Dominant and recessive mutations in the Raf-like kinase HT1 gene completely disrupt stomatal responses to CO2 in Arabidopsis

Highlight Loss-of-function and gain-of-function ht1 Arabidopsis mutants have completely disrupted CO2 responses due to reduced and enhanced kinase activities, respectively.


Introduction
Plants control CO 2 uptake for photosynthesis and regulate transpirational water loss through stomatal pores. The two guard cells forming each pore are able to sense environmental signals and endogenous stimuli, integrate this information and optimize stomatal aperture size (Hetherington and Woodward, 2003;Shimazaki et al., 2007;Kim et al., 2010;Kollist et al., 2014). Stomata open in response to low CO 2 concentration ([CO 2 ]) to prevent a decrease in CO 2 uptake and close at high [CO 2 ] in order to maintain high values of water-use efficiency. Stomata are thought to respond to the intercellular [CO 2 ] (C i ) changes caused by photosynthesis and transpiration (Mott, 1988;Assmann, 1999). The global annual mean concentration of atmospheric CO 2 has steadily increased from a pre-industrial value of about 280 ppm to 395 ppm as of 2013 (National Oceanic and Atmospheric Administration data; http://www.esrl.noaa.gov/gmd/ccgg/ insitu.html). This rise in [CO 2 ] has caused a significant decrease in stomatal conductance in many species on a global scale (Medlyn et al., 2001) and affects plant ecosystems (Ainsworth and Rogers, 2007;Keenan et al., 2013). Understanding the molecular basis for stomatal sensitivity to atmospheric CO 2 will improve our ability to predict future ecosystem responses; however, the detailed mechanisms by which CO 2 effects changes in stomatal aperture remain largely unknown.
Stomatal pore size is regulated by the turgor pressure of the guard cells. Stomatal opening is driven by an increase in guard-cell turgor when plasma membrane H + -ATPases are activated (Shimazaki et al., 2007). The activated H + -ATPases induce membrane hyperpolarization, thereby facilitating K + entry, which in turn causes solute influx followed by water uptake into the guard cells (Schroeder et al., 1984). Elevated [CO 2 ] has been suggested to inhibit proton pumps (Edwards and Bowling, 1985) and activate anion channels and K + out efflux channels in the guard cells (Brearley et al., 1997;Roelfsema et al., 2002;Raschke et al., 2003). These changes result in chloride release from the guard cells, membrane depolarization, the loss of guard-cell turgor, and thus stomatal closure .
Recently, mutant screening and functional characterization have led to the identification of plant mutants and genes involved in CO 2 signaling (Israelsson et al., 2006;Negi et al., 2014). The first Arabidopsis thaliana mutant with a defective stomatal CO 2 response, ht1 (high leaf temperature 1), was isolated by analysing leaf temperature changes using thermography (Hashimoto et al., 2006). HT1 encodes a protein kinase mainly expressed in the guard cells, and the two allelic mutations, ht1-1 and ht1-2, cause reduced and no kinase activity, respectively. These altered activities are correlated with unusual stomatal CO 2 responses: stomatal opening in response to low [CO 2 ] is impaired in both mutants; the ht1-1 mutant has a reduced CO 2 response; and the ht1-2 mutant has a severely impaired CO 2 response leading to constitutively high-[CO 2 ] induced stomatal closure. In Arabidopsis, disruption of two carbonic anhydrases, βCA1 and βCA4, also leads to reduced changes in stomatal aperture in response to [CO 2 ] changes . The triple mutant ca1ca4ht1-2 has an impaired response to CO 2 similar to that of ht1-2, indicating that HT1 is epistatic to the genes for these carbonic anhydrases . Elevated intracellular bicarbonate and CO 2 levels activate S-type anion channels including SLAC1 in the guard cells Xue et al., 2011). The SLAC1 anion channel is required for ABA-and CO 2 -induced stomatal closure (Negi et al., 2008;Vahisalu et al., 2008). The OST1 protein kinase has been isolated as an ABA-signaling regulator and shown to activate SLAC1 anion channels (Mustilli et al., 2002;Geiger et al., 2009;Lee et al., 2009). Recent findings have shown that the loss-of-function alleles of OST1 are impaired in the HCO 3 − activation of anion channels, suggesting that OST1 is an ABA and CO 2 signaling component (Xue et al., 2011). Tian et al. (2015) reported that a MATE-type transporter, RHC1, is activated by bicarbonate and functions upstream of HT1. Furthermore, HT1 directly phosphorylates OST1 and inhibits OST1-induced activation of SLAC1 (Tian et al., 2015).
In this study, we demonstrate that not only loss-of-function but also gain-of-function ht1 mutations completely disrupt CO 2 -regulated stomatal aperture changes. Collectively, these mutants are the most severely compromised phenotypes for CO 2 -signaling among the mutants reported to date. This finding indicates that CO 2 signaling pathways associated with HT1 have not been completely explained yet.

Plant material and growth conditions
The Arabidopsis thaliana wild type (WT) accessions used in this study were derived from the Columbia (Col-0) background unless otherwise noted. EMS-mutagenized Col M 2 seeds were purchased from Lehle Seeds (Round Rock, TX, USA). We obtained ht1-6 [stock number CS93263, Col erecta (Col er) background] from the Arabidopsis TILLING project (http://tilling.fhcrc.org) and the ht1-7 T-DNA insertional mutant line [FLAG_446H04, Wassilewskija (Ws) background] from the Versailles Arabidopsis Stock Center (http://dbsgap.versailles.inra.fr/publiclines/). Arabidopsis seeds were surface-sterilized and grown on solid 1/2 MS medium for 18 d in a growth chamber [constant white light of 80 µmol m −2 s −1 at 22 °C, 60% relative humidity (RH)]. The plants were then transplanted into pots with vermiculite and grown for 3 d. These 3-week-old plants were then used for experiments unless otherwise noted.
Thermal imaging Thermal imaging of plants was performed as described previously (Hashimoto et al., 2006). The 3-week-old plants were transferred to a growth cabinet (constant white light of 40 µmol m −2 s −1 at 22 o C, 40% RH) equipped with an automatic CO 2 control unit (FR-SP, Koito). Thermal images of plants were captured under different [CO 2 ] conditions using a thermography apparatus (TVS-8500, Nippon Avionics). Images were fed into a Windows-based computer and were analysed using the software GTStudio (Nippon Avionics). The image processing program Image J (http://imagej.nih.gov/ij/) was also used to quantify leaf temperatures. Details of replication are given in the figure captions.

Stomatal conductance
Gas exchange was measured on the aerial tissues of 24-d-old seedlings using a portable gas exchange system (GFS3000, Heinz Walz, Effeltrich, Germany) equipped with a 3010-A Arabidopsis chamber (Monda et al., 2011). The GFS3000 system was connected to a computer equipped with data acquisition software (GFS-Win). The cuvette for Arabidopsis conditions was set at a light intensity of 200 µmol m −2 s −1 , provided by a special artificial light (LED-Array/ PAMFluorometer 3055-Fl, Heinz Walz, Effeltrich, Germany), with relative humidity and air temperature being set at 50% and 22 o C, respectively. All measurements were made every 60 s. Details of replication are given in the figure captions.
Stomatal aperture response analyses Stomatal aperture measurements were performed as described previously (Hashimoto et al., 2006). Three-week-old plants were incubated at the selected [CO 2 ] in a growth cabinet. Abaxial epidermal peels of the plants were taken from the sixth or seventh leaf and were used immediately to measure stomatal apertures. Leaves from 4-to 5-week-old plants were floated on solutions containing 30 mM KCl, 1 mM CaCl 2 and 5mM MES-KOH, pH 6.15, and were incubated in a growth chamber. ABA from a stock solution in dimethyl sulfoxide (DMSO) was added to the solution after 2 h of illumination, and stomatal apertures were measured 2 h later from epidermal peels using a digital camera attached to a microscope (BH2, OLYMPUS, Tokyo Japan).

Preparation of recombinant proteins
The His-tagged recombinant HT1 (His-HT1) and the HT1 protein with the ht1-3 mutation (His-HT1 R102K ) were expressed and purified from E. coli as described previously (Hashimoto et al., 2006). NdeI sites were introduced in front of the ATG start codon of HT1 and HT1 with the ht1-3 mutation by PCR using each cDNA as a template. The constructs were then ligated in-frame into the pET-28a (+) vector (Novagen) and were confirmed by DNA sequencing. BL21(DE3) cells transformed with pET-28a (+) constructs were induced with 1 mM IPTG for 16 h at 25 °C. His-tagged proteins were purified on nickel columns (Amersham Biosciences). Purified Histagged proteins were recognized specifically by anti-His-probe antibodies (Toyobo) in an immunoblot analysis.

In vitro phosphorylation assay
The kinase assay was performed as described previously (Hashimoto et al., 2006). For the His-HT1 or His-HT1 R102K kinase assay, purified recombinant proteins (1 µg) were incubated in a reaction buffer (25 mM Tris, pH 7.5, 10 mM MgCl 2 ) with 1 mM CaCl 2 or calcium chelators (1mM EGTA, 20 µM BAPTA) in the presence of 0.6 µCi [γ-32 P]ATP at 30 °C for 15 min. The reaction was also performed with 100 µM kinase inhibitors (GW5074, ML-9, K252a, staurosporine and genistein) in the reaction buffer. A negative control containing 20 units CIAP (calf intestinal alkaline phosphatase) was also used in this assay. These reactions were stopped by the addition of SDS-loading buffer, and the proteins were resolved on a 12% SDSpolyacrylamide gel. The proteins were visualized by Coomassie staining with InstantBlue (Ex-pedeon) to verify equal loading, and the kinase activities were detected by autoradiography. Phosphorylation activities of HT1 and its mutants were determined in 10 µl of the kinase reaction buffer using 0.15 µg casein as a substrate under the same reaction conditions. ImageJ software was used to quantify gel bands from the SDS-polyacrylamide gels and the kinase assays.

Transgenic plants
The HT1 genomic region (nucleotides 54586 to 58950 of BAC F24O1) containing At1g62400 was amplified by PCR from the genomic DNA of the ht1-3 mutant using the oligonucleotide primers 5´-CTTCTCTAAGCTTTCGATGCAAACCA-3´ and 5´-GATGTATTGCAAGAGCTGATCAATTGGGTCATGAGA CGAC-3´ and was then inserted into the pGEM-T Easy Vector (Promega). A SalI-MunI fragment including the HT1 genomic sequences with the ht1-3 mutation was cloned into the SalI/ EcoRI site of the T-DNA vector pBI101. For 35S:GFP-HT1, a modified GFP ORF fragment with a glycine linker obtained by PCR using primers 5´-ACCATGGTGAGCAAGGGCGA-3´ and 5´-ACATATGAGCACCTCCACCTCCCTTATACAGC TCGTC-3´ (the glycine linker site is underlined) was inserted into the pGEM-T Easy vector (Promega) to produce pG-GFP1. The full-length HT1 cDNAs were amplified using Pfu DNA polymerase (Stratagene) with the oligonucleotides 5´-CCATATGTCTGGTTTATGTTTCA-3´ and 5´-CCAACGCGTTGGTGTACATCAATAAAGTATCATTATA TATC-3´, and were inserted into the pGEM-T Easy vector to produce pG-HT1-C. The NdeI-BstXI fragment of pG-HT1-C was inserted into the pG-GFP1 to produce pG-HT1-GFP. The NcoI-BsrGI fragment of pG-HT1-GFP was inserted into the NcoI/BsrGI site of pKS(+)GFP (Sugimoto et al., 2007) to produce KS-35S-HT1-GFP. The ApaI-SmaI fragment of KS-35SHT1-GFP containing the CaMV 35S promoter and HT1-GFP translation fusion was inserted into the ApaI/SmaI site of pPZP2H-lac. Transgenic Arabidopsis plants were generated by Agrobacterium tumefaciensmediated transformation.

Confocal laser scanning microscopy
Leaf specimens were observed using a fluorescence microscope (IX70, Olympus) equipped with a spinning-disc confocal laser scanning unit (CSU10, Yokogawa) and a cooled CCD camera head system (CoolSNAP HQ2, Photometrics), as previously described (Hashimoto-Sugimoto et al., 2013). To determine the location of GFP-tagged HT1, serial optical sections of whole guard cells were obtained at 1 μm intervals. To localize plasma membranes with FM4-64 staining, the leaves were treated with 33 µM FM4-64 for 10 min. GFP and FM4-64 fluorescence were detected with appropriate optical settings as previously described (Hashimoto-Sugimoto et al., 2013). To observe plasmolysed guard cells, leaves were mounted in 0.4 M mannitol for 30 min and then observed.

Isolation and characterization of ht1 alleles
In order to isolate the CO 2 -signaling genes, we screened for mutants with altered stomatal CO 2 responses by monitoring leaf temperature changes using thermography, since these are indicators of changes in stomatal aperture (Hashimoto et al., 2006). To date, we have isolated five ht1 alleles using the thermal screening technique (ht1-1, ht1-2, ht1-3, ht1-4, and ht1-5; Fig. 1A and Supplementary Fig. S1 at JXB online). All five of the mutant alleles were ethyl methansulfonate (EMS)-induced mutations of an M 2 population and were found to contain single base-pair alterations in the HT1 gene (see below). In addition, we obtained the ht1-6 mutant, which has a missense mutation, from the Arabidopsis TILLING project, and a T-DNA insertional mutant line, ht1-7 (FLAG_446H04), from the INRA collection (see below and Supplementary Fig. S1). All mutants except ht1-3 had higher leaf temperatures under low [CO 2 ], and leaf temperature changes in response to CO 2 were altered in a manner similar to that in ht1-1 or ht1-2 (Hashimoto et al., 2006) (Fig. 1A and Supplementary Fig. S1). In contrast, ht1-3 had constitutively lower leaf temperatures even in high [CO 2 ] (Fig. 1A

HT1 is a Raf-like Group C MAPKKK
MAPKKKs (mitogen-activated protein kinase kinase kinases), which are involved in various physiological, developmental and hormonal responses, are divided into three groups (Groups A to C) (Ichimura et al., 2002). The HT1 kinase At1g62400 is, from its amino acid sequence, predicted to be a Raf-like Group C MAPKKK (Ichimura et al., 2002), and few functions of Group C MAPKKKs in Arabidopsis have yet been determined. To examine the properties of HT1 kinase, we performed an in vitro kinase assay using calcium chelators and protein kinase inhibitors ( Fig. 2A-C). We generated a His-tagged HT1 protein using E. coli. This recombinant protein (His-HT1) was capable of autophosphorylation and casein phosphorylation ( Fig. 2A). Phosphorylation activities of HT1 were completely lost when the assay was performed in a buffer without magnesium ions (-Mg 2+ ). In the absence of Mg 2+ , the phosphorylation level was almost identical to that of the negative control, in which phosphorylated proteins were dephosphorylated by CIAP ( Fig. 2A-C). Calcium ions did not affect the HT1 kinase activity because the calcium chelators EGTA and the more effective BAPTA did not inhibit phosphorylation. Furthermore, these phosphorylation levels were not significantly different from those containing calcium ions ( Fig. 2A-C). Few differences in the HT1 phosphorylation activity were observed in the presence of ML-9 or genistein ( Fig. 2A-C). ML-9 is an inhibitor of myosin light chain kinase (MYLK) and calmodulin kinase (CaMK), and genistein is a tyrosine kinase inhibitor. K252a and staurosporine, which are broad-spectrum inhibitors of protein kinases including Ser/Thr kinase, partially decreased the HT1 phosphorylation activity ( Fig. 2A-C). The most effective inhibitor of HT1 kinase activity was GW5074, a Raf-1 kinase inhibitor. The HT1 autophosphorylation activity and casein phosphorylation activity at the presence of GW5074 were, respectively, 35% and 38% compared with the BAPTA treatment ( Fig. 2A-C). Taken together, these results indicate that HT1 kinase is a Ca 2+ -independent Raf-like MAPKKK with Ser/Thr kinase activity.

Intracellular localization of HT1
To determine the subcellular localization of HT1, five independent transgenic lines of 35S::GFP-HT1 were analysed by confocal laser scanning microscopy. GFP fluorescence was evenly distributed around the periphery of the guard cells in the 35S::GFP-HT1 plants (Fig. 2D). We stained live leaf tissue of 35S::GFP-HT1 plants with the lipophilic dye FM4-64, which produces a bright red fluorescence in plasma membranes just after application (Fischer-Parton et al., 2000;Bolte et al., 2004). Confocal microscopy analysis revealed a precise overlap of GFP-HT1 fluorescence with FM4-64 fluorescence immediately after staining (Fig. 2D). To confirm this localization, the cotyledons of the transgenic plants were plasmolysed with 0.4 M mannitol for 30 min. GFP fluorescence in the plasmolysed guard cells was observed in Hechtian strands, parts of the plasma membrane connected to the cell wall (Fig. 2E,  arrowheads), further supporting the localization of HT1 in the plasma membranes. The secondary-structure prediction programs SOSUI (http://harrier.nagahama-i-bio.ac.jp/sosui/sosui_ submit.html) (Hirokawa et al., 1988) and TMHMM (v. 2.0; http://www.cbs.dtu.dk/services/TMHMM/) (Krogh et al., 2001) indicated that HT1 kinase has no transmembrane domains. Overall, the results therefore suggest that HT1 associates with plasma membranes.

In vitro kinase assay using HT1 R102K
To examine whether the ht1-3 mutation affects the expression level of the HT1 gene, transcript levels were analysed by qRT-PCR (Fig. 3A). The HT1 mRNA abundance in mature leaves of ht1-3 seedlings was not significantly different from that of WT or other ht1 mutants (ht1-1, ht1-2 and ht1-4) (Fig. 3A) We have previously demonstrated that site-directed mutagenesis of ht1-1 (His-HT1 R211K ) severely reduces phosphorylation activity, and that the analogous His-construct of ht1-2 (His-HT1 Δ136-149 ), which contains a 14-amino-acid deletion, disrupts phosphorylation activity (Hashimoto et al., 2006). Thus, the kinase activities of the WT HT1 and its two mutants are strongly linked to the phenotypes observed in whole plants. This result led us to expect that the ht1-3 mutation may increase phosphorylation activity. To verify this hypothesis, we produced His-tagged recombinant HT1 protein with the ht1-3 mutation (His-HT1 R102K ) and compared the HT1 kinase activities by in vitro kinase assays (Fig. 3B). The autophosphorylation activity in His-HT1 R102K was significantly enhanced compared with that of His-HT1 (Fig. 3C). On the other hand, the phosphorylation activity of His-HT1 R102K to the universal substrate casein was not significantly different from that of His-HT1 (Fig. 3C). The inhibitory effect of kinase inhibitors on His-HT1 R102K kinase activity was also similar to that of His-HT1 (see Supplementary Fig. S3).

The ht1-3 mutation replaces Lys with Arg at 102 -a not highly conserved region
We aligned and compared the amino acid sequences of several Raf-like MAPKKKs, characterized the HT1 sequences, and mapped the ht1 mutation sites. Protein kinases have eleven conserved subdomains. One of these, subdomain I, has the consensus sequence Gly-X-Gly-X-X-(Gly)-X-Val, and the glycine-rich motif forming a loop is thought be involved in anchoring ATP (Hanks et al., 1988). The consensus sequences are not clearly conserved in the HT1 protein kinase or in some other kinases, including OsILA1 (increased leaf angle 1), a recently identified kinase of MAPKKKs in rice (Ning et al., 2011) (Fig. 4B). However, there are regions similar to the consensus sequences with a highly conserved Gly that could correspond to subdomain I [(I) in Fig. 4B]. The other subdomains (II to XI) of protein kinases were highly conserved in HT1, and recombinant HT1 protein had phosphorylation activity (Hashimoto et al., 2006) (Figs 2A  and 4B).
We have previously reported that recessive mutations of ht1-1 and ht1-2 cause reduced and no phosphorylation activity, respectively (Hashimoto et al., 2006). The ht1-1 mutation causes a single amino acid substitution (R211K) at a highly conserved Arg in subdomain VI (Fig. 4). The ht1-2 mutation results in the deletion of amino acids (ht1-2; Δ136-149) in the highly conserved regions of the catalytic domains corresponding to subdomains III and IV (Hashimoto et al., 2006) (Fig. 4). We found that the mutations of ht1-4 and ht1-6 contained G to A transitions at positions 858 and 851 that are predicted to result in changes from Gly to Glu at position 232 and Asp to Asn at 230, respectively (Fig. 4). These are invariant Gly and Asp residues within subdomain VII. The ht1-5 mutation contains a C to T transition at position 1023 that is predicted to result in a stop codon at amino acid 287 (Q287stop); thus, the mutation would lead to a truncated HT1 protein lacking subdomains X and XI (Fig. 4). The site of a T-DNA insertion in ht1-7 (FLAG_446H04) was found to be in the second exon and is predicted to result in disruption of the HT1 gene (Fig. 4). All these recessive mutations either convert an amino acid (ht1-1, ht1-4 and ht1-6), lead to deletion of amino acids (ht1-2 and ht1-5), or disrupt the T-DNA insertion (ht1-7) at highly conserved regions in the catalytic domain.
The ht1-3 mutation was found to contain a G to A transition at position 305 that results in an Arg being replaced by Lys at amino acid 102 (R102K) (Fig. 4). The position is not highly conserved among protein kinases; for example, some plant kinases such as FsPK1 and its homologous protein GmPK6 have His at this position (Fig. 4B). Furthermore, a variety of amino acids other than Arg, including His, Asn and even Lys, can be present at this position in kinases of  -1, ht1-2, ht1-3 and ht1-4) plants. The UBQ10 gene was used as an internal standard for cDNA amounts. Error bars on each column indicate the standard deviation from four biological replicates. The statistical significance was determined by a one-way ANOVA with Tukey-Kramer multiple comparison tests. The same letter indicates no significant difference (P>0.05). (B) In vitro kinase assays using recombinant HT1 and HT1 R102K with or without casein. Coomassie staining of the His-fusion protein and casein served as the loading controls. (C) Quantified autophosphorylation (Auto) and casein phosphorylation (Casein) corrected for protein content as quantified by Coomassie staining and calculated as values relative to HT1 phosphorylation activity. Error bars on each column indicate the standard deviation from seven biological replicates. ***, indicates a significant difference (P<0.001), from the HT1 phosphorylation levels as assessed by a paired t-test. animals and fungi (Hanks et al., 1988). Arg and Lys residues are structurally similar and are classified as basic amino acids. This result led us to wonder why this amino acid replacement caused such a drastic change in stomatal response to CO 2 .

Structural model of the HT1 kinase
To gain insights into the functional meaning of the ht1-3 mutation, we constructed a 3D structural model of the HT1 kinase and examined the mutated site. The 3D structural model of 273 amino acids of the HT1 catalytic region (residues 73-345) was generated by homology modeling using Modeller (Fiser and Šali, 2003) (Fig. 5). The template used to build the HT1 kinase model was derived from the crystal structure of the CTR1 kinase domain (chain A, Protein Data Bank code 3ppz) (Mayerhofer et al., 2011). Since the ligand of CTR1 in 3ppz is staurosporine, an ATP analog, AMP-PNP was docked at the ATP binding site of HT1 by superimposing the main chain structure on that of another HT1 analog, cAMP-dependent protein kinase (pdb code 4dfx) that is bound to AMP-PNP (Fig. 5). CTR1 is a negative regulator of the ethylene response pathway in Arabidopsis and is a member of the Raf-like MAPKKKs of Group B that have extended N-terminal domains (Kieber et al., 1993;Ichimura et al., 2002). HT1 kinase has a short N-terminal domain that is unlike that of CTR1; however, the catalytic domains of HT1 are relatively similar to those of CTR1. Notably, R102 of HT1 corresponds with the Arg present at position 567 in CTR1 (Fig. 4B).
Lys 113 is an invariant residue in subdomain II that is essential for phosphorylation activity in protein kinases; a replacement of this amino acid by Trp abolishes kinase activity (Hanks et al., 1988;Hashimoto et al., 2006). The highly conserved residues in subdomain VI (corresponding to R211; ht1-1) and subdomain VII (corresponding to D230; ht1-6, and G232; ht1-4) have been implicated in ATP binding (Hanks et al., 1988). Consistent with these reports, four amino acids (Lys 113, R211, D230, and G232) are located  (Hanks et al., 1988). The consensus amino acids (Gly-X-Gly-X-X-X-X-Val) of subdomain I are not evident in AtHT1 and OsILA1, but there is an invalid Gly (indicated by an asterisk) in the potential subdomain I that is indicated as (I). Previously reported positions of subdomain I in FsPK1, GmPK6, OsILA1, and AtCTR1 are indicated with red letters. The positions of amino acid changes or the T-DNA insertion sites for the seven ht1 alleles are indicated on the sequences of AtHT1. AtHT1, Arabidopsis thaliana high leaf temperature 1 (accession no. Q2MHE4); FsPK1, Fagus sylvatica protein kinase 1 (accession no. CAC09580); GmPK6, Glycine max protein kinase 6 (accession no. NP_001238530); AtCTR1, Arabidopsis thaliana constitutive triple response 1 (accession no. NP_195993); OsILA1, Oryza sativa increased leaf angle 1 (accession no. NP_001058617). The positions of amino acids changes found in the ht1 alleles are indicated above the alignments. close to the above-mentioned ATP analog (AMP-PNP) (Fig. 5). Deletions of the amino acids caused by the mutations ht1-2 (Δ136-149) or ht1-5 (Δ287-C) also are likely to destroy the kinase structure and lead to loss of function. The ht1-1 and ht1-2 loss-of-function mutations of HT1 reduce or disrupt HT1 kinase activity and result in constitutive stomatal closure even under low [CO 2 ] (Hashimoto et al., 2006). This result is consistent with the finding that all recessive ht1 alleles cause a deletion or replacement of the highly conserved amino acids at kinase catalytic domains and result in plants with higher leaf temperatures even in low [CO 2 ] ( Fig. 1 and Supplementary Fig. S1).
On the other hand, the ht1-3 mutation site (R102) is not a highly conserved residue, and the replacement of Arg102 by Lys seems not to affect the HT1 kinase structure severely. In the model of HT1 protein structure, Arg102 seems to protrude from the surface of the kinase. When the Arg is substituted with Lys, the side chain becomes shorter, and thus the positive charge stretches less outward (Fig. 5). The R102K change may therefore affect HT1 interactions with its targets.
CO 2 responses in the ht1-3 and ht1-2 mutants were completely defective The dominant ht1-3 mutation leads to stomatal opening due to guard cell insensitivity to high [CO 2 ] and, therefore, ht1-3 and the other loss-of-function ht1 alleles are likely to have opposite effects on stomatal opening or closing in response to CO 2 . To examine the nature of the ht1-3 mutation, we compared the CO 2 , light and ABA responses of the ht1-3 mutant with the responses of the severe loss-of-function mutant ht1-2. First, we investigated how stomatal conductance responded to changing [CO 2 ] in the leaves of ht1-3, ht1-2 and WT plants (Fig. 6A). In the WT, increasing [CO 2 ] from 350 ppm to 700 ppm induced a distinct decrease in stomatal conductance, and a subsequent decrease of [CO 2 ] from 700 to 100 ppm induced a large increase in stomatal conductance (Fig. 6A, left). In contrast, stomatal conductance remained at a lower level in ht1-2 and a higher level in ht1-3, with little change in stomatal conductance in response to CO 2 (Fig. 6A, left). Interestingly, ht1-3 still had an extremely small, inverse response to CO 2 , as did ht1-2 (Fig. 6A, right) . These inverse responses may be the result of the activity of a counterbalancing regulator of the normal CO 2 -induced stomatal response. These observations suggest that these mutations severely disrupt stomatal CO 2 signaling.
Next, we analysed light-induced stomatal responses in ht1-3 mutants (Fig. 6B). Irradiation resulted in a large increase in stomatal conductance in the WT, a smaller increase in ht1-3, and a much smaller increase in ht1-2 (Fig. 6B, left). In contrast to their inverse responses to changing [CO 2 ], the ht1-2 and ht1-3 mutants did respond to changes in light intensity in the same direction as the WT responses (Fig. 6B, right).
The ht1-3 mutant did not have the wilted phenotype under normal conditions; however, when grown without water for a week, the mutant started to exhibit a wilted phenotype compared with the WT (see Supplementary Fig. S4). As drought stress triggered a wilted phenotype in the ht1-3 plants, it was possible that the mutant plants might be insensitive to ABA. Production of the phytohormone ABA is triggered by desiccation and induces stomatal closure, and ABA-insensitive mutants have a wilted phenotype under dry conditions. We found that the degree of stomatal closure induced by ABA in the ht1-2 and ht1-3 mutants was similar that of the WT ( Fig. 6C; WT, ht1-2 or ht1-3; P<0.01, for controls vs. ABA; Welch's t-test), suggesting that ABA responses in these mutants are normal. This finding indicates that the wilted phenotype observed in ht1-3 was due to insensitivity to CO 2 but not to ABA.

HT1 is a Group C Raf-like MAPKKK essential for the stomatal CO 2 response
We performed phosphorylation assays with several kinase inhibitors and confirmed that HT1 is a Group C Raf-like MAPKKK. Almost all of the MAPKKKs that have been functionally characterized are members of either Group A or B. In contrast, most Group C members have been described only from genomic sequence analyses (Ichimura et al., 2002). FsPK1, an ABA-induced protein kinase, is classified in subgroup C5, the same group as HT1. FsPK1 has Ca 2+ -dependent protein kinase activity that is inhibited by the kinase inhibitors staurosporine and genistein, indicating dual activities (Ser/Thr and Tyr protein kinases) (Lorenzo et al., 2003). OsILA1, also a member of Group C, is a key factor for regulating mechanical tissue formation at the leaf lamina joint; this protein has Ser/Thr kinase activity but no Tyr kinase activity (Ning et al., 2011). An in vitro kinase assay revealed that HT1 could be a Raf-related MAPKKK with Ca 2+ -independent Ser/Thr kinase activity (Fig. 2A).
These findings demonstrate that the features of this protein kinase and others vary among the subgroups of Group C MAPKKKs. HT1 has phosphorylation activity, and mutations inducing changes in the highly conserved amino acids in the catalytic domain impair its activity (Hashimoto et al., 2006). All six ht1 recessive mutation sites were expected to alter the highly conserved amino acid residues that play a critical role in phosphorylation activity, and resulted in lossof-function phenotypes (Figs 1, 4 and Supplementary Fig.  S1). These results indicate that the kinase activity of Raf-like MAPKKK HT1 is important for stomatal CO 2 responses.

HT1 is localized on plasma membranes
OST1 acts as an ABA-activated SnRK2-type protein kinase (Mustilli et al., 2002), and phosphorylates and activates the S-type anion channel SLAC1 (Geiger et al., 2009;Lee et al., 2009). OST1 and SLAC1 have also been reported to be involved in elevated [CO 2 ]-induced stomatal closure (Negi et al., 2008;Vahisalu et al., 2008;Xue et al., 2011). A recent study has indicated that a MATE-type transporter, RHC1, could interact with HT1 and overcome HT1 inhibition of downstream SLAC1 activation by OST1 under high-bicarbonate conditions (Tian et al., 2015). The subcellular localization of RHC1 was reported to be the plasma membranes, whereas OST1 localizes to nuclei and the cytosol (Fujita et al., 2009;Tian et al., 2015). We showed that HT1 is associated with plasma membranes, indicating that it could transduce CO 2 signals from plasma membrane-resident RHC1 to OST1 in the cytosol.

HT1 is a master regulator in the stomatal CO 2 response
Similar to recessive loss-of-function ht1-2 mutants, the ht1-3 plants had defects in their CO 2 response; however, the ht1-3 plants showed a functional stomatal response to light and a normal response to ABA, indicating a CO 2 -specific role for HT1 ( Fig. 6) (Hashimoto et al., 2006). CO 2 -induced stomatal conductance changes in ht1-2 and ht1-3 plants were disrupted, and the stomatal conductance in ht1-2 and ht1-3 remained low and high, respectively (Fig. 6A). Interestingly, we found slightly inverse responses to CO 2 in both alleles (Fig. 6A). No other mutants have been reported that show such severe and specific damage in their CO 2 responses. A crucial role of the HT1 kinase in CO 2 signaling is also supported by the observation that many of the CO 2 -signaling mutants we have been isolating by thermal imaging screens are HT1 mutant alleles.

HT1 is partially involved in the light-signaling pathway
The HT1 loss-of-function and gain-of-function mutations brought about a reduced response to light (Fig. 6). This result indicates that the functions of HT1 partially share the light-signaling pathway. Red light induces stomatal opening by reducing the intracellular [CO 2 ] caused by photosynthesis (Roelfsema et al., 2002(Roelfsema et al., , 2006; however, other research has reported that red light can induce stomatal opening when the intracellular [CO 2 ] was constantly maintained (Messigger et al., 2006;Lawson et al., 2008). A recent study has reported that the ht1-2 mutant was impaired in red light-induced stomatal opening (Matrosova et al., 2015). In our study, light induced a rise in stomatal conductance in ht1-3 that was larger than that of ht1-2, although both mutants completely lost their stomatal response to CO 2 (Fig. 6B). Therefore, analysis of the red light-induced stomatal opening response in ht1-3 should provide more clear information about the contribution of HT1 to the red light signaling pathway and the roles of the reduced [C i ]-dependent and the [C i ]-independent pathways.

ht1-3 is a dominant mutation leading to enhanced autophosphorylation activity
The abundance of HT1 mRNA in ht1-3 plants was not significantly different from that in the wild type (Fig. 3A), suggesting the dominant mutation may affect post-transcriptional regulation. In the HT1 protein structure model, Arg102 seems to protrude from the surface of the kinase (Fig. 5). Arg and Lys have similar positively charged residues, but Lys has a shorter side chain and thus the positive charge would not stretch outward as much (Fig. 5). This explanation suggests that the ht1-3 mutation might affect the kinase's interaction with its targets. It may be hypothesized that the ht1-3 mutation influences kinase activity because Arg102 may be located close to the kinase active site (Fig. 5); however, the kinase assays revealed that the HT1 R102K was not affected in its ability to phosphorylate casein (a universal substrate for a wide range of kinases) in vitro (Fig. 3C). This result indicates that the dominant mutation does not enhance kinase activity by itself. In contrast, HT1 autophosphorylation activity was significantly increased by the ht1-3 mutation (Fig. 3C). This result suggests that the ht1-3 mutation enhances the formation of HT1 oligomers and/or the efficiency of self-phosphorylation, and then the activated HT1 R102K kinase can interact and phosphorylate target proteins. Future studies searching for direct HT1 targets and using phosphorylation profiling by means of activated HT1 R102K kinase may allow further isolation of the CO 2 signaling factors, and thus result in more detailed elucidation of the CO 2 signaling pathways.