Transcription factor HAT1 is phosphorylated by BIN2 kinase and mediates brassinosteroid repressed gene expression in Arabidopsis

Authors

  • Dawei Zhang,

    1. College of Life Science, Ministry of Education Key Laboratory for Bio-Resource and Eco-Environment, State Key Laboratory of Hydraulics and Mountain River Engineering, Sichuan University, Chengdu, China
    2. Department of Genetics, Development and Cell Biology, Iowa State University, Ames, IA, USA
    Search for more papers by this author
  • Huaxun Ye,

    1. Department of Genetics, Development and Cell Biology, Iowa State University, Ames, IA, USA
    Search for more papers by this author
  • Hongqing Guo,

    1. Department of Genetics, Development and Cell Biology, Iowa State University, Ames, IA, USA
    Search for more papers by this author
  • Abbagail Johnson,

    1. Department of Genetics, Development and Cell Biology, Iowa State University, Ames, IA, USA
    Search for more papers by this author
  • Meishan Zhang,

    1. Department of Genetics, Development and Cell Biology, Iowa State University, Ames, IA, USA
    2. Department of Agronomy, Jilin Agricultural University, Changchun, China
    Search for more papers by this author
  • Honghui Lin,

    Corresponding author
    1. College of Life Science, Ministry of Education Key Laboratory for Bio-Resource and Eco-Environment, State Key Laboratory of Hydraulics and Mountain River Engineering, Sichuan University, Chengdu, China
    Search for more papers by this author
  • Yanhai Yin

    1. Department of Genetics, Development and Cell Biology, Iowa State University, Ames, IA, USA
    Search for more papers by this author

Summary

Plant steroid hormones, brassinosteroids (BRs), play essential roles in modulating cell elongation, vascular differentiation, senescence and stress responses. BRs signal through plasma membrane-localized receptor and other components to modulate the BES1/BZR1 (BRI1-EMS SUPPRESSOR 1/BRASSINAZOLE RESISTANT 1) family of transcription factors that modulate thousands of target genes. Arabodopsis thaliana homeodomain-leucine zipper protein 1 (HAT1), which encodes a homeodomain-leucine zipper (HD-Zip) class II transcription factor, was identified through chromatin immunoprecipitation (ChIP) experiments as a direct target gene of BES1. Loss-of-function and gain-of-function mutants of HAT1 display altered BR responses. HAT1 and its close homolog HAT3 act redundantly, as the double mutant hat1 hat3 displayed a reduced BR response that is stronger than the single mutants alone. Moreover, hat1 hat3 enhanced the phenotype of a weak allele of the BR receptor mutant bri1 and suppressed the phenotype of constitutive BR response mutant bes1-D. These results suggest that HAT1 and HAT3 function to activate BR-mediated growth. Expression levels of several BR-repressed genes are increased in hat1 hat3 and reduced in HAT1OX, suggesting that HAT1 functions to repress the expression of a subset of BR target genes. HAT1 and BES1 bind to a conserved homeodomain binding (HB) site and BR response element (BRRE) respectively, in the promoters of some BR-repressed genes. BES1 and HAT1 interact with each other and cooperate to inhibit BR-repressed gene expression. Furthermore, HAT1 can be phosphorylated and stabilized by GSK3 (GLYCOGEN SYNTHASE KINASE 3)-like kinase BIN2 (BRASSINOSTEROID-INSENSITIVE 2), a well established negative regulator of the BR pathway. Our results thus revealed a previously unknown mechanism by which BR signaling modulates BR-repressed gene expression and coordinates plant growth.

Introduction

Plant steroid hormones called brassinosteroids (BRs) control plant growth and developmental processes such as cell expansion and division, senescence, vascular development, photomorphogenesis and stress responses. BR biosynthesis- or perception-defective mutants display dwarf phenotypes due to reduced cell elongation (Clouse, 1996; Li and Chory, 1999; Krishna, 2003), while the gain-of-function mutant bes1-D displays opposite phenotypes, including long hypocotyls, petiole and narrower leaves (Yin et al., 2002).

The BR signaling pathway is well established (Li, 2010; Clouse, 2011; Gudesblat and Russinova, 2011; Yang et al., 2011; Ye et al., 2011; Choudhary et al., 2012; Guo et al., 2013; Zhu et al., 2013). BRs are perceived by plasma-membrane-localized and leucine-rich repeat (LRR) receptor kinase BRI1 (BRASSINOSTEROID-INSENSITIVE1) (Li and Chory, 1997; Hothorn et al., 2011; She et al., 2011). In the absence of BRs, BKI1 (BRI1 KINASE INHIBITOR 1) binds to BRI1 and inhibits its function (Wang and Chory, 2006). The absence of BR signaling leads to active BIN2, which then phosphorylates and negatively modulates the functions of the BES1/BZR1 family of transcription factors through multiple mechanisms (Choe et al., 2002; Li and Nam, 2002; Pérez-Pérez et al., 2002; Wang et al., 2002; Yin et al., 2002; Zhao et al., 2002; Li, 2010). In the presence of BRs, BRI1 binds BRs through an extracellular domain, leading to the release of BKI1 (Wang et al., 2011) and the association with a co-receptor, BAK1 (BRI1-ASSOCIATED RECEPTOR KINASE 1) (Li et al., 2002; Nam and Li, 2002; Oh et al., 2010; Gou et al., 2012). BRI1 phosphorylates BSKs (BR SIGNALING KINASES), which are believed to activate protein phosphatase BSU1 (BRI1-SUPPRESSOR 1) (Mora-Garcia et al., 2004; Tang et al., 2008; Kim et al., 2009; Sreeramulu et al., 2013). Dephosphorylation of BIN2 by BSU1 was proposed to inhibit BIN2 function (Kim et al., 2011), which leads to the accumulation of unphosphorylated BES1/BZR1 in the nucleus to control BR target genes (Li, 2010; Clouse, 2011).

BIN2 is a GSK3 (GLYCOGEN SYNTHASE KINASE 3)-like kinase, which phosphorylates serine and threonine residues in consensus sequence S/TXXXS/T. Phosphorylation by GSK3 kinases has been found to control the ubiquitination and proteolysis of a number of important signaling proteins or transcription factors, such as β-catenin in the WNT signaling pathway (Xu et al., 2009). Arabidopsis has 10 GSK3-like kinases, also known as AtSKs (Arabidopsis SHAGGY-like protein kinases), which can be classified into four subgroups (Jonak and Hirt, 2002). BIN2, one of the three members in group II, is a negative regulator in the BR signaling pathway (Li et al., 2001; Li and Nam, 2002; Yan et al., 2009). In addition to phosphorylating BES1/BZR1, recent studies have demonstrated that BIN2 can phosphorylate and modulate many more signaling components involved in BR or other signaling pathways (Vert et al., 2008; Gudesblat et al., 2012a,b; Kim et al., 2012; Tong et al., 2012; Khan et al., 2013).

Genome-wide microarray analyses performed in Arabidopsis have indicated that BRs up- or down-regulate thousands of genes (Goda et al., 2004; Nemhauser et al., 2004; Guo et al., 2009; Li et al., 2010; Sun et al., 2010; Yu et al., 2011). BES1 and BZR1 target genes were identified based on chromatin immunoprecipitation coupled with Arabidopsis tiling arrays (ChIP-chip) and gene expression studies (Sun et al., 2010; Yu et al., 2011). BES1 and BZR1 target thousands of genes that are involved in plant growth and in interactions with other signal pathways. BES1/BZR1 targets include about 200 transcription factors, which probably mediate downstream BR target gene expression. Several BES1/BZR1 targeted transcription factors have been indeed found to mediate BR responses and interactions and other signaling pathways (Yin et al., 2005; Yu et al., 2008; Li et al., 2009, 2010, 2012; Bai et al., 2012; Gallego-Bartolome et al., 2012; Oh et al., 2012).

BES1/BZR1 can interact with other transcription regulators to induce BR target gene expression (Yin et al., 2005; Yu et al., 2008; Li et al., 2009, 2010, 2012; Bai et al., 2012; Gallego-Bartolome et al., 2012; Oh et al., 2012). BES1 can also modulate BR-repressed genes, but mechanisms are not well established. It was recently reported that myeloblastosis family transcription factor-like 2 (MYBL2), a BES1 target gene that encodes a single MYB domain transcription factor, interacts with BES1 to repress some BR-repressed gene expression (Ye et al., 2012). As the loss-of-function mybl2 mutant has a relative weak BR response phenotype, we proposed that there are other BES1 co-repressors that act to inhibit BR-repressed genes.

Arabodopsis thaliana homeodomain-leucine zipper protein 1 (HAT1) belongs to the homeodomain-leucine zipper (HD-ZIP) family of transcription factors that play important roles in plant development and in the response to the environment (Harris et al., 2011). The homeodomain (HD) is responsible for specific DNA binding and the closely associated leucine zipper (LZ) acts as a dimerization motif. HD-ZIP proteins can bind to partially inverted repeats such as CAAT(A/T)ATTG (BS1 site), CAAT(C/G)ATTG (BS2 site) or a slightly modified version TAAT(C/T)ATTA for AtHB2/HAT4 (Harris et al., 2011). HAT1 and its close homologs belong to Class II HD-ZIP, which has nine members in Arabidopsis. Several members of the family, HAT1, HAT4/AtHB2 and AtHB4, are induced by shade avoidance (FR-rich light) (Ariel et al., 2007; Harris et al., 2011). HAT2 is induced by auxin and promotes cell elongation in hypocotyls and leaf petioles when overexpressed (Sawa et al., 2002). Over-expression of other members such as HAT1 and HAT4 seems to have a similar effect in promoting cell elongation (Steindler et al., 1999; Ohgishi et al., 2001; Sawa et al., 2002; Ciarbelli et al., 2008). AtHB4 was also reported to modulate auxin, BRs and gibberellin responses (Sorin et al., 2009). Several Class II HD-ZIP proteins are repressors of gene expression; over-expression of ATHB2, ATHB4, HAT1 or HAT2 was shown to repress the gene expression of other HD-Zip class II members (Ohgishi et al., 2001; Sawa et al., 2002; Ciarbelli et al., 2008; Sorin et al., 2009).

In this study, we found that HAT1 is a BES1 target gene and that HAT1 functions as a BES1 co-repressor. HAT1 appears to function redundantly with other family members such as HAT3 to mediate BR responses. We further found that BIN2 phosphorylates HAT1 and probably stabilizes HAT1. Unlike MYBL2, which does not bind DNA itself, HAT1 binds to DNA elements in BR-repressed gene promoters and cooperates with BES1 to inhibit BR-repressed gene expression.

Results

HAT1 is a direct target of BES1

Recent ChIP-chip studies have indicated that HAT1 and HAT3 were repressed by BL (Brassinolide, the most active BR) and were direct targets of BES1 and BZR1 (Sun et al., 2010; Yu et al., 2011). To confirm these results, we performed quantitative reverse transcription polymerase chain reaction (qRT-PCR) experiments (Figure S1). HAT1 expression was decreased to about 35% in wild-type (WT) seedlings after BL treatment (Figure S1a). In addition, in bes1-D, in which the BES1 protein level is increased, the expression of HAT1 is reduced to 18% without exogenous BL and even more reduced with BL treatment. The expression of HAT3 is similarly regulated (Figure S1b). We also performed ChIP experiments to confirm that HAT1 is a direct target of BES1. The ChIP assay was done using anti-BES1 antibody and a control antibody. TA3, a retrotransposable element, was used as the internal control (Li et al., 2009). In the promoter of HAT1, there are two putative BES1 binding sites. One is BRRE at −941 bp site and the other is the E-box at −1441 bp. ChIP-qPCR was performed (Figure S1c) and the results indicated that BES1 was enriched significantly at BRRE site rather than at E-box or 3′ untranslated region. The results confirm that BES1 binds to the HAT1 promoter at the BRRE site to represses HAT1 expression. Previous studies have also indicated that HAT1 is a direct target of BZR1 and its expression is regulated by BZR1 (Sun et al., 2010).

HAT1 is positively involved in BR responses

To investigate the role of HAT1 in the BR response, we obtained T-DNA insertion alleles of both HAT1 and HAT3, hat1 and hat3, respectively. Then we created the double mutant hat1 hat3. Knockout lines showed an altered BR response as measured by hypocotyl elongation assays in the absence or presence of BL (Figure 1a). While hat1 and hat3 single mutants did not have clear reduced hypocotyl elongation in response to BL, the hat1 hat3 double mutant had a significantly reduced BR response (Figure 1a,b). We also measured the sensitivities of the mutants to brassinazole (BRZ), a BR biosynthesis inhibitor (Asami et al., 2000) in seedlings grown in the dark (Figure S2a,b) or weak light (Figure S2c,d). The hart1 hat3 double mutant is more sensitive to BRZ than WT and hat1, hat3 single mutants. These results indicated that HAT1 and HAT3 function redundantly to modulate BR-regulated growth.

Figure 1.

HAT1 is a positive regulator in the BR pathway. (a) Hypocotyl lengths of 2-week-old seedlings grown in the light in the absence or presence of 100 nm BL (brassinolide, the most active BR) were measured. (b) BL-induction fold in hypocotyl lengths is calculated for indicated genotypes.(c) Two-week-old light-grown WT, bes1-D and bes1-D hat1 hat3 seedlings grown on half-strength MS plates. Average hypocotyl lengths are indicated (in mm). (d) Two-week-old light-grown WT, bri1-5 and bri1-5 hat1 hat3 seedlings grown on half-strength MS plates. Average hypocotyl lengths are indicated (in mm). Averages and standard deviations were calculated from 15 to 30 seedlings. The significance of difference was analyzed by Student's t-test (**P < 0.01).

To further confirm that HAT1 and HAT3 were involved in the BR response, we generated a triple mutant of bes1-D hat1 hat3, which had shorter hypocotyls at seedling stage compared with bes1-D (Figure 1c). We also generated the bri1-5 hat1 hat3 triple mutant (Figure 1d). bri1-5 has a mutation in the extracellular domain of the BR receptor BRI1, and displays a semi-dwarf phenotype (Noguchi et al., 1999). The bri1-5 hat1 hat3 mutant displayed a significantly enhanced dwarf phenotype, compared with bri1-5, with reduced hypocotyl and leaf petiole length, and rounder, curlier and darker-green leaves under light conditions (Figures 1d and 2a,b). Although bri1-5 (Wassilewskija, WS) and hat1 hat3 (Col-0) are in different ecotype backgrounds, the enhancement of the bri1-5 dwarf phenotype by hat1 hat3 mutant is comparable with the increased sensitivity of hat1 hat3 to BR biosynthesis inhibitor BRZ (Figure S2), a finding that suggested that the enhanced phenotype is probably not derived from the ecotypes. Taken together, these results suggest that HAT1 and HAT3 play a positive role in BR-regulated cell elongation.

Figure 2.

Phenotypes of bri1-5 hat1 hat3 and HAT1 over-expression transgenic plants. (a, b) bri1-5 and bri1-5 hat1 hat3 plants at 4-week (a) and adult stages (b). (c) The phenotypes of WT and three different HAT1 over-expression (OX) lines. The average petiole lengths of the third leaves for Wt and three OX lines are indicated (mm). (d) HAT1 protein was detected by western blotting with anti–GFP antibody. Different HAT1 protein levels in different HAT1OX lines. Phosphorylated HAT1–GFP (see Figure 4) was detected by anti–GFP antibody and the sample loading is indicated by a control antibody. Averages and standard deviations were calculated from 15 to 30 seedlings. The significance of difference was analyzed by Student's t-test (**P < 0.01).

Consistent with the conclusion, transgenic plants that over-expressed green fluorescent protein (GFP)-coupled HAT1–GFP under the cauliflower mosaic virus 35S promoter (HAT1OX) displayed enhanced BR response phenotypes, with longer leaf petioles and narrower leaf blades (Figure 2c). Western blotting using an anti-GFP antibody showed that HAT1–GFP accumulated in the HAT1OX plants (Figure 2d). Based on the above results, we conclude that HAT1 is a BES1 direct target gene and plays a positive role in the BR signaling pathway.

BIN2 interacts with HAT1 and phosphorylates HAT1

BIN2 is a GSK3-like kinase and acts as a negative regulator in the BR pathway. Most known GSK3 substrates contain repeats of a short consensus sequence, S/TxxxS/T (S/T corresponds to Ser or Thr and x denotes any other residue) (Woodgett, 2001). HAT1 protein contains more than 20 predicted BIN2 phosphorylation sites (Figure 3a), a situation that suggests that HAT1 is a BIN2 substrate. We tested if BIN2 could directly interact with HAT1. Glutathione S-transferase (GST)–BIN2 and maltose binding protein (MBP)–HAT1 fusion proteins were expressed in E. coli and purified. First, GST–BIN2 was used to pull-down MBP–HAT1 protein, which can be detected by an anti-MBP antibody. While GST protein alone did not pull-down MBP–HAT1, GST–BIN2 did, indicating an interaction between BIN2 and HAT1 (Figures 3b, S3 and S4). To identify the domains in HAT1 that are required for the interaction, we examined the interactions between GST–BIN2 and a series of truncated HAT1 proteins. While deletions to amino acid 134 and 191 in HAT1 had no effect on the GST–BIN2 interaction, deletion to amino acid 233 of HAT1 abolished the interaction with BIN2 (Figure 3b and S3a). Taken together, a LZ motif in HAT1 mediates the interaction between HAT1 and BIN2.

Figure 3.

BIN2 interacts with and phosphorylates HAT1. (a) Schematic representation of predicated HAT1 protein structure. *Putative BIN2 phosphorylation sites. The N-terminal, putative PEST domain (59–91), homeodomain (HD) (134–191) and leucine zipper (LZ) (192–233) are indicated. (b) BIN2 interacts with HAT1 in GST pull-down assay. GST, GST–BIN2 and MBP-tagged full-length or different truncated HAT1 were used in this assay. HAT1 was detected by western blotting with anti-MBP antibody. (c, d) BiFC analysis of BIN2 and HAT1 interactions with fusions to N- and C-terminal fragments of YFP, respectively. The constructs were expressed in tobacco leaves and the reconstitution of YFP is determined. (e) BIN2 phosphorylates HAT1 in vitro. MBP–BIN2 (1 μg) was used to phosphorylate 1 μg MBP or the indicated amount of MBP-BES1 (top panel) or MBP–HAT1 (middle panel). BIN2 phosphorylation of HAT1 is decreased with the increasing concentrations of BIN2 inhibitor bikinin and 50 ng each of BIN2 and HAT1 protein (bottom panel). Phosphorylated proteins are indicated (P), and BIN2 (P) indicates BIN2 autophosphorylation.

The interaction between BIN2 and HAT1 was further confirmed in planta. Bimolecular fluorescence complementation (BiFC) assay was performed in tobacco leaves. BIN2 fused with the C-terminus of yellow fluorescent protein (BIN2–cYFP) interacted with HAT1 fused with the N-terminus of YFP (HAT1–nYFP), leading to the reconstruction of YFP, as fluorescence was detected in cells co-transformed with the corresponding constructs (Figure 3d). No fluorescence was detected in the negative control, in which cYFP plasmid was co-transformed with the HAT1–nYFP plasmid (Figure 3c). These results indicated that HAT1 and BIN2 interact with each other in vitro and in vivo.

In vitro kinase assays were performed with BIN2 and HAT1. We found that BIN2 can phosphorylate HAT1 (Figure 3e middle panel) as well as BES1 (Figures 3e top panel, S5 and S6). BIN2 inhibitor bikinin (De Rybel et al., 2009) inhibited the phosphorylation of HAT1 and BIN2 autophosphorylation (Figures 3e bottom panel and S3b). The results demonstrated that BIN2 can phosphorylate HAT1 protein in vitro.

BIN2 phosphorylation stabilizes HAT1

When using anti–GFP antibody to test the expression of HAT1–GFP in transgenic plants (Figure 2d), two HAT1–GFP bands were detected. To investigate whether HAT1 exists as phosphorylated and/or unphosphorylated forms in plants, HAT1–GFP was immunoprecipitated from HAT1–GFPOX transgenic plants and followed by calf intestinal phosphatase (CIP) treatment (Figures 4a, S7 and S8). The top two bands disappeared and a new lower band, probably the unphosphorylated form of HAT1, appeared after CIP treatment. The results indicated that HAT1 exists mostly as phosphorylated forms in plants and the unphosphorylated HAT1 is barely detectable under normal growth conditions.

Figure 4.

Detection of HAT1 protein levels and forms in plants. (a) HAT1 exists in plants as phosphorylated forms. Immunoprecipitated HAT1 from transgenic plants was treated without or with CIP and detected by western blotting with anti–GFP antibody. Phosphorylated (P) and unphosphorylated HAT1 are indicated. (b) Effects of BL or BRZ on HAT1 level. Two-week-old seedlings grown on half-strength MS plates without or with 100 nm BL or 1000 nm BRZ were used to prepare protein to detect HAT1 (top) and a control protein (bottom) by western blotting using anti–GFP antibody or a control antibody. Protein levels were quantified using ImageJ software (http://rsbweb.nih.gov/ij/index.html) and HAT1 mRNA levels were quantified by qRT-PCR. (c–e) The effects of BL, bikinin on HAT1 (c), BES1 (d) and the control (e) protein accumulation. Two-week-old seedlings grown on half-strength MS plates were treated without (top, mock), with 1 μm BL (middle) or with 100 μm bikinin (bottom) for indicated time and used to prepare protein to detect corresponding protein by western blotting using anti–GFP antibody (c), anti-BES1 antibody (d) or the control antibody (e). (f) Proteasome inhibitor MG132 can inhibit BL-induced degradation of HAT1. Two-week-old seedlings grown on half-strength MS plates were treated without or with 1 μm BL as well as with both BL and 30 μm MG132 for 2 h and used to prepare protein to detect HAT1 (top) and a control protein (bottom).

The phosphorylation of HAT1 by BIN2 promoted us to test the HAT1 protein forms and levels in plants in response to BR treatment. Because there is low or no detectable unphosphorylated HAT1 in plants, we hypothesized that the phosphorylated forms of HAT1 are stable and that the unphosphorylated form is unstable. We treated transgenic plants with or without BL or BRZ. The HAT1 protein level was reduced in plants grown in medium with BL and increased in plants grown in medium with BRZ (Figures 4b and S7). As HAT1–GFP expression was under the control of a constitutive CAMV 35S promoter and the HAT1 mRNA did not change significantly among different treatments (Figure 4b), we concluded that BL can destabilize and BRZ can stabilize HAT1 protein.

To confirm this, we examined HAT accumulation with short-term BL treatments. We found that HAT1 protein increases in the liquid MS (Murashige and Skoog) medium without BL treatment (Figures 4c top panel and S9). In contrast, HAT1 protein clearly decreased in relation to the mock treatment in the presence of 1 μm BL (Figures 4c middle panel, S7 and S9). As expected, BL induces accumulation of unphosphorylated BES1 (Figure 4d middle panel). We also treated plants with the BIN2 inhibitor bikinin and found that phosphorylated HAT1 protein level decreased significantly compared with the mock control (Figures 4c bottom panel and S7), while BES1 was dephosphorylated and accumulated with bikinin treatment (Figure 4d bottom panel). It is worth noting that although HAT1 appears to increase at 1 h after bikinin treatment, its relative level actually decreases after normalizing with the mock control in which HAT1 increased more at the same time point (Figure S9). These results suggest that phosphorylated HAT1 by BIN2 is stable in the absence of BL and in unphosphorylated HAT1 it is unstable.

MG132, a proteasome inhibitor, was used to test if BL-induced decrease of HAT1 was at least partially caused by proteasome-mediated degradation. When we treated plants with BL and MG132 together, the HAT1 protein level significantly increased and the unphosphorylated HAT1 also appeared (Figures 4f and S7). Taken together, our results demonstrated that BIN2 phosphorylates HAT1, and that BIN2 phosphorylation appears to stabilize the HAT1 protein.

HAT1 interacts with BES1 in vitro and in vivo

Previous results have indicated that BES1 represses MYBL2 expression and that BES1 interacts with MYBL2 protein to inhibit BR-repressed gene expression (Ye et al., 2012). We tested if HAT1 also interacts with BES1 to down-regulate BR-repressed genes.

First, we performed GST pull-down assays with several truncated GST–BES1 and MBP–HAT1 (Figures 5a and S10). As shown, GST–BES1 can pull-down MBP–HAT1 significantly compared with GST. While truncated BES1 with deletions of amino acids (aa) 89–140 still interacted with HAT1, BES1 deletion up to aa 198 reduced the interaction and deletion up to aa 272 largely abolished the interaction. The results suggested that aa 140–272 of BES1 are important for interaction with HAT1.

Figure 5.

HAT1 interacts with BES1 in vitro and in vivo (a) BES1 interacts with HAT1 by GST pull-down assay. GST, MBP–HAT1 and GST-tagged full-length or different truncations of BES1 were used in this assay. HAT1 was detected by western blotting with anti-MBP antibody. (b) HAT1 interacts with BES1. GST, GST–BES1 and MBP-tagged full-length or different truncations of HAT1 were used in this assay. HAT1 was detected by western blotting with anti-MBP antibody. (c, d) BiFC analysis of HAT1 and BES1 interactions with HAT1 and BES1 fused to N- and C-terminal fragments of YFP, respectively. The constructs were expressed in tobacco leaves and the reconstitution of YFP is determined. (e) Schematic representation of BES1 structure and corresponding domain involved in the interaction with HAT1. *Putative BIN2 phosphorylation site. DBD (BES1 DNA binding domain) and CTD (BES1 C -terminal domain) are indicated.

To test which domain of HAT1 is involved in the interaction with BES1, GST pull-down assays were performed with GST–BES1 and several truncated MBP–HAT1 (Figure 5b). When HAT1 was deleted up to aa 135, the interaction still existed. But HAT1 with deletion up to aa 192 did not interact with BES1. The region from aa 135 to aa 192 is the homeodomain (HD) in HAT1, which probably mediates the interaction between HAT1 and BES1.

To confirm the in vivo interaction of BES1 and HAT1, we performed a BiFC experiment with HAT1 fused to N-terminal YFP (HAT1-nYFP) and BES1 fused to C-terminal YFP (BES1-cYFP). When the constructs were co-transformed into tobacco leaves, strong fluorescence signals were observed in the nuclei (Figure 5d). As a negative control, we co-transformed HAT1-nYFP and cYFP and there was no detectable fluorescence signal (Figure 5c). These results indicated that HAT1 interacts with BES1 in vivo.

HAT1 binds to a conserved HB site in target gene promoters

To investigate how HAT1 controls BR-repressed gene expression, we examined the promoter of DWF4, a well established BES1 target gene that is also regulated by HAT1 (see next section). In the approximately 1800 bp DWF4 promoter region, there are four BRRE sites for BES1/BZR1 binding (Figure 6a). There is a putative HB-binding site (TAATAATTA) close to the −1780 bp BRRE site (Harris et al., 2011). Electrophoretic mobility shift assays (EMSAs) were carried out to test if HAT1 directly binds to BR target gene promoters with BES1. We designed several DNA probes that contained both BRRE and HB sites or a mutated BRRE with five of the six nucleotides changed or a mutated HB site with eight of the nine nucleotides changed (Figure 6a). The probes were used with the recombinant MBP-BES1 and MBP–HAT1 proteins in the binding assay. BES1 was able to bind to the WT probe, and the binding was competed off by unlabeled WT probe, but not by the mutant 1 probe in which BRRE was mutated (Figure 6b lanes 2–6).

Figure 6.

HAT1 and BES1 can bind to promoters of target genes simultaneously. (a) Schematic representation of DWF4 promoter showing four BRRE and one HB-binding sites. Sequences of wild-type (WT) probe with both BRRE and HB sites and various mutant probes were shown. (b) Electrophoretic mobility shift assays (EMSAs) with BES1 and HAT1 proteins using WT and mutant probes. The BES1, HAT1 and BES1+HAT1 complexes are indicated by arrows. The triangles represent competition with increasing concentrations (50× and 100×) of unlabeled WT or mutant probes.

HAT1 was also able to bind to the WT probe and the binding was also competed off by unlabeled WT probe, but not by the mutant 2 probe in which the putative HB site was mutated (Figure 6b lanes 7–11). Moreover, BES1 and HAT1 were able to bind the WT probe together and the binding was competed off by unlabeled WT probe but not by the mutant 3 probe in which both BRRE and HB sites were mutated (Figures 6b lanes 12–16 and S11). The results demonstrated that HAT1 binds to a conserved HB site in the target gene promoters, together with BES1 binding to BRRE, to control gene expression.

HAT1 and BES1 act cooperatively to inhibit BR-repressed gene expression

It has been shown that HAT1 can function as a transcription repressor (Sorin et al., 2009). The binding of HAT1 and BES1 to HB-binding site and BRRE in the promoter of DWF4, a BR-repressed gene (Figure 6), prompted us to test if HAT1 functions to suppress BR-repressed genes. The expression of six BR-repressed genes, which are also direct targets of BES1 (Ye et al., 2012), were examined in WT, hat1 hat3 and HAT1OX in the absence or presence of BL. The expression of three of the six examined genes increased significantly in hat1 hat3 and decreased in HAT1OX (Figure 7a). Interestingly, all the three gene promoters contained putative HB-binding sites (Figure S12).

Figure 7.

HAT1 and BES1 act cooperatively in inhibiting BR-repressed gene expression. (a) The expression of BR-repressed genes are increased in hat1 hat3 and reduced in HAT1OX both in the absence and presence of BR. qRT-PCR was performed using 4-week-old plants treated with or without 1 μm BL. UBQ5 was used as the reference gene. Data are mean + standard deviation (SD). The averages and SD were from three biological repeats. The significance of difference was analyzed by Student's t test (*P < 0.05, **P < 0.01). HAT1 and BES1 cooperate to inhibit DWF4 gene expression. (b) Transient transfection assays were performed in tobacco leaves with a DWF4promoter-LUC reporter gene co-transfected with BES1 and/or HAT1 via Agrobacterium. The relative expression levels were normalized with total protein. The average and standard deviations (SDs) were from three biological repeats. The significant difference was analyzed by Student's t-test (*P < 0.05, **P < 0.01). (c) A model of HAT1 function in BES1-repressed gene expression. HAT1 and BES1 bind to HB sites and BRRE, respectively, in the gene promoters to repress a subset of BR target genes.

We then performed transient gene expression with DWF4 promoter::luciferase (LUC) reporter gene in tobacco leaves. While expression of BES1 and HAT1 repressed the expression of DWF4pro:LUC gene expression, the expression of the reporter gene was further reduced when both BES1 and HAT1 were co-expressed (Figure 7b). Taken together, our results demonstrated that HAT1 and BES1 bind to corresponding sites and function synergistically in inhibiting BR-repressed gene expression (Figure 7c).

Discussion

Although the BR signal transduction pathway is well established, the mechanisms and network through which BRs control thousands of genes, have just begun to be revealed. In this study, we established that HAT1, a direct target gene of BES1, is a BES1 corepressor in the regulation of BR-repressed gene expression. HAT1 and its close homolog HAT3 function redundantly to promote BR-regulated growth. We also found that HAT1 is phosphorylated by BIN2 kinase and that BIN2 phosphorylation appears to stabilize HAT1 protein, unlike most other known BIN2 substrates. Our results thus provided significant insight into the integration of BR signaling and BR-regulated gene expression.

BRs are known to activate and repress approximately equal numbers of genes. While BES1 interacts with other transcription factors, histone modifying enzymes and transcription elongation factor to activate many BR-induced genes (Li, 2010; Clouse, 2011; Ye et al., 2011), very little information is known about how BR-repressed genes are regulated. We previously have shown that BES1 binding to the BRRE site interacts with one of its target gene product, MYBL2, to repress some BR-repressed genes (Ye et al., 2012). In this study, we have shown that BES1 functions cooperatively with HAT1 to repress BR-repressed gene expression. Unlike MYBL2, which does not bind DNA, HAT1 can bind to specific DNA sequences (HD binding sites) and cooperate with BES1 bound to the BRRE site (Figures 7 and S12). It is likely that BES1 and HAT1 bound to their corresponding sites interact with each other to inhibit BR-repressed gene expression.

Our results have identified an additional substrate for BIN2 kinase, a well established negative regulator in the BR signaling pathway. Recent studies have demonstrated that in addition to the originally identified BES1 and BZR1, BIN2 has additional substrates that play various roles in BR-regulated processes. BIN2 phosphorylates the BES1/BZR1 family of transcription factors and inhibits their functions through several mechanisms, including protein degradation, reduced DNA binding, and/or cytoplasmic retention by 14-3-3 proteins (de Vries, 2007; Li, 2010; Clouse, 2011; Ye et al., 2011). BIN2 can also phosphorylate ARF2 and inhibit its DNA binding (Vert et al., 2008). More recently, it has been reported that the BIN2 phosphorylation of YDA (mitogen-activated protein (MAP) kinase kinase kinase), MKK4 (MAP kinase kinase 4) and SPEECHLESS (SPCH), a basic helix-loop-helix transcription factor, to regulate stomatal development (Casson and Hetherington, 2012; Gudesblat et al., 2012a,b; Kim et al., 2012; Khan et al., 2013). In rice, transcription factor dwarf and low-tillering (DLT) is phosphorylated by GSK2, one of the orthologs of BIN2, and the function of phosphorylated DLT is inhibited by targeted protein degradation or other mechanisms (Tong et al., 2012). These BIN2 substrates are all phosphorylated and inhibited by BIN2. Very recently, MYBL2 has been identified to interact with BES1 and facilitate BES1 in down-regulating BR-repressed gene expression upon phosphorylation and stabilization by BIN2 (Ye et al., 2012). In this study, we found that HAT1, is also a substrate of BIN2, and BIN2 phosphorylation stabilizes HAT1. Our results reinforce the idea that BIN2 phosphorylation can have different consequences in protein stabilities. While BIN2 phosphorylation destabilizes some proteins (BES1, DLT, and SPCH), it can stabilize others (MYBL2 and HAT1). In addition to stabilizing HAT1 protein, BIN2 phosphorylation can affect other HAT1 functions such as DNA binding and interaction with BES1, similar to multiple regulations of BES1/BZR1 by BIN2. These possibilities remain to be determined in future studies.

Our results also imply that BIN2, under specific condition, can be a positive regulator in the BR pathway as it can phosphorylate and stabilize MYBL2 and HAT1, both of which function as positive regulators of BR-regulated growth. While negative regulation of positive regulators BES1/BZR1 by BIN2 phosphorylation can explain the overall dwarf phenotype of bin2-1 (gain-of-function mutant in BIN2 that display a dwarf phenotype), the positive regulation of BIN2 phosphorylation on HAT1 and MYBL2 may explain why gain-of-function bin2 mutants are somewhat different than bri1 mutants (Li et al., 2001; Li and Nam, 2002; Yan et al., 2009).

Our results defined the function of HAT1 and its close homologs HAT3 in BR-regulated gene expression and BR-regulated plant growth. Several lines of evidence support roles for HAT1 and/or HAT3 in BR signaling. First, HAT1 and HAT3 are both direct target genes for BES1/BZR1. Second, HAT1 interacts with BES1 to repress BR-repressed gene expression. Consistent with the role in BR-repressed gene expression, the hat1 hat3 knockout mutant displayed reduced sensitivity to BRs and HAT1OX displayed an enhanced BR signaling phenotype. Finally, the modulation of HAT1 by BIN2 kinase further suggests that HAT1 and it homologs form part of a BR signaling network to control BR-regulated gene expression and BR responses. Several members of class II HD-ZP are induced by shade avoidance and mediate stem cell elongation (Harris et al., 2011). The regulation at least two of these members by BR signaling suggests that there are potential links between BR signaling and shade avoidance responses.

Based on our results, we propose that HAT1 is regulated by BR signaling through BIN2 kinase and BES1 to fine-tune BR-regulated gene expression (Figure S13). At a low level of BR signaling, active BIN2 kinase inhibits BES1 and promotes HAT1 functions, and HAT1 thus plays an important role in BR-repressed gene expression. In contrast, high-level of BRs lead to accumulation of BES1 and decreased levels of HAT1 (due to lack of BIN2 phosphorylation and transcriptional repression by BES1). HAT1 and its homologs thus can be considered as a ‘buffer’ system for BR-repressed gene expression, a finding that explains the fact that BR-repressed genes are up-regulated, but their repression fold is not reduced in the hat1 hat3 double mutant (Figure 7a). We propose that under a given BR level, the balanced actions of BES1 and HAT1 are important for proper BR-regulated growth.

In summary, our results have identified HAT1 as a BES1 corepressor, which help BES1 to inhibit BR-repressed genes and thus acts as a positive regulator in the BR pathway. We also establish that HAT1 is phosphorylated by BIN2 kinase and that BIN2 phosphorylation may stabilize HAT1 protein. Our study, therefore, provides previous unknown mechanisms for BR-repressed gene expression and the integration of BR signaling and the BR transcriptional network.

Experimental Procedures

Plant materials and growth conditions

Arabidopsis thaliana ecotype Columbia (Col-0) was used as the WT control. The bes1-D mutant in Col-0 background has been described previously (Gonzalez-Garcia et al., 2011). T-DNA insertion mutants hat1 and hat3 were obtained from ABRC (Arabidopsis Biological Resource Center) (Alonso et al., 2003), corresponding to line SALK_059835 and SALK_056541. All the plants were grown on half-strength MS plates and/or in soil under long-day conditions (16 h light/8 h dark) at 22°C.

Hypocotyl length measurements

Seeds sterilized by 70% (v/v) ethanol and 0.1% (v/v) Triton X-100 were grown on half-strength MS medium supplemented with 1% phytoagar. After 3 days of incubation at 4°C, seeds were irradiated by white light for 12 h to promote germination and then incubated in the dark or under specific light conditions. BRZ and BL were added to the half-strength MS agar medium during the assays.

Plasmid construction

The DNA primers used in this studies are listed in Table S1. For GFP-tagged transgenic plants, the HAT1 genomic sequence was cloned from WT plants and fused with a GFP tag into the pZP211 vector (Hajdukiewicz et al., 1994). For the recombinant protein and GST pull-down assay, the HAT1 coding region was amplified from Col-0 cDNA and various deletion constructs were incorporated into the pETMAL vector. BIN2, BES1 and truncated BES1 fragments were cloned into pET42a(+) (Novagen, http://www.merckmillipore.com). For the BiFC assay, constructs of the N- or C-termini of YFP were used, which has been described previously (Yu et al., 2008). The coding regions of HAT1 and BIN2 were inserted into the YFP-C and YFP-N construct, respectively. The coding regions of HAT1 and BES1 were inserted into YFP-N and YFP-C construct, respectively.

Transgenic plants

The construct of HAT1–GFP driven by 35S promoter were transformed into Agrobacterium tumefaciens (stain GV3101), which were used to transform plants by the floral dip method. Transgenic lines were selected on half-strength MS medium that contained 50 μg ml−1 kanamycin. Transgene expression was analyzed by western blotting.

Biomolecular fluorescence complementation (BiFC)

BiFC analysis was performed as described (Gallego-Bartolome et al., 2012). The assay was repeated more than three times with similar results.

GST pull-down assay

HAT1 and HAT1 fragments fused with MBP were purified with amylose resin (NEB). BIN2, BES1 and BES1 fragments fused with GST were purified with glutathione beads (Sigma, http://www.sigmaaldrich.com). GST pull-down assays were performed as described (Yin et al., 2002). The assays were repeated three times with similar results.

Gene expression analysis

Total RNA was extracted and purified from 4-week-old plants of different genotypes using the RNeasy Mini Kit (Qiagen, http://www.qiagen.com). Mx4000 multiplex quantitative PCR (qPCR) system (Stratagene, http://www.stratagene.com) and SYBR Green PCR Master Mix (Applied Biosystems, http://www.appliedbiosystems.com.cn) were used in quantitative real-time PCR (real-time qPCR) analysis. DWF4 promoter (972 bp including 5′-UTR) were cloned and used to drive luciferase reporter gene expression. HAT1 and BES1 coding regions driven by CaMV 35S promoter were cloned into the pZP211 vector. Tobacco leaf transient assay (Antony et al., 2010) was used to examine the repression effect of HAT1 on reporter gene expression in the presence or absence of BES1. Equal amounts of Agrobacterium cells was infiltrated into tobacco leaves. The luciferase activities were measured from protein extracts from triplicate samples using a Berthold Centro LB960 luminometer (https://www.berthold.com) with luciferase assay system (Promega, http://www.promega.com). The luciferase levels were normalized by the total protein from each sample. The average and standard deviations were from three biological repeats.

Phosphorylation assays

The in vitro kinase assay was performed as previously described (Yin et al., 2002). MBP, MBP-BES1, and MBP–HAT1 were incubated with MBP–BIN2 kinase in 20 μl of kinase buffer [20 mm Tris (pH 7.5), 100 mm NaCl, and 12 mm MgCl2] and 10 μCi 32P-ATP. After incubation at 37°C for 60 min, the reactions were stopped by adding 20 μl of 2× sodium dodecyl sulfate (SDS) buffer and boiling for 5 min. Proteins were resolved by polyacrylamide gel electrophoresis (PAGE) and phosphorylation was detected by exposing the dried gel to an X-ray film. HAT1 protein was immunoprecipitated from transgenic plants and treated with or without calf alkaline phosphatase (CIP) as described (Yin et al., 2002). Proteins from 35S::HAT1–GFP transgenic plants were used for phosphatase (CIP; NEBhttp://www.neb.com) treatments at 37°C 1 h. The assays were repeated three times with similar results.

Electrophoretic mobility shift assays (EMSAs)

EMSAs were performed as described previously (Yin et al., 2005). Briefly, oligonucleotide probes were synthesized, annealed, and labeled with 32P-γ-ATP using T4 nucleotide kinase (NEB). The binding reactions were carried out in 20 μl binding buffer [25 mm HEPES-KOH pH 8.0, 50 mm KCl, 1 mm dithiothreitol (DTT) and 10% glycerol] with approximately 1 ng probe (10 000 cpm) and recombinant proteins purified from E. coli. After 30 min incubation on ice, the reactions were resolved by 5% native polyacrylamide gels with 1× TGE buffer (6.6 g L−1 Tris, 28.6 g L−1 glycine, 0.78 g L−1 EDTA, pH 8.7) and exposed to a phosphorimaging screen. The assays were repeated three times with similar results.

Acknowledgements

We thank Ana Cano-Delgado for providing the bes1-D mutant on a Col-o background. This work was supported by United States National Science Foundation (NSF) grants (IOS-1122166 and IOS-1257631) and Plant Science Institute (to YY), the National Science Foundation of China (31070210, 91017004 and 30970214), the Doctoral Foundation of the Ministry of Education (20110181110059), the National Key Basic Research ‘973’ Program of China (2009CB118500) to HL. DZ is supported by a fellowship from the Chinese Scholarship Council (CSC).

Ancillary