Arabidopsis thaliana Columbia-0 (Col-0) and bak1-3 (Salk_034523, hereafter bak1) were used as the wild-type plant and bak1 mutant, respectively, in all experiments. Arabidopsis seeds were sterilized with 75% ethanol containing 0.05% Tween-20, washed twice with 95% ethanol, and grown on selective 1/2 MS (Duchefa, Netherlands) plates containing 0.8% phytoagar. Arabidopsis was grown at 22°C, and Nicotiana benthamiana was grown at 28°C under long-day conditions (16 h light/8 h dark).

For Arabidopsis mesophyll cell protoplast preparations, we followed previously described methods (Yoo et al., 2007) with minor modifications. Fourth leaves were used from 3- to 4-week-old plants before bolting. Leaf strips were incubated for 30 min in an enzyme solution [20 mM MES, pH 5.7, 1.5% (w/v) cellulase R10 (Yakult Pharmaceutical, Japan), 0.4% (w/v) macerozyme R10 (Yakult Pharmaceutical, Japan), 0.4 M mannitol, 20 mM KCl, 10 mM CaCl2, and 0.1% BSA], and incubated for 3.5 h in the dark. The digestion was terminated by adding an equivalent volume of W5 solution (2 mM MES pH 5.7, 154 mM NaCl, 125 mM CaCl2, and 5 mM KCl). The protoplast solution was filtered through a 100 μm mesh, centrifuged at 100 × g for 2 min at 4°C, and resuspended in MMG solution (0.4 M mannitol, 4 mM MES, pH 5.7, 15 mM MgCl₂). Protoplast transfection was mediated by PEG solution (40% PEG4000, 0.2 M mannitol, 100 mM CaCl₂).

For plant transformation, transgenes were transformed into Agrobacterium tumefaciens strain GV3101. Plant transformation was performed using the floral dipping method (Clough and Bent, 1998).

The constructs pET28a-ABI1, pET28a-BAK1KD, pET28a-BAK1KD(K317E), pGEX-OST1, pGEX-OST1(G33R), and pPZP222-BAK1:BAK1-FLAG have been described previously (Shang et al., 2016). For pull-down assays, open reading frames (ORFs) encoding full-length ABI1, PYR1, PYL4, and the kinase domain of BAK1 were amplified and cloned into pGEX, pET28a, and pRSET vectors. For protoplast transfection, ORFs encoding full-length PYR1 and PYL4 were amplified and cloned into p326 vectors containing 35S:HA. Full-length ABI1 cDNA was amplified and cloned into a p326-35S:GFP vector to generate p326-35S:ABI1-GFP. pAVA393-35S:FLAG-PYR1 was generated by substituting the GFP sequence with the FLAG-PYR1 fragment in the pAVA393 vector. To generate the p326-RD29B:LUC plasmid, 1.7 kb of RD29B promoter sequence was amplified and cloned into a p326-LUC vector. To generate PYR1/Col-0, PYR1DD/Col-0 and PYR1AA/Col-0 transgenic plants, the ORF of PYR1 was amplified and ligated to the commercial pENTR–TOPO-D vector (Invitrogen, United States), followed by an LR reaction with the pEG202 binary vector using gate-way cloning. Site-direct mutagenesis was performed using the indicated primers following the manufacturer’s instructions (Agilent Technologies, United States). For yeast transformation, ORFs of PYR1 variants and ABI1 were amplified and cloned into pAS2 and pACT2 vectors to generate pAS2- or pACT2-PYR1 variants and pAS2-ABI1. For bimolecular fluorescence complementation (BiFC), the ORF of PYR1 was cloned into p326-YFP^C vectors to obtain p326-PYR1, and the other constructs were generated as described in our previous study (Shang et al., 2016). The primers used are shown in Supplementary Table 1.

Agrobacterium tumefaciens GV3101 colonies containing the indicated constructs were inoculated overnight in selective LB media containing 20 μM acetosyringone. The harvested Agrobacterium cells were resuspended in infiltration media (100 mM MgCl₂, 10 mM MES, pH 5.6, 100 μM acetosyringone) to OD₆₀₀ values of 0.6 to 0.8. The same volume of Agrobacterium cells containing p19 and indicated transgenes were mixed and incubated for 2 h at 25°C. Four-week-old Nicotiana benthamiana leaves were subsequently infiltrated. After 48 h of incubation, 30 μM ABA or an equivalent volume of ethanol was injected into the infiltrated leaves and incubated for 3 h. Leaf tissues were harvested for co-immunoprecipitation, or subjected to epidermal peels. YFP signals were examined using an LSM700 confocal microscope (Zeiss, Germany).

The purified recombinant protein (3 μg) was incubated in kinase buffer (20 mM HEPES, pH 7.5, 125 mM NaCl, 10 mM MgCl₂, 5 mM MnCl₂) and 10 μM ATP containing 1 μCi [γ-³²P]ATP (3000 Ci/mmol, IZOTOP, Japan) for 30 min at 25°C. The reactions were stopped by the addition of 2 × SDS sample buffer. The phosphorylated proteins were electrophoresed on a 10% SDS-PAGE gel. The total protein content of the gel was visualized by CBB staining, and the phosphorylated proteins were detected using an FLA 7000 Bio-Imaging Analyzer (Fuji, Japan).

Four-week-old plant protoplasts were isolated and transfected with reporter and effector genes. The p326-RD29B:LUC plasmid was used as a reporter gene, and p326-REN, the Renilla luciferase (REN) gene driven by the 35S promoter, served as an internal control. The effector genes were as follows: p326-35S:HA-PYR1, p326-35S:HA-PYR1DD, p326-35S:HA-PYR1AA, p326-35S:HA-BAK1, and p326-35S:ABI1-GFP. A total of 10 μg of DNA was used for each protoplast transfection using a ratio of 5:4:1 for the effector: reporter: internal control. For ABA treatments, transfected protoplasts were incubated overnight in a growth room and exposed to 30 μM ABA for 3 h before harvesting. Firefly luciferase activities were quantified using a dual luciferase assay kit (Promega, United States) with at least three replicates per transfection.

For binding assay of ABA for ABA receptors, we followed previously described methods (Zhang et al., 2002; Liu et al., 2007) with minor modifications. Each set of purified PYR1 alone or in combination with PYR1 variants, including PYR1, PYR1DD, and PYR1AA, with BAK1 was incubated in ABA-binding buffer [50 mM Tris, pH 8.0, 50 mM NaCl, 10 mM MgCl₂, 5% glycerol, 0.025% BSA, 100 ng/mL unsaturated phosphatidyl choline, 0.1% Tween-20, and 1× complete protease inhibitor (Roche Applied Sciences, United States)] in a 30 μL reaction mixture. Tritium-labeled ABA {(±)-ABA [3H] (10–20 Ci/mmol), American Radiolabeled Chemicals, Inc. United States} was added to initiate the reaction. After incubation on ice for 1 h, the reaction mixture was filtered through a GF/C membrane (Whatman, Inc. United States), and washed three times with 5 mL of binding buffer. Bound radioactivity was measured in 5 mL of scintillation cocktail solution using a liquid scintillation counter (Tri-Carb 2910TR; Perkin Elmer, United States).

Transfected protoplasts were incubated with or without ABA (30 μM)-containing MMG solution for 3 h. And 10-day-old transgenic seedlings were treated with or without ABA (30 μM) for 3 h by spraying (2 ml/1/2 MS plate). Total proteins were extracted in extraction buffer containing 50 mM Tris (pH 7.5), 150 mM NaCl, and 0.2% NP-40. Protein extracts were immunoprecipitated with each specific antibody listed below at 4°C overnight, and immunoprecipitants were pulled down using Protein A-Sepharose beads (GE Healthcare, Sweden). The co-immunoprecipitated proteins were detected by western blotting using the following antibodies: anti-FLAG Ab (1:3,000; Abcam, United Kingdom), anti-GFP Ab (1:3,000; Abcam, United Kingdom), anti-HA Ab (1:3,000; MBL, Japan), anti-Phospho-Threonine Ab (1:1000, Cell Signaling, United States), anti-Rubisco antibody (1:5,000, Abbkine, China) and anti- BAK1 Ab (Yun et al., 2009) (1:500). Signals from secondary antibodies were visualized using the ECL Plus detection reagent (GE Healthcare, Sweden).

Glutathione (GSH) agarose resin-bound GST-tagged kinase domain of BAK1 (BAK1KD) was incubated with HIS-tagged PYR1 recombinant proteins at 4°C for 1 h. The resins were then washed thrice with GST washing buffer (50 mM Tris, 0.1 M NaCl), and the samples were analyzed by SDS-PAGE. Proteins were detected with western blotting using an anti-GST antibody (1:3,000; Abcam, United Kingdom) and anti-HIS antibody (1: 5,000; MBL, Japan). The same process was performed for pull-down assays between GST-ABI1 and HIS-PYR1, HIS-PYR1DD, or HIS-PYR1AA.

The AH109 yeast cell line was used for the yeast two-hybrid assay. The constructs were transformed into AH109 cells and selected on selective media without Trp and Leu. The interactions were detected on selective media lacking Trp, Leu, and His with or without 50 μM ABA. Images were captured 2 days after incubation on selective media at 30°C.

Stomatal apertures were measured according to methods described by Shang et al. (2016). Cotyledons from 10-day-old seedlings were incubated in stomatal opening solution (50 mM KCl, 10 μM CaCl₂, 10 mM MES, pH 6.15) for 2 h. Then, 1 μM ABA was added to the opening solution and incubated for an additional 2 h. Images of stomata were captured using a DM2500 microscope (Leica, Germany), and the width and length of stomatal apertures were measured.

After adding 5 μM ABA to 1/2 MS media-containing plates with seeds, they were exposed to far-red light for 5 min, followed by a red light for 5 min, and then grown under long-day conditions. Germinated seeds were counted at the same time daily for 1 week, and the germination ratio was calculated.

Seedlings were grown vertically on 1/2 MS media containing the indicated concentrations of ABA, and root lengths were measured after 8 days of growth.

Each seed was sown in soil pots saturated with equal volumes of water and grown for 4 weeks under long-day conditions (16 h light/8h dark) at 22°C. The humidity was maintained at 40% by controlled irrigation using a portable hygrometer. After a 4-week growth period, irrigation was interrupted for 7 days and resumed. The number of plants that resumed growth was observed 2 days after re-watering commenced.

RNA was isolated from 10-day-old seedlings. First-strand cDNAs were synthesized using RNA treated with RNase-free RQ1 DNases (Promega, United States) with SuperScript III-MMLV reverse transcriptase (Invitrogen, United States) and the oligo(dT)₁₅ primer. Quantitative RT-PCR was performed using SYBR Green PCR Master Mix (Applied Biosystems, United States). UBQ5 (At3g62250) served as the control. The primers used are shown in Supplementary Table 1.

BAK1 is known to interact not only with ligand-binding proteins but also with other proteins (Ma et al., 2016). Previously, we reported that BAK1 interacts with OST1 to regulate ABA signaling in guard cells (Shang et al., 2016). Here, we further investigated whether BAK1 interacts with PYR1, which is critical for triggering ABA signaling, and if so, whether the presence of ABA affects this interaction. We first performed an in vitro GST pull-down assay and observed a direct interaction between BAK1 and PYR1 (Figure 1A). To confirm these interactions, we performed bimolecular fluorescence complementation (BiFC) analyses for BAK1 and PYR1, BAK1 and PYL4, a close homolog of PYR1, and AtSERK4 and PYR1. We observed that YFP signals were reconstituted near the plasma membrane from BIFC pairs of BAK1 and PYR1, BAK1 and PYL4, similar to the BAK1/OST1 interaction shown as a positive control. We also detected YFP signals from a pair of AtSERK4, the closest homolog of BAK1, and PYR1. However, we did not observe any YFP signals from a PYR1/BRI1 pair (Figure 1B). To further investigate BAK1 and PYR1 plant interactions, we conducted co-immunoprecipitation analysis in plants overexpressing FLAG-PYR1. We determined that ABA induces BAK1 and PYR1 interactions. The basal level of PYR1 was co-immunoprecipitated with BAK1 even in the absence of ABA, and ABA treatment increased the amount of PYR1 bound to BAK1 (Figures 1C,D). In addition, we examined where PYR1 was localized in plants overexpressing PYR1-GFP. We observed GFP signals in the outlines of both leaf guard cells and pavement cells and minor signals throughout the cytosol of both cells (Supplementary Figure 1). As BAK1 is a plasma membrane-localized protein, interactions with BAK1 usually appear in or near the plasma membrane. Therefore, these results suggest that the interaction between BAK1 and PYR1 occurs in the plasma membrane and that this interaction can be promoted in the presence of ABA in plants.

As BAK1 is a plasma membrane-localized kinase, the interaction between BAK1 and PYR1 prompted us to check whether BAK1 phosphorylates PYR1. In vitro kinase assays showed that BAK1 phosphorylated PYR1 (Figure 2A). In comparison, BAK1 did not phosphorylate MPK1, MPK2 or MPK3, randomly selected as proteins to be tested for the putative substrate of BAK1 (Supplementary Figure 2). These results indicated that BAK1 specifically phosphorylated PYR1. To further confirm that BAK1-mediated PYR1 phosphorylation also occurs in plants in response to ABA, we transfected the FLAG-PYR1 construct in protoplasts prepared from Col-0 wild-type or bak1 mutant plants and immunoprecipitated PYR1. Phosphorylated PYR1 in each plant was analyzed by western blotting using anti-phosphothreonine antibodies. ABA-induced phosphorylation of PYR1 was detected in wild-type protoplast. However, it was barely detected in the bak1 mutant protoplasts (Figure 2B). These results showed that BAK1 is required for the phosphorylation of PYR1.

Based on this result, we next attempted to determine the phosphorylation sites of PYR1 by BAK1. Although in vivo phosphomapping using transgenic seedlings overexpressing FLAG-PYR1 was attempted, we repeatedly failed to obtain meaningful results. Thus, as an alternative way, we directly examined the amino acid sequence of PYR1 to identify sites where BAK1 can phosphorylate, because PYR1 is a small protein comprising 191 amino acids. From the alignment of protein sequences for PYR1 and its 13 homologs, we found that there were five completely conserved serine residues (S32, S85, S92, S122, and S142) and two highly conserved threonine residues (T137 and T156) (Supplementary Figure 3a). Thus, we generated PYR1 proteins with a point mutation at each of these seven sites and performed an in vitro kinase assay with BAK1. Compared with wild-type PYR1, the PYR1 variants with T137 or S142 substituted with alanine showed significantly reduced PYR1 phosphorylation by BAK1 (Supplementary Figure 3b and Figure 2C). When both T137 and S142 sites were substituted with alanine (PYR1AA), the overall phosphorylation levels of PYR1AA by BAK1 were much lower in vitro than wild-type PYR1, but not eliminated (Figure 2D). To further confirm that these two sites on PYR1 can also be phosphorylated in plants, we generated transgenic plants by expressing the FLAG-PYR1AA construct and examined the phosphorylation status of PYR1 in comparison with that of transgenic plants expressing FLAG-PYR1 by western blotting with anti-phosphothreonine antibodies. We found that phosphorylated PYR1 was more than twofold higher in transgenic plants expressing PYR1 in response to ABA. In comparison, the relative level of phosphorylated PYR1 was much lower regardless of the presence of ABA in the transgenic plants expressing FLAG-PYR1AA (Figures 2E,F). In addition, we confirmed that BAK1 also interacted with PYL4 in a co-immunoprecipitation assay using protoplasts transfected with both BAK1 and PYL4 (Supplementary Figure 4a). Using an in vitro kinase assay, we found that PYL4 was phosphorylated by BAK1 (Supplementary Figure 4b). Therefore, we transfected wild-type protoplasts with the mutated HA-PYL4 construct, in whichT156 and S161 residues of PYL4, corresponding to T137 and S142 of PYR1, were substituted with alanine (PYL4AA). We then examined the phosphorylation levels of them. Similar to PYR1AA, the phosphorylation level of PYL4AA did not respond to ABA, whereas PYL4 phosphorylation increased in response to ABA (Figures 2G,H). Taken together, these results indicate that T137 and S142 of PYR1 and the corresponding residues of PYL4 are important residues phosphorylated by BAK1 in plants. However, the possibility of the presence of additional phosphosites in these ABA receptors cannot be excluded.

To investigate whether the phosphorylation of PYR1 at the T137 and S142 sites is important in ABA signaling activation, we constructed three PYR1 variants: one with only PYR1 and two others with both T137 and S142 sites mutagenized to aspartic acid to generate PYR1DD, or to alanine to generate PYR1AA. We then performed a reporter assay using the luciferase gene driven by the ABA-inducible RD29B promoter. We observed an ABA-induced increase in luciferase activity due to RD29B promoter activation in PYR1- and PYR1DD-transfected wild-type protoplasts, but not in PYR1AA-transfected protoplasts (Figure 3A). In contrast, ABA activation of reporter gene activity was observed only in PYR1DD-transfected bak1 protoplasts, not in the wild-type PYR1- or the PYR1AA-transfected bak1 protoplasts (Figure 3B). These results suggest that BAK1-mediated PYR1 phosphorylation on T137 and S142 is required to activate ABA signaling. To further validate this assumption in plants, transgenic plants overexpressing FLAG-tagged PYR1, PYR1DD, or PYR1AA were generated. Compared with the wild-type, the overall phenotypes did not vary under normal conditions (Supplementary Figure 5a). We confirmed PYR1 overexpression in each line compared to wild-type plants (Supplementary Figure 5b). Next, we examined the expression of genes RD29B, RD29A, and COR47, which are known to be induced by ABA and various abiotic stresses (Gilmour et al., 1992; Yamaguchi-Shinozaki and Shinozaki, 1993). The PYR1-overexpressing transgenic plants showed increased gene expression in response to ABA, although each gene showed a different degree of increased expression. Compared with these patterns, in PYR1DD-overexpressing plants, ABA-induced increased expression of these genes appeared greater than in PYR1-overexpressing plants. In contrast in PYR1AA-overexpressing transgenic plants these phenomena were highly attenuated (Figures 3C–E).

Overexpression of PYR1 or PYR1DD also changed the effect of ABA on various physiological aspects of plants. We examined the effects of ABA-induced stomatal closure. Compared to wild-type or PYR1-overexpressing transgenic plants, in PYR1DD-overexpressing plants, ABA-induced stomatal closure was stronger, whereas it was less prominent in PYR1AA-overexpressing plants (Figure 4A). PYR1 overexpression confers increased drought tolerance in plants (Cao et al., 2013; Kim, 2014; Zhao et al., 2016). As the efficiency of stomatal movement is closely correlated with drought stress responses, we further examined the phenotypic alterations in these transgenic plants under drought stress conditions. The transgenic plants overexpression PYR1 and PYR1 variants were grown for 4 weeks in soil with the same relative humidity, then stopped watering for 7 days. Then, we watered the plants again and allowed them to grow for an additional 2 days. We observed that PYR1- and PYR1DD-overexpressing transgenic plants were more resistant to drought and subsequently showed better growth resumption after re-watering than wild-type plants. PYR1AA-overexpressing plants showed a similar response to that of the wild-type plants (Figure 4B). Next, we examined whether the effects of ABA on seed germination and root growth patterns were affected by the overexpression of PYR1 or PYR1DD. Compared to ABA responses of the wild-type and abi5-7 seeds, overexpression of PYR1 or PYR1DD resulted in significantly increased inhibition of seed germination, indicating that these transgenic plants were ABA hypersensitive. Conversely, transgenic plants overexpressing PYR1AA showed little difference from the wild-type plants (Figure 4C). In addition, ABA-induced root growth inhibition was more robust in PYR1DD-overexpressing plants than in wild-type plants. PYR1AA-overexpressing plants showed partial insensitivity to ABA, which was weaker than the abi5-7 mutant (Figure 4D). Taken together, these results suggest that, although the extent of each physiological phenomenon varied, PYR1DD-overexpressing plants showed hypersensitivity when treated with exogenous ABA. Therefore, ABA signaling might be more activated in PYR1DD-overexpressing plants than in the other tested lines. Our next question was whether the PYR1DD-overexpressing plants could overcome or bypass the need for BAK1 in ABA signaling. To perform this experiment, we generated bak1 transgenic plants overexpressing the PYR1DD construct and examined ABA-induced stomatal closure. PYR1DD overexpression considerably recovered the ABA sensitivity of bak1, although it was not completely recovered compared to Col-0 wild-type plants (Figure 4E). Taken together, these results indicate that phosphorylation of PYR1 at T137 and S142 sites by BAK1 in response to ABA increases ABA responses in plants.

Our current results suggest that BAK1 promotes ABA signaling by phosphorylating PYR1, but the mechanisms underlying this remain ambiguous. To understand the underlying mechanisms, we examined whether BAK1 affects the ABA-binding capacity of PYR1. We performed an in vitro ABA-binding assay with recombinant proteins using [³H]-labeled ABA. BAK1 itself did not bind to ABA, similar to the BSA control. PYR1 exhibited increased ABA-binding activity as the concentration of radiolabeled ABA increased. Interestingly, when PYR1 was pre-incubated with BAK1 in kinase assay buffer, the ABA-binding capacity of PYR1 was higher than that of PYR1 alone (Figure 5A). As BAK1 enhanced the binding capacity of PYR1 to ABA, we further examined whether it was mediated by PYR1 phosphorylation by BAK1. Therefore, we examined the ABA-binding capacities of PYR1DD and PYR1AA. Compared with PYR1, PYR1DD showed increased ABA-binding capacity, whereas PYR1AA showed reduced ABA-binding capacity (Figure 5B). Similar to PYR1, the ABA-binding activity of PYL4 increased in the presence of BAK1 (Figure 5C), and PYL4DD, in which T156 and S161 of PYL4 were substituted with aspartic acid, showed higher ABA-binding activity than PYL4 (Figure 5D). Taken together, these results indicate that BAK1-mediated phosphorylation of PYR1 at T137 and S142 sites increases the binding capacity of PYR1 for ABA to trigger ABA signaling.

Among the 14 members of PYR1/PYLs possessing ABA binding capacity, PYR1, PYL1, PYL2, and PYL3 exist as dimers in the absence of ABA (Melcher et al., 2009; Nishimura et al., 2009; Santiago et al., 2009; Yin et al., 2009). Many hormone binding studies have shown that monomeric ABA receptors such as PYL5, PYL8, and PYL9 have lower K_d values for ABA than dimeric receptors (Miyazono et al., 2009; Yin et al., 2009; Szostkiewicz et al., 2010; Nakagawa et al., 2014). Thus, despite the conclusion that PYR1DD increased ABA binding capacity, we speculated that there are additional underlying reasons for promoting ABA signaling by BAK1, because compared with PYL4, monomeric PYL4DD exhibited increased ABA-binding capacity. ABA induces the dissociation of dimeric receptors (Dupeux et al., 2011; Shukla et al., 2019) and bind to the hydrophobic cavity of the receptors (Melcher et al., 2009; Nishimura et al., 2009). We hypothesized that PYR1 phosphorylation by BAK1 might contribute to dimeric PYR1 dissociation. In this case, PYR1DD was more likely to exist as a monomer. To test this possibility, we performed a yeast two-hybrid assay. We confirmed each pair of constructs were transformed successfully, and that considerable amount of PYR1, PYR1DD, and PYR1AA were expressed in yeast (Supplementary Figures 6a,b). We found that each pair of PYR1 or PYR1AA interacted with each other. In these systems, ABA did not seem to affect PYR1 and PYR1AA dimerization. In contrast, it was difficult to detect an interaction between PYR1DD, even in the absence of ABA (Figure 6A). This result showed that PYR1DD exists as a monomer and does not interact with each other. It is also possible that PYR1DD may exist in a state of higher dissociation capability, even if it consists of a dimeric complex. To further determine the effects of ABA in plants, we generated constructs including PYR1, PYR1DD, and PYR1AA fused to either HA- or FLAG-tag and infiltrated them as pairs into tobacco leaves to express proteins, followed by ABA treatment. Co-immunoprecipitation analyses were performed using total protein extracts from each tobacco leaf. We observed that although immunoprecipitated band intensities for PYR1, PYR1DD, and PYR1AA were similar, the co-immunoprecipitated band intensity markedly decreased for PYR1DD but increased for PYR1AA, compared with PYR1. ABA treatment decreased PYR1 co-immunoprecipitation, consistent with previous report that ABA induces PYR1 monomerization (Dupeux et al., 2011; Shukla et al., 2019). Compared with PYR1, weaker co-immunoprecipitation signals were observed in PYR1DD even in the absence of ABA, while the amount of co-immunoprecipitated PYR1AA was unchanged with or without ABA (Figures 6B,C). These results suggest that less dimer formation occurred in PYR1DD than in PYR1. Therefore, the increased ABA binding capacity of PYR1DD is likely due to an increased proportion of monomeric ABA receptors.

Structural analyses have revealed that once ABA induces dissociation of the PYR1 dimer, ABA-bound PYR1 forms complexes with PP2Cs such as ABI1 and HAB1 in a 1:1 ratio (Melcher et al., 2009; Nishimura et al., 2009; Santiago et al., 2009; Yin et al., 2009). Interfaces for homodimerization or PP2C binding in ABA receptors largely overlap, suggesting that the two events compete. Because BAK1 phosphorylates PYL4 (Supplementary Figure 4b), which usually exists in a monomeric state (Ma et al., 2009), and the ABA-binding capacity of PYL4 was also increased in the presence of BAK1 or in the form of PYL4DD (Figures 5C,D), we speculated that there may be other functions of BAK1 besides dissociating PYR1 into a monomer at a higher proportion. Therefore, we investigated whether BAK1 affects the binding of ABI1 to PYR1. To test this hypothesis, we performed a yeast two-hybrid assay for ABI1 and PYR1 variants. PYR1 and PYR1DD were strongly bound to ABI1, whereas PYR1AA showed very weak binding to ABI1 in yeast. Interestingly, we found that the interaction between PYR1DD and ABI1 occurred even in the absence of ABA (Figure 7A). To further validate this result, we performed both in vitro GST pull-down assays by incubating GST-ABI1 and HIS-PYR1 variants (Figure 7B) and a transient expression assay in N. benthamiana leaves, in which ABI1-YFP and HA-PYR1 variants were transfected (Figures 7C,D). In both experiments, we observed a basal level of interaction between PYR1 and ABI1 even without ABA treatment, and that ABA treatment further increased the interactions between PYR1 and ABI1. However, such events were hardly detectable between PYR1AA and ABI1 compared to wild-type PYR1. In contrast, PYR1DD showed stronger binding activity to ABI1 than PYR1, even in the absence of ABA. These results indicate that PYR1DD is a favorable binding form for ABI1. To validate this further in plants, we transfected wild-type and bak1 mutant protoplasts with FLAG-PYR1 and ABI1-GFP constructs and performed co-immunoprecipitation analyses. The amount of PYR1/ABI1 complex increased when wild-type plants were treated with ABA. However, in the bak1 mutant, ABA did not induce increased PYR1/ABI1 complex formation, although the basal level of the PYR1/ABI1 complex remained (Figures 7E,F).

It is well known that PYR1 binding to ABI1 relieves ABI1-mediated OST1 inhibition, resulting in OST1 activation (Soon et al., 2012). Thus, we performed an in vitro kinase assay to further investigate whether BAK1 induces OST1 activation by facilitating the interaction between PYR1 and ABI1 in the presence of ABA. The addition of ABI1 greatly reduced the autophosphorylation activity of OST1 in the absence of ABA, which is consistent with the fact that ABI1 inhibits the kinase activity of OST1 (Vlad et al., 2009). In the presence of PYR1 together with ABI1 under ABA-treated conditions, PYR1-ABA-ABI1 complex were formed, and resulting free OST1 activate kinase activity, leading to the autophosphorylation of the OST1. This effect was maximized when PYR1DD was present (Figures 7G,H). Given that same amount of OST1 exist, it was not sure how this happened. There were several reports about autophosphorylation sites of the OST1 from in vitro and in vivo studies (Belin et al., 2006; Umezawa et al., 2009; Vlad et al., 2010). However, it has not been known that how many sites among them are phosphorylated at the same time under certain circumstances. Therefore, it cannot be ruled out the possibility that OST1 incubated with PYR1DD and ABI1 in the presence of ABA was placed in an environment capable of exhibiting more stable activity compared with OST1 alone. Taken together, these results suggest that, when ABA is present, BAK1-mediated phosphorylation of PYR1 facilitates PYR1 dimer dissociation and increased recruitment of ABI1 to PYR1, thereby resulting in OST1 activation.