A Raf-like protein kinase BHP mediates blue light-dependent stomatal opening

Stomata in the plant epidermis open in response to blue light and affect photosynthesis and plant growth by regulating CO2 uptake and transpiration. In stomatal guard cells under blue light, plasma membrane H+-ATPase is phosphorylated and activated via blue light-receptor phototropins and a signaling mediator BLUS1, and H+-ATPase activation drives stomatal opening. However, details of the signaling between phototropins and H+-ATPase remain largely unknown. In this study, through a screening of specific inhibitors for the blue light-dependent H+-ATPase phosphorylation in guard cells, we identified a Raf-like protein kinase, BLUE LIGHT-DEPENDENT H+-ATPASE PHOSPHORYLATION (BHP). Guard cells in the bhp mutant showed impairments of stomatal opening and H+-ATPase phosphorylation in response to blue light. BHP is abundantly expressed in the cytosol of guard cells and interacts with BLUS1 both in vitro and in vivo. Based on these results, BHP is a novel signaling mediator in blue light-dependent stomatal opening, likely downstream of BLUS1.


Presumption of the potential inhibitor targets in Arabidopsis.
In mammalian cells, Tyrphostin 9, Sphingosine, GW5074, and BML-265 inhibited platelet-derived growth factor receptor kinase (PDGFRK), protein kinase C (PKC), c-Raf, and epidermal growth factor receptor kinase (EGFRK), respectively. We predicted the inhibitor targets in Arabidopsis by BLAST search based on amino acid sequences of the mammalian kinases. We found that all of the Arabidopsis kinases similar to PDGFRK, c-Raf, and EGFRK belong to a Raf-like kinase subfamily of the MAPKKK family (Supplementary Table 1). In Arabidopsis, the kinase subfamily contains 48 members 24,25 (Supplementary Fig. 1a). However, the physiological functions of the members are largely unknown 26 . We suspected that some members in the subfamily mediate blue light-dependent stomatal opening. To investigate this, we first focused on the expression levels of the subfamily genes in guard cells because most signaling components for stomatal movements, such as ABA INSENSITIVE 1 (ABI1), ABI2, OPEN STOMATA 1 (OST1), OPEN STOMATA2/ARABIDOPSIS H + -ATPase 1 (OST2/AHA1), PHOT1, HIGH LEAF TEMPERATURE 1 (HT1), SLOW ANION CHANNEL-ASSOCIATED 1 (SLAC1), and BLUS1 are highly expressed in guard cells [27][28][29] . Thus, we searched for guard cell-rich Raf-like kinases in a public microarray database eFP browser (http://bar.utoronto. ca/efp/cgi-bin/efpWeb.cgi?dataSource = Guard_Cell). Among 48 genes of the members, 12 were highly expressed in guard cells ( Supplementary Fig. 1b). It had been demonstrated that one of the 12 members, HT1, regulates stomatal movements in response to CO 2 and the ht1 mutant exhibits normal stomatal opening in response to blue light 30,31 . Therefore, we obtained T-DNA inserted knockout mutants of the other 11 members, excluding HT1 (Supplementary Fig. 2 and Fig. 2b).

Identification of BHP that regulates blue light-dependent stomatal opening.
We determined the blue light-dependent stomatal opening in the knockout mutants ( Supplementary Fig. 3). Although most of the mutants showed stomatal opening similar to WT, two knockout mutants of the C1 group members, GABI_626D02 and SALK_002267, lost stomatal opening in response to blue light ( Fig. 2d and Supplementary Fig. 3). These results indicate that two genes of the Raf-like kinases, At4g18950 and At1g14000, are required for blue light-dependent stomatal opening. We named At4g18950 as BLUE LIGHT-DEPENDENT H + -ATPASE PHOSPHORYLATION (BHP). T-DNA of the GABI_626D02 (bhp-1 mutant) was inserted into the ninth exon of the BHP gene (Fig. 2a). BHP mRNAs and BHP proteins were not detected in the bhp-1 mutant (Fig. 2b,c), indicating that the bhp-1 is a null mutant. Blue light-dependent increase in stomatal conductance was also reduced by about 60% in the bhp-1 mutant (Fig. 2f,g). We next determined the blue light-dependent phosphorylation of the guard cell H + -ATPase using the immunohistochemical method in WT and bhp-1 epidermis (Fig. 2h). As expected, the phosphorylation of H + -ATPase in guard cells was lost in the bhp-1 mutant. This result was similarly obtained by immunoblotting using guard cell protoplasts (GCPs) (Fig. 2i). The amount of H + -ATPase in bhp-1 guard cells was similar to that in WT. Consistent with the results, another allelic knockout mutant, bhp-2, also lost stomatal opening in response to blue light ( Supplementary Fig. 4). The impairment of the stomatal opening in the bhp-1 mutant was completely recovered by introduction of the WT BHP gene (Fig. 3a,b). These results demonstrate that the stomatal phenotype is derived from the BHP gene knockout.
Next, we investigated whether or not stomatal opening requires the protein-kinase activity of BHP. We expressed an inactive kinase of BHP D299N 32,33 under the control of the BHP promoter in the bhp-1 mutant (Fig. 3c,d). The mutant BHP D299N protein in transgenic plants did not restore stomatal opening in response to Effects of four inhibitors on blue light-dependent H + -ATPase phosphorylation in Arabidopsis guard cells. Immunohistochemical detection of the phosphorylated H + -ATPase was performed using anti-pThr antibody in the leaf epidermis 23 . The fluorescent images (a) and relative intensities of the fluorescent signals (b) are shown. Epidermal fragments from dark-adapted plants were illuminated with red light (50 μ mol m −2 s −1 ) for 20 min (Red), after which blue light was superimposed on the red right (10 μ mol m −2 s −1 ) for 2.5 min (Blue). Each kinase inhibitor was administered at 50 μ M to the blue light (Fig. 3d). These results suggest that BHP functions as a protein kinase upstream of H + -ATPase phosphorylation and positively regulates blue light-dependent stomatal opening. Therefore, we predicted that the selected inhibitors suppress H + -ATPase phosphorylation through the inhibition of BHP. To examine the effects of the inhibitors on BHP kinase activity, we attempted to measure activity using recombinant GST-BHP and BHP-GST produced by both E. coli and in vitro translation systems. However, we could not detect BHP activity in the systems or estimate the effects of inhibitors on kinase activity.
The H + -ATPase activator fusicoccin (FC) irreversibly induces phosphorylation of the penultimate Thr of H + -ATPase and stomatal opening, even in the dark 34 . Stomata in both WT and bhp-1 mutant showed large opening in response to FC (Fig. 2e). Furthermore, FC similarly induced phosphorylation of the guard cell H + -ATPase in both bhp-1 mutant and WT (Fig. 2h). These results indicated that BHP regulates phosphorylation of a penultimate Thr of the H + -ATPase in response to blue light, but is not involved in FC-induced phosphorylation.
In the knockout mutant of At1g14000, SALK_002267, the phosphorylation of the guard cell H + -ATPase was normally observed in response to blue light and FC ( Supplementary Fig. 5a) and the amount of H + -ATPase was not significantly different between WT and mutant guard cells ( Supplementary Fig. 5b). Stomata openings in SALK_002267 were smaller than those in WT in response to FC ( Supplementary Fig. 5c). The results indicate that a Raf-like kinase At1g14000 is involved in stomatal opening, but is not required for H + -ATPase phosphorylation.

Characterization of BHP.
BHP is a member of the C1 group of the Raf-like kinase subfamily, and the functions of BHP remain unclear. BHP has an ankyrin repeat domain that acts in protein-protein interactions in the N-terminus and a protein kinase domain in the C-terminus (Fig. 4a). We found that the BHP homologous proteins are widely conserved in land plants such as Populus trichocarpa, Ricinus communis, Vitis vinifera, Oryza sativa, Sorghum bicolor, Physcomitrella patens, and Selaginella moellendorffii (Fig. 4b), and may have an identical function in stomatal opening.
We determined the subcellular localization of the BHP protein using transgenic plants expressing the BHP-GFP gene under control of the 35S promoter. The fluorescence signal from BHP-GFP was observed in the cytosol of guard cells (Fig. 4c). We note that the signal from GFP alone was observed in both the cytosol and nucleus. RT-PCR analysis was performed using total RNA from rosette leaves (leaves), GCPs, mesophyll cell protoplasts (MCPs), inflorescence stems (stems), petioles, and roots (Fig. 4d). The BHP mRNAs were ubiquitously detected in all these tissues, including guard cells. We further determined the BHP expression using transgenic plants expressing the GUS reporter gene under control of the native BHP promoter (Fig. 4e). In accord with the results of RT-PCR, GUS staining was observed in guard and mesophyll cells, leaves, petioles, and roots. Especially, the stain was strongly observed in vascular tissues and guard cells ( Fig. 4e; ii, iii, iv, v). Together, these results suggested that BHP functions not in the nucleus, but in the cytosol of guard cells, to induce stomatal opening.
To investigate whether BHP functions generally in phots-mediated blue light responses, we examined other blue light responses, such as phototropism, chloroplast movement, and leaf flattening, in the bhp-1 mutant ( Supplementary Fig. 6) 35 . The bhp-1 mutant exhibited similar responses as in WT plants. These results indicate that BHP specifically mediates stomatal opening among phots-mediated blue light responses despite its ubiquitous expression.
Functional location of BHP in signaling for stomatal opening. To investigate the position of BHP in guard cell signaling, we examined the protein-protein interactions between BHP and the known signaling components for stomatal opening using an in vitro pull-down assay. Recombinant GST-BHP was used as bait and incubated with recombinant FLAG-BLUS1, FLAG-PP1, or microsomal proteins from the Arabidopsis etiolated seedlings. BHP did not interact with phot1 and the H + -ATPase from microsomal membranes, but interacted with purified BLUS1 and PP1 in vitro ( Fig. 5a and Supplementary Fig. 7). We further confirmed the in vivo interactions among the proteins by bimolecular fluorescence complementation (BiFC) assay in Nicotiana benthamiana leaves ( Fig. 5b and Supplementary Fig. 8a). In agreement with the results of pull-down assays, BHP interacted with BLUS1 in plant cells. Furthermore, we found that BHP interacts with phot1 and phot2. In contrast, BHP does not interact with the H + -ATPase isoforms, AHA1 and AHA2, although AHA1 and AHA2 show a positive interaction with each other in the present experimental conditions ( Supplementary Fig. 8b).
BLUS1 is a substrate for phot kinases in guard cells 18 . Therefore, BHP may be a substrate of BLUS1 kinase. To test this possibility, we attempted to detect the phosphorylation of BHP by BLUS1 in vitro using recombinant proteins produced by both E. coli and in vitro translation systems. However, we could not detect BLUS1 activity in the systems nor evaluate the target relationship between BLUS1 and BHP. epidermis under red light illumination. DMSO was used as a solvent control. The signal intensity was expressed as the ratio of the signal from the red light with DMSO or blue light with each inhibitor to that from blue light with DMSO. Scale bar represents 50 μ m. Data indicate means ± SD (n = 3) with measurement of 30 stomata in each sample. *indicates values that statistically differ from blue light sample without inhibitor (Student's t test; *p < 0.01). (c) Effects of the four inhibitors on blue light-dependent stomatal opening. The epidermal fragments were pre-treated with each inhibitor at 10 μ M for 20 min in the dark, and then illuminated with or without mixed light (red light at 50 μ mol m −2 s −1 and blue light at 10 μ mol m −2 s −1 ) for 3 h. Values indicate means ± SD (n = 3); measurement of 30 stomata in each experiment. * indicates values that statistically differ from the light sample without inhibitor (Student's t test; *p < 0.01). (d) Effects of the four inhibitors on phot1 activity. Blue light-dependent autophosphorylation of phot1 was determined by protein blotting using GST-14-3-3 protein (GF14φ ) as a probe 32 . Etiolated seedlings were pre-treated with each of the inhibitors at 10 μ M for 30 min and then illuminated with or without blue light at 100 μ mol m −2 s −1 for 1 min.

Discussion
To identify a novel protein kinase involved in blue light-dependent stomatal opening, we first screened kinase inhibitors that suppress both the blue light-dependent phosphorylation of guard cell H + -ATPase and stomatal opening. We searched the potential Arabidopsis targets of the inhibitors based on amino acid sequence homology with mammalian targets and identified BHP. Based on phenotypic analyses of the bhp mutants and transgenic bhp-1 plants expressing WT and inactive BHP, we concluded that BHP functions as a novel positive regulator of blue light-dependent stomatal opening and BHP kinase activity is required for this response.
In the bhp-1 mutant, blue light-dependent phosphorylation of the H + -ATPase was strongly impaired, but FC-induced phosphorylation was similar as in WT (Fig. 2h). Identical protein kinase is most likely to phosphorylate the H + -ATPase in response to both blue light and FC 34 . These results suggest that BHP does not directly phosphorylate the H + -ATPase. In agreement with this, BHP did not interact with the H + -ATPase both in vitro and in vivo (Supplementary Fig. 7b and 8). There may be an unknown protein kinase that directly phosphorylates the H + -ATPase in stomatal opening.
The interaction of BHP with phots was observed in vivo but not in vitro ( Supplementary Figs 7a and 8). Thus, BHP and phots may be in the same complex and interact via other scaffold proteins in plant cells. Interaction of BHP with PP1 could not be evaluated by our BiFC assay, because interaction of PP1 with a regulatory subunit, INHIBITOR3, was not observed in this experiment (see Methods). Because BHP interacted with PP1 in vitro ( Supplementary Fig. 7c), BHP may also interact with PP1 in guard cells. Clarification of the relationship between BHP and the known components, and identification of the unknown binding partner of BHP, will clarify the actual role of BHP in blue light signaling in guard cells.
The promoter GUS and RT-PCR analyses indicated that BHP expression is ubiquitous (Fig. 4d,e). The present results indicate that BHP specifically mediates stomatal opening among the phots-mediated blue light responses ( Supplementary Fig. 6). The identity of guard cell-signaling may be determined by BLUS1, which is expressed only in guard cells 18 . BHP may be involved in other physiological responses in tissues other than guard cells. Further investigations are needed to clarify the physiological roles of BHP in other tissues.
BHP belongs to the Raf-like kinase subfamily in the MAPKKK family, which is different from the typical MAPKKKs involved in the MAPK cascade. However, functions of the Raf-like kinases remain unclear 36,37 . MAPKKK family is divided into three clades (A to C), and members of clade B and C belong to the Raf-like kinases in Arabidopsis 25 . The C clade of Raf-like kinases (C1 to C7 groups) is the largest clade in the MAPKKK family, and BHP and At1g14000 belong to the C1 group. In the present study, we showed that both proteins mediate blue light-dependent stomatal opening. Moreover, HT1 (C5 group) acts as a negative regulator of high CO 2 -induced stomatal closure 30 . These results suggested that C clade members of Raf-like kinases control stomatal aperture in response to light and CO 2 to adapt plant transpiration rates to the environment.
BHP and At1g14000 function differently in blue light-dependent stomatal opening; At1g14000 did not mediate blue light-dependent phosphorylation of the H + -ATPase without affecting the H + -ATPase amount ( Supplementary Fig. 5a,b). Stomatal opening by FC was reduced in SALK_002267 compared with WT ( Supplementary Fig. 5c), suggesting that signaling for stomatal opening is decreased downstream of H + -ATPase activation in the mutant guard cells. Accordingly, previous studies indicated that At1g14000 (VH1-INTERACTING KINASE; VIK) is involved in auxin and brassinosteroid signaling, and functions in glucose import into the vacuole at the tonoplast 38,39 . Ion transports into the vacuole have crucial roles in increasing guard-cell volume for stomatal opening 40 . These results suggested that BHP functions in the guard cell cytosol to regulate H + -ATPase activity, and VIK may function in the guard cell tonoplast to regulate stomatal opening. There may be unknown C-clade members of Raf-like kinases that function in guard cells. However, further analyses are required to understand the Raf-like kinase functions and signaling for stomatal movement. Stomatal aperture and conductance. Stomatal aperture was measured using epidermis from the rosette leaves according to previous methods 23,32 . Epidermal fragments were incubated in reaction buffer 1 containing 5 mM MES/bistrispropane (pH 6.5), 50 mM KCl, and 0.1 mM CaCl 2 and illuminated with light (red light at 50 μ mol m −2 s −1 and blue light at 10 μ mol m −2 s −1 ) or treated with FC (10 μ M) in the dark for 3 h (Figs 1c,2d,e and 3b,d; Supplementary Figs 3 and 4c). The protein kinase inhibitors (10 μ M) were pretreated with the epidermis 30 min before light illumination. To determine stomatal opening by FC in SALK_002267, the epidermis was incubated in reaction buffer 2 containing 5 mM MES/bistrispropane (pH 6.5), 10 mM KCl, and 0.1 mM CaCl 2 with or without FC at the indicated concentrations in the dark for 3 h (Supplementary Fig. 5c).

Methods
Stomatal conductance in intact leaves of WT and bhp-1 was determined according to previous methods 5,32 .

Isolation of guard and mesophyll cell protoplasts. Guard and mesophyll cell protoplasts (GCPs and
MCPs) were isolated from rosette leaves of Arabidopsis plants, as described previously 6 with minor modifications. We used Macerozyme R-10 (Yakult) instead of pectolyase Y-23 in the enzyme mixture. The obtained protoplasts were suspended in 400 mM mannitol and 1 mM CaCl 2 , and were kept on ice in the dark until use. Detection of phot1 autophosphorylation in etiolated seedlings. Three-day-old etiolated seedlings were grown as described in a previous report 32 . Fifty etiolated seedlings were incubated in reaction buffer 1 with or without each kinase inhibitor (10 μ M) in the dark for 30 min. The seedlings were kept in the dark or irradiated  Antibodies. Polyclonal antibodies against the conserved catalytic domain of AHA2 (Anti-H + -ATPase), phosphorylated penultimate Thr (Anti-pThr), and the N-terminus of phot1 (Anti-phot1) were used for immunostaining and immunoblotting, as reported previously 5,23,41 . Anti-FLAG antibody was purchased (Sigma-Aldrich) and used. Anti-BHP or anti-GST antibody was raised against the recombinant BHP or GST protein as an antigen in a rabbit (Medical & Biological Laboratories). Full-length cDNA of BHP was amplified by PCR with primers 5′ -CGGGATCCATGGAAGAGGATTATCAACAGC-3′ and 5′ -CGGGATCCTCACAAATGTGAACCGGATG-3′ and cloned into the BamHI site of the pGEX-2T vector (GE Healthcare). The resulting construct or empty vector was transformed into the E. coli BL21 strain. The recombinant BHP protein was expressed as a fusion protein to the glutathione S-transferase (GST-BHP). The fusion protein or GST protein was purified using the glutathione-Sepharose 4B (GE Healthcare). The BHP protein was obtained from the sepharose beads by digestion with thrombin to cut off the GST. GST proteins were eluted from the boundary beads by addition of elution buffer including 50 mM Tris-HCl (pH 8.0) and 10 mM glutathione. The proteins were used for the immunization as antigens.

Preparation of microsomes. The microsomes from etiolated seedlings of
Immunoblotting using GCP proteins. Immunoblots to determine phosphorylation and the amount of guard cell H + -ATPase were performed according to previous methods 23 with minor modifications. GCPs were prepared as described above. After light illumination of GCPs, the phosphorylation reaction of H + -ATPase was terminated 2.5 min after the start of blue light illumination. The guard cell proteins (50 μ g) were solubilized by addition of GCPs-solubilization buffer containing 5 mM MOPS-KOH, 1.75 mM EDTA, 1 mM phenylmethylsulfonyl fluoride (PMSF), 20 μ M leupeptin, 1.5% SDS, 15% sucrose, 5% 2-mercaptoethanol, 0.01% Coomassie Brilliant Blue, and 15 mM Tris-HCl (pH 8.0) at room temperature. The protein samples were subjected to SDS-PAGE and then immunoblotted using anti-pThr and anti-H + -ATPase. To detect BHP proteins in GCPs, we performed immunoblotting using anti-BHP antibodies as above. For the immunoblotting of BHP, the nitrocellulose membrane was blocked and incubated with the antibody in Blocking One (Nacalai Tesque) instead of a 5% skim milk solution. The anti-BHP antibody was used at a 2,000-fold dilution in the primary antibody. Phosphorylation and the total amount of the H + -ATPase were determined by immunoblotting using anti-pThr and H + -ATPase antibodies, respectively. Experiments repeated three times on independent occasions gave similar results.
In vitro pull-down assay. The GST-BHP protein was expressed in E. coli as described above. FLAG-tagged BLUS1 and -PP1 [TYPE ONE PROTEIN PHOSPHATASE 4; TOPP4 (At2g39840)] were generated as recombinant proteins from E. coli ( Fig. 5a and Supplementary Fig. 6c). Microsomal proteins from the etiolated Arabidopsis seedlings were used as a source of the proteins ( Supplementary Fig. 6a,b). The BLUS1 and TOPP4 cDNA fragments were amplified by RT-PCR from the Col cDNAs using primers 5′ -CGGAATTCCATGGCTCGGAACAAGCTCGAGTTC -3 ′ an d 5 ′ -C G G A AT TC T TA AC C C A A A AC AC TATC T T TATC AG -3 ′ for B LU S 1 an d 5 ′ -CGGAATTCCATGGCGACGACGACGACG-3′ and 5′ -CGGAATTCTCAAATCTTTGTGGACATCATGAACTTG -3′ for TOPP4. The PCR products were cloned into the EcoRI site of the pFLAG-MAC expression vector (Sigma-Aldrich), and plasmids were introduced into the E. coli BL21 strain. E. coli was grown in LB broth at 37 °C for 16 h and the culture was diluted three times with new LB broth. The culture was further grown at 20 °C for 2 h in the presence of 0.5 mM isopropyl-β -D-thiogalactopyranoside. E. coli cells were collected and suspended in Tris-buffered saline (TBS) containing 5 mM dithiothreitol (DTT), 1 mM PMSF, and 20 μ M leupeptin, and were then disrupted by sonication on ice. After centrifugation at 10,000 × g for 10 min at 4 °C, each supernatant from E. coli lines was mixed and added with 0.5% TritonX-100. The mixtures were reacted with 30 μ L of glutathione-Sepharose 4B (GE Healthcare) for 30 min at 4 °C. The beads were washed three times with TBS and then solubilized by the addition of 20 μ L of SDS sample buffer [5% (w/v) SDS, 30 mM Tris-HCl pH 8.0, 3 mM EDTA, 30% (w/v) sucrose, 0.012% (w/v) Coomassie Brilliant Blue, 15% (v/v) 2-mercaptoethanol]. The solubilized samples were subjected to SDS-PAGE on a 10% acrylamide gel and immunoblotted using anti-GST or FLAG-antibody (Sigma-Aldrich).
For the pull-down assay using phot1 and the H + -ATPase, microsomal membranes were prepared as described above and suspended in TBS containing 5 mM DTT, 1 mM PMSF, 20 μ M leupeptin, and 0.5% TritonX-100. The microsomal proteins were reacted with the GST-14-3-3 protein (GF14φ ) or GST-BHP on glutathione-Sepharose 4B for 2 h at 4 °C. After washing the beads, samples were solubilized in 20 μ L of SDS sample buffer and SDS-PAGE was performed on a 9% acrylamide gel. Immunoblotting was performed using anti-GST, -phot1, and -H + -ATPase antibodies.
Expression of BHP mRNA by RT-PCR. Total RNAs were extracted from rosette leaves, GCPs, MCPs, stems, petioles, and roots of 5-week-old Col plants using NucleoSpin RNA Plant (Takara). First strand cDNA was synthesized from the RNAs using PrimeScript II 1st strand cDNA Synthesis Kit (Takara). The BHP cDNA fragment was amplified from the cDNAs by PCR using primers 5′ -ATGGAAGAGGATTATCAACAGC-3′ and 5′ -TCACAA ATGTGAACCGGATG-3′ .TUB2 fragment was amplified using primers 5′ -CATTGTTGATCTCTA AGATCCGTG-3′ and 5′ -TACTGCTGAGAACCTCTTGAG-3′ , and was used for an internal standard. The primers for amplification of the other genes of Raf-like kinases are listed in Supplementary Table 2.
Functional complementation experiment. Genomic fragments of the BHP gene (5,974 bps), including 5′ and 3′ noncoding regions, were amplified by PCR from genomic DNA of Col using primers 5′ -CGGGATCCTGGGTGAAAGTTGACGAACATACTC-3′ and 5′ -CGGGATCCCCTCGATCAATA AAGGTCGGCGATC-3′ . The DNA fragment was cloned into the BamHI site of the pCAMBIA1300 vector (CAMBIA). To generate a transgenic plant expressing the inactive kinase of the BHP D299N protein, single amino acid substitution was introduced. The aspartic acid is a binding site of Mg 2+ -ATP in protein kinases 33 and substitution of the aspartic acid with asperagine leads to inactivation of protein kinases 18,32 . Nucleotide substitution was introduced into the BHP gene in the pCAMBIA1300 vector as templates for inverse PCR. Inverse PCR was performed using the oligonucleotide primers 5′ -AACTTTGGAGTAAGCAAGCTTG-3′ and 5′ -TGCAACTTTCAGATGCCCTG-3′ . The PCR mixtures were treated with DpnI to digest the template DNA and then with T4 polynucleotide kinase to phosphorylate the 5′ -ends of the PCR products. The phosphorylated linear DNAs were self-ligated. The vectors were transformed into the Agrobacterium tumefaciens GV3101 strain. Agrobacterium was transformed into the bhp-1 mutant by the floral dip method 42 . Transgenic plants were selected by resistance against hygromycin and used for phenotypic analysis of stomatal opening.

Subcellular localization of BHP.
To analyze the subcellular localization of BHP in guard cells, we generated the transgenic plants expressing BHP-GFP under the control of 35S promoter. The full-length cDNA of BHP was amplified by PCR using primers 5′ -CATGCCATGGAAGAGGATTATCAACAGC-3′ and 5′ -CATGCCATGGCCAAATGTGAACCGGATGATG-3′ , and was cloned into the NcoI site of the CaMV35S-sGFP(S65T)-NOS3′ vector 43 . The cDNA of the BHP-GFP was amplified by PCR from the resulting vector using primers 5′ -CATATGCCCGTCGACATGGAAGAGGATTATCAACAGC-3′ and 5′ -TCAGAATTCGGATCCTTACTTGTACAGCTCGTCCATGCC-3′ and cloned into the pRI 101-AN DNA vector (Takara) using the In-Fusion cloning system (Clontech). The resulting construct was introduced into Arabidopsis plants (Col) using an Agrobacterium-mediated transformation method 42 . The transgenic plants were selected based on the kanamycin resistance of the introduced T-DNA. GFP fluorescent signal from guard cells was detected in the epidermis of rosette leaves using a confocal laser microscope (FV10i; Olympus).
Expression pattern of BHP determined by a promoter-GUS assay. The promoter region of BHP gene (2,139 bps) was amplified by PCR from the genomic DNA of Col using primers 5′ -CAGGCATGCAAGCTTT G G G T G A A A G T T G A C G A A C ATA C T C -3 ′ a n d 5 ′ -A G T C A G AT C TA C C AT T C C A A C A AAAACCCTAAAACCCTGAT-3′ . The PCR product was cloned into pCAMBIA1303 vector (CAMBIA) using the In-Fusion cloning system. The resulting vector was introduced into Arabidopsis plants (Col) as described above. The transgenic plants were selected by hygromycin resistance and were used for histochemical GUS staining.