Salt Stress Triggers Phosphorylation of the Arabidopsis Vacuolar K⁺ Channel TPK1 by Calcium-Dependent Protein Kinases (CDPKs)

Abstract

14–3–3 proteins play an important role in the regulation of many cellular processes. The Arabidopsis vacuolar two-pore K⁺ channel 1 (TPK1) interacts with the 14–3–3 protein GRF6 (GF14-λ). Upon phosphorylation of the putative binding motif in the N-terminus of TPK1, GRF6 binds to TPK1 and activates the potassium channel. In order to gain a deeper understanding of this 14–3–3-mediated signal transduction, we set out to identify the respective kinases, which regulate the phosphorylation status of the 14–3–3 binding motif in TPK1. Here, we report that the calcium-dependent protein kinases (CDPKs) can phosphorylate and thereby activate the 14–3–3 binding motif in TPK1. Focusing on the stress-activated kinase CPK3, we visualized direct and specific interaction of TPK1 with the kinase at the tonoplast in vivo. In line with its proposed role in K⁺ homeostasis, TPK1 phosphorylation was found to be induced by salt stress in planta, and both cpk3 and tpk1 mutants displayed salt-sensitive phenotypes. Molecular modeling of the TPK1-CPK3 interaction domain provided mechanistic insights into TPK1 stress-regulated phosphorylation responses and pinpointed two arginine residues in the N-terminal 14–3–3 binding motif in TPK1 critical for kinase interaction. Taken together, our studies provide evidence for an essential role of the vacuolar potassium channel TPK1 in salt-stress adaptation as a target of calcium-regulated stress signaling pathways involving Ca²⁺, Ca²⁺-dependent kinases, and 14–3–3 proteins.

INTRODUCTION

Abiotic stress such as elevated salt levels affects plant growth and productivity. To survive salt stress and to reproduce, plants have to accommodate a stress physiology. A basic requirement for plants to adapt to salt stress is the maintenance of cellular potassium homeostasis. This is achieved by pumping out Na⁺ or dumping the toxic cation into the vacuole (Munns and Tester, 2008). Accordingly, ion transporters and channels in the plasma and vacuolar membrane represent key factors in determining salt and osmotic stress tolerance (Apse et al., 1999). To maintain electroneutrality, long-term sodium uptake requires counterfluxes of appropriate anions or efflux of potassium from the vacuole (Rodriguez-Navarro and Rubio, 2006). Salt stress, particularly sodium stress, activates the salt-overly-sensitive (SOS) system in plants leading to sodium sequestration from the cytosol (Zhu, 2003). In experiments, sudden exposure of plants to salt stress triggers a transient rise in free cytosolic calcium (Knight et al., 1997), which can be perceived by the SOS system (Qiu et al., 2002). In general, these calcium signals could be sensed by calcium-binding proteins including calmodulin, the Calcineurin B-like sensors with their interacting kinases (CBLs/CIPKs), as well as calcium-dependent protein kinases (CDPKs or CPKs in Arabidopsis) (Dodd et al., 2010; Kudla et al., 2010). The latter two transmit the signal into phosphorylation cascades capable of modulating gene expression and target protein activity (Curran et al., 2011). In line with the existence of up to four calcium-binding EF-hands in CDPKs for intramolecular calcium perception (Wernimont et al., 2010), these kinases were found to be involved in calcium-mediated abiotic stress responses (Zhu et al., 2007; Mehlmer et al., 2010; Franz et al., 2011; Wurzinger et al., 2011). In this context, it should be mentioned that CDPKs and CBL/CIPKs, through their interaction with ion channels and transporters, seem to represent part of membrane-delimited plant stress responses (Hedrich and Kudla, 2006; Wan et al., 2007; Geiger et al., 2009, 2011; Tsay et al., 2011; Hubbard et al., 2012). Like salt stress, osmotic stress adaption requires action on the level of the plasma membrane as well as the vacuolar membrane.

Arabidopsis TPK1 represents the founding member of potassium-selective vacuolar K⁺ (VK) channels (Ward and Schroeder, 1994; Allen and Sanders, 1996; Gobert et al., 2007;Isayenkov et al., 2010; Maathuis, 2011). The TPK1 channel is activated by elevated cytosolic calcium concentrations and, moreover, the open probability of this calcium-activated potassium channel is boosted by interaction with GRF6, a 14–3–3 protein (GF14-lambda; Bihler et al., 2005; Gobert et al., 2007; Latz et al., 2007). 14–3–3 proteins constitute important hubs within the Arabidopsis interactome and, as such, function not only as regulators of key metabolic points, but in addition mediate red light signaling, defense-related cell death, stomatal opening, or abiotic stress responses (Paul et al., 2012, and references therein; Tseng et al., 2012). Interaction of TPK1 with GRF6 is dependent on the phosphorylation of a serine residue in the cytosolic N-terminus of the vacuolar potassium channel. Here, we report that this residue is phosphorylated during salt-stress signaling by CDPKs. The interaction of TPK1 with CPK3 is stabilized by two conserved arginine residues within the channels’ 14–3–3 binding motif. Phenotypes of tpk1 and cpk3 mutants underpinned the relevance of this vacuolar K⁺ channel kinase interaction for Arabidopsis thaliana in balancing cytosolic potassium homeostasis and thereby overcoming episodes of salt stress.

RESULTS

TPK1 Is Phosphorylated by Calcium-Dependent Protein Kinases

Phosphorylation of the tandem-pore potassium channel TPK1 at serine 42 in its N-terminal 14–3–3 consensus binding motif is a prerequisite for 14–3–3 protein binding (Latz et al., 2007). In order to identify the kinase in question, cellular extracts from Arabidopsis seedlings were tested for TPK1 phosphorylation capability. For this purpose, the N-terminus of TPK1 (amino acids 1–79) covering the 14–3–3 binding motif was recombinantly expressed in Escherichia coli, purified and subjected to in vitro phosphorylation by Arabidopsis leaf extracts after cell fractionation. Thereby, the cytoplasmic protein fraction, rather than the microsomal part, was identified to exhibit highest activity in phosphorylating the TPK1 N-terminus (Supplemental Figure 1). Interestingly, TPK1 substrate phosphorylation efficiency seemed to depend on the presence of divalent ions such as calcium (Supplemental Figure 1). The presence of the chelator EDTA (Ethylenediaminetetraacetic acid) in the soluble kinase fraction dramatically reduced phosphorylation efficiency. These subcellular fractionation and phosphorylation properties suggested that a putative soluble, calcium-dependent protein kinase is capable of phosphorylating the 14–3–3 binding motif of TPK1.

The family of CDPKs in A. thaliana comprises 34 members. While some CDPKs have been shown to localize in the cytoplasm, others exhibit a more complex localization pattern including (partial) membrane localization (Lu and Hrabak, 2002; Dammann et al., 2003; Benetka et al., 2008; Mehlmer et al., 2010). Based on the localization characteristics determined for the as-yet unknown kinase that can phosphorylate TPK1, we thus focused on calcium-dependent kinases, which are located primarily in the cytoplasm and showed overlapping expression profiles with the TPK1 channel. Among the Arabidopsis CDPKs, CPK1, 3, 4, 5, 11, 12, and 29 met the search criteria (Hruz et al., 2008), and were thus further investigated for their ability to phosphorylate TPK1. To this end, the corresponding cDNAs were cloned, expressed in E. coli, and employed in in vitro phosphorylation assays using the TPK1 N-terminus as a substrate. Among the CDPKs tested, we found CPK3, CPK4, CPK5, and CPK11 capable of efficiently phosphorylating the purified, recombinant TPK1 N-terminal peptide in a calcium-dependent manner (Figure 1). Compared with CPK3, TPK1 phosphorylation by CPKs 4 and 11 was similarly effective, while CPK5 exhibited slightly lower efficiency. These results were obtained with at least three independent experiments and were corroborated in Western blots using a commercially available phosphorylation state-specific 14–3–3-binding site antibody (Supplemental Figure 2). In contrast, CPK12 did not phosphorylate the TPK1 peptide under conditions tested in this study (Supplemental Figure 3). Hence, CPK3, CPK4, CPK5, and CPK11 were assigned as candidate kinases responsible for the phosphorylation of TPK1 in Arabidopsis.

Figure 1.

In Vitro Kinase Assay with Purified Recombinant CDPKs. Comparison of the phosphorylation activity of recombinant CPK3, CPK4, CPK5, and CPK11 towards the TPK1 N-terminus. To test for calcium dependency, Ca²⁺ was added to 0.1 mM (+), or was replaced by 2 mM EGTA in samples without Ca²⁺ (−).

(A) Autoradiography showing auto-phosphorylation of the CDPKs in the upper area (as indicated by the arrows).

(B) Coomassie-staining of the same gel to demonstrate the similar loading of proteins. Quantification of TPK1 phosphorylation was normalized to protein amounts of the different kinases and the substrate (see Coomassie staining in Figure 2B) as well as specific activities of the kinases (tested with Histone as generic substrate), deduced from the level of calcium-dependent auto-phosphorylation (upper bands in (A)). *CPK3 (59 kDa), CPK4 (56 kDa), CPK5 (62 kDa), CPK11 (56 kA); ^# N-terminal GST-fusion of the first 81 amino acids of TPK1. Note that the band shift, which is visible for the CDPKs, is due to the presence of EGTA in these samples, a well-known phenomenon for calcium-binding proteins (Kameshita and Fujisawa, 1997).

CPK3 Is Constitutively Co-Expressed with TPK1 and Activated by Calcium

Among the CDPKs investigated in this study, Arabidopsis CPK3 is well characterized in terms of biotic or abiotic stress responses (Mori et al., 2006; Kanchiswamy et al., 2010). Recently, CPK3 has also been shown to play an important role in salt-stress acclimation in Arabidopsis and a number of membrane proteins have been identified as potential CPK3 targets (Mehlmer et al., 2010). To test for the presence of CPK3 in TPK1-expressing tissues, we analyzed the transcriptional regulation of TPK1 and CPK3 in response to salt stress. In line with transcriptome data available (Kreps et al., 2002;Hruz et al., 2008), however, TPK1 and CPK3 under the given experimental settings appeared not to be transcriptionally regulated in 10-day-old seedlings of wild-type plants (Col0) as well as cpk3 knockout mutants (Figure 2A). In contrast, CPK29, which was barely detectable under control conditions, was strongly induced within 30 min after application of salt stress (Figure 2A). This behavior remained unchanged in the cpk3 knockout line, indicating that CPK3 is not affecting transcriptional regulation of CPK29, but may rather function as posttranslational regulator of TPK1 (together with CPK4 and CPK11) in the early stress response. Thus, on the basis of its proposed physiological role in stress responses and overlapping expression patterns for further in-depth analysis of TPK1 regulation, in the following, we focused on CPK3.

Figure 2.

CPK3 Expression and Activation by Ca²⁺.

(A) Analysis of tpk1, cpk3, and cpk29 expression by semi-quantitative RT-PCR in response to salt stress in 10-day-old seedlings of wild-type (Col0) or cpk3 knockout plants. Seedlings were grown on ½ Murashige–Skoog medium for 10 d and treated with 150 mM NaCl for the times indicated. RT–PCR analysis was performed as described in the Methods section. mRNA levels of Actin (act3) served as loading control. (B) Ca²⁺-dependent phosphorylation of the TPK1 N-terminus by the calcium-dependent kinases CPK3 and CPK29. Kinase activity could be detected in the range of 2–160 μM free calcium for CPK3 and between 0.3 and 160 μM free calcium for CPK29. Both kinases exhibit a maximal phosphorylation activity at approximately 17 μM free calcium.

All CDPKs except for CPK25 possess four calcium-binding EF-hands located in their C-terminus (Cheng et al., 2002). In line with these functional domains, calcium overlay assays using recombinant kinases CPK3 and CPK29 demonstrated efficient calcium-binding down to about 2 μM free calcium (Supplemental Figure 4). We thus went on determining the Ca²⁺ dependency of kinase activities towards the TPK1 N-terminal substrate (Figure 2B). Based on in vitro phosphorylation assays, maximal enzyme activity around concentrations of 17 μM free Ca²⁺ was monitored with both kinases. However, the phosphorylation of TPK1 was already detectable at concentrations as low as 2 μM and 0.3 μM free calcium for CPK3 and CPK29, respectively (Figure 2B). Using optimal assay conditions, we now quantified exemplarily CPK3-TPK1 interaction by means of two different methods. First, the Michaelis–Menten binding constant (K_m) and the turnover number (k_cat) were calculated from in vitro kinase assays using a set of six different GST–TPK1 substrate concentrations (Figure 3A and 3B). Thereby, a K_m value of about 4 μM and a turnover number k_cat of about 9 × 10⁻³ s⁻¹ was determined, which well corresponds to the wide range of K_m values reported for kinase-substrate reactions (Songyang et al., 1996). Since the Michaelis–Menten constant is only an estimate for the affinity between the kinase and its substrate under given conditions, we performed additional Microscale Thermophoresis (MST) analyses to determine in solution the affinity for interaction between CPK3 and the TPK1 substrate (Wienken et al., 2010). Based on experiments with 15 substrate concentrations of highly purified GST–TPK1 in the presence of recombinant CPK3 thermophoresis, the data allowed calculation of an apparent affinity of CPK3 to the TPK1 N-terminus of 30 ± 0.9 μM (Figure 3C and 3D).

Figure 3.

In Vitro Interaction Analysis.

(A) Quantitative analysis of the Michaelis–Menten kinetics of the CPK3–TPK1 interaction reveals a K_M of 3.9 ± 0.6 μM or 4.9 μM (according to analysis using the Hanes plot (B). The substrate conversion rate v_max is calculated to 2.7 ± 0.2 × 10⁻³ μM s⁻¹ (for [E_total] = 0.3 μM) providing a turnover number k_cat of 9 × 10⁻³ s⁻¹.

(C) Binding curve as determined from interaction analysis of CPK3 and TPK1 by Microscale Thermophoresis. Analysis of the concentration-dependent thermophoresis and temperature jump of the CPK3 GST–TPK1 interaction yields an affinity of 30 ± 0.9 μM. The normalized fluorescence is plotted for different concentrations of GST–TPK1 (substrate) in the presence of a constant concentration of CPK3 (35 nM), the latter of which was labeled with the fluorescence dye NT-647 via amino-coupling.

(D) The thermophoretic depletion was measured via the fluorescence of the dye-labeled CPK3 kinase in a spot heated by an IR laser. At time point t = 5 s, the IR laser was switched on, the fluorescence decreased as the temperature increases; at time point t = 20 s, the IR laser was switched off again. The diffusion of the dye-labeled CPK3 (in the presence of varying concentrations of its substrate GST–TPK1) was analyzed, providing the affinity of the kinase–substrate interaction.

Ser-42 of the 14–3–3 Binding Motif Is the Sole N-Terminal Target of CPK3 in TPK1

Having characterized CPK3 as a possible TPK1 phosphorylating kinase, we tested whether the 14–3–3 binding motif is phosphorylated specifically. Taking advantage of an antibody which recognizes site-specific phosphorylated 14–3–3 binding motifs, we performed in vitro kinase assays using purified recombinant GST–TPK1 N-terminus and CPK3 (Figure 4A). As expected, the phospho-14–3–3 specific antibody recognized TPK1 in the presence of active CPK3 only (Figure 4A, middle panel). As only trace amounts of CPK3 were used in these assays (see Coomassie stain Figure 4A, left panel), the kinase was detected using a specific anti-CPK3 antibody (Figure 4A, right panel). These results provided evidence that phosphorylation of TPK1 by CPK3 is addressing the 14–3–3 binding domain. In order to identify the residue(s) targeted by the kinase, we performed luminescence-based in vitro kinase assays. As a substrate, we used either recombinant TPK1 N-terminus or a chemically synthesized peptide (encompassing the 14–3–3 binding motif) in the presence of Ca²⁺ and Mg²⁺–ATP (Figure 4B). Please note that this assay is based on quantifying ATP consumption, and is thus detecting total kinase activity in the assay. In the absence of the TPK1 substrate, kinase activity was already visible due to the well-known CPK3 autophosphorylation capacity (Figure 4B, CPK3). In the presence of the TPK1 N-terminus, however, CPK3-dependent phosphorylation increased about seven-fold (Figure 4B, TPK1). A point mutation of Ser-42 to alanine within the 14–3–3 binding motif of TPK1 (Figure 4B, TPK1 S42A) reduced the kinase activity down to the CPK3 autophosphorylation level. Similar results were obtained using the corresponding chemically synthesized peptides of TPK1 (Figure 4B, TPK1pep and TPK1pep-S42A). Our observation that signal intensities obtained with the TPK1–S42A mutant compare to those of CPK3 alone indicates that Ser-42 in the 14–3–3 binding motif in the cytosolic TPK1 N-terminus very likely represents the sole phosphorylation site addressed by CPK3.

Figure 4.

Ser-42 Represents the N-Terminal Target of CPK3 in the TPK1 Channel.

(A) Detection of 14–3–3 binding motif phosphorylation upon in vitro phosphorylation of recombinant TPK1 N-terminus without (−) and with (+) recombinant CPK3 in the presence of Ca²⁺. Kinase assays were performed as described for Figure 1. The specific detection of CPK3 protein and phosphorylation of the 14–3–3 binding motif in TPK1 was conducted by Western blot using the specific antibodies described in the Methods section.

(B) Luminescence-based kinase assay performed with 20 μg TPK1 N-terminal protein or peptide (pep): wild-type (TPK1), S42A mutant (TPK1 S42A). TPK1 phosphorylation was performed in 10 μM Mg²⁺–ATP, 1 μM Ca²⁺, using 0.5 μg CPK3 in 100 mM HEPES–KOH (pH 7.8). Note that CPK3 plus TPK1–S42A mutant protein/peptides exhibit phosphorylation signals comparable to CPK3 alone. Relative kinase units denote the amount of ATP consumed in a given reaction normalized to ATP-containing buffer without kinase added. Bars mark the mean of three independent measurements with the errors showing standard error of the mean.

CPK3 Co-Localizes and Specifically Interacts with TPK1 In Vivo

Following biochemical characterization of TPK1–CPK3 interaction, we tested the physiological relevance of this channel–kinase pair by means of in vivo studies. At first, we examined the subcellular distribution of the channel and the kinase. Following transient cotransformation of Allium cepa epidermal cells by particle bombardment with TPK1::GFP, fluorescence was emitted by the tonoplast (Figure 5A). In contrast, fluorescence originating from CPK3::RFP-fusion constructs was predominately associated with the cytoplasm (Figure 5A). Co-localization studies based on a GFP/RFP overlay, however, showed that a significant fraction of CPK3 was located at the tonoplast though (Figure 5A, lower panel). Considering that only a transient interaction would be required for phosphorylation, it seems likely that in vivo CPK3 phosphorylates TPK1 in a fashion similar to the in vitro assays (Figure 1). Yet, it is well known that N-terminal myristoylation of CPK3 leads to a partial association of this calcium-dependent kinase with the plasma membrane (Stael et al., 2011). In order to prove a specific CPK3-TPK1 interaction in vivo, we employed bimolecular fluorescence complementation (BiFC, SPLIT-YFP; Gehl et al., 2009). We co-expressed the N-terminal portion of CFP (nCFP) fused to the N-terminus of CPK3 and the C-terminal portion of CFP (cCFP) fused to the N-terminus of TPK1. As a result, in transiently transformed tobacco epidermal leaf cells (48 h following infiltration), we observed reconstitution of the CFP fluorescence signal (Figure 5B). In contrast—despite similar expression levels (see Figure 5B, lower panel)—CPK3 did not interact with KAB1 (Figure 5C), a putative potassium channel beta subunit, previously identified as a potential CPK3 target (Mehlmer et al., 2010). The in vivo TPK1/CPK3 association initially identified with transiently transformed tobacco cells was confirmed by related studies with Arabidopsis mesophyll protoplasts (Supplemental Figure 5). Following gentle osmotic lysis of protoplasts and release of intact vacuoles, the interaction between kinase and channel could be associated with the tonoplast. Similar TPK1 interaction was obtained with CKP5 and CPK29 (Supplemental Figure 5). However, we never observed an interaction of TPK1 with the unrelated, calcium-independent kinase OST1 (open stomata 1). Taken together, our data point towards specific association in vivo between a given calcium-dependent kinase and TPK1.

Figure 5.

CPK3 Co-Localization and Interaction with TPK1 In Vivo.

(A) Subcellular localization studies of TPK1::GFP (green) in the vacuolar membrane and CPK3 as RFP-fusion protein (red) in the cytoplasm of onion epidermal cells. Co-localization of both proteins is indicated by the yellow sphere close to the vacuolar membrane in the overlays. The lower left images in both panels represent bright field images while lower right images represent a merge of TPK1–GFP and CKP–RFP confocal images. Bimolecular fluorescence complementation of N-terminal fusions of CPK3 and TPK1 (upper panel) or KAB1 (lower panel) with N- and C-terminal parts of CFP, respectively, in tobacco epidermal leaf cells 2 d after infiltration.

(C) Western blot analysis of expression levels shows comparable signal intensities of TPK1 and KAB1 under the different experimental conditions employed to quantify BiFC efficiency.

Two Arginine Residues in the TPK1 Amino Terminus Stabilize Interaction with CPK3

The conserved structure among CDPKs exhibiting 44–95% amino acid sequence identity in their kinase domains (Supplemental Figure 5A; Cheng et al., 2002), combined with their ubiquitous expression throughout the plant’s organs and developmental stages, gives rise to the question of how specificity of downstream signaling is generated (Ludwig et al., 2004). To gain a more mechanistic understanding of the channel–kinase interaction, we generated a structural model for the Arabidopsis CPK3 kinase domain and the TPK1 N-terminus and modeled their molecular interaction. Taking advantage of a recently solved CDPK structure (Wernimont et al., 2010), we initially performed homology modeling of the CPK3 kinase domain by using the crystal structure of Toxoplasma gondii CDPK1 (PDB entry 3HX4) as a template. Based on this structure, docking and modeling of the CPK3 complex with the N-terminal fragment of TPK1 bound in its substrate-binding pocket were performed (Supplemental Figure 6A and Figure 6A-6B). To explore whether the observed kinase specificity for TPK1 could be explained on a structural basis, we first performed a multiple sequence alignment for the kinase domain of the CDPKs investigated (Supplemental Figure 7A). Analysis of the level of conservation of the amino acid residues in the kinase domain with the software Consurf (Ashkenazy et al., 2010) revealed a high level of conservation of surface-accessible residues close to the active site around Asp202 of CPK3 (Figure 7A). Asp 202 transfers the γ-phosphate of the kinase-bound ATP to the substrate (Figure 7D). Moreover, our model suggests that the substrate-binding cleft extends towards the C-termini of helices α2 and α4. Remarkably, this binding epitope structurally resembles a known complex of cAMP-dependent kinase bound to its inhibitor PKI5-24 (Narayana et al., 1997). We therefore superimposed our CPK3 model onto this structure using the inhibitor PKI5-24 as a template to model the TPK1 substrate (Figure 7A-7C). Secondary structure prediction performed with the complete TPK1 N-terminus showed that the energetically most favorable structure between Ser30 to Ser40 represents an α-helical structure (Supplemental Figure 7B). Following manual alignment and replacement of the PKI5-24 peptide residues by Asn26 to Arg45 of TPK1, side chain-rotamer searches and several steps of energy refinement ensured that the substrate interacts tightly with the kinase domain and bad van-der-Waals contacts were absent from our final model of the 14–3–3 binding site. In this model, residues Asn26 to Arg33 appeared as short α-helix similar to the PKI peptide, whereas residues Arg38 to Arg45 formed an extended strand-like structure, fitting perfectly in the substrate-binding cleft of the kinase (Figure 7A-7D). Ser42, which is known to be the site of phosphorylation in TPK1, is found in close proximity of the active site residue Asp202 of CPK3 in our model. The C-terminal end of the α-helix of the TPK1 substrate contains multiple positively charged amino acids, mainly arginines, which possibly interact via ionic and polar bonds with the highly conserved acidic residues Glu163 and Asp166 of the helix α2 and the substrate-binding loop in the kinase domain (Figure 7C and 7D).

Figure 6.

Salt Triggers Phosphorylation of TPK1 and Phenotype of Mutants.

(A) Western blot analysis of TPK1 phosphorylation derived from plants treated with cold, heat, 300 mM sorbitol, or 150 mM NaCl for 2 h.

(B) Western blot analysis of TPK1 phosphorylation using antibody anti-BAD 136 in wild-type (WT) plants and tpk1-1 knockout plants under control conditions (basal mineral solution = MS+1 mM KCl) or treated with MS + 150 mM KCl, MS + 150 mM NaCl, or MS + 300 mM sorbitol as indicated.

(C) Germination rates of Arabidopsis seedlings under different conditions of salt stress. Plants were grown on 0.8 % agarose plates containing mineral solution (MS, see the Methods section), in the presence of 150 mM NaCl/1 mM K⁺ and 150 mM NaCl/50 μM K⁺ to mimic reduced supply of K⁺. Germination was scored at day 3 after dissemination and normalized to germination rates obtained in basal medium supplemented with 1 mM KCl. Data represent the mean of n = 3 ± SE.

(D) Expression levels of CPK3 and TPK1 in the used plant lines. Single asterisks denote differences between Col0 and mutants at the 0.01 significance level and double asterisks at the 0.001 significance level.

Figure 7.

Molecular Modeling of CPK3–TPK1 Interaction and CPK3 Phosphorylation Sites.

(A) Surface representation of the CPK3 with the substrate peptide region of TPK1 (indicated in green) bound into the putative substrate-binding cleft. The surface of CPK3 is color-coded on the basis of amino acid sequence conservation as analyzed by the Consurf software (http://consurf.tau.ac.il). Invariant regions are colored in dark red, decreasing sequence conservation is then marked in lighter shades of red, gray areas indicate regions that do not show sequence conservation above average, whereas blue regions mark areas where amino acid sequence is increasingly variable between different CDPKs.

(B) Surface representation of a model of CPK3 with TPK1 N-terminus (carbon atoms marked in cyan) bound color-coded based on amino acid polarity.

(C) Ribbon representation of the kinase–substrate complex of CPK3 with the N-terminus of TPK1 (residues Asn26 to Arg45). The upper lobe is colored in magenta; the lower lobe of the kinase is shown in green. The ATP molecule and the magnesium ions required for phosphate transfer are shown as sticks and spheres, respectively. The TPK1 substrate peptide contains numerous arginine residues suggesting an interaction with charge-complementary residues in the CPK3 kinase. Conserved aspartate and glutamate residues of CPK kinases are shown as sticks.

(D) Ribbon representation of a close-up view of the CPK3 catalytic site and the phosphorylated Ser42 in the TPK1 peptide. The autophosphorylation site Ser242 of CPK3 is marked by an arrow. Potential hydrogen bonds between the substrate TPK1 and the kinase CPK3 are indicated by black stippled lines.

(E) Scheme of the 14–3–3 binding sequence in the TPK1 N-terminus highlighting the mutagenized arginine residues.

(F) Quantification of BiFC interaction signals of nCFP–CPK3 with wild-type cCFP-TPK1 and the mutated form obtained from infiltrated tobacco leaves.

(G) Expression levels of infiltrated CPK3 and TPK1 detected by Western blot.

Finally, we tested the proposed model for CPK3–TPK1 interaction experimentally. Therefore, we performed mutant-based quantitative BiFC assays in transiently transformed tobacco leaves, testing the predicted mode of ionic interactions between the conserved acidic residues in the active site of CPK3 with the basic residues in the 14–3–3 binding motif in the N-terminus of TPK1. For this purpose, TPK1 point mutants in which arginine 39 within the 14–3–3 motif and the preceding arginine 35 were mutated to alanine (see Figure 7D) and co-expressed with CPK3. Quantification of the BiFC signals arising from the different combinations was achieved by co-infiltration of CPK3 with wild-type and mutated TPK1 constructs in different sectors of the same leaf (Figure 7E and 7F; Bhat et al., 2006). Thereby, we found that the double mutant TPK1–R35A/R39A reproducibly showed about half of the CPK3 interaction signal intensities compared to the wild-type potassium channel (Figure 7E). Given that both mutant and wild-type channel express to the same amount (Figure 7F), these studies point towards a critical role of both arginine residues in the channel–kinase interaction. We also tested whether kinase autophosphorylation would feedback on CPK3–TPK1 interaction and substrate phosphorylation. As discussed below, however, and shown in Supplemental Figure 8, mutation of the identified residue Ser242 did not affect TPK1 phosphorylation by CPK3 (see also Supplemental Figures 6B and 9).

Salt Stress Induces Phosphorylation of TPK1 In Planta

In plants facing salt stress, the SOS2/3 system is known to regulate the plasma membrane and tonoplast Na⁺/H⁺ exchangers SOS1 and NHX1 in a calcium-dependent manner (Qiu et al., 2004). In line with the SOS model, calcium levels rise in response to salt stress (Knight et al., 1997), activating the sodium exchangers and allowing plants to cope with sodium chloride loads. To test whether phosphorylation of TPK1 is also triggered by salt stress, we applied 150 mM sodium chloride to 4-week-old Arabidopsis plants and subsequently analyzed the phosphorylation status of TPK1. Phosphorylation status was probed using a phosphorylation-state-specific antibody, which recognizes the N-terminal 14–3–3 binding motif in TPK1 (see also Figure 4). This commercially available antibody was originally selected against phosphorylated Ser¹³⁶ in mammalian BAD protein; however, the phosphorylated binding epitope is identical to that of TPK1. With plants challenged for 8 h with various stresses, phosphorylation of TPK1 was analyzed with Western blots of protein extracts from leaves (Figure 6A). Treatments with sorbitol, cold, or heat stress, known to elicit calcium responses as well, resulted in only weak blot signals. Pronounced TPK1 phosphorylation, however, was detected in response to sodium chloride exposure. To reassess the Western blot data, we tested TPK1 antibody specificity in vivo using hydroponically cultured tpk1-3 knockout plants. The finding that a salt-stress-induced phosphorylation signal appeared only in the blot analysis of control plants but was absent from Western analyses of tpk1-1 mutant plants demonstrated that the antibody detected in vivo phosphorylated TPK1 specifically (Figure 6B). Notably, in these experiments, signal strength was higher with sodium chloride compared to potassium chloride stress.

Salt Stress Impairs Germination of tpk1 and cpk3 Knockout Plants

To test the assumed role of TPK1 and CPK3 in salt tolerance, we characterized the germination efficiency in response to sodium (salt) and potassium (starvation) stress of tpk1 and cpk3 knockout mutants as well as of overexpressor lines in comparison to wild-type plants (Figure 6C). T-DNA insertion lines that have been described in detail by Mehlmer et al. (2010) for the cpk3 mutant and Gobert et al. (2007) for the tpk1 mutants were obtained from the Salk collection (http//signal.salk.edu; Alonso et al., 2003). In addition, all plant lines used were tested for cpk3 and tpk1 expression by semi-quantitative RT–PCR using actin (ACT3) as control (Figure 6D). When Arabidopsis seedlings were germinated on basal medium without additional sodium chloride, all lines displayed a germination efficiency of 100% (data not shown). In contrast, when facing 150 mM NaCl, the cpk3 knockout line (cpk3-2) and the knockdown line (cpk3-1) were clearly impaired in their germination efficiency. A CPK3-overexpressing line (cpk3-3) exhibiting elevated CPK3 protein levels due to promoter-based T-DNA insertion (Mehlmer et al., 2010) displayed strongly enhanced germination rates on high-salt medium (Figure 6C, middle panel). Likewise, the tpk1 knockout line tpk1-3 was impaired in germination on plates containing 150 mM NaCl. Two TPK1 overexpression lines TKP1 ox3 and TPK1 ox8, however, displayed increased germination rates on high-salt medium (Figure 6C, middle panel). The second T-DNA insertion line tpk1-1 appeared not to be affected in germination on such sodium chloride-supplemented medium. This at-first-sight unexpected result could be explained by the fact that the latter mutant was identified as a partial knockout (Figure 6D). However, under conditions combining sodium stress with potassium deficiency (only 50 μM K⁺ present in the medium), the effect of CPK3 overexpression (cpk3-3) was stronger and the importance of TPK1 became much more obvious. Now, under combined stress conditions, both tpk1 lines, the tpk1-3 knockout line as well as the tpk1-1 knockdown line, appeared severely impaired in germination. In contrast, the weaker TPK1-overexpressing line TPK1 ox8 exhibited comparable germination efficiency to the wild-type, but the stronger overexpressing line (TPK1 ox3) showed significantly improved germination efficiency compared to wild-type (Figure 6C, right panel). In summary, our data indicate that a loss-of-function mutation in TPK1 as well as in CPK3 renders seedlings more sensitive to salt stress, supporting our initial observation that different CDPKs are able to phosphorylate TPK1 in planta. Accordingly, a knockout of only a single CDPK (i.e. CPK3) has only little effect in planta, whereas kinase overexpression increases salt tolerance, even more so, when salt stress meets potassium starvation.

DISCUSSION

Calcium-Dependent Kinases: Target Specificity by Calcium Affinity

Changes in cytosolic calcium levels are among the first plant cell-derived signals detectable in response to a variety of biotic and abiotic stresses (Knight et al., 1997; Harper and Harmon, 2005; Hepler, 2005; Garcia-Brugger et al., 2006;Oldroyd and Downie, 2006; Mahajan et al., 2008; Dodd et al., 2010). Plants have evolved a unique set of calcium-dependent kinases such as CIPKs and CDPKs, allowing the decoding of these calcium signals (Harper et al., 2004; Batistic and Kudla, 2008; Luan, 2009). Members of the CBL/CIPK proteins have been shown to comprise essential components of the SOS pathway regulating salinity tolerance as well as adaptation to potassium or nitrate depletion in Arabidopsis (Hedrich and Kudla, 2006; Mahajan et al., 2008, and references therein;Tsay et al., 2011). Likewise, CDPKs, directly binding calcium via EF-hands, have been associated with plant stress physiology and ABA-mediated signaling pathways (Ludwig et al., 2004; Mori et al., 2006; Wan et al., 2007; Tena et al., 2011). In context of the latter, guard cell plasma membrane anion channels were recently shown to represent targets of CDPKs (Geiger et al., 2010, 2011). Our results here show that the ubiquitously expressed vacuolar potassium channel TPK1 represents a novel downstream target of distinct calcium-dependent kinases addressed during salt-stress-evoked calcium signaling. The A. thaliana kinases CPK3, CPK4, CPK5, CPK11, and CPK29 were shown to phosphorylate AtTPK1 in vitro (see Figures 1 and 2). In accordance with DNA array analyses, CPK3, just like TPK1, was found widely expressed but unaffected by salt stress (Figure 2A; Kreps et al., 2002;Hruz et al., 2008). This suggests that CPK3 as a potential regulator of TPK1 is already present when plants experience stress onset. In contrast to CPK3 and in line with a possible role in long-term acclimation to salt stress, CPK29 transcripts are rare under pre-stress conditions but increase strongly with salt treatment. Both kinases require calcium ions for efficient TPK1 phosphorylation and—according to calcium overlay assays—they exhibited comparable Ca²⁺ binding affinities. Kinase activity of CPK3, however, mandated micromolar calcium concentrations for substrate phosphorylation while CPK29 showed clear TPK1 phosphorylation already at sub-micromolar concentrations of Ca²⁺ (see Figure 2B). This could indicate that the immediate first response to salt stress involving CPK3 would require higher Ca²⁺ levels, and later stages following an increase in CPK29 transcript and protein kinase number could take place already at lower Ca²⁺, allowing pre-stressed plants covering a wide dynamic range of Ca²⁺ signals. A behavior similar to the CPK3/29 pair was shown for the CPK21/23 in activating the guard cell anion channel SLAC1 (Geiger et al., 2010; Stange et al., 2010). CPK3 is reminiscent of CPK23, a kinase showing a Ca²⁺-independent core activity and a Ca²⁺-inducible component with low Ca²⁺ affinity (Geiger et al., 2010). Recent studies could show that different calcium affinities are indeed crucial for full channel activation (Scherzer et al., 2012). On the basis of EF-hand mutants, it turned out that previously ineffective kinases converted into potent activators of the SLAC1 anion channel. It thus seems that the interaction spectrum of calcium-dependent kinases with their targets is determined by changing cytosolic calcium concentrations and calcium affinities as well as co-expression in individual tissues and cell types.

Besides calcium affinities and calcium dependency of target phosphorylation, little is known about the affinity of CDPKs towards their substrates. Our analysis of TPK1 interaction with CPK3 revealed an apparent affinity of about 5–30 μM for the kinase towards the vacuolar channel (see Figure 3C and 3D). This value is reminiscent of the often observed low affinity of other kinases for their protein substrates and is considerably lower compared to the previously reported tight interaction of the vacuolar potassium channel with the 14–3–3 protein GRF6 (Latz et al., 2007). The fact that distinct CDPKs can phosphorylate TPK1 while others cannot suggests that, in spite of low substrate affinity, structural constraints determine substrate specificity. Here, our modeling approach predicted that ionic and polar bonds between positively charged amino acids of the channel substrate and negatively charged residues in the binding loop of the kinase domain would stabilize the TPK1–CPK3 interaction. Indeed, mutations in two conserved arginine residues within the 14–3–3 binding domain of TPK1 significantly impaired interaction with the kinase, providing evidence for the validity of our model. While the model allows insight into the structural basis of the observed CPK3–TPK1 interaction, substrate specificity of different CDPKs towards TPK1 as shown above cannot be extracted from modeling data. We thus wondered whether determinants for substrate specificity could be located in sites/domains not included in the model, namely calcium affinity in the calcium-binding domains or, as reported previously, sites susceptible for kinase autophosphorylation (Hegeman et al., 2006). Several phosphorylation sites are predicted for CPK3 (Supplemental Figure 6A) and we could detect in vitro autophosphorylation of CPK3 at multiple sites using purified recombinant kinase. Six different protein spots appeared after 2-D separation by isoelectric focusing (IEF) and subsequent SDS–PAGE (Supplemental Figure 6B). Further analysis of excised spots by tandem mass spectrometric analysis revealed that they indeed correspond to differently phosphorylated forms of CPK3 (Supplemental Figure 6B). After phosphopeptide enrichment by IMAC chromatography and SDS–PAGE, we were able to identify a phosphorylation of Ser242 by its neutral loss of phosphoric acid in the indicated peptide fragmentation pattern (Supplemental Figure 6B). Although Serine 242 is located in a highly conserved stretch in very close proximity to the active site Asp202 (indicated by the arrow in Figure 7D) and is invariant in all CDPKs, Ser²⁴² mutants to either alanine or aspartate did not abolish phosphorylation of TPK1-NT by CPK3 (Supplemental Figure 9), suggesting that Ser242 phosphorylation is neither essential for kinase activation nor required for substrate recognition and binding.

Based on the low affinity of CPK3 towards TPK1 and the fact that, due to its N-terminal myristoylation, CPK3 associates with both the plasma membrane and the vacuolar membrane, TPK1 most likely does not represent the sole membrane protein target of this kinase. Accordingly, studies by Mori et al. (2006) with a cpk3/cpk6 double mutant in response to ABA showed impaired regulation of guard cells’ plasma membrane slow-type anion channel together with Ca²⁺-permeable channels. Moreover, a recent phosphoproteome study identified several vacuolar membrane proteins as CPK3 targets (Mehlmer et al., 2010). It is thus tempting to speculate that decoding of stress-related calcium signals via members of the CDPK family coordinates ion fluxes across the plasma membrane and tonoplast. This view somehow resembles the situation of CBL interacting protein kinases of the CIPK family in activating targets located in different subcellular compartments (Kudla et al., 2010). In this context, recent studies have shown that, on the basis of lipid modification, the amino terminus of CBL3 entirely defines a tonoplast targeting domain sufficient to direct this calcium sensor to the vacuolar membrane (Batistič et al., 2010). Moreover, vacuolar CBL2 targeting requires S-acylation of three cysteine residues in its N-terminus (Batistic et al., 2012). On the other hand, plasma membrane targeting might result from specific complex formation between kinases and targets. The latter seems to hold true for TPK1-mediated localization of CPK3 to the vacuolar membrane (see Figure 5A), for example, which, together with lipid modifications, allows for cellular control ion of homeostasis and water status (Figure 8; Voelker et al., 2010; Munemasa et al., 2011).

Figure 8.

Schematic View of TPK1-Mediated K⁺-Homeostasis during Salt Stress.

Upon salt stress, sodium ions may enter the cytoplasm of root cells via non-selective cation channels (NSC), for example, such as HKT1 or cyclic nucleotide-gated (CNG) channels (Corratge-Faillie et al., 2010). Proton-coupled antiporters in the plasma membrane or the vacuolar membrane such as SOS1 or NHX, respectively, exclude toxic sodium ions from the cytoplasm. Mechanical forces generated by osmotic stress could provide means for activation of calcium-permeable channels of an as-yet unknown nature. Upon calcium binding, TPK1 is activated immediately upon stress onset to release potassium ions into the cytosol. During prolonged salt stress, phosphorylation of TPK1 by CPK3, for example, could provide the basis for interaction with 14–3–3 proteins and thus sustained potassium efflux in order to maintain high cytosolic K⁺/Na⁺ ratio.

Regulation of Potassium Homeostasis under Salt Stress

Plants respond to sodium-based salt stress by activating proton-driven sodium antiporters in the plasma and vacuolar membrane in order to balance the cytosolic K⁺/Na⁺ ratio (Mahajan et al., 2008). Under salt stress, the vacuole provides a major storage compartment for sodium ions. In addition, the central organelle represents a major calcium reservoir (Conn et al., 2011). The latter is very likely engaged with the generation of salt-stress-induced calcium signals. In this context, the AtTPC1 channel has been discussed in calcium-induced calcium release from the vacuole into the cytoplasm (Ward and Schroeder, 1994; Peiter et al., 2005; Hedrich, 2012). The AtTPC1 gene encodes the non-selective slowly activating vacuolar (SV) cation channel permeable for potassium, sodium, and even calcium ions (Sanders et al., 1995; Ivashikina and Hedrich, 2005; Peiter et al., 2005; Hedrich and Marten, 2011). The fact, however, that tpc1-2 mutants and TPC1-overexpressing plants exhibit normal calcium responses towards various stresses, including salt stress, thus challenges the in vivo role of AtTPC1 in the generation of calcium signals (Ranf et al., 2008). In contrast, the fou2 allele of TPC1 exhibits a K⁺-starvation phenotype. As a matter of fact, this tpc1 mutant builds up threefold higher vacuolar Ca²⁺/K⁺ ratios compared to wild-type plants. This implicates that this mutant might suffer from over-optimal vacuole-derived cytosolic calcium loads in response to wounding for instance and probably also to salt stress (Bonaventure et al., 2007; Beyhl et al., 2009). Still, the AtTPC1 channel might serve as an important pathway for sodium uptake into the vacuole well in line with the proposed role of its animal homologs (Wang et al., 2012). Interestingly, the cytosolic Ca²⁺ affinity of CPK3 is in the same range as that of the SV channel (Scherzer et al. 2012). Elevated sodium or calcium concentrations in the vacuole shift the voltage dependence of AtTPC1 towards positive vacuolar membrane potentials, rendering the contribution of this channel for potassium release and thus potassium homoeostasis under salt stress unlikely (Dadacz-Narloch et al., 2011; Hedrich and Marten, 2011). However, vacuolar potassium (VK) channels, such as TPK1—like other members of its family—resemble voltage-independent so-called ‘open rectifiers’ and, as such, TPK1 provides a selective pathway for potassium fluxes across the vacuolar membrane (Ward and Schroeder, 1994; Pottosin et al., 2003; Becker et al., 2004; Gobert et al., 2007; Hedrich, 2012). This channel possesses EF-hands in its C-terminal end and requires elevated cytosolic calcium concentrations for activation (Hedrich and Neher, 1987; Bihler et al., 2005; Gobert et al., 2007; Latz et al., 2007). Recently, it was shown that, besides calcium, membrane stretch and osmotic gradients can additionally alter TPK activity in many plant species (Maathuis, 2011). Thus, besides osmotic stress, sodium chloride-induced elevation of cytosolic calcium directly activates TPK1, leading to instant release of potassium ions into the cytosol (see Figure 8; Bihler et al., 2005; Latz et al., 2007; Isayenkov et al., 2010). It is conceivable that sustained salt stress would activate CDPK-controlled signaling pathways including phosphorylation of TPK1. Although the amino acid composition in the vicinity of the 14–3–3 binding motif of the TPK1 channel (R-K-R-R-L-R-R-S-R-S₄₂-A-P-R-G-D) points to at least another likely phosphorylation motif, Ser42 appeared to be the only residue targeted by CDPKs (see Figure 4B). Phosphorylation of Ser42 results in interaction of TPK1 with 14–3–3 proteins, such as GRF6. Interaction with GRF6 was shown to potentiate TPK1 activity (Latz et al., 2007). This would result in sustained potassium efflux from the vacuole to the cytosol (Figure 8) and provide means for salt-stress adaptation.

METHODS

Plant Growth Conditions

Plants were grown in soil in a growth chamber using a 16-h/8-h day/night regime, 26/22°C day/night temperature, and a photon flux density of 150 μmol m⁻² s⁻¹. Fully developed young leaves of 3–6-week-old plants were used for protoplast isolation, transformation, protein extraction, and kinase assays.

Germination Assay

Approximately 100 sterilized Arabidopsis seeds were disseminated on 0.7 % agarose plates containing 1 mM KCl plus basal mineral nutrients 2.5 mM NaNO₃, 2.5 mM Ca(NO₃)₂, 2 mM NH₄H₂PO₄, 2 mM MgSO₄, 0.1 mM FeNaEDTA, 25 μM CaCl₂, 25 μM H₃BO₃, 2 μM ZnSO₄, 2 μM MnSO₄, 0.5 μM CuSO₄, 0.2 μM Na₂MoO₄, and 0.01 μM CoCl₂, adjusted to pH 5.7 with NaOH (Hirsch et al., 1998). Potassium chloride in control plates or for K⁺ starvation was adjusted to 1 mM or 50 μM, respectively. To expose seeds and seedlings to conditions of salt stress, the total sodium content was adjusted by adding NaCl as indicated in the figures. Seeds on agar plates were stratified for 2 d and subsequently incubated under the conditions mentioned above. Germination was scored at day 3 following transfer of the plates to standard growth conditions (see above). All seeds were harvested from plants grown in parallel under the same conditions and all experiments were carried out in two independent repetitions, using eight replicates of each line.

RT–PCR

Semi-quantitative RT–PCR was carried out as described in detail (Doczi et al., 2007). Briefly, plant material was mixed with 130 μl RNA extraction buffer (1% SDS, 10 mM Na₂EDTA, 200 mM sodium acetate, pH 5.2), 130 μl phenol (pH 4.0), and 50 μl sea sand and ground. RNA was subsequently extracted with phenol/chloroform/iso-amyl alcohol and digested with RNAse-free DNAse (RQ1 DNAse, Promega). Concentration and purity of RNA were determined at 260-nm and 280-nm wavelength. 2 μg RNA were used for reverse transcription with M-MLV reverse transcriptase (Promega). PCR amplification of target genes was performed using GoTaq DNA polymerase (Promega) and products were separated employing 1.5% agarose gels. Primers used: At4g23650, CPK3-1: 5′-AGA TGT TCG CCG TGA AGT CC-3′ and CPK3-2: 5′-ACG GAT GAT TTA GCA CTT CCG-3′. At1g76040, CPK29-1: 5′-CCT TGA AGG TGA TAG CGG A-3′ and CPK29-2: 5′-GCA AAA GCA TTC GTC ACT GT-3′. At2g37620, ACT3-1: 5′-ATG GTT AAG GCT GGT TTT GC-3′ and ACT3-2: 5′-AGC ACA ATA CCG GTA GTA CG-3′. At5g55630, TPK1-1: CCA GTG GTG TGG TAG ATG CTC TCT and TPK1-2: ACT CTG CAG CTC CAA CAA CTC CAT C.

Protein Expression and Purification

For kinase assays, recombinant CDPKs and the TPK1 N-terminus (amino acids 1–79) were expressed as GST-fusion proteins in E. coli BL21(DE3)pLys or BL21TUNER(DE3) cells. GST-fusion constructs were obtained by cloning the coding sequences for kinases into the vector pGEX–4T3 and the TPK1 N-terminal peptide was cloned into the expression vector pGEX–6P1.

For the Michaelis–Menten analysis of TPK1 phosphorylation, recombinant purified CPK3 was obtained from a GST-fusion protein. A. thaliana CPK3 was cloned between the EcoRI and SalI restriction sites of the expression vector pGEX–6P-1 (GE Healthcare) and the expression construct was transformed into the E. coli strain BL21TUNER(DE3). The transformed bacteria were then grown in LB-medium with ampicillin at 37°C to an optical density of OD₆₀₀ 0.8 at which the cells were cooled down to 18°C. Protein expression was induced by adding 0.1 mM IPTG and expression was continued overnight. Cells were harvested and re-suspended in PBS, supplemented with protease inhibitor cocktail (Calbiochem), and lysed by sonification. After centrifugation (20 000 g, 60 min, 4°C), the recombinant protein present in the supernatant was subjected to column purification at 4°C using glutathione sepharose (Amocol). GST–CPK3 fusion protein was eluted using 10 mM glutathione (reduced); protein-containing fractions were pooled and dialyzed at 4°C against 100 vol. 50 mM Tris-HCl, pH 7.0, 150 mM NaCl, 1 mM DTT. The GST–CPK3 fusion was cleaved for 12 h at 4°C using PreScission protease in a ratio of 15 μg GST-PreScission protease per milligram fusion protein. Final purification of the CPK3 protein was performed by another GST affinity chromatography, which removed uncleaved fusion protein GST as well as the PreScission protease. To achieve sufficient purity, repetitive chromatography runs using Glutathion sepharose 4B (GE Healthcare) were carried out.

Western Blot Analysis and Immunoprecipitation

For extraction of total proteins, Arabidopsis leaves were batch-frozen in liquid N₂ and ground to powder. Leaf powder (1 g) was subsequently re-suspended in 1 ml of RIPA extraction buffer, containing 1% Triton X100, 0.1% SDS, 2 mM EDTA, 10 mM Na₂HPO₄, and 10 mM NaH₂PO₄ (pH 7.2) and a protease inhibitor cocktail (Complete Mini EGTA-free, Roche) according to the manufacturer’s protocol. The same buffer without Triton X100 was used for the preparation of microsomal fractions. The extraction mixture was incubated for 1 h at 4°C. To remove insoluble material, the resulting extracts were centrifuged at 15 000 g for 10 min.

For microsomal fractions, the supernatant was centrifuged at 100 000 g for 1 h and the pellet was re-suspended in RIPA buffer. Proteins (30 μg) were loaded on a 12% acrylamide gel, size-fractionated, and electro-transferred to PVDF membranes (Amersham). PVDF membranes were incubated for 16 h at 4°C in PBST buffer, containing 150 mM NaCl, 0.1% Tween 20, 40 mM Na₂HPO₄, and 8 mM NaH₂PO₄ (pH 7.0) and subsequently incubated with anti-BAD antibody for 1 h at room temperature. Blots were washed in PBST for 10 min and incubated with secondary horseradish peroxidase-conjugated antibodies for 1 h at room temperature. Subsequently, the blots were again washed with PBST four times for 10 min and bound antibodies were detected using SuperSignal West Pico chemiluminescent substrate (Pierce, www.piercenet.com) followed by exposure to X-ray film (Kodak-Biomax Light, Rochester, USA).

For in vitro and in planta analyses of the phosphorylation status of the TPK1 14–3–3 binding motif, a phospho-specific TPK1 antibody that was originally derived for phospho-specific detection of phosphorylated human BAD protein (anti-BAD, Ab-136, Acris, Herford, Germany) and shares the same epitope with the 14–3–3 motif in TPK1 was used (Latz et al., 2007). CPK3 was detected using a peptide antibody directed against the C-terminal 15 amino acids of CPK3 described before (Mehlmer et al., 2010), and specific phosphorylation of the 14–3–3 binding site in the TPK1 N-terminus was detected using a phosphorylation state-specific commercial antibody from Cell Signaling Technology (#9601, www.cellsignal.com) according to the manufacturer’s instructions. HA epitopes and Flag epitopes were detected using a commercial HA antibody (Roche Cat. No. 11867423001) and a commercial FLAG antibody (Sigma Life Science F3165–0.2MG) according to the manufacturer’s instructions, respectively. For detection of the secondary antibody, the SuperSignal WestPico Chemiluminescent Substrate (Pierce) was used. Afterwards, detection membranes were stained with Coomassie brilliant blue R-250.

Kinase Assay and Michaelis–Menten Analysis

20 μg purified recombinant GST::TPK1 N-terminus was incubated with 10 μM Mg²⁺–ATP, 0.5 mM DTT, 0.5 μg purified kinase (CDPKs) and different amounts of CaCl₂ in 50 mM HEPES (pH 7.8) for 0.5 h at 30°C. Phosphorylation of TPK1 was determined by Western blot analysis as described above. The Kinase-Glo™ Luminescence Kinase Assay (Promega) was carried out according to the manufacture’s protocol. For autoradiography, 6 μg of GST-tagged TPK1 N-terminus were incubated for 30 min at room temperature with 1 μg of the respective kinase in kinase buffer (40 mM HEPES, pH 7.8, 10 mM MgCl₂, 5 mM EGTA, 10 μM ATP) containing 1 μCi γ-P³²–ATP and optionally 10 mM CaCl₂. SDS-loading buffer was added to stop the reaction and the samples were loaded to an 8–16% pre-casted gradient gel (Thermo Scientific). After Coomassie-staining and de-staining, incorporated radioactivity was detected using a phospho-imager.

For the Michaelis–Menten analysis, samples of 10 μl with different concentrations of highly purified GST–TPK1 NT protein (range 0.4–13.8 μM) were mixed with a fixed amount of purified CPK3 (0.3 μM) in kinase buffer (40 mM HEPES, pH 7.4, 10 mM MgCl₂, 10 mM CaCl₂, 5 mM EGTA). The kinase reaction was started by addition of ATP (final concentration 8.5 μM unlabeled ATP and 1 μCi [γ-³²P]ATP). To determine reaction velocities, samples were incubated at 21°C for 4, 8, 16, and 32 min. The samples were then transferred onto Whatman P81 ion exchange papers and the reaction was stopped by transferring the paper spots into 50 ml 75 mM H₃PO₄; excess of ATP was removed by washing the spots five times. The filter paper spots were subsequently washed with acetone and dried. Phosphotransfer was determined in cpm by measuring the ³²P-Cerenkov counting using a MicroBeta JET analyzer (PerkinElmer). For quantification of the transferred, phosphate-defined amounts [γ-³²P]ATP were measured; to subtract the autophosphorylation effects of CPK3, samples with and without substrate were analyzed. All experiments were performed in triplicate.

Microscale Thermophoresis

Binding interactions between CPK3 and GST-TPK1 were measured using MST as described (Wienken et al., 2010). Briefly, recombinant purified CPK3 protein dissolved in PBS was mixed with labeling buffer (provided in the Monolith NT Protein labeling kit) to yield a 20-μM protein solution buffered at a neutral pH with amine-free buffer chemicals. 250 μl of the protein solution were then mixed with 250 μl of a 40-μM dye solution (NT-647); the NHS-activated dye was incubated with the CPK3 protein for 60 min at room temperature. Un-reacted free dye was subsequently removed by size exclusion chromatography using PBS buffer according to the manufacturer’s recommendation. For the measurement, samples were prepared consisting of 35 nM NT-647-labeled CPK3 protein and varying concentrations of GST-TPK1. For data acquisition, a buffer consisting of 30 mM Tris-HCl, pH 7.8, 100 mM NaCl, 2.5 mM CaCl₂, 2.5 mM MgCl₂, 3 mM TCEP, 5 mM glutathione (reduced), 1 mM ATP, 0.1% (w/v) BSA, and 0.05% (v/v) Triton X-100 was used throughout. The samples were filled into Monolith NT hydrophilic capillaries and a titration series comprising 15 substrate concentrations (120, 100, 85, 70, 60, 45, 30, 15, 7.5, 3.75, 1.88, 0.94, 0.47, 0.23, and 0.12 μM) was used for analysis. Parameters for Thermophoresis acquisition were set to LED power 15% (fluorescence excitation), IR laser power 30%, laser on-time 15 s, laser off-time 5 s. The K_D values for the CPK3-TPK1 interaction were obtained by fitting the fraction of bound CPK3 to the quadratic solution of the binding reaction equilibrium, derived from the law of mass action and with K_D as a single free parameter.

Calcium overlay

Recombinant CPK3 and CPK29 were expressed as GST-fusion proteins in E. coli BL21 cells and purified by affinity chromatography. For dot blot analysis, 15 μg of the purified recombinant kinases along with appropriate controls (BSA and calmodulin, Sigma) were spotted onto a PVDF membrane. The membrane was equilibrated in binding buffer containing 10 mM Imidazol/HCl (pH 6.8), 60 mM KCl, and 5 mM MgCl₂ overnight. Blots were incubated for 30 min in fresh buffer containing the indicated calcium concentrations prepared form a stock solution (100 μM CaCl₂ plus 2 μCi/ml ⁴⁵Ca²⁺ (Amersham). Subsequently, the membrane was washed with 70% ethanol, air-dried, and exposed to X-ray film for 1 d. In experiments in which a defined calcium concentration was used, this was determined using the program WEBMAXC (www.stanford.edu/~cpatton/webmaxcE.htm). For quantification, the dot intensity was analyzed using a phospho-imager.