserved cys-63 of BZR1 and cys-84 of BES1 are essential for functions in various BR-mediated responses. Next, we examined the effects of cysteine residues on BZR1 regulation of the expression of target genes. BZR1 is known to activate growth-related genes, such as PRE1, EXP8, and IAA19, to promote cell elongation. The expression levels of PRE1, EXP8, and IAA19 were increased in bzr1-1D-Ox plants compared with that in wild-type plants, but were less increased in bzr1-1D 5S -Ox than in bzr1-1D-Ox (Fig. 3g). BZR1 is also known to repress BR biosynthesis genes, including CPD and DWF4, for feedback regulation of BR levels. Transcription levels of CPD and DWF4 significantly decreased in bzr1-1D-Ox seedlings, while they were only slightly repressed in the bzr1-1D 5S -Ox seedlings, suggesting that the mutations of cysteines impaired BZR1 activity (Fig. 3g). To more directly examine the functions of the oxidized residues in BZR1 regulation of gene expression, we performed transient gene expression assays using Arabidopsis mesophyll protoplasts. Co-expression of bzr1-1D significantly increased the expression of luciferase reporter driven by the PRE5 promoter, whereas bzr1-1D C63S only marginally increased PRE5 promoter activity, and bzr1-1D 5S lost the transcriptional activation activity (Fig. 3h). These results indicate that mutagenesis of oxidized cysteine residues decreased the transcriptional activity of BZR1.
H 2 O 2 is required for the transcriptional activity of BZR1. We carried out RNA-Seq analysis to determine whether H 2 O 2 affects the BZR1-mediated gene expression in Arabidopsis genome. Wild-type and bzr1-1D mutant seedlings were grown in the dark for 6 days on the medium containing BR biosynthesis inhibitor PPZ, and/or NADPH oxidase inhibitor DPI, and/or H 2 O 2 . RNA-Seq analysis identified 4428 genes affected more than 1.5-fold by bzr1-1D comparing with wild-type in the presence of PPZ (BZR1regulated genes in the mock condition). In the presence of PPZ and DPI, 2458 genes were affected by bzr1-1D (BZR1-regulated genes in the DPI condition), while 5152 genes were affected by bzr1-1D in the presence of PPZ, DPI, and H 2 O 2 (BZR1-regulated genes in the DPI plus H 2 O 2 condition) ( Fig. 4a (Fig. 4a, b). Scatter plot analysis showed that the slope of the trend line in the plot comparing BZR1regulated genes in the mock condition with that in the DPI condition was only 0.67, but was restored to 0.87 in the plot comparing the BZR1 response genes in the mock condition with that in the DPI plus H 2 O 2 condition (Fig. 4c, d). The overall effect of BZR1 on gene expression was significantly reduced by DPI treatment, but this reduction was effectively reversed by H 2 O 2 cotreatment, suggesting H 2 O 2 plays a positive role in BZR1regulated gene expression. c Immunoblot analysis of phosphorylated and dephosphorylated BZR1 in 5-day-old dark-grown pBZR1:BZR1-YFP using anti-YFP antibody. Actin bands showed protein loadings. d-g In vitro pull-down assays showed that H 2 O 2 enhanced, but Cys-63 mutation to Ser reduced, the BZR1 binding to PIF4. e, g show the quantification of pull-down assays (normalized to input) in the d and f, respectively. Error bars represented the s.d. of three independent experiments. *p < 0.05, as determined by a Student's t-test. h, i rBiFC confocal images showed that H 2 O 2 enhanced, but mutation of cysteines reduced the interaction between BZR1 and PIF4 in plants. The fluorescent signals of YFP (BiFC) and RFP (reference) were determined along a line drawn on the confocol images using ImageJ software. Error bars, s.d. (n = 50 images). Different letters above the bars indicated statistically significant differences between the samples (two-way ANOVA, p < 0.05). Scale bar, 10 μm. j Co-immunoprecipitation (CoIP) assays showed that H 2 O 2 increased the interaction between BZR1 and PIF4 in plants. The Arabidopsis mesophyll protoplast transformed with BZR1-Myc and/or PIF4-YFP were treated with mock solution or 100 μM H 2 O 2 for 1 h, then immunoprecipitation was performed using GFP-Trap agarose beads, and the immunoblots were probed using anti-Myc and anti-YFP antibodies To further define H 2 O 2 -dependent BZR1-regulated genes, we analyzed the genes differentially expressed in bzr1-1D mutant with or without DPI treatment, and identified 4054 genes affected >1.5-fold by DPI (Fig. 4e and Supplementary Data 4). Among the 4428 genes affected by bzr1-1D in comparison with wild-type, 2200 genes (49.7%) were also affected by DPI treatment, which we named H 2 O 2 -dependent BZR1-regulated genes, and the other 2228 genes were named H 2 O 2 -independent BZR1regulated genes (Fig. 4e). Among the 2200 H 2 O 2 -dependent BZR1-regulated genes, 2053 genes (93.3%) were affected in the opposite way by DPI treatment and bzr1-1D vs. wild-type, confirming that DPI inhibits the transcriptional activity of BZR1 (Fig. 4f).
Gene ontology (GO) analysis showed that the H 2 O 2 -dependent BZR1-regulated genes play dominant roles in BZR1-mediated biological processes and cellular activities (Fig. 4g). For example, genes involved in cell growth and auxin responses were highly enriched in H 2 O 2 -dependent BZR1-activated genes. In contrast, the genes involved in ROS production, photosynthesis, and chloroplasts were highly enriched in H 2 O 2 -dependent BZR1repressed genes. Genes involved in gibberellic acid (GA) and light responses were highly enriched in both H 2 O 2 -dependent BZR1activated and BZR1-repressed genes (Fig. 4g). Previous studies showed that BZR1 interacts with PIF4 and ARF6 to form BZR1-PIF-ARF module to regulate cell elongation downstream of multiple hormonal and environmental signals 31,32 . GO analysis showed that PIF-induced genes and IAA3-repressed genes were highly enriched in the H 2 O 2 -dependent BZR1-activated genes, while PIF-repressed genes were highly enriched in the H 2 O 2dependent BZR1-repressed genes, suggesting H 2 O 2 plays important roles in the regulation of BZR-PIF-ARF common target gene expression ( Supplementary Fig. 7a, b).  (Fig. 5a, b). Treatment with BR also caused dephosphorylation of BZR1, but co-treatment with H 2 O 2 or DPI had no obvious effects on the levels of phosphorylated and unphosphorylated BZR1 (Fig. 5c). These results indicated that disruption of H 2 O 2 homeostasis does not have a significant effect on BR levels or BR signaling upstream of BZR1.
To determine whether H 2 O 2 affects the DNA-binding ability of BZR1, we performed DNA-protein pull-down and chromatin immunoprecipitation quantitative PCR (ChIP-qPCR) assays. The results showed that BZR1 specifically binds to the promoter of SAUR15, which is a known BZR1-binding target gene. H 2 O 2 treatment did not enhance in vitro binding activity of BZR1 to the promoter of SAUR15 in the absence or presence of PIF4 and ARF6 ( Supplementary Fig. 8a-c). Similarly, ChIP-qPCR analysis showed that H 2 O 2 did not have a significant effect on BZR1 DNA-binding capacity in plants ( Supplementary Fig. 8d), suggesting that H 2 O 2 does not affect DNA-binding ability of BZR1. In contrast, H 2 O 2 significantly improved the transcriptional activity of BZR1, PIF4, and ARF6 in protoplast transient assays ( Supplementary Fig. 8e).
The enrichment of PIF and ARF-bound/regulated genes among the H 2 O 2 -dependent BZR1-regulated genes (Supplementary Fig. 7a, b) suggests that H 2 O 2 may affect the interactions between BZR1 and PIF4 or ARF6. To test this hypothesis, we analyzed the effects of H 2 O 2 on the binding affinity of BZR1 to PIF4 and ARF6. The in vitro pull-down assays showed that glutathione-S-transferase (GST)-PIF4 interacted with MBP-BZR1, while H 2 O 2 treatment dramatically increased the interaction between BZR1 and PIF4 (Fig. 5d, e). Mutagenesis of cysteine to serine residues reduced the binding ability of BZR1 to PIF4 (Fig. 5f, g). Using GST-ARF6, we observed that H 2 O 2 increased the interaction between BZR1 and ARF6 ( Supplementary  Fig. 9a-d). These results suggested that H 2 O 2 -induced oxidation enhances BZR1-binding affinity with its partner, PIF4 and ARF6 in vitro.
To verify the H 2 O 2 effects on the interactions of BZR1 with PIF4 and ARF6 in plants, we analyzed these interactions using a newly developed ratiometric bimolecular fluorescence complementation (rBiFC) assay 33 . In this assay, BZR1 or BZR1 5S and PIF4 or ARF6 were simultaneously cloned into a single vector backbone containing a monomeric red fluorescent protein driven by the 35S promoter as an internal marker for expression control and ratiometric analysis. Consistent with previous results, BZR1 interacted with PIF4 and ARF6 in the epidermal cells of tobacco leaves, but this interaction was significantly enhanced by H 2 O 2 treatment (Fig. 5h, i and Supplementary Fig. 9e, f). However, coexpression of BZR1 5S and PIF4 or ARF6 resulted in very weak YFP signals, and H 2 O 2 treatment had no significant effect on the interactions between BZR1 5S and PIF4 or ARF6 (Fig. 5h, i and Supplementary Fig. 9e, f). We further tested the effect of H 2 O 2 on the interactions of BZR1 with its partner PIF4 and ARF6 in plants using co-immunoprecipitation assays. The results showed that H 2 O 2 clearly enhanced the interactions between BZR1 and PIF4 or ARF6 in plants (Fig. 5j and Supplementary Fig. 9g). Together, these results demonstrated that H 2 O 2 -induced oxidation of BZR1 promotes its interactions with PIF4 and ARF6.
TRX-h5 is a redox mediator of BZR1. To counter the effects of oxidation-facilitated BZR1 activation, reducing agents must be engaged to catalyze the switch of BZR1 from oxidized to reduced form. In mammals, the cytosolic/nuclear thioredoxin TRX1 has been well characterized as a master transcriptional regulator through direct or indirect interactions with different transcription factors 3 . The plant h-type thioredoxin members have been reported to localized in cytoplasm and nucleus, and behave similarly to the mammalian TRX1 system 34 . To test whether members of TRX-h family interact with and mediate the reduction of BZR1, we performed the BiFC assay in tobacco leaves by co-expressing BZR1 or BES1 fused to the carboxyl-terminal half of yellow fluorescent protein (BZR1-cYFP or BES1-cYFP) and members of TRX-h proteins fused to the amino-terminal half of YFP (nYFP). Strong YFP fluorescent signals were observed in the nucleus when TRXh5-nYFP was co-transformed with BZR1-cYFP or BES1-cYFP, indicating a specific TRXh5 interaction with BZR1 and BES1 ( Fig. 6a and Supplementary Fig. 10). In vitro pull-down assays showed that GST-TRXh5 directly interacted with MBP-BZR1, but not with MBP alone (Fig. 6b). Gene expression analysis using the Arabidopsis microarray data displayed in the eFP browser indicated very similar ubiquitous expression patterns of TRXh5 and BZR1 ( Supplementary Fig. 11a, b), suggesting they may function together to regulate plant growth and development. Consistent with the BiFC and in vitro assays, co-immunoprecipitation assays showed that TRXh5 interacts with BZR1 in plants (Fig. 6c). To examine whether TRXh5 can catalyze the reduction of BZR1, we carried out the in vitro reduction reaction. Recombinant BZR1 proteins were first oxidized by H 2 O 2 treatment, and then incubated with recombinant TRXh5 and NTR isoform A (NTRA). The result showed that TRXh5 caused a decrease of the level of oxidized BZR1, suggesting TRXh5 catalyzes BZR1 reduction (Fig. 6d).
To determine whether the thioredoxin system regulates the function of BZR1 in plants by regulating its oxidized state, we characterized the phenotype of TRXh5 knock out mutant (trxh5-1), and transgenic plants overexpressing TRXh5 (TRXh5-Ox). Microscopic observation showed that trxh5-1 mutant exhibited higher frequency of QC cell division than wild-type plants (Fig. 6e). On the other hand, overexpression of TRXh5 partially suppressed the BR-induced and BZR1-induced QC cell division (Fig. 6f, g). TRXh5 overexpression not only partly restored the sensitivity of bzr1-1D to PPZ in the dark (Fig. 6h and Supplementary Fig. 12a, b), but also attenuated the large-leaf phenotype of bzr1-1D under light ( Supplementary Fig. 12c-g). Furthermore, in transient assays TRXh5 reduced the transcription activity of BZR1 (Fig. 6i). Together, these results demonstrate that TRXh5 negatively regulates BR signaling by changing the oxidation state of BZR1.

Discussion
Among ROS, H 2 O 2 in particular plays indispensable roles in a wide range of plant growth and developmental processes, as well as in stress responses. H 2 O 2 is referred to as a second messenger integrating a large number of signals to orchestrate correct cellular responses. However, the molecular targets and underlying mechanism of regulation have remained unclear. In this study, we demonstrated that BR increases cellular level of H 2 O 2 , which causes oxidation of BZR1 at a conserved cysteine residue; this oxidation enhances BZR1 transcriptional activity by promoting its interaction with ARF6 and PIF4. The oxidized BZR1 is reduced by thioredoxin TRXh5. Such redox regulation of BZR1 plays an important role in BR signal transduction (Fig. 7).
Several studies have reported that BR-induced H 2 O 2 accumulation is necessary for numerous BR-mediated biological processes, including stomatal movement 15,16 , salt tolerance 35 , responses to heat and oxidative stresses 36,37 . In this study, we showed that BR-induced H 2 O 2 production is also critical for BR promotion of cell elongation and root QC cell division. Chemical scavenging of H 2 O 2 , silencing RBOHD and RBOHF expression, and overexpression of catalase genes, all compromised BR-and/ or activated BZR1-induced cell elongation and QC cell division. DPI treatment decreased BR sensitivity in a dose-dependent manner in the hypocotyl elongation assays. These results indicated that H 2 O 2 is essential for diverse BR-mediated plant biological processes, and H 2 O 2 plays an important role in BR signal transduction.
Multiple components of the BR signaling pathway have been reported to be the modification targets of ROS 24,38 . BAK1 was shown to be glutathionylated by glutaredoxin AtGRXC2 via a thiol-dependent reaction with GSSG, which inhibits the kinase activity 38 . Nitric oxide inhibits the kinase activity of BIN2 by Snitrosylation of cys-162, which is conserved in GSK3 kinase family 24 . However, these studies only provided the in vitro data, and whether redox regulates the kinase activity of BAK1 and BIN2 in plants still remains unknown.
In the present study, we demonstrate, through several lines of evidence, that H 2 O 2 is involved in BR signal transduction through oxidation of the master transcription factor BZR1. First, chemical scavenging of H 2 O 2 suppressed the long hypocotyl and high frequency of QC cell division in bin2bil1bil2 mutants and wildtype plants treated with GSK3 inhibitors bikinin or LiCl. Further, the BZR1ΔDM transgenic plants, which express a mutant BZR1 protein that cannot be phosphorylated by BIN2, exhibited DPI inhibition of cell elongation and QC cell division, suggesting that DPI inhibits BR responses independent of BIN2/GSK3s. Second, H 2 O 2 and its biosynthesis inhibitor DPI had no significant effect on the protein level or on the phosphorylation status of BZR1, confirming that H 2 O 2 does not affect BR signaling upstream of BZR1. Third, BIAM labeling assays showed that H 2 O 2 effectively induced the oxidation of BZR1 in vivo and in vitro. Finally, mutation of the H 2 O 2 -oxidized cysteine residues abolished in vitro H 2 O 2 oxidation of BZR1 and abrogated BZR1's activities in activating gene expression and promoting cell elongation and QC cell division. Together, these results demonstrated that H 2 O 2 contributes to BR signal transduction downstream of GSK3 kinase and through oxidation of the transcription factor BZR1.
Transcriptomic analyses further support that the transcriptional activity of BZR1 is under redox regulation and that BZR1 is a major factor mediating transcriptional responses to H 2 O 2 . Under an H 2 O 2 -deficient condition, the amplitude of BZR1regulated gene expression was significantly reduced and approximately 64% of the genes were no longer altered in bzr1-1D, while adding H 2 O 2 back to the H 2 O 2 -deficient condition partially restored the responsiveness to bzr1-1D for about 58% of the BZR1-regulated genes. On the other hand, 2200 (54%) of the 4054 DPI-altered genes were also altered by bzr1-1D, with 93.3% of them affected oppositely by DPI and bzr1-1D. Such strong correlation between DPI and BZR1 effects of transcriptome supports that BZR1 is a major factor for H 2 O 2 modulation of gene expression under our experimental conditions. GO comparison of BZR1-regulated genes under oxidized and reduced states suggests that the redox regulates BZR1 to modulate many cellular functions. H 2 O 2 -dependent BZR1-repression genes are enriched in photosynthesis and ROS production. BZR1 repressed the expression of several members of thioredoxins, glutaredoxins, and peroxidase family genes, including TRXf1, TRXm1, TRXm2, TRXm4, TRXy2, TRXz, GRX2, and PRXQ via an H 2 O 2 -dependent manner (Supplementary Data 1). Such BZR1 repression of ROS-related genes may contribute to feedback regulation and redox homeostasis. H 2 O 2 enhances BZR1 activation of genes involved in cell wall biogenesis and responsive to GA, auxin, and light, consistent with its function in promoting BR-induced cell elongation. The H 2 O 2 -dependent BZR1regulated genes are enriched with PIF4-regulated or IAA3regulated genes. Such significant overlaps with auxin and light responses are consistent with the recent finding that auxin, light, and BR regulate shoot cell elongation through a central module of BZR1 interaction with the PIF4 and ARF6 factors 31 . Indeed, we found that H 2 O 2 oxidation of BZR1 enhances its interaction with PIF4 and ARF6. As such, BR and H 2 O 2 control BZR1 activity through distinct mechanisms of enhancing nuclear localization/ DNA binding and interaction with partners, respectively. Such a specific mechanism of H 2 O 2 regulation is likely to allow redox regulation of subsets of BR-regulated processes.
H 2 O 2 -mediated oxidative modification is a reversible process, and in many cases the reduction of oxidized cysteine residues is carried out by the thioredoxin system 34,39 . Thioredoxins are typically small proteins that are conserved in all free-living organisms and are able to catalyze the reduction of the disulfide bond in their substrates and the denitrosylation of potential protein-SNO modifications 34 . Detailed characterization of TRXh5 in this study reveals that the growth-promoting hormone BR represses the expression of TRXh5 through BZR1 (Supplementary Fig. 13a, b) and that TRXh5 negatively regulates BR signal transduction by catalyzing BZR1 oxidation-to-reduction switch. Previous studies have shown that TRXh5 regulates SAdependent plant immunity by selective protein denitrosylation 25 exchange and nuclear localization to activate immune responses. As such, TRXh5 is activated by pathogens but repressed by BR, and it activates immunity and represses growth, and thereby functions as a major node of crosstalk that mediates a trade-off between growth and immunity in plants.
The proper balance among stem cell maintenance, proliferation, and differentiation is a key aspect of development in multicellular organisms, and is controlled by gradient signals such as BR and ROS 41,42 . BR and activated BZR1 are patterned in the Arabidopsis root tip, with low levels of nuclear-localized BRactivated BZR1 in the stem cell niche to maintain stem cell and meristem size, and high levels of nuclear BZR1 in the epidermal cells in the transition-elongation zone to promote cell elongation 41 . H 2 O 2 also accumulates mainly in the expanding cells of the elongation zone to promote cellular differentiation, whereas superoxide mainly accumulates in the dividing cells of the meristem zone to maintain stem cell 42 . BR treatment not only induces the nuclear localization of BZR1 and the accumulation of H 2 O 2 , but also reduces the content of superoxide, as the BR-deficient mutant det2-9 was recently shown to over accumulate superoxide in the meristem zone, which leads to the inhibition of root growth 43 . Overall, the distribution pattern of H 2 O 2 is similar to that of nuclear BZR1 ( Supplementary Fig. 14a, b), consistent with H 2 O 2 modulating BZR1 activity in root growth regulation. The gradient pattern of H 2 O 2 depends on the basic/helix-loop-helix transcription factor UPBEAT1 (UPB1), which regulates the expression of peroxidases and root growth rate 42 . While the role of H 2 O 2 in QC function remains unclear. BZR1 has been shown to promote QC division in response to DNA damage stress 20 . The H 2 O 2 enhancement of the transcription activity of BZR1 could potentially provide dual regulation of root growth by BR and redox signals, important for optimal growth according to endogenous and environmental signals.
The redox regulation of BZR1 by H 2 O 2 and TRXh5 establishes a molecular framework of H 2 O 2 crosstalk with BR signaling. As, H 2 O 2 is involved in numerous plant physiological processes and stress responses, this mechanism of crosstalk is likely important for coordinating plant growth with environmental adaptation.
Microscopy analysis. Phenotypic analysis of roots were carried out with modified Pseudo-Schiff Propidium Iodide (mPS-PI) staining method 44 . Seedlings of wildtype and various mutants for analysis of QC cell division were grown on ½ MS basal salt medium supplemented with 1% sucrose for 4 days in the continues light, and then transferred to ½ MS liquid medium containing different hormones and chemicals as indicated for another 1 day under light conditions. Whole seedlings were fixed in fixative buffer (50% methanol and 10% acetic acid) and stained using the mPS-PI staining method. The cellular organization of root tips was analyzed from longitudinal optical sections obtained using an LSM700 laser scanning confocal microscope (Zeiss). At least 50 roots were analyzed for each treatment in each experiment and each experiment was repeated independently at least three times. The statistical significance of differences between means was determined using the Student's t-test.
BIAM labeling assay. MBP and MBP-BZR1 proteins were purified from E. coli using amylose resin (NEB), and then treated with different concentrations of H 2 O 2 at room temperature for 15 min. The proteins were precipitated by adding one volume of acetone at −20 ℃ for 20 min and centrifuged at 5000 × g for 5 min. The pellets were washed three times with 50% acetone and dissolved in 500 µl labeling buffer (50 mM MES-NaOH, pH 6.5, 100 mM NaCl, 1% TritonX-100, 100 µM BIAM), and then incubated at room temperature in the dark for 1 h. The labeling reactions were terminated by the addition of β-mercaptoethanol to a final concentration of 20 mM. The reaction mixtures were precipitated by adding one volume of acetone at −20 ℃ for 20 min and centrifuged at 5000 × g for 5 min. The pellets were dissolved in 50 µl SDS sample buffer, and subjected to separate on SDS-PAGE. Proteins labeled with BIAM were detected with HRP-conjugated streptavidin 30 (Cell Signaling, Cat: 7075S, 1:5000 dilution). An antibody against MBP was used to show the total MBP or MBP-BZR1 proteins (NEB, Cat: 8038L, 1:5000 dilution).
In vivo and in vitro biotin-switch assays. The slightly modified biotin switch assays 30,45 that selectively add biotin to reduced or oxidized thiol groups were used to analyze the redox status of BZR1 proteins in Arabidopsis plants or purified recombinant BZR1 proteins treated with H 2 O 2 .
For in vivo oxidation analysis, the p35S:BZR1-YFP transgenic seedlings were grown in ½ MS liquid medium containing 1% sucrose for 7 days under 16-h light/ 8-h dark condition, and then treated with or without 100 nM BL, and/or 1 mM H 2 O 2 for another 3 h under light conditions. Seedlings were harvested and ground to fine powder in liquid nitrogen. Proteins were extracted in EBR buffer (20 mM HEPES, pH 8.0, 40 mM KCl, 5 mM EDTA, 0.5% TritonX-100, 1% SDS, 1 mM PMSF, and 1× protease inhibitor cocktail).
For in vitro oxidation analysis, the maltose-binding protein-BZR1 fusion (MBP-BZR1) protein was purified from E. coli, and then treated with 1 mM H 2 O 2 or 1 mM DTT at room temperature for 15 min. The proteins were precipitated by adding one volume of acetone at −20 ℃ for 20 min and centrifuged at 5000 × g for 5 min. The pellet was washed three times with 50% acetone and dissolved in 500 µl EBR buffer.
To detect the oxidized form of BZR1, the plant extracts or the recombinant BZR1 proteins were incubated with 100 mM NEM in the EBR buffer at room temperature for 30 min with frequent vortexing to block free thiol. The samples were precipitated with one volume of acetone and washed three times with 50% acetone. The pellets were dissolved in 500 µl EBR buffer plus 20 mM DTT, and then incubated at 37 ℃ for 30 min to reduce the oxidized thiols. DTT was removed by protein precipitation and pellets were resuspended in 500 µl EBR buffer. The supernatant was labeled with 100 µM BIAM at room temperature for 1 h in the dark, and proteins were precipitated with one volume of acetone to remove free BIAM.
The BIAM-treated proteins were then dissolved in 250 µl EBR buffer and diluted by 750 µl NEB buffer (20 mM HEPES, pH 8.0, 40 mM KCl, 5 mM EDTA, 0.25% TritonX-100, 1 mM, PMSF and 1× protease inhibitor cocktail). After centrifugation at 10,000 × g for 5 min, the supernatant was added to 40 µl streptavidin beads (Biomag) and incubated at 4 ℃ for overnight. Beads were washed five times with NEB buffer, and proteins were eluted by 50 µl 2× SDS sample buffer and the samples were separated on 8% SDS-PAGE gels. An aliquot of proteins before incubation with streptavidin beads were also analyzed as total BZR1 protein. The gel blots were probed with anti-GFP antibody (homemade, 1:5000 dilution) or anti-MBP antibody (NEB, Cat: E8038L, 1:5000 dilution). The uncropped gel/blot images were presented in the Supplementary Fig. 15.
Mass spectrometric analysis of oxidized residues in BZR1. Briefly, purified MBP-BZR1 protein labeled with biotin-IAM was digested in-gel by trypsin, and then analyzed by liquid chromatograph tandem mass spectrometry (LC-MS/MS) using a ThermoFisher Orbitrap Fusion mass spectrometer. MS1 scans were acquired in the Orbitrap MS using a mass resolution of 240,000, and MS2 scans were obtained from Orbitrap MS with a resolution of 30,000. The raw data from mass spectrometry was searched against the NCBI Arabidopsis protein database using the Protein Prospector searching software (http://prospector.ucsf.edu/ prospector/mshome.htm), with a 10-ppm tolerance of the precursor and a 20-ppm tolerance for higher energy collision conditions (HCD) MS/MS. Cysteine biotinylation (C14H22N4O3S, mass change = 326.14) were included in the search.
RNA-Seq. Wild-type Col-0 and bzr1-1D were grown on ½ MS medium containing 2 µM PPZ, and/or 1 µM DPI, 0.3 mM H 2 O 2 for 6 days in the dark. Total RNA was extracted with Trizaol RNA extraction kit (Transgene), and the mRNA sequencing libraries construction and sequencing on the BGISEQ-500 platform were performed at Beijing Geonomics Institute. The sequence reads were mapped to the Arabidopsis genome using HISAT and Bowtie2 software, and differential gene expression was analyzed using Noiseq software. Differentially expressed genes were defined by a 1.5-fold expression difference with a possibility >0.8. The accession number for the RNA-Seq data in the Gene Expression Omnibus database is GSE110488.
Data availability. All sequencing data that support the findings of this study have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus (GEO) and are accessible through the GEO Series accession number GSE110488. All data relevant to this study are provided in the manuscript and its supplementary files or are available from the corresponding author upon request.