Abscisic acid-dependent multisite phosphorylation regulates the activity of a transcription activator AREB1

Abstract

bZIP-type transcription factors AREBs/ABFs bind an abscisic acid (ABA)-responsive cis-acting element named ABRE and transactivate downstream gene expression in Arabidopsis. Because AREB1 overexpression could not induce downstream gene expression, activation of AREB1 requires ABA-dependent posttranscriptional modification. We confirmed that ABA activated 42-kDa kinase activity, which, in turn, phosphorylated Ser/Thr residues of R-X-X-S/T sites in the conserved regions of AREB1. Amino acid substitutions of R-X-X-S/T sites to Ala suppressed transactivation activity, and multiple substitution of these sites resulted in almost complete suppression of transactivation activity in transient assays. In contrast, substitution of the Ser/Thr residues to Asp resulted in high transactivation activity without exogenous ABA application. A phosphorylated, transcriptionally active form was achieved by substitution of Ser/Thr in all conserved R-X-X-S/T sites to Asp. Transgenic plants overexpressing the phosphorylated active form of AREB1 expressed many ABA-inducible genes, such as RD29B, without ABA treatment. These results indicate that the ABA-dependent multisite phosphorylation of AREB1 regulates its own activation in plants.

The phytohormone abscisic acid (ABA) plays important roles in seed maturation and dormancy and is also involved in the adaptation of vegetative tissues to abiotic environmental stresses, such as drought and high salinity. ABA promotes stomatal closure in guard cells and regulates the expression of many genes, the products of which may function in dehydration tolerance in both vegetative tissues and seeds. Many ABA-inducible genes contain a conserved element named ABA-responsive element (ABRE) (PyACGTGG/TC) in their promoter regions. The ABRE functions as a cis-acting element and is involved in ABA-responsive gene expression (for reviews, see refs. 1 and 2).

The RD29B promoter region contains two ABREs, and analyses with the ABA-deficient and insensitive mutants aba1 and abi1 revealed that the drought-inducible expression of RD29B is controlled mainly by ABA (3–6). Yeast one-hybrid screening with the RD29B promoter including ABREs enabled us to clone three independent cDNAs encoding ABRE-binding proteins (AREB1, AREB2, and AREB3) in Arabidopsis (7). Each AREB protein contained a single bZIP-type DNA-binding domain, and expression of AREB1 and AREB2 was up-regulated by ABA, drought, and high-salinity stresses, shown to function as trans-acting activators by using transient expression in protoplasts (7). Choi et al. (8) also reported the cloning of four independent cDNAs for ABRE-binding factors (ABF1, ABF2, ABF3, and ABF4) from Arabidopsis. ABF2 and ABF4 were identical to AREB1 and AREB2, respectively.

In the Arabidopsis genome, 75 distinct bZIP transcription factors exist, and nine members are classified as a homologous subfamily of AREBs that contain three N-terminal (C1, C2, and C3) and one C-terminal (C4) conserved domains (9–12). Most of the AREB subfamily proteins are involved in ABA-responsive signal-transduction pathways in vegetative tissues or seeds (7, 8, 13). Orthologues of AREB subfamily proteins were also reported in rice (TRAB1) (14) and in barley (HvABI5) (15) and were also involved in ABA signal-transduction pathways. Arabidopsis AREB1/ABF2, AREB2/ABF4, and ABF3 were mainly expressed in vegetative tissues but not in seeds (7, 8), whereas ABI5 and EEL were expressed during seed maturation and/or germination (9, 13, 16). It is possible that redundancy and tissue-specific expression of these genes may be important for their functions.

Kang et al. (17) reported that overexpression of ABF3 and ABF4/AREB2 resulted in ABA-hypersensitive phenotypes in germination and seedling growth stages in Arabidopsis. These transgenic plants also showed improvement of drought stress tolerance, suggesting that AREB/ABF proteins are involved in ABA response and stress tolerance in plants. However, AREB1 and AREB2 require ABA for their maximum activation, as shown by their low transactivation abilities in protoplasts prepared from the ABA-deficient aba2 mutant (7). We have shown that the ABA-responsive 42-kDa kinase activities phosphorylate conserved regions of AREBs, suggesting that ABA-dependent phosphorylation may be involved in activation of the AREB subfamily proteins (7). Phosphorylation/dephosphorylation-regulated events were reported to play important roles in ABA signaling; SNF1-related protein kinase homologues, ABA-activated protein kinase (AAPK) in Vicia faba (18), and SRK2E/OST1 in Arabidopsis (19, 20) modulate ABA-dependent stomatal closure. ABI1 and ABI2, of which dominant-negative mutation causes ABA-insensitive mutants abi1 and abi2, encode a type 2C protein phosphatase (4, 5, 21). Because null mutations of ABI1 and ABI2 resulted in ABA hypersensitivity, ABI1 and ABI2 negatively regulate ABA-dependent responses (22).

Here, we report that the ABA-activated 42-kDa kinase activity phosphorylates Ser/Thr residues in the conserved R-X-X-S/T sites of AREB1. Amino acid substitution of the Ser/Thr residues to Ala and Asp resulted in suppression and high transactivation activity, respectively. A phosphorylated active form of AREB1 was obtained by substitution of Ser/Thr to Asp in all conserved R-X-X-S/T sites. We show that transgenic plants overexpressing the phosphorylated active form AREB1 express not only ABA-inducible genes, such as RD29B, but also seed-specific genes without ABA treatment. We also discuss the activation mechanisms of AREB1 by ABA-dependent phosphorylation in plants.

Results

Effects of Amino Acid Substitutions of Putative Phosphorylation Target Sites on ABA-Dependent Phosphorylation of AREB Proteins.

ABA-dependent 42-kDa protein kinases phosphorylate three conserved regions of AREB (C1, C2, and C3), which contain possible target sequences for Ser/Thr protein kinases (7). In the conserved regions of AREB, the most common sequences were R-X-X-S/T (C1-1, C2-1/2, C3-1, and C4-1, putative targets for CDPK etc.) and S/T-X-X-E/D (C1-2, C2-3, C3-1/2, for CK II) (Fig. 1A).

Fig. 1.

Putative target sites of protein kinases within AREB conserved regions and ABA-dependent phosphorylation of recombinant AREB proteins by in-gel kinase activity assay. (A) The conserved regions, C1, C2, C3, and C4 contain protein kinase target sequences. In the conserved regions of AREBs, the most common sequences were R-X-X-S/T (C1-1, C2-1/2, C3-1, and C4-1, putative targets for CDPK etc.) and S/T-X-X-E/D (C1-2, C2-3, C3-1, and C3-2, for CK II). Color lines with letters a (red), b (blue), and c (green) indicate the polypeptides used for the in-gel kinase activity assay, which provides data in B–D. (B) ABA-dependent phosphorylation of recombinant AREB proteins by the in-gel kinase activity assay. T87 cell extracts treated with or without 50 μM ABA treatment were resolved on 10% polyacrylamide gel containing recombinant AREB1a and AREB2a polypeptides shown diagrammatically in A. The protein kinase activities were analyzed as described in Materials and Methods. (C) The wild-type and mutated recombinant AREB1b polypeptide shown diagrammatically in A were used as substrates. (D) The wild-type and mutated recombinant AREB1c (shown diagrammatically in A) polypeptides were used as substrates.

To investigate which target sequence(s) were phosphorylated, an in-gel kinase assay was carried out by using amino acid-substituted recombinant AREB polypeptides as substrates. When a recombinant AREB1 polypeptide AREB1a S26A (Ser 26-to-Ala substitution) was used, ABA-dependent protein kinase activity disappeared; however, a T31A substitution in AREB1a did not affect its phosphorylation (Fig. 1B). A recombinant AREB2a polypeptide with a substitution of S39A could not be phosphorylated, but a T44A substitution of AREB2a did not affect its phosphorylation (Fig. 1B). When AREB1b was tested with a single substitution S86A or S94A, phosphorylated bands were evident; however, they disappeared with double substitutions S86A and S94A (Fig. 1C). AREB1c with a substitution of T135A was not phosphorylated (Fig. 1D). These results suggest that Ser/Thr residues at R-X-X-S/T in the conserved regions of AREB were phosphorylated by ABA-dependent 42-kDa protein kinase activity in T87 cell extracts, but not at S/T-X-X-E/D or other sites.

Properties of Phosphorylation Activities.

The 42-kDa protein kinase activity was inhibited by a protein kinase inhibitor staurosporine (Fig. 2A). Staurosporine suppressed transactivation activities of the AREB proteins in Arabidopsis leaf protoplasts (7) but not by ≈10 mM BAPTA (EGTA derivative Ca²⁺-specific chelating reagent) treatment (Fig. 2B). Upon addition of 5 mM CaCl₂ before the phosphorylation reaction, additional kinase activity appeared near 60 kDa, and its M_r was comparable with AtCDPK (23). These results suggest that the 42-kDa kinase activity phosphorylating the AREB polypeptides does not require Ca²⁺ for its activation.

Fig. 2.

Properties of AREB1 protein fragment phosphorylation. (A) Effects of protein kinase inhibitor staurosporine on phosphorylation of the recombinant AREB1a polypeptide. Staurosporine (20 and 100 nM) was added to reaction mixture. (B) Effects of BAPTA on phosphorylation of the recombinant AREB1b polypeptide. Na-BAPTA (5 mM final concentration) was added to reaction buffer and equilibrated for 30 min before addition of radiolabeled ATP. In the far right lane, additional kinase activity appeared near 60 kDa (see Results for details). (C) Stress-dependent phosphorylation of the recombinant AREB1b polypeptide. Protein extracts prepared from T87 cells treated for 30 min with 50 μM ABA (Ab), 0.5 M NaCl (Na), and 0.8 M mannitol (Os, high osmolality) and at low temperature (Lt, 4°C) or untreated (Ct) were used for in-gel kinase activity assay. The recombinant AREB1b polypeptide was used as a substrate. (D) ABA-activated SnRK2-type protein kinases phosphorylate the recombinant AREB1b polypeptide. Protein extracts prepared from transgenic T87 cells overexpressing each SnRK2-GFP fusion protein under control of the CaMV 35S promoter were used for in-gel kinase activity assay. Phosphorylated bands derived from SnRK2-GFP fusion proteins were indicated by circles. Arrowheads indicate the position of 42 kDa in A–D.

Stress-responsive phosphorylation of AREBs was examined with an in-gel protein kinase assay. Phosphorylation of the AREB1b polypeptide was also detected with 0.5 M NaCl and 0.8 M mannitol (30 min) treatments but not by low temperature (Fig. 2C). The NaCl- and mannitol-activated kinases were smaller in M_r than the ABA-activated kinase, suggesting that additional kinases exist for the phosphorylation of AREB1b in response to salt and/or osmotic stresses.

ABA-Activated SNF1-Related Protein Kinase (SnRK)2 Protein Kinases Phosphorylate the AREB1 Polypeptide.

The SnRK2 family, including AAPK in fava bean, SRK2A-J (SnRK2.1–10) in Arabidopsis, and SAPK1–10 in rice, has been recently identified as protein kinases activated by hyperosmotic stresses and/or ABA treatment (18–20, 24–26). Because the M_r of these SnRK2-type protein kinases is ≈42 kDa and they do not require Ca²⁺ for activation, we examined the phosphorylation of a recombinant AREB polypeptide by Arabidopsis SnRK2 protein kinases, including SRK2C (SnRK2.8), SRK2D (SnRK2.2), SRK2E/OST1 (SnRK2.6), SRK2F (SnRK2.7), and SRK2I (SnRK2.3), which are activated by ABA (20, 24). We performed an in-gel protein kinase assay using these SnRK2-GFP fusion proteins overexpressed in Arabidopsis T87 cultured cells (Fig. 2D). The AREB1b polypeptides were phosphorylated by all these ABA-activated SnRK2-GFP proteins, but the amino acid-substituted AREB1b (S86A) and AREB1b (S94A) polypeptides were not (see Fig. 7, which is published as supporting information on the PNAS web site; data not shown). Also, an activated form of human MAP kinase 2, which is one of the commercially available recombinant proteins in Escherichia coli (Calbiochem #454854), did not phosphorylate the AREB1b polypeptide, which SnRK2 kinases can do (see Fig. 8, which is published as supporting information on the PNAS web site). These data suggest that the SnRK2-type protein kinases can phosphorylate the AREB1 proteins and that the target sequences of SnRK2 were R-x-x-S/T sites of AREB1.

Transactivation Activity of the Amino Acid-Substituted AREB1 Proteins.

We performed transient transactivation experiments using T87 cells to analyze the effect of amino acid substitution in the phosphorylation sites of AREB1 on its transactivation activity. Protoplasts were cotransfected with a GUS reporter gene fused to the 77-bp fragment containing two ABRE motifs and each effector plasmid (Fig. 3A). Amino acid substitution of each R-x-x-S/T phosphorylation target site in the N-terminal three conserved regions of AREB1 to Ala (Fig. 3A; S26A, S86A, S94A, and T135A) decreased GUS reporter gene expression in response to ABA, whereas mutations in the C-terminal conserved region (Fig. 3A; S413A) did not alter GUS gene expression. Amino acid substitution of two or three phosphorylation sites decreased the expression of the GUS gene more strongly than that of a single phosphorylation site (Fig. 3B; M1 and M2). When four phosphorylation sites, Ser/Thr in C1, C2, and C3 conserved regions were substituted to Ala, GUS expression in response to ABA was suppressed almost completely (Fig. 3B).

Fig. 3.

Transient transactivation analysis of amino acid-substituted AREB1. Transient transactivation of the RD29B promoter-GUS fusion gene by AREB1 and its amino acid-substituted proteins by using Arabidopsis protoplasts prepared from T87 suspension cells. The reporter GUS gene driven by the 77-bp DNA fragment of the RD29B promoter containing two ABREs was transfected into T87 protoplasts with each effector plasmid or the vector plasmid (pBI35SΩ) as a control. To normalize for transfection efficiency, the CaMV 35S promoter/luciferase plasmid was cotransfected in each experiment. Each effector plasmid contains the CaMV 35S promoter and TMV Ω sequence fused to wild-type or amino acid-substituted AREB1 ORF. Bars indicate the SD of triplicates. Transformed protoplasts were incubated with (open) or without (solid) 50 μM ABA in culture medium, at 22°C for 16–20 h in the dark in all transient transactivation experiments shown. (A) Effects of each single substitution of the R-X-X-S/T sites. (B) Effects of multiple amino acid substitutions (Ser/Thr to Ala). M1; S94A, M2; S(86, 94)A, M3; S(26, 86, 94)A, M4; S(26, 86, 94)A and T135A, M5; S(26, 86, 94, 413)A and T135A. (C) Effects of single or multiple amino acid substitutions (Ser/Thr to Asp). M6; S94D, M7; S(26, 86, 94, 413)D, M8; S(26, 86, 94, 413)D and T135D.

On the other hand, amino acid substitutions of Ser/Thr residues of the phosphorylation sites to Asp, which provides a negative charge and mimics their phosphorylated status, resulted in higher expression of the reporter gene without ABA application (Fig. 3C). When any one of these phosphorylation sites was substituted, ABA-independent activation of GUS expression increased partially (Fig. 3C; M6). Substitution in several target sites further increased GUS expression without ABA application, and substitutions of Ser/Thr to Asp residues in all five R-x-x-S/T sites resulted in almost complete activation without ABA (Fig. 3C). These results suggest that ABA-dependent phosphorylation of the Ser/Thr residues in the AREB1 conserved regions regulates AREB1 activation.

Overexpression of the Phosphorylated Active Form of AREB1 in Arabidopsis.

To analyze the role of multisite phosphorylation of AREB1, the phosphorylated active form of AREB1 with amino acid substitution of the Ser/Thr residues of all conserved R-x-x-S/T sites to Asp was overexpressed in Arabidopsis plants under the control of the enhanced CaMV35S promoter (27). After quantitative RT-PCR analysis, three lines of 35S:AREB1pa-1, 35S:AREB1pa-2, and 35S:AREB1pa-3 were selected for further analysis, which showed weak, moderate, and strong transgene expression, respectively (Fig. 4A). We also analyzed RD29B expression levels and found that these transgenic plants accumulated the RD29B mRNA without ABA treatment, and levels of its expression were correlated with the expression levels of the transgene (Fig. 4A).

Fig. 4.

Effects of overexpression of the phosphorylated active form of AREB1 on plant growth and expression of the RD29B gene. (A) Three and one independent lines of transgenic plants harboring 35S:AREB1pa and the pBI121 vector (pBI121) were grown on germination media agar plates, respectively. The relative amount of the AREB1 and RD29B mRNAs in the transgenic plants was analyzed by quantitative RT-PCR (amount of the most abundant point was indicated as 100). Total RNA was prepared from 3-week-old seedlings of transgenic plants. Pictures of 3-day-old (B) and 9 day-old (C) seedlings. (D) Total RNA was prepared from 3-week-old seedlings of transgenic plants treated with or without 50 μM ABA. The relative amount of the RD29B mRNA in the transgenic plants was analyzed by quantitative RT-PCR and was indicated after considering the nontreated vector control as “1.”

We subsequently observed germination and growth of the 35S:AREB1pa plants grown on germination media agar plates and compared them with control plants carrying the pBI121 vector (pBI121) at 3 and 19 days after sowing. The germination rates of these transgenic plants were lower than those of the control plants, and this reduction correlated with levels of mRNA accumulation of AREB1 among the three transgenic lines (Fig. 4B). Most of the transgenic plants grew normally, but 35S:AREB1pa-3 overexpressing the transgene revealed slight growth retardation at 19 days (Fig. 4C).

We then analyzed the effect of ABA treatment on RD29B expression in the 35S:AREB1pa plants in comparison with control plants carrying the pBI121 vector and the 35S:AREB1wt plants overexpressing the wild-type AREB1 cDNA. Expression of RD29B was higher in 35S:AREB1pa than 35S:AREB1wt without ABA treatment and RD29B expression increased by ABA treatment (Fig. 4D). Levels of RD29B expression in the ABA-treated 35S:AREB1pa plants were similar to 35S:AREB1wt plants but were higher than those in the pBI121 plants (Fig. 4D).

The Active Form of AREB1 Can Induce Expression of Target Genes in an ABA-Deficient Mutant aba2-1.

To clarify whether the phosphorylated active form of AREB1 can induce ABRE-dependent gene expression without ABA in plants, we generated transgenic plants overexpressing the phosphorylated active form and wild type of AREB1 by using an ABA-deficient mutant aba2-1 (35S:AREB1wt/aba2 and 35S:AREB1pa/aba2). RD29B expression was detected without ABA treatment in 35S:AREB1pa/aba2 but not in 35S:AREB1wt/aba2 (Fig. 5A). These results clearly indicated that the active form of AREB1 induced expression of the downstream gene without ABA.

Fig. 5.

Effects of overexpression of the phosphorylated active form of AREB1 in the ABA-deficient aba2–1 mutant. (A) Total RNA was prepared from 3-week-old seedlings of transgenic aba2–1 mutants harboring 35S:AREB1pa, 35S:AREB1wt, and pBI121. Quantitative RT-PCR was performed as described in Fig. 4D. (B) Effects of dehydration stress on the plants. Three-week-old transgenic plants were grown on germination media agar plates, transferred to filter papers and left for 30 min. (C) Electrolyte leakage was measured before and after dehydration treatment of 3-week-old plants removed from agar plates, on filter paper for 30 min, as described in ref. 38. Bars indicate SDs. (D) Photographs of the guard cells were taken through a color laser 3D profile microscope (Keyence, Osaka) as described in ref. 35. (Scale bars, 50 μm.)

We subsequently analyzed the effect of overexpression of the active form of AREB1 on plant stress tolerance. The aba2-1 mutant 35S:AREB1wt/aba2 and two lines of 35S:AREB1pa/aba2 plants were removed from the agar plates and kept on dry plastic plates for 30 min (15% relative humidity). The aba2-1 and 35S:AREB1wt/aba2 plants wilted after 30 min, whereas 35S:AREB1pa/aba2 did not (Fig. 5B). Electrolyte-leakage analysis is a sensitive measure of loss-of-membrane integrity and is commonly used to assay osmotic injury (28). In our study, after 30 min of desiccation, aba2-1 and 35S:AREB1wt/aba2 plants exhibited 51% and 64% electrolyte leakage, respectively (Fig. 5C). For 35S:AREB1pa-a/aba2 and 35S:AREB1pa-b/aba2 plants, the ion leakage was 27% and 30%, respectively. Under normal conditions, all plant lines exhibited a nearly equivalent electrolyte leakage percentage (≈15%). These data indicated that electrolyte leakage of plants overexpressing the active form of AREB1 was reduced significantly after a dehydration stress treatment.

Stomatal closure under dehydration conditions is a crucial process regulated by ABA (1). We therefore analyzed stomatal closure in the guard cells of the same plant lines and found that overexpression of the active form of AREB1 did not affect the stomatal closure (Fig. 5D), suggesting that ABA-dependent stomatal closure was not involved in the AREB1-dependent signal-transduction pathway.

Microarray Analysis of the Transgenic Arabidopsis Plants Overexpressing the Phosphorylated Active Form of AREB1.

To analyze genes affected by overexpression of the phosphorylated active form of AREB1 in Arabidopsis, we compared the expression profiles in two independent lines of 3-week-old 35S:AREB1pa plants under unstressed conditions with that of vector control plants by using Agilent Technologies (Palo Alto, CA) Arabidopsis 22K oligonucleotide array. Genes with expression increased >3 times that of the average of two independent 35S:AREB1pa lines listed in Table 1, which is published as supporting information on the PNAS web site.

Quantitative RT-PCR was performed to confirm the expression of several genes identified by the microarray analysis (Fig. 6). Many genes for LEA-class proteins including RD29B and RAB18 were overexpressed in the 35S:AREB1pa plants, even without ABA treatment (Table 1 and Fig. 6A). As compared with vector control plants, expression of these genes was not increased in 35S:AREB1wt without ABA but was induced to higher levels by ABA treatment (Fig. 6A). These LEA-class genes contain at least two ABRE motifs in their promoter regions (Table 1), and these data collectively suggest that the LEA-class genes are candidates for direct targets of AREB1.

Fig. 6.

Quantitative RT-PCR analysis of up-regulated genes in transgenic plants overexpressing the constitutive active form of AREB1. mRNA accumulation of up-regulated genes identified in microarray analysis were confirmed by using quantitative RT-PCR. (A) Expression of genes for LEA-class proteins [At5g52300 (RD29B), white; At3g17520, gray; At5g66400 (RAB18), black]. (B) Expression of seed-specific genes. [At2g05580, white; At5g50600 (steroleosin), gray; At3g27660 (oleosin), black]. The x axis indicates the relative amount of mRNA after considering the most abundant point as “100.”

On the other hand, several genes for seed storage proteins were also overexpressed in the transgenic plants (Table 1). These genes were not induced by ABA in the vector control plants (Fig. 6B). Some of the genes were categorized in VP1 “AND” ABA coregulated genes (12), indicating that the active form of AREB1 also activates the expression of genes involved in seed development in the vegetative tissues. Most of these genes contain the Sph sequence, which is a binding site of ABI3/VP1, in their promoter regions (Table 1).

Discussion

Although the expression of AREB1 is induced by ABA application, overexpression of AREB1 is not sufficient to activate ABRE-dependent gene expression. On the basis of recent reports that AREB1 and its homologues, such as ABI5, TRAB1, and TaABF, are phosphorylated in vitro or in vivo (7, 29–31), phosphorylation of AREB1 may be involved in the posttranscriptional modification. Because of the occurrence of highly conserved target sites for putative protein kinases in the conserved regions of AREB proteins, we analyzed ABA-dependent phosphorylation of these sites of AREB1 (Fig. 1). We found that ABA-activated 42-kDa kinase activity phosphorylates Ser/Thr residues of the R-X-X-S/T sites in the conserved regions of AREB1, and conversion of Ser/Thr to Ala in these R-X-X-S/T sites suppressed transactivation activity of AREB1. In contrast, amino acid substitutions of the Ser/Thr residues to Asp conferred the high transactivation activity without exogenous ABA application both in Arabidopsis protoplasts and in transgenic plants (Figs. 3 and 4). In amino acid substitutions of the Ser/Thr residues to Ala and Asp, multiple substitutions of these sites resulted in almost complete suppression and activation of transactivation activity in transient assay, respectively (Fig. 3). These results revealed that the ABA-dependent multisite phosphorylation of AREB1 activates ABRE-dependent gene expression.

We demonstrated that a 42-kDa ABA-activated protein kinase phosphorylates R-X-X-S/T sequences in the conserved regions of AREB1 by using an in-gel kinase activity assay (Fig. 1) (7). The M_r of this kinase is close to the predicted molecular masses of the SnRK2 proteins (from 38 to 43 kDa), which are known to be activated by ABA or hyperosmolarity (20, 24, 32). In this work, we showed that ABA-activated SRK2C, SRK2D, SRK2E, SRK2F, and SRK2I (24) can phosphorylate the AREB1 polypeptide by using in-gel kinase activity assay (Fig. 2). However, SRK2E/OST1 is expressed in guard cells and the vascular system and mainly regulates ABA-dependent stomatal closure (19, 20). On the other hand, SRK2C is mainly expressed in root tips, and its overexpression in transgenic plants increased expression of stress-inducible genes, such as RD29A, COR15A, and DREB1A, but not target genes of AREB1, such as RD29B (32). Temporal and spatial expression analysis of the SnRK2 genes may be important for further elucidation of AREB protein activation in response to ABA. Recently, Kobayashi et al. (26) reported that the rice SAPK8, SAPK9, and SAPK10 that are activated by ABA can phosphorylate an ABRE-binding factor TRAB1. The SnRK2-mediated activation of transcription by ABREs may be important in rice and Arabidopsis.

Transient transactivation analysis using amino acid substituted AREB1 as effectors, clearly indicated that multiple phosphorylation of the R-X-X-S/T sites of the conserved regions activates AREB1 (Fig. 3C). In animal cells, the activity of many transcription factors is regulated by multisite phosphorylation, such as heat-shock factor1 (HSF1), p53, activating transcription factor 2 (ATF2), and nuclear factor of activated T cells (NFAT) (33). Multisite phosphorylation of ATF2 and p53 affects their stability, and, in the case of NFAT, a high-phosphorylation state prevents nuclear localization. The transcriptional activity of HSF1 is regulated by multisite phosphorylation. Moreover, phosphorylation of several transcription factors facilitates protein–protein interaction and affects DNA-binding activity. Thus, phosphorylation can directly modulate the activity of transcription factors.

Microarray analysis using transgenic plants overexpressing the active form of AREB1 showed that not only ABA-inducible genes but also seed-specific genes were overexpressed in the vegetative tissues (Table 1). Although these seed-specific genes are strictly regulated to express in seeds and not in vegetative tissues in Arabidopsis, overexpression of the active form of AREB1 overcomes its repression in the vegetative tissues. Some of these genes were categorized as VP1- “AND” ABA-dependent genes (12), suggesting that ABI3/VP1 regulation of these genes is important for their specific expression in seeds. The ABI3 protein is a seed-specific transcriptional activator and interacts with ABI5. Ectopic expression of ABI3 from a constitutive promoter confers the ability to accumulate seed-specific transcripts in response to exogenous ABA to vegetative tissues of Arabidopsis (34). However, the phosphorylated active form of AREB1 can induce ABA-dependent and seed-specific genes ectopically in the vegetative tissues without contribution of ABI3 (Table 1 and Fig. 6). Thus, interactions between AREB1 and some specific proteins through multisite phosphorylation may be involved in the regulation of these genes.

Materials and Methods

Plant Materials and Generation of Transgenic Plants.

Arabidopsis plants and T87 suspension cultured cells were grown, transformed, and treated as described in refs. 20 and 35. ORFs of the wild-type or amino acid-substituted AREB1 were cloned into the binary vector pBE2113Not as described in ref. 35.

In-Gel Kinase Assay.

The ABA-dependent phosphorylation of the AREB polypeptides was analyzed by an in-gel kinase activity assay, and protein extracts from ABA-treated or nontreated T87 cells were prepared as described in ref. 7. GST-fused AREB proteins were expressed in Escherichia coli as described in ref. 7.

Transient Transactivation Experiment with T87 Protoplasts.

Effector plasmids used in the transient transactivation experiment were constructed by cloning ORFs for the wild-type or amino acid-substituted AREB1 made with mutagenic PCR primers into the NotI site of the plant expression vector pBI35SΩ as described in ref. 7. A transactivation experiment with T87 cells was performed as described in refs. 36 and 37.

Quantitative RT-PCR Analysis and Primers.

Total RNA was prepared from 3-week-old seedlings by using TRIzol reagent (Invitrogen) according to the manufacturer’s protocol. cDNA was subsequently synthesized from 1 μg of total RNA by using ReverTraAce reverse transcriptase (Toyobo, Osaka) and random hexamers as primers. Gene expression was analyzed by real-time quantitative PCR with a LightCycler (Roche Diagnostics) and SYBR Premix ExTaq kits (Takara Bio, Otsu, Japan). Values were normalized with the amount of 18S ribosomal RNA as an internal control. Primers used in this work were listed in Table 2, which is published as supporting information on the PNAS web site.

Microarray Analysis and Data Mining.

Total RNA was isolated from 3-week-old plants and used for the preparation of Cy5- and Cy3-labeled cDNA probes. Each experiment was repeated three times, and only genes showing a signal intensity >1,000 in at least one experiment were analyzed further. An Arabidopsis 2 Oligo Microarray kit (Agilent Technologies) was used to compare profiles of transcripts that were induced or reduced in the transgenic plants harboring 35S:AREB1pa as described in ref. 35.

Abbreviations

ABA

abscisic acid

ABRE

ABA-responsive element

SnRK

SNF1-related protein kinase