Identification of Features Regulating OST1 Kinase Activity and OST1 Function in Guard Cells^,

Abstract

The phytohormone abscisic acid (ABA) mediates drought responses in plants and, in particular, triggers stomatal closure. Snf1-related kinase 2 (SnRK2) proteins from several plant species have been implicated in ABA-signaling pathways. In Arabidopsis (Arabidopsis thaliana) guard cells, OPEN STOMATA 1 (OST1)/SRK2E/SnRK2-6 is a critical positive regulator of ABA signal transduction. A better understanding of the mechanisms responsible for SnRK2 protein kinase activation is thus a major goal toward understanding ABA signal transduction. Here, we report successful purification of OST1 produced in Escherichia coli: The protein is active and autophosphorylates. Using mass spectrometry, we identified five target residues of autophosphorylation in recombinant OST1. Sequence analysis delineates two conserved boxes located in the carboxy-terminal moiety of OST1 after the catalytic domain: the SnRK2-specific box (glutamine-303 to proline-318) and the ABA-specific box (leucine-333 to methionine-362). Site-directed mutagenesis and serial deletions reveal that serine (Ser)-175 in the activation loop and the SnRK2-specific box are critical for the activity of recombinant OST1 kinase. Targeted expression of variants of OST1 kinase in guard cells uncovered additional features that are critical for OST1 function in ABA signaling, although not required for OST1 kinase activity: Ser-7, Ser-18, and Ser-29 and the ABA-specific box. Ser-7, Ser-18, Ser-29, and Ser-43 represent putative targets for regulatory phosphorylation and the ABA-specific box may be a target for the binding of signaling partners in guard cells.

Drought is a major environmental constraint and plants have developed several protective strategies to survive this stress. Most plants first respond to drought by closing their stomata to prevent water loss via transpiration. Abscisic acid (ABA) is a major phytohormone that mediates drought responses in plants and, particularly, reduction of stomatal conductance through its action on guard cells (Tardieu et al., 1992). Regulation of potassium and anion channels in response to ABA leads to a rapid decrease in osmotic pressure and, consequently, a massive efflux of water from guard cells. This decrease in turgor results in stomatal pore closure. If the mechanism of the stomatal closure is now quite well described (Roelfsema and Hedrich, 2005), a thorough understanding of the signal transduction pathway of ABA in guard cells clearly requires further research. Many second messengers and genes have been shown to be involved in ABA signal transduction, but little evidence exists concerning molecular interactions between these components (Hetherington, 2001; Schroeder et al., 2001; Himmelbach et al., 2003). Moreover, although an ABA receptor was recently characterized (Razem et al., 2006), it is not active in guard cells and the molecular mechanisms of ABA perception in these cells remain unknown.

Previous studies reported the involvement of Snf1-related kinase 2 (SnRK2) proteins in ABA signal transduction pathways. First, PKABA1 was implicated in ABA signal transduction in wheat (Triticum aestivum; TaPKABA1) and barley (Hordeum vulgare; HvPKABA1) seeds (Anderberg and Walker-Simmons, 1992; Gomez-Cadenas et al., 1999). Then, the fava bean (Vicia faba) SnRK2 protein AAPK was shown to be activated by ABA in guard cells (Li et al., 2000) and to control stomatal response to ABA. More recently, its ortholog in Arabidopsis (Arabidopsis thaliana), OPEN STOMATA 1 (OST1)/SRK2E, has been identified as a key component of ABA signal transduction in guard cells (Mustilli et al., 2002; Yoshida et al., 2002). Mutations in the OST1 gene severely impair both stomatal closure and inhibition of stomatal opening induced by ABA. In contrast, ost1 mutants display wild-type ABA responsiveness in seeds. Thus, they are the first recessive mutants that specifically affect ABA signal transduction in guard cells.

Several studies have reported the phosphorylation of substrates by ABA-activated SnRK2 in different plant species. They include bZIP transcription factors, such as TaABF from wheat (Johnson et al., 2002), TRAB1 from rice (Oryza sativa; Kagaya et al., 2002), and AREB1 from Arabidopsis (Furihata et al., 2006), or RNA-binding proteins such as VfAKIP1 from fava bean (Li et al., 2002). In contrast, OST1 has been positioned as the most upstream positive regulator known in the ABA-signaling pathway in Arabidopsis guard cells (Mustilli et al., 2002) and, to our knowledge, nothing is known about the identity of upstream components targeting activation of SnRK2 in plants. Understanding the mechanisms of OST1 activation in response to ABA in guard cells is a major question.

Many kinases are themselves regulated by phosphorylation and dephosphorylation events. Recent studies reported the importance of phosphorylation in rice SnRK2 activation in response to osmotic stresses (Kobayashi et al., 2004). Moreover, OST1 kinase activation in response to ABA is suppressed in the dominant abi1-1 mutant (Mustilli et al., 2002), indicating that the protein phosphatase 2C ABI1 (Koornneef et al., 1984; Leung et al., 1994; Meyer et al., 1994) negatively regulates ABA signal transduction upstream of OST1. A recent study indicates that ABI1 may bind OST1 protein, suggesting that it could directly modulate its phosphorylation status (Yoshida et al., 2006). SnRK2 kinases also have a very acidic carboxy-terminal region with unknown function. Recent works on Arabidopsis and rice SnRK2 families reported that three of 10 proteins in each species are strongly activated in response to ABA in cultured cell protoplasts, whereas all of them are activated in response to osmotic stress (Boudsocq et al., 2004; Kobayashi et al., 2004). Domain-swapping experiments using rice SAPKs have demonstrated that grafting the noncatalytic C-terminal region from ABA-activated SnRK2 (Glu-254 to Met-372 from SAPK8) onto others (SAPK2 catalytic domain) is sufficient to confer ABA responsiveness (Kobayashi et al., 2004). Hence, phosphorylation and dephosphorylation events, as well as binding of partners to the regulatory carboxy terminus of the protein, are probably both involved in OST1 activation. We decided to investigate the relationship between OST1 structure and its function in guard cells, focusing on these two types of regulation.

Many previous studies using SnRK2 proteins from different plant species, including soybean (Glycine max), fava bean, tobacco (Nicotiana tabacum), wheat, or Arabidopsis, report that expression of recombinant proteins in Escherichia coli results in inactive proteins (Yoon et al., 1997; Li et al., 2000; Johnson et al., 2002; Mustilli et al., 2002; Kelner et al., 2004). Authors hypothesized that SnRK2 activity requires activation by specific signaling cascades in plants. A single study on rice SnRK2 OsREK (SAPK3) reported some autophosphorylation of the recombinant kinase partially purified from E. coli extracts (Hotta et al., 1998), suggesting that active SnRK2 can be obtained. Production of SnRK2 in E. coli would make it easier to obtain large amounts of pure proteins and thus to perform biochemical studies than by using immunoprecipitated proteins from plant systems. Moreover, production of pure recombinant proteins in E. coli would allow the study of their biochemical kinase activity independently of plant factors that might be coimmunoprecipitated.

Here we demonstrate that the recombinant fusion protein 10xHis-OST1 produced in E. coli and purified under native conditions is active and is able to autophosphorylate. This allows the identification of phosphorylated Ser residues representing putative targets of upstream kinases or phosphatases. The importance of these residues is investigated both in vitro to study biochemical activity and in Arabidopsis guard cells to analyze their impact on OST1 function in signal transduction. Truncated versions also allow us to investigate the role of the C terminus. We report here the identification of conserved features critical for OST1 kinase activity and function in guard cells.

RESULTS

Recombinant OST1 Protein Is an Active Kinase and Autophosphorylates

To study the kinase activity of recombinant OST1, a vector was constructed that allowed expression of a 10xHis N-terminal-tagged OST1 protein in E. coli. Figure 1A shows the different steps of 10xHis-OST1 production and purification under native conditions. Large amounts (>20 mg) of highly purified (>95%) recombinant OST1 were obtained. Assays for kinase activity show that this recombinant kinase is able to efficiently phosphorylate generic substrates such as histone (Fig. 1B) and myelin basic protein (data not shown). Moreover, the occurrence of a band corresponding to the size of OST1 (about 50 kD) suggests that it is able to autophosphorylate. This is confirmed by kinase assays performed in the absence of a substrate (Fig. 1C). In addition, a modified kinase assay, using either [α-³³P]ATP as a substrate or with the addition of unlabeled ATP 1 min prior to stopping the reaction, shows that the signal at OST1 size is due to autophosphorylation rather than binding of radioactive ATP (Fig. 1C). In contrast, the mutated recombinant version 10xHis-OST1^G33R corresponding to the protein encoded by the ost1-2 allele (Mustilli et al., 2002) purified in the same conditions is totally inactive (Fig. 1B).

Figure 1.

OST1 recombinant protein is active and autophosphorylates. A, Different steps of 10xHis-OST1 production and native purification, SDS-PAGE, and Coomassie staining. L, Ladder; Ni, noninduced bacteria lysate; 3 h, bacteria lysate 3 h after isopropylthio-β-galactoside induction; S, soluble fraction; E, elution. B, In vitro kinase activity (top) of the native recombinant 10xHis-OST1 (OST1) and the mutated version 10xHis-OST1^G33R (G33R). C, In vitro kinase activity (top) of the native recombinant 10xHis-OST1 without (−) or with (+) histone III-S substrate and with usual [γ-³³P]ATP (γ), [α-³³P]ATP (α), or [γ-³³P]ATP followed by a pulse of cold ATP 1 min before stopping the reaction (cold). B and C, Arrowheads indicate autophosphorylation (OST1) and phosphorylation of histone III-S substrate (H). The bottom image shows the protein level by immunodetection using antibodies directed against the poly-His tag (WB).

We tested whether OST1 activity is modulated by second messengers involved in ABA signal transduction or ABA itself. Changes of external pH (from 7–8.5) or calcium concentrations (from 0–500 μm CaCl₂) did not significantly alter OST1 activity in vitro (data not shown). Finally, we demonstrated that ABA does not directly regulate OST1 in vitro activity (data not shown).

We are thus able to purify under nondenaturing conditions a recombinant OST1 protein kinase and we demonstrate that it is active alone without any additional factor. This provides us with a powerful tool to investigate structure-function relationships of OST1 protein.

Phosphorylation Is Required for OST1 Activity

Active OST1 kinase is slightly shifted to higher apparent molecular mass compared to the inactive OST1^G33R mutant (Fig. 1B). In rice, such shifts have been correlated with phosphorylations of SnRK2 kinases (Kobayashi et al., 2004). Moreover, active recombinant OST1 is able to autophosphorylate. These results lead us to test the importance of OST1 phosphorylation status for its activation in guard cells. Using an OST1-specific antiserum, we immunoprecipitated OST1 from extracts of purified guard cells treated or not with ABA. Figure 2 shows that ABA-induced kinase activity is immunoprecipitated from wild-type guard cells. This kinase activity is absent from srk2e extracts, demonstrating that it corresponds to OST1. It is abolished when immunoprecipitated proteins are dephosphorylated before the kinase assay (Fig. 2). This indicates that OST1 phosphorylation is critical for its activity in guard cells and leads us to analyze differences of phosphorylation status between active and inactive recombinant proteins.

Figure 2.

In planta OST1 phosphorylation is required for its activity. Kinase assay of proteins immunoprecipitated from GCP extracts using a serum that specifically recognizes OST1. Protoplasts were treated for 30 min with 30 μm ABA (+) or with ethanol solvent (−). Immunoprecipitated proteins were treated with a calf intestinal phosphatase (+CIP) or were submitted to the same treatment without any phosphatase (−CIP). The faint band at high M_r corresponds to autophosphorylation and the high-intensity band at low M_r represents histone phosphorylation.

Ser-7, Ser-18, Ser-29, Ser-43, and T Loops Are Targets of Autophosphorylation of OST1 in Vitro

Mass spectrometry (MS) studies were performed to detect phosphorylated residues in the recombinant OST1 protein. The average mass of 10xHis-OST1 measured by matrix-assisted laser-desorption ionization (MALDI)-MS analysis was about 400 D above the theoretical value calculated from the sequence. However, the pseudomolecular ion cluster was broad and unresolved. More convincing indications of phosphorylation were obtained by means of electrospray ionization (ESI)-MS analysis of recombinant proteins 10xHis-OST1 and 10xHis-OST1^G33R. Seven different protein species were separated for active OST1 with masses ranging between 46,658 and 47,151 D (Table I). A comparison between the theoretical mass of the recombinant protein (46,342 D, taking into account excision of the N-terminal Met) and experimental masses indicates that these different species correspond to isoforms of the protein carrying from four to 10 phosphate groups. Indeed, the average masses of the seven detected protein species differ by 80 D (i.e. the mass of one added phosphate group). The analysis of OST1^G33R led to a major peak at 46,436 D that matches the theoretical mass of unphosphorylated OST1^G33R and a minor one that probably corresponds to a degradation product. We may thus conclude that 10 residues of 10xHis-OST1 recombinant protein are targets of autophosphorylation.

Table I.

Accurate mass determination indicates multiple phosphorylation of recombinant OST1

ESI-LC-TOF mass measurements for 10xHis-OST1 and mutant version 10xHis-OST1^G33R. The third column indicates the experimental average mass of the various isoforms found in each sample. The fourth one indicates for each isoform the calculated number of phosphate groups based on the difference between experimental and theoretical masses (79.96 D per phosphate group).

Recombinant Protein	Theoretical Mass	Experimental Masses	Calculated No. of Phosphate Groups
	D	D
10xHis-OST1	46,342	46,658 ± 6	4
		46,740 ± 5	5
		46,820 ± 8	6
		46,899 ± 4	7
		46,978 ± 5	8
		47,060 ± 5	9
		47,151 ± 7	10
10xHis-G33R	46,442	40,299 ± 7	xa
		46,436 ± 8	0

^aThis isoform does not correspond to a full-length protein.

In a second stage, tandem MS (MS/MS) analysis of a tryptic digest of recombinant OST1 was used to map target residues. A combination of MALDI-quadrupole time-of-flight (Q-TOF) and nano-liquid chromatography (LC)-ESI-Q-TOF analysis led to 68% sequence coverage of the fusion protein; this low score is explained by the fact that the [280–369] tryptic peptide accounting alone for 22% of the sequence was not detected. Nine potential phosphorylation sites were detected by a combination of exact mass measurement of precursor ions and observation of neutral loss (H₃PO₄) from these ions (Supplemental Table I). Four phosphorylated sites belong to the N-terminal tag (residues -25, -24, -10, and -9) and are not relevant for OST1 function. Ser-7, Ser-18, Ser-29, and Ser-43 belonging to the catalytic domain are modified by phosphate groups (Fig. 3). Peptide ¹⁷⁵STVGTPAYIAPEVLLK¹⁹⁰ corresponds to the activation loop (also called activation segment or T loop) that is conserved in most kinases. The MS/MS spectrum of this peptide ruled out the modification of Thr-179, but did not allow the assignment of the phosphate group to Ser-175 or Thr-176. No trace of a Ser-175 and Thr-176 doubly modified peptide was found, and incomplete coverage of the protein sequence presumably hindered identification of the tenth phosphorylated residue.

Figure 3.

Phosphorylated residues identified on recombinant OST1 by MS. Arrows indicate positions of identified phosphorylated residues in OST1 sequence. The catalytic domain and carboxy-terminal region are represented by bold and italic letters, respectively. Shaded sequences are not covered by MS analysis. Underlined stretches represent a conserved ATP-binding site GXGXXG (ATP) and activation segment (T loop). The black circle shows Gly-33, which results in an inactive protein when mutated to Arg.

Thus, MS pointed out six residues that are putatively important for OST1 activity regulation by phosphorylation, namely, Ser-7, Ser-18, Ser-29, Ser-43, Ser-175, and Thr-176 (Fig. 3).

Ser-175 and Thr-176 Are Critical for OST1 Biochemical in Vitro Activity

In a first step, the requirement of each of the candidate phosphorylated residues identified for kinase activity of the recombinant fusion protein was tested. Each was substituted, either for an Ala to prevent phosphorylation or for an Asp to mimic constitutive phosphorylation. All point mutants (OST1^S7A, OST1^S7D, OST1^S18A, OST1^S18D, OST1^S29A, OST1^S29D, OST1^S43A, OST1^S43D, OST1^S175A, OST1^S175D, OST1^T176A, and OST1^T176D) were produced as recombinant 10xHis N-terminal-tagged proteins in E. coli. Circular dichroism measurements in far UV (200–250 nm) revealed that none of the mutations drastically modified OST1 secondary structure (mostly α helices; data not shown). Figure 4A shows that mutations of Ser-7, Ser-18, Ser-29, and Ser-43 do not critically affect kinase activity. Indeed, quantification of three independent assays shows that activity ratios for the corresponding mutants are within the range 0.6 to 1.3 compared with wild-type OST1. Phosphorylation on these four residues is thus not required for OST1 kinase activity in vitro.

Figure 4.

Phosphorylation of Ser-175 is required for OST1 kinase activity. A, Point mutations on Ser-7, Ser-18, Ser-29, and Ser-43. B, Point mutations on Ser-175 and Thr-176. A and B, Each residue was mutated to Ala or Asp. As in Figure 1, the top band corresponds to autophosphorylation (OST1) and the bottom band corresponds to phosphorylation of histone III-S (H). The bottom sections (WB) show the protein level by immunodetection using antibodies directed against the poly-His tag.

In contrast, both mutations of Ser-175 greatly affect ability of the kinase to phosphorylate a substrate (Fig. 4B). In this case, it would appear that substitution of Ser-175 for an Asp does not mimic constitutive phosphorylation of the residue because both mutants are equally impaired in kinase activity. The mutation Thr-176 to Asp is also critical for OST1 activity because we can only detect a slight band of autophosphorylation. However, the mutation of the same Thr to Ala has no effect on the kinase activity, demonstrating that phosphorylation of Thr-176 is not required for OST1 activity. Two hypotheses may account for these results: Either the Thr to Asp mutation does, in this case, mimic constitutive phosphorylation and phosphorylation of this residue inactivates the kinase, or the Thr-176 D mutation results in a misfolded protein when expressed in E. coli, which could explain the very low activity. At this stage, we cannot discriminate between the two hypotheses.

We conclude that Ser-175 is critical for OST1 biochemical activity in contrast to Ser-7, Ser-18, Ser-29, and Ser-43. However, these residues might still be targets of regulatory phosphorylation events in planta.

In Planta Studies Focus on OST1 Function in Guard Cells

To test the functionality of OST1 point mutants in plants, we generated a binary vector driving the expression of the protein of interest in fusion with a triple hemagglutinin (3xHA) N-terminal tag under the control of the OST1 promoter. The use of the gateway cloning system and of seed-specific green fluorescent protein expression as the selection marker (Bensmihen et al., 2004) allowed us to quickly generate and sort numerous lines in the srk2e background (Yoshida et al., 2002).

The uidA gene was first inserted in this vector to test OST1 promoter specificity. Histochemical β-glucuronidase (GUS) staining in leaves (Fig. 5A) and all aerial organs (data not shown) confirmed that this vector allows strong and specific expression of the protein of interest in guard cells. This is in agreement with the results reported previously (Mustilli et al., 2002).

Figure 5.

OST1 expression is targeted to guard cells where it is activated by ABA and limits transpiration. A, Histochemical detection of GUS activity in the leaf of a representative line srk2e/PRO_OST1:3xHA-GUS. s, Stomata; v, vascular tissue. B, In-gel kinase assay of guard cell total protein extracts with histone III-S as the substrate. The arrowhead indicates the 42-kD band corresponding to OST1. C, Infrared thermography false color picture of detached leaves from Col and srk2e plants. The histogram represents the quantification of this picture (mean ± sd).

An in-gel kinase assay (Fig. 5B) was performed on total protein extracts from Columbia-0 (Col-0) and srk2e guard cells. This experiment confirmed that OST1 is strongly activated by ABA in wild-type guard cells, whereas OST1 kinase activity is totally absent in the srk2e mutant background. In agreement with these results, the srk2e mutant shows increased water loss due to defects in stomatal closure. This phenotype can be monitored by infrared thermography (Merlot et al., 2002) as shown in Figure 5C.

Guard cell-specific expression in the srk2e null mutant allows us to investigate OST1 function in the signaling pathways leading to stomatal closure, while avoiding possible side effects caused by ectopic expression from a 35S promoter or the presence of an inactive OST1 protein. OST1 function was investigated by testing mutant complementation using the infrared thermography approach on detached leaves to monitor stomatal aperture.

Ser-7, Ser-18, Ser-29, Ser-43, and Ser-175 Play Distinct Roles in OST1 Function in Planta

To investigate in planta the function of the Ser and Thr identified as putative phosphorylation targets, the point-mutated versions of OST1 were introduced into the srk2e null background and several independent lines were selected for all constructs. GUS and OST1 lines were used as negative and positive controls for mutant complementation, respectively. Infrared images of detached leaves with wild-type (Col-0) and mutant (srk2e) control leaves on each image were used to quantify the leaf temperature. Results in Figure 6B show that GUS is unable to complement the srk2e mutant, whereas the wild-type warm phenotype is restored when OST1 is expressed in srk2e guard cells. This figure shows one representative line for each of the residues targeted by site-directed mutagenesis. Similar results were obtained when the residues were mutated to Ala or Asp and in several lines for each substitution (Supplemental Fig. 2).

Figure 6.

Several phosphorylable Ser residues are required for OST1 function in planta. A, In vitro biochemical activity of recombinant proteins produced in E. coli: quantification of one to four kinase assays as shown in Figure 4 (0 = no activity; dotted line = OST1 activity). B, Quantification of infrared thermography images of detached leaves. Black, white, and gray bars, respectively, represent wild-type (Col-0), mutant (srk2e), and one representative line srk2e/PRO_OST1:3xHA-X (where X = uidA, OST1, or a point-mutated version of OST1) for each construct (mean ± sd, n = 3 or 4 leaves). C, In planta activation of fusion proteins in response to ABA. In vitro kinase assay on proteins immunoprecipitated from plants treated (+) or not (−) by 30 μm ABA for 3 h. Values represent normalized activity of each version: quantification of the gel using ImageJ, subtraction of the background (srk2e+ = 0), and values normalized (OST1+ = 1).

Both mutations affecting Ser-175 totally abolish OST1 ability to complement the srk2e mutant in agreement with the decreased kinase activity of the corresponding recombinant proteins (Fig. 6, A and B; Supplemental Fig. 2). In the case of Thr-176, both mutants (Thr to Ala and Thr to Asp) fully restore the wild-type phenotype when expressed in the mutant background (Supplemental Fig. 2). This Thr is thus not necessary for OST1 function. The lack of in vitro activity reported for the OST1^T176D recombinant protein is most likely due to incorrect protein folding in E. coli.

Ser-7, Ser-18, Ser-29, and Ser-43 are not required for OST1 kinase activity in vitro. Complementation assays show that mutations of Ser-7, Ser-18, or Ser-29 greatly impair the ability of OST1 to complement the srk2e mutant phenotype (Fig. 6B; Supplemental Fig. 2). Direct stomatal aperture measurements confirm that guard cells expressing point-mutated versions of Ser-7, Ser-18, or Ser-29 in the srk2e background do not respond to ABA (data not shown). To determine whether Ser-7, Ser-18, and Ser-29 mutations impair OST1 activation by ABA in planta, we tested the activity in response to ABA of the corresponding versions. Fusion proteins were immunoprecipitated from plants treated or not with ABA, using an antibody directed against the 3xHA tag. Kinase assays show that mutations on Ser-7, Ser-18, and Ser-29 do not affect activation of the kinase in response to ABA (Fig. 6C). In contrast, mutations of Ser-43 do not prevent srk2e complementation by OST1. Interestingly, the OST1^S43A version immunoprecipitated from plants is not activated by ABA, but rather is constitutively active (Fig. 6C). We conclude that Ser-7, Ser-18, and Ser-29 are critical for OST1 function in the signaling pathway leading to stomatal closure, whereas Ser-43 is involved in negative regulation of OST1 activity in the absence of ABA.

The C-Terminal Domain Affects OST1 Function through Two Conserved Motifs

To better understand how the carboxy-terminal domain affects OST1 function, we investigated the structure-function relationship of this domain. Alignment of 10 Arabidopsis SnRK2s, shown in Figure 7, reveals a highly conserved feature among the 10 proteins between the Gln-303 and the Pro-318. We named it the SnRK2-specific box. This box is also conserved in all rice SAPKs (Supplemental Fig. 1). A second region after Ser-332, which we named the ABA-specific box, is conserved specifically among the three strongly ABA-activated kinases: OST1, OSKL2, and OSKL3 (Fig. 7), as well as in ABA-activated rice SAPK (Supplemental Fig. 1). Based on these observations, we decided to dissect the function of these two domains by generating five truncated versions of the OST1 protein: OST1Δ280, which is lacking the C-terminal region, OST1Δ302, which is lacking the SnRK2 box, OST1Δ320, OST1Δ331, and OST1Δ348, which is lacking one-half of the ABA box (Fig. 7). Truncated versions of OST1 were expressed in E. coli as recombinant proteins. Circular dichroism measurements in far UV (200–250 nm) on recombinant proteins revealed that none of the truncations drastically modified OST1 secondary structure (mostly α helices; data not shown). In a second step, truncated versions of OST1 were introduced in plants in the srk2e background.

Figure 7.

SnRK2 C terminus contains two conserved boxes. Alignment of the C-terminal regions of the 10 Arabidopsis SnRK2s. Boxes show two conserved features: the SnRK2-specific box, which is conserved among all 10 kinases, and the ABA-specific box only found in strongly ABA-responsive kinases. Arrows indicate the end of the five truncated versions we have generated.

In vitro kinase assays of recombinant proteins (Fig. 8) show that OST1Δ280 and OST1Δ302 truncated proteins have no detectable kinase activity, whereas OST1Δ320 activity is similar to the full-length OST1 protein. Hence, the SnRK2-specific domain is necessary for OST1 kinase activity and its conservation suggests that it is responsible for the activation of all SnRK2 proteins. Longer truncated versions are all active when produced in E. coli. The second half of the C-terminal region, after Ala-320, is thus not necessary for OST1 kinase activity. The efficiency of substrate phosphorylation, however, differs between versions, suggesting a role of this domain in modulating OST1 kinase activity.

Figure 8.

Truncation of the SnRK2-specific feature abolishes recombinant kinase activity in vitro. The top band corresponds to autophosphorylation and the bottom one to phosphorylation of histone III-S (H). The bottom image indicates protein level by immunoblot against the His tag (WB). OST1, Full-length recombinant protein. Δ348, Δ331, Δ320, Δ302, and Δ280 refer to the position of the last residue of each truncated version.

Results of in planta complementation assays are shown in Figure 9B and Supplemental Figure 3. As expected, OST1Δ280 and OST1Δ302, which do not display kinase activity in vitro, do not complement the srk2e phenotype. More surprisingly, truncation after Ala-320, which results in an active protein when expressed in E. coli, totally abolishes activation of the kinase by ABA in planta and the ability of OST1 to complement the srk2e mutant (Fig. 9, B and C). Likewise, OST1Δ331, which retains substantial kinase activity in vitro, is not activated by ABA in planta and is not able to complement the mutant. Accordingly, the stomata of srk2e plants expressing OST1Δ331 do not respond to ABA (Fig. 10). This indicates that the ABA-specific domain is very important for OST1 activation by ABA and function in plants. In contrast, OST1Δ348 fully complements the srk2e transpiration phenotype and restores ABA responsiveness in stomatal bioassays (Fig. 10). OST1Δ348 displays stronger activation by ABA than OST1 in plants (Fig. 9), although it shows a lower kinase activity in vitro than the full-length protein. The Asp-348 to Met-362 stretch may be involved in a mechanism that negatively regulates OST1 activity. We conclude that the Leu-333 to Asp-348 stretch within the ABA-specific box is critical for the activation and function of OST1 in response to ABA in guard cells. This led us to check whether ABA could directly regulate OST1 activity. However, in vitro treatment of OST1 protein immunoprecipitated from plants by 10 μm ABA did not induce kinase activity (data not shown).

Figure 9.

Impact of C-terminal deletions on OST1 in planta activity. A, In vitro biochemical activity of recombinant proteins produced in E. coli: quantification of one to three kinase assays as shown in Figure 8 (0 = no detectable activity; dotted line = OST1 activity). B, Histograms represent quantification of infrared thermography pictures of detached leaves. Black, white, and gray bars, respectively, represent the wild type (Col-0), mutant (srk2e), and one representative line srk2e/PRO_OST1:3xHA-X (where X = OST1 or a truncated version of OST1) for each construct (mean ± sd, n = 3 or 4 leaves). C, In planta activation of fusion proteins in response to ABA. In vitro kinase assay on proteins immunoprecipitated from plants treated (+) or not (−) by 30 μm ABA for 3 h. Values represent normalized activity of each version: quantification of the gel using ImageJ, subtraction of the background (srk2e+ = 0), and values normalized (OST1+ = 1).

Figure 10.

The ABA box is required for OST1 function in response to ABA in guard cells. Stomatal aperture in darkness (black bars) and after 3 h of light without (white bars) or with 10 μm ABA (hatched bars) in the incubation medium (mean ± sem; n > 120 stomata). One representative line is shown for each of the constructs: srk2e, srk2e complemented by the 3xHA-OST1 construct (OST1), and srk2e expressing either 3xHA-OST1Δ331 (Δ331) or 3xHA-OST1Δ348 (Δ348) fusion proteins.

DISCUSSION

OST1 protein kinase is a positive regulator of stomatal closure in guard cells (Mustilli et al., 2002; Yoshida et al., 2002). Here, we present results leading to a better understanding of how the structure of OST1 protein affects its function in guard cell-specific signal transduction.

Several previous works reported that recombinant SnRK2s were inactive (Yoon et al., 1997; Li et al., 2000; Johnson et al., 2002; Mustilli et al., 2002; Kelner et al., 2004). We successfully produced in E. coli and purified a recombinant SnRK2 protein in a native active form. This discrepancy is probably due to the use of different tags. Indeed, former attempts used glutathione S-transferase or short 6xHis tags in N- or C-terminal positions. In our case, a short 6xHis tag fused at the N terminus did not allow purification of native recombinant proteins (data not shown). Only the construct with the gateway cassette allowed us to purify the 10xHis-tagged protein, probably thanks to the short linker encoded by this cassette, which provides a slightly longer tag.

The expression of SnRK2 recombinant proteins in E. coli allowed us to identify target residues of autophosphorylation in vitro. Using MS, we found four unequivocal target residues: Ser-7, Ser-18, Ser-29, and Ser-43, as well as a fifth peptide carrying one phosphate group for which we could not distinguish whether Ser-175 or Thr-176 was the target residue. This peptide corresponds to the OST1 activation loop (or T loop). Activation loops of many kinases have been extensively studied and their phosphorylation is often necessary for kinase activity of the protein (Johnson et al., 1996; Adams, 2003). In the SnRK3 protein SOS2, for instance, mimicking constitutive phosphorylation of one specific residue of the T loop (Thr-168 mutated to Asp) results in a constitutive active protein when expressed in E. coli, whereas the native protein is inactive (Guo et al., 2001). Previous results on rice SAPK have shown that this mechanism cannot be simply extended to SnRK2 proteins (Kobayashi et al., 2004). Indeed, substitution of Ser (Ser-158) or Thr (Thr-159, Thr-162) residues of the T loop for an Asp inactivates the SAPK2 kinase when expressed in rice cultured cell protoplasts.

Production of active recombinant proteins in E. coli allowed us to investigate the importance of this activation loop and other residue targets of autophosphorylation by testing in vitro the activity of versions of the protein in which each has been mutated. Our results are in agreement with reports on immunoprecipitated SAPKs (Kobayashi et al., 2004). Indeed, the nonmutated OST1 protein produced in E. coli is active and mutations on Ser-175 in the T loop greatly affect this activity. In addition, our results show that the mutation Thr-176 to Ala results in a fully active recombinant protein. Phosphorylation of Thr-176 is thus not required for activity of the OST1 recombinant protein. This is further supported by srk2e mutant complementation by OST1^T176A and OST1^T176D (Fig. 6). In contrast, Ser-175 is critical for kinase activity and corresponds most likely to the phosphorylated residue of the T loop identified using MS. Accordingly, neither OST1^S175A nor OST1^S175D is able to complement the srk2e stomatal phenotype.

Production of recombinant proteins also allowed us to investigate the structure-function relationships of the regulatory C terminus of the OST1 protein. Former studies on rice have demonstrated that the SnRK2 family may be divided into three subclasses and only rice kinases from subclass III (SAPK8, SAPK9, and SAPK10) were strongly activated in response to ABA (Kobayashi et al., 2004). In Arabidopsis, previous reports have also shown that strong activation by ABA is restricted to members of the same subclass (OST1/SnRK2-6, OSKL2/SnRK2-3, and OSKL3/SnRK2-2 [Boudsocq et al., 2004]). Alignment of the carboxy-terminal domain of all SnRK2s reveals two conserved boxes within the two domains recently defined (Yoshida et al., 2006): the SnRK2-specific box shared by all SnRK2 kinases and the ABA-specific box specifically conserved in subclass III of ABA-activated SnRK2. We demonstrate here that the SnRK2-specific box is necessary for recombinant OST1 kinase activity. Indeed, whereas the OST1Δ320 version is fully active, removal of this box in OST1Δ302 results in an inactive kinase. We propose, therefore, that the SnRK2 box is important for general intramolecular mechanisms of SnRK2 activation rather than specific low-humidity ABA-independent activation (Yoshida et al., 2006).

We also identify four phosphorylated Ser residues (Ser-7, Ser-18, Ser-29, and Ser-43) and another box (ABA specific) between Leu-333 and the end of the protein (Met-362), which are not required for recombinant OST1 kinase activity. These features represent good candidates as regulatory sites for OST1 function in plants. Mechanisms of activation of OST1 in plants probably involve some of these features through binding of upstream partners or phosphorylations.

OST1 is expressed in vascular tissues of all plant organs, but its function in this context is unknown. In contrast, OST1 is a key regulator of stomatal movements. Because stomata are very specialized structures, it is likely that the mechanisms of OST1 regulation in guard cells differ with respect to other cell types. We use the OST1 promoter to focus our assay for in planta OST1 function on guard cell-signaling pathways regulating stomatal aperture. GUS histochemical staining (Fig. 5A) confirmed that the OST1 promoter principally drives expression in guard cells (Mustilli et al., 2002). In our staining conditions, expression in vascular tissues is hardly detectable. Moreover, we investigated specific guard cell function of OST1 by monitoring transpiration by infrared thermography as well as direct measurements of stomatal aperture. Our functional analysis of OST1 structure function in guard cells complements previous studies using suspension cells overexpressing SnRK2 proteins (Boudsocq et al., 2004; Kobayashi et al., 2004) or studies of OST1 in roots, where its function remains unknown (Yoshida et al., 2006). It allows the identification of key structural features that might be relevant for OST1 modulation in the context of stomatal regulation.

Expression in the srk2e mutant of truncated OST1 versions that are inactive when produced in E. coli (OST1Δ280 and OST1Δ302), as well as point mutations of Ser-175, which greatly affect in vitro activity of OST1 recombinant protein, does not allow mutant complementation. These results indicate that the function of OST1 in guard cells involves its kinase activity rather than a scaffolding of transduction components.

Among the mutations that do not impair recombinant OST1 protein kinase activity, mutated versions on Ser-43 are able to complement the srk2e mutant for guard cell phenotype. Interestingly, the OST1^S43A mutant kinase is constitutively active in plants. We propose that phosphorylation of this Ser, which is perfectly conserved in SnRK2 and SnRK3 families, represses OST1 activity in the absence of ABA. Mutations on the three other identified Ser residues affect the ability of OST1 to complement the mutant, but not its activation by ABA in plants. We suggest that Ser-7, Ser-18, and Ser-29 may be targets of phosphorylation and dephosphorylation events in the guard cell ABA-specific signaling pathway, and participate in the regulation of OST1 protein stability, target recognition, or docking of OST1 to a specific transduction complex. Ser-7, in the N-terminal extension peptide specific to OST1, and Ser-18, which links this N-terminal peptide with the catalytic domain (Fig. 3), are specific of the subclass III SnRK2 proteins as defined previously (Kobayashi et al., 2004). This supports their implication in regulating OST1 function in the guard cell-signaling pathway. Moreover, difficulties in purifying a native active OST1 protein when fused with an N-terminal tag support the importance of the N-terminal conformation for OST1 activity. Finally, it is in agreement with a recent work (Yoshida et al., 2006) that shows that this N-terminal extension enhances OST1 kinase activity, but does not participate in activation by ABA or osmotic stress. We propose that phosphorylation of this peptide may regulate OST1 function independently of its activation by changing the conformation of the N terminus.

Domain-swapping experiments between ABA-responsive and non-ABA-responsive members of the rice SnRK2 family have shown that the carboxy-terminal domain confers ABA responsiveness (Kobayashi et al., 2004). More recently, two subdomains (I and II) were identified within the C terminus of Arabidopsis SnRK2 (Yoshida et al., 2006). This laboratory proposed a model in which domain II would be responsible for ABA activation and domain I for activation by an ABA-independent pathway. Our results show that domain I is critical for OST1 kinase activity. In contrast, we agree that domain II is not necessary for kinase activity, but is critical for OST1 function in guard cells because we demonstrate here that integrity of the ABA-specific box is absolutely required for the activation of OST1 and its function in guard cell responses to ABA. Taken together, these results suggest that the Leu-333 to Asp-348 motif is responsible for the binding of regulatory components positively regulating OST1 kinase activity in the ABA-signaling pathway. In contrast, our data suggest that the Asp-348 to Met-362 end is involved in the negative regulation of OST1 activity in response to ABA. This is in agreement with the binding of ABI1 on the ABA box to negatively regulate OST1 function in the ABA-signaling pathway in guard cells (Yoshida et al., 2006).

Combining studies of OST1 kinase activity in vitro and OST1 function in guard cells, we could identify two classes of regulatory features. Ser-175 and the SnRK2-specific box are critical for kinase activity and the mechanisms involved are probably shared by all kinases of the SnRK2 family, whatever the signaling pathway in which they act. In contrast, Ser-7, Ser-18, and Ser-29 and the ABA-specific box are required for OST1 function in ABA responses of guard cells. In addition, our results suggest that Ser-43 is involved in repression of OST1 activity in the absence of ABA. These Ser residues may be targets of upstream phosphorylation or dephosphorylation events, and the ABA-specific motif may bind specific regulatory components of the pathway. These phosphorylations and interactions with upstream partners may regulate the function of OST1 kinase in guard cells by modulating its activity, stability, or recruitment of specific substrates.

MATERIALS AND METHODS

Plant Material and Culture Conditions

The srk2e (Yoshida et al., 2002) mutant carries a T-DNA insertion in the first intron of the OST1 gene and is derived from the Arabidopsis (Arabidopsis thaliana) accession Col. srk2e seeds were kindly provided by Dr. Kazuo Shinozaki. Plants were grown routinely in a greenhouse (22°C with a 16-h light period) on soil irrigated with mineral nutrients. Transgenic plants used for immunoprecipitation of fusion proteins were grown in vitro in 0.5× Murashige and Skoog liquid medium (Murashige and Skoog basal salt mixture; Sigma) supplemented with 0.5% Suc, 0.5 g L⁻¹ MES, and B5 vitamins on an orbitary shaker (110 rpm) at 21°C with a 16-h light period.

Isolation and ABA Treatment of Guard Cell Protoplasts

Guard cell protoplasts (GCPs) were prepared essentially as reported (Pandey et al., 2002). GCPs were then washed twice in 0.35 m mannitol, 10 mm KCl, and 1 mm MES-NaOH, pH 6.1, and incubated in the same solution with ABA or its solvent ethanol as a control. Then, GCPs were collected by centrifugation (20,000g, 2 min), frozen in liquid nitrogen, and stored at −80°C.

Stomatal Aperture Measurements

Leaves from 4- to 5-week-old plants (grown in 8 h of light at 22°C and 16 h of darkness at 20°C; 70% relative humidity) were harvested in darkness at the end of the night. Abaxial epidermis was incubated for 30 min in darkness in 30 mm KCl and 10 mm MES-KOH, pH 6, at 20°C. Samples were then transferred to light for 3 h with or without 10 μm ABA in the bath solution. Stomatal apertures were measured with an optical microscope (Optiphot-2; Nikon) fitted with a camera lucida and a digitizing table (TG 1017; Houston Instruments) linked to a personal computer.

Construction of Entry Vectors

The OST1 cDNA was amplified by PCR from a pBS-OST1 (Mustilli et al., 2002) vector using high-fidelity polymerase and oligonucleotides GWOST1-F and GWOST1-R (Supplemental Table II). A BP Gateway recombination was performed with the pDONR201 vector (Invitrogen) as described in the manufacturer's instructions. The same method was used to generate entry clones for truncated versions using the 5′ primer GWOST1-F and 3′ primers GWOST1ΔC-R, 3′GWOST1/G302, 3′GWOST1/A320, 3′GWOST1/G331, and 3′GWOST1/D348 (Supplemental Table II) for OST1ΔC, OST1/G302, OST1/A320, OST1/G331, and OST1/D348 versions, respectively. Point mutations were generated directly in the OST1 entry vector, using the QuickChange site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions using mutagenic oligonucleotides (Supplemental Table II). All entry clones were sequenced and displayed only the introduced mutation.

Expression in Escherichia coli and Native Purification

OST1 variants were introduced using LR recombination in a pET16b (Novagen) modified to a Gateway destination vector (Invitrogen). This modification was made by insertion of an RfC cassette (Invitrogen) into a blunt-ended BamH1 site. The recombined expression vectors were introduced into the Escherichia coli Rosetta (DE3) pLysS strain (Novagen). Bacteria were grown to OD 0.6 at 600 nm and 0.4 mm isopropylthio-β-galactoside were added to induce the production of 10xHis-tagged proteins during 3 h at 25°C. Cells were collected by centrifugation and then lysed in the lysis buffer using both 1 mg mL⁻¹ lysozyme and sonication. After DNAse 1 (5 μg mL⁻¹) and RNAse A (10 μg mL⁻¹) treatment, a second centrifugation was performed to remove the insoluble fraction. Native purification was performed using nickel-nitrolotriacetic acid agarose resin (Qiagen) in columns according to the manufacturer's instructions. Buffers were composed of 50 mm Na₂HPO₄ and 300 mm NaCl (pH 7.5) with 20 mm, 50 mm, and 1 m imidazole, respectively, for lysis, wash, and elution buffers. Proteins were concentrated and dialyzed using Vivaspin 2-mL concentrators with a 10,000-D cutoff (Vivascience) and finally stored at −20°C in a 50 mm Na₂HPO₄, 55% (v/v) glycerol buffer.

Antibody Production

To detect and immunoprecipitate OST1 from wild-type plants, we produced an OST1-specific antibody (OST1^N) against a synthetic peptide corresponding to the N terminus of the protein (MDRPAVSGPMDLC). The peptide was coupled to the keyhole limpet hemocyanin through cystein 13 and polyclonal antiserum was raised in rabbit (Eurogentec).

Immunoprecipitation of Endogenous or Transgenic Proteins from Plants

Whole plants were ground using a Mixer Mill MM 301 (Retsch). Proteins were extracted in an immunoprecipitation buffer: 20 mm HEPES, pH 7.5, 2 mm EDTA, 2 mm EGTA, 4 mm dithiothreitol, 10 mm NaF, 50 mm β-glycerophosphate, 0.1 mm orthovanadate, 1× protease inhibitor cocktail (Roche), 200 mm NaCl, and 0.5% (v/v) Triton X-100, 0.5% (v/v) NP40. After centrifugation (4°C, 15 min, 14,000 rpm) the supernatant was recovered and 500 μL (2 μg μL⁻¹ proteins) were incubated during 30 min with protein A-Sepharose 50% slurry (Sigma) in immunoprecipitation buffer to deplete proteins binding aspecifically to protein A-Sepharose. After a second centrifugation, the supernatant was saved and incubated for 1 h with 1.5 μL of monoclonal anti-HA antibody (Sigma) or 2 μL of OST1^N serum, followed by a 20-min incubation with 30 μL of protein A-Sepharose 50% slurry. This slurry was washed three times in 1 mL immunoprecipitation buffer and three times in 1 mL kinase (see below; Figs. 6 and 9) or phosphatase buffer (50 mm Tris-HCl, pH 8, 1 mm MgCl₂, 1× protease inhibitor cocktail; Roche; Fig. 2). Protein A-Sepharose was finally resuspended, adding 25 μL of the same buffer.

For phosphatase treatment of immunoprecipitated proteins, 30 μL of the protein A-Sepharose slurry were incubated at 37°C during 16 h with 10 units of calf intestinal alkaline phosphatase (New England Biolabs).

MS Analyses

Full-length recombinant proteins purified from E. coli were subjected to MS analyses before any kinase assay. They were analyzed by MALDI-MS (reflex III; Bruker) in the linear mode using sinapinic acid as a matrix and bovine serum albumin for external calibration (singly and doubly charged species).

ESI-MS (Q-TOF micro; Waters) of OST1 isoforms was achieved after online fast desalting on a C18 column (300-μm i.d.). Multiply charged species of horse heart myoglobin were used for external mass calibration.

Nano-LC-ESI-MS/MS spectra of tryptic peptides were acquired with a Q-TOF micro (Waters) instrument interfaced to a CapLC chromatographic system using a C18 column, 75-μm i.d. (Waters). Accurate mass analysis was obtained by means of singly and doubly charged ion species of reference peptides as lock masses (lockspray system; Waters).

MALDI-MS and MALDI-MS/MS data were acquired with a Q-TOF Ultima mass spectrometer (Waters). The matrix was α-cyano-4-hydroxy-cinnamic acid. Monoisotopic masses were corrected by using the pseudomolecular ion of GluFibrinopeptide as a lock mass. Peptide ions found with masses in excess of 79.97 D or multiples of this value relative to theoretical masses were selected for MS/MS analysis and searched for neutral loss of phosphoric acid (97.977 D).

In Vitro Kinase Assays

Phosphorylation assays on recombinant proteins were performed by incubation for 45 min at room temperature of 100 ng kinase and 200 ng histone III-S substrate (Sigma) in 12.5 μL of 20 mm HEPES, pH 7.5, 0.5% (v/v) Triton X-100, 2 mm MnCl₂, 1× protease inhibitor cocktail (Roche), 10 mm NaF, and 5 mm β-glycerophosphate with 5 μCi of [γ-³³P]ATP (3,000 Ci mmol⁻¹). Reaction was stopped by adding 12.5 μL Laemmli 2× (Laemmli, 1970) supplemented with 100 mm EDTA and heating at 95°C for 5 min. Phosphorylation assays on immunoprecipitated proteins were performed in the same conditions using 15 μL of protein A-Sepharose slurry in a final reaction volume of 25 μL.

Proteins were separated by SDS-PAGE using a 10% (w/v) acrylamide gel and transferred to a nitrocellulose membrane. Radioactivity was detected on the dried membranes using a Storm imaging system (Molecular Dynamics). The same membrane was used, after scanning, for immunodetection of recombinant proteins (see below). Quantification of the activity was performed using the public domain image analysis program ImageJ, version 1.32j (http://rsb.info.nih.gov/ij). Radioactive bands were quantified using the plot lanes function on the image of the scan and data were normalized using the level of recombinant kinases quantified by the same method on immunoblot images.

In-Gel Kinase Assay

Protein extraction and in-gel protein kinase assay were performed as described (Droillard et al., 2000). Protein extracts (20 μg) were separated by SDS-PAGE using a 10% (w/v) acrylamide gel embedded with 0.5 mg mL⁻¹ histone III-S (Sigma) as substrate. After renaturation, kinase activity was assayed in 40 mm HEPES, pH 7.5, 2 mm dithiothreitol, 5 mm MnCl₂, 2 mm EGTA, 0.1 mm orthovanadate, 25 μm cold ATP, and 80 μCi [γ-³³P]ATP per gel. After an extensive wash in 5% (w/v) TCA and 1% (w/v) disodium-pyrophosphate solution, gels were dried on Whatman 3MM paper and analyzed using a Storm imager (Molecular Dynamics).

Generation of Transgenic Lines

The OST1 promoter excised from the pOST1∷GUS vector (Mustilli et al., 2002) was cloned in a pBluescript vector upstream of a 3xHA tag followed by a nopaline synthase terminator excised from pFP101 (Bensmihen et al., 2004). This combination was then transferred into the pFP100 plasmid carrying a Pro_At2S3:green fluorescent protein selection marker (Bensmihen et al., 2004). The RfA cassette (Invitrogen) was introduced in 3′ of the 3xHA tag to generate a destination vector we named p$POHA. Expression vectors were obtained by LR recombination between entry vectors and p$POHA according to the manufacturer's instructions. Transformation was performed by floral dip using the AGL.0 Agrobacterium tumefaciens strain (Lazo et al., 1991). Seeds were selected using a Leica MZFLIII epifluorescence stereomicroscope with excitation at 470 nm and emission band-pass (525 ± 50 nm) filter.

Histochemical GUS Staining

A transcriptional fusion between the OST1 promoter and uidA was generated by recombination of the pENTR-gus vector (Invitrogen) with our p$POHA vector. We performed GUS histochemical staining on five srk2e/Pro_OST1:3xHA-gus lines as described (Jefferson et al., 1987).

Infrared Thermography Complementation Assays

Thermal imaging was performed on leaves 5 to 10 min after they were detached from 3-week-old plants grown in the greenhouse. The Col-0 and srk2e controls were present on each image with transgenic lines (three to four leaves from independent plants for each line). Images were obtained using a Thermacam PM250 infrared camera (Inframetrics), saved on a PCMCIA memory card, and analyzed using the public domain image analysis program ImageJ, version 1.32j (http://rsb.info.nih.gov/ij) as described previously (Merlot et al., 2002). First, a false color scale was calibrated using the defined temperature range. Then, the complete surface of the different leaves of one line was selected and the average temperature was calculated applying the calibration line on all pixels of this surface.

Immunoblot Analyses

3xHA-tagged proteins extracted from plants were separated by SDS-PAGE using a 10% (w/v) acrylamide gel, transferred on nitrocellulose membranes, and detected by monoclonal anti-HA antibody (1:10,000; Sigma) and 10xHis-tagged recombinant proteins used for in vitro kinase assays were detected by monoclonal anti-His antibody (1:10,000; Sigma). Then they were revealed by a horseradish peroxidase-conjugated anti-mouse IgG (1:10,000; Sigma) or anti-rabbit IgG (1:10,000) using the chemiluminescence ECL plus kit (Amersham Biosciences). All western-blot analyses were performed as described (Sambrook et al., 1989) with 0.3% (v/v) Tween 20 and 0.5% (w/v) fat-free milk powder in all incubation buffers. Quantification was performed as described for in vitro kinase assays.