Autophosphorylation Affects Substrate-Binding Affinity of Tobacco Ca²⁺-Dependent Protein Kinase1

Autophosphorylation could prevent the excessive phosphorylation of substrates and alter the substrate preference of Ca²⁺-dependent protein kinases.

Abstract

Protein kinases regulate diverse physiological processes. Because many kinases preserve inherent autophosphorylation capability, autophosphorylation appears to be one of the most important mechanisms for cellular signaling. However, physiological functions of autophosphorylation are still largely unknown, other than the self-activation by phosphorylation of activation loop in the catalytic domain. REPRESSION OF SHOOT GROWTH (RSG) is the transcription factor involved in gibberellin (GA) feedback regulation. The tobacco (Nicotiana tabacum) Ca²⁺-dependent protein kinase, NtCDPK1, phosphorylates RSG, resulting in the negative regulation of RSG. NtCDPK1 was previously shown to be autophosphorylated in a Ca²⁺-dependent manner. Here, we investigated the functional importance of autophosphorylation in NtCDPK1. Ser-6 and Thr-21 were identified as autophosphorylation sites of NtCDPK1. Autophosphorylation not only reduced the binding affinity of NtCDPK1 for RSG, but also inhibited the homodimerization of NtCDPK1. Furthermore, autophosphorylation decreased the phosphorylation efficiency of RSG yet increased that of myelin basic protein. Ser-6 and Thr-21 of NtCDPK1 were phosphorylated in response to GAs in plants. The substitution of these autophosphorylation sites with Ala enhanced the NtCDPK1 overexpression-induced sensitization of seeds to a GA biosynthetic inhibitor during germination. These results suggest new functions of autophosphorylation in CDPKs, namely, autophosphorylation can prevent the excessive phosphorylation of substrates and alter the substrate preference of CDPKs.

Phosphorylation is one of the most common posttranslational modifications of proteins in cells. Phosphorylation plays a key role in the modulation of diverse protein functions including enzymatic activity, interactions, conformation, localization, stability, and cross talk with other posttranslational modifications, which in turn regulates broad physiological function. Therefore, the catalytic activity and substrate specificity of protein kinases must be precisely controlled. Although protein kinases are regulated by various modes, a major regulatory mechanism is through phosphorylation; the same modification they catalyze. Some kinases require specific upstream kinases for activation; however, many kinases preserve inherent autophosphorylation capability. Thus, autophosphorylation appears to be a crucial reaction that triggers cellular signaling in response to internal and external stimuli.

Classical protein kinases have a canonical catalytic domain of 250 amino acids in length, which consists of a small N-terminal lobe of β-sheets and a larger C-terminal lobe of α-helices (Ubersax and Ferrell, 2007). The C-terminal lobe contains the activation loop that serves as a phosphorylation-dependent activation switch in many protein kinases. The protein substrate binds along the cleft between the two lobes, and a set of conserved residues within the catalytic domain catalyzes the transfer of the phosphate of ATP to the Ser, Thr, or Tyr residue of the substrate. Structurally related kinases must recognize their specific substrates among many proteins capable of phosphorylation to avoid unwanted crosstalk. The first level of substrate specificity is caused by the interaction of the active site of the kinase with the amino acid sequences surrounding the phosphorylation site of the substrate. Although all classical protein kinases share a common fold in catalytic domains, they differ in terms of the charge and hydrophobicity of the surface residues (Brown et al., 1999). Secondly, conserved docking motifs on the substrate that interact with specific regions of the kinase domain, may increase the selectivity of the kinase substrate (Sharrocks et al., 2000).

REPRESSION OF SHOOT GROWTH (RSG) is a tobacco (Nicotiana tabacum) transcriptional activator with a basic Leu zipper domain. RSG regulates the transcription of gibberellin (GA) biosynthetic genes (Fukazawa et al., 2000, 2010). The 14-3-3 proteins bind to RSG depending on the RSG phosphorylation of Ser-114 and sequester RSG in the cytoplasm, resulting in the reduction of expression of target genes (Igarashi et al., 2001; Ishida et al., 2004). GA levels regulate the intracellular localization of RSG; that is, RSG is translocated into the nucleus in response to a reduction in GA levels; GA treatment could reverse this nuclear accumulation (Ishida et al., 2004). We identified a Ca²⁺-dependent protein kinase, NtCDPK1, as an RSG kinase that promotes 14-3-3 binding of RSG by phosphorylation of Ser-114 in RSG (Ishida et al., 2008). NtCDPK1 interacts with RSG in a Ca²⁺-dependent manner and specifically phosphorylates Ser-114 in RSG in response to GAs. Thus, RSG and NtCDPK1 were shown to be involved in the feedback regulation of a GA biosynthetic gene NtGA20ox1.

CDPKs are unique Ca²⁺ decoders that are only found in plants and some protozoans, including malarial parasites (Harper and Harmon, 2005). CDPKs are sensor responders that have both Ca²⁺ sensing function and kinase activity within a single gene product (Harper et al., 1991; Suen and Choi, 1991). Among Ca²⁺-binding sensory proteins in plants, CDPKs are thought to play central roles in Ca²⁺ signaling because protein kinase C and conventional Ca²⁺/calmodulin-dependent protein kinase (CaMK), which represent the two major types of Ca²⁺-regulated kinases in animal systems, are missing from Arabidopsis (Arabidopsis thaliana; Hrabak et al., 2003). CDPKs have been shown to play important roles in various physiological processes, including growth, development, and responses to biotic and abiotic stresses in plants (Cheng et al., 2002; Schulz et al., 2013).

CDPKs are Ser/Thr protein kinases that are composed of a variable N-terminal domain, a catalytic domain, a junction domain containing an autoinhibitory segment, and a calmodulin-like domain (Harper et al., 2004). Although amino acid sequences of a catalytic domain, a junction domain, and a calmodulin-like domain are highly conserved among CDPK isoforms, the variable N-terminal domain of CDPKs is diversified not only in amino acid sequence but also in length, ranging from 25 to 180 amino acids in Arabidopsis (Takahashi and Ito, 2011). Although little is known about the functions of the variable N-terminal domain of CDPKs, we have demonstrated that the variable N-terminal domain is involved in substrate recognition (Ito et al., 2010, 2011). The autoinhibitory segment blocks the substrate binding site of the catalytic domain in the absence of Ca²⁺. Ca²⁺ binding triggers the translocation of all the regions downstream of the catalytic domain (i.e. the junction domain and calmodulin-like domain) to a new position roughly 135° clockwise from the substrate-binding site. This movement frees the important catalytic residue to interact with ATP, allowing the activation loop to assume the active orientation and removes occlusion of the autoinhibitory segment from the substrate-binding state (Wernimont et al., 2010, 2011). We recently reported that 14-3-3 interacts with NtCDPK1 in the presence of Ca²⁺ (Ito et al., 2014b). RSG and 14-3-3 bound to the variable N-terminal domain and the catalytic domain of NtCDPK1, respectively, and NtCDPK1 formed a heterotrimer with RSG and 14-3-3. The 14-3-3 protein was transferred from NtCDPK1 to the immediate product, phosphorylated RSG, suggesting that NtCDPK1 not only phosphorylates the substrate RSG but may also play a role as a scaffold that promotes the binding between phosphorylated RSG and 14-3-3 (Ito et al., 2014b).

Autophosphorylation at the activation loop is a fundamental and conserved mechanism in the activation of many protein kinases (Giese et al., 1998; Lochhead et al., 2005). For example, cAMP-dependent protein kinase A is activated by phosphorylation at Thr-197 in the activation loop (Johnson et al., 1996). The equivalent position in CDPKs is Asp or Glu that mimics a phosphoactivated state, suggesting that CDPKs exhibit full activity without phosphorylation in the activation loop. Thus, the catalytic activity of CDPKs is exclusively regulated by the concentration of Ca²⁺. In contrast, functions of autophosphorylation outside the activation loop remain mostly unclear. An exception is Ca²⁺/CaMKII. CaMKII undergoes intersubunit autophosphorylation at Thr-287 (or Thr-286; specific numbering is isoform dependent), resulting in Ca²⁺/calmodulin-independent activity (Hudmon and Schulman, 2002). Thr-287 lies within the autoinhibitory segment of CaMKII, and autophosphorylation at Thr-287 produces Ca²⁺-autonomous activity by preventing reassociation of the catalytic domain and the autoinhibitory segment. Phylogenetic analyses have proposed that the CDPK gene family arose through the fusion of a CaMK and a calmodulin (Harper et al., 1991; Cheng et al., 2002). Although the autoinhibitory segment of CDPKs shows similarity to that of CaMKII, Thr/Ser is missing at the corresponding position in CDPKs, indicating the lack of a Ca²⁺-independent active state by autophosphorylation. Previously, we found that NtCDPK1 is phosphorylated in response to GAs in plants and autophosphorylated in vitro (Ishida et al., 2008; Ito et al., 2014b).

The aim of our study was to determine the functional importance of autophosphorylation in NtCDPK1. We found that autophosphorylation sites of NtCDPK1 are located in the variable N-terminal domain that is involved in the recognition of its specific substrate RSG. The autophosphorylation of NtCDPK1 reduced both RSG binding and RSG phosphorylation while it increases the phosphorylation of myelin basic protein. Our results suggest that the substrate preference of NtCDPK1 is altered by autophosphorylation.

RESULTS

Ser-6 and Thr-21 Are the Autophosphorylation Sites of NtCDPK1

To examine the functional importance of the autophosphorylation of NtCDPK1, we first determined the autophosphorylation sites in vitro using tandem mass spectrometry. Glutathione S-transferase-fused NtCDPK1 (GST-NtCDPK1 wild type) was autophosphorylated and digested with trypsin or V8 protease. The mass spectrometry analysis showed the multiple peaks that resulted from potential phosphopeptides (Supplemental Fig. S1). These peptides were subjected to tandem mass spectrometry and the obtained spectra were analyzed by MASCOT (http://www.matrixscience.com/). The results showed that one peptide corresponds to a 13 to 28 amino acid sequence of the variable N-terminal domain of NtCDPK1 and Thr-21 is autophosphorylated.

The mass spectrometry analysis showed 79% coverage of the amino acid sequence of NtCDPK1, therefore, the missing fragments may include phosphorylated peptides. To examine whether the other autophosphorylation sites exist in NtCDPK1, we constructed a mutant version of NtCDPK1 in which Thr-21 was replaced with Ala (GST-NtCDPK1 T21A). To evaluate the phosphorylation state of autophosphorylated NtCDPK1 T21A, we used phosphate-binding tag SDS-PAGE (Phos-tag SDS-PAGE), which visualizes phosphorylated proteins as slower migration bands (Kinoshita et al., 2006). We used GST-NtCDPK1 dephosphorylated by λ phosphatase as a control for unphosphorylated NtCDPK1 as recombinant NtCDPK1 was already phosphorylated in Escherichia coli (Ito et al., 2014b). Phos-tag SDS-PAGE showed that autophosphorylated NtCDPK1 T21A migrates slower than dephosphorylated NtCDPK1 but faster than autophosphorylated NtCDPK1 wild type (Fig. 1A). This result suggested that NtCDPK1 contains at least one autophosphorylation site in addition to Thr-21.

Figure 1.

Ser-6 and Thr-21 are the autophosphorylation sites of NtCDPK1. A, Autophosphorylation level of NtCDPK1 decreased by S6A, T21A, S280A, and S494A mutation. GST-NtCDPK1 wild type (WT) was autophosphorylated or dephosphorylated by λ-phosphatase. Mutant versions of GST-NtCDPK1 were reacted in an autophosphorylation buffer. The autophosphorylation status of NtCDPK1 proteins was examined by Phos-tag SDS-PAGE. Autophosphorylated NtCDPK1 T21A migrated slower than autophosphorylated NtCDPK1 S6A due to some unknown factor. Proteins were visualized by CBB staining. This experiment was repeated three times with similar results. B, The catalytic activity of NtCDPK1 significantly decreased by S280A and S494A mutation. The kinase activities of GST-NtCDPK1 proteins were examined using MBP-RSG as a substrate. D219N mutant version of NtCDPK1 (a catalytically inactive form) was used as a negative control. Aliquots of reactions were subjected to SDS-PAGE followed by immunoblot analysis with anti-pS114 antibody, which specifically recognizes the phosphorylated Ser-114 of RSG. This experiment was repeated three times with similar results. C, NtCDPK1 S6A/T21A was not autophosphorylated. Ser residue in a polylinker between GST and NtCDPK1, which is already phosphorylated in E. coli, was substituted with Ala. This GST-NtCDPK1 was used in all the following experiments. The phosphorylation status of GST-NtCDPK1 proteins were examined in the same way as described in A. The nonspecific autophosphorylation of NtCDPK1 S6A, T21A, and S6A/T21A was only slightly detected (arrowhead). This experiment was repeated three times with similar results.

NtCDPK1 contains four Ser and five Thr residues in the remaining 21% of amino acid sequence that was not detected by mass spectrometry. To reveal the autophosphorylation sites other than Thr-21, five possible phosphorylation sites were selected using a phosphorylation site prediction tool. Ser-6 in the variable N-terminal domain, Ser-280 and Thr-299 in the catalytic domain, and Thr-434 and Ser-494 in the CaM-like domain were replaced with Ala, respectively. Phos-tag SDS-PAGE showed that the autophosphorylation level of NtCDPK1 was decreased by a S6A, S280A, or S494A mutation, but not by T299A or S434A mutation (Fig. 1A). To exclude the mutations that affect catalytic activity, we next performed an in vitro kinase assay using RSG as a substrate. The phosphorylation of RSG was detected using an antipS114 antibody, which specifically recognizes the phosphorylated Ser-114 of RSG (Ishida et al., 2004). Asp-219 of NtCDPK1 corresponds to the essential amino acid for the catalytic activity (Hanks and Hunter, 1995; McCubbin et al., 2004). D219N mutant version of GST-NtCDPK1 was used as a negative control for the kinase assay. As shown in Figure 1B, the S6A mutant version of NtCDPK1 phosphorylated RSG, yet the D219N mutant version did not. S280A and S494A mutant versions of NtCDPK1 only slightly phosphorylated RSG. These results suggested that the S280A or S494A mutation significantly decreased the catalytic activity of NtCDPK1 but not the S6A mutation. Therefore, Ser-6 in the variable N-terminal domain was shown to be an autophosphorylation site of NtCDPK1.

Furthermore, mass spectrometry showed that a Ser residue in a polylinker between GST and NtCDPK1 was the phosphorylated site in E. coli as described above. Therefore, the following experiments were performed using the GST-fused NtCDPK1 in which the Ser in the polylinker was substituted with Ala. A S6A/T21A mutant version of NtCDPK1 (GST-NtCDPK1 S6A/T21A) was constructed to confirmthat no autophosphorylation site existed other than Ser-6 and Thr-21 in NtCDPK1. Phos-tag SDS-PAGE showed that the mobility of GST-NtCDPK1 S6A/T21A was comparable to that of dephosphorylated GST-NtCDPK1 wild type (Fig. 1C). Furthermore, the autophosphorylation of NtCDPK1 was confirmed by autoradiography using [γ-³²P]ATP. Radioactivity was detected in NtCDPK1 wild type but not in NtCDPK1 S6A/T21A (Supplemental Fig. S2). These results suggested that Ser-6 and Thr-21 are the only autophosphorylation sites in NtCDPK1.

Autophosphorylation Is Not Required for 14-3-3 Binding to NtCDPK1

We found that 14-3-3 binds to NtCDPK1 by a new mode (Ito et al., 2014a, 2014b). The 14-3-3 proteins bind to phosphorylated motifs containing phospho-Ser residues, therefore, we supposed that autophosphorylation of NtCDPK1 could be involved in binding to 14-3-3. To examine this, we performed a pull-down assay using GST-NtCDPK1 S6A/T21A mutant version of NtCDPK1. As shown in Figure 2, His-14-3-3 bound to GST-NtCDPK1 S6A/T21A in the presence of both Ca²⁺ and ATP, indicating that 14-3-3-binding does not require the phosphate group in NtCDPK1 generated by autophosphorylation but depends on the conformational change of NtCDPK1 induced by Ca²⁺ and ATP. This result is consistent with the previous observations that 14-3-3 binds to the catalytic domain of NtCDPK1 (Ito et al., 2014b), and the autophosphorylation of NtCDPK1 was not necessary for 14-3-3-binding when RSG was bound to NtCDPK1 in advance (Ito et al., 2014a). Taken together, our results suggested that autophosphorylation of NtCDPK1 does not create a 14-3-3-binding site.

Figure 2.

14-3-3 binds to NtCDPK1 S6A/T21A in the presence of Ca²⁺ and ATP. GST-NtCDPK1 wild type (WT) and S6A/T21A were immobilized on glutathione beads and incubated with His-14-3-3 in the presence or absence of Ca²⁺ and ATP. NtCDPK1-bound proteins were subjected to SDS-PAGE, followed by immunoblot analysis with anti-14-3-3 antibody for the detection of His-14-3-3. The autophosphorylation status of NtCDPK1 proteins was examined by Phos-tag SDS-PAGE. GST-NtCDPK1 proteins were visualized by CBB staining. Ca²⁺-bound NtCDPK1 migrated slower than Ca²⁺-free NtCDPK1. This experiment was repeated three times with similar results.

The Autophosphorylation of Ser-6 and Thr-21 Decreases the Binding Affinity of NtCDPK1 for RSG

We previously revealed that the binding affinity of NtCDPK1 for RSG decreased when NtCDPK1 was autophosphorylated (Ito et al., 2014b). To examine whether the S6A/T21A mutation affects the binding affinity of NtCDPK1 for RSG, an in vitro pull-down assay was performed (Fig. 3). GST-NtCDPK1 wild type and GST-NtCDPK1 S6A/T21A were absorbed onto glutathione beads. When GST-NtCDPK1 proteins were reacted in an autophosphorylation buffer, the binding affinity of GST-NtCDPK1 wild type for maltose-binding protein (MBP)-RSG decreased, whereas GST-NtCDPK1 S6A/T21A did not. This result suggested that the autophosphorylation of Ser-6 and Thr-21 negatively regulates the RSG binding of NtCDPK1.

Figure 3.

RSG binding affinity of NtCDPK1 decreased by the autophosphorylation of Ser-6 and Thr-21. A, GST-NtCDPK1 proteins were immobilized on glutathione beads and incubated with MBP-RSG. NtCDPK1-bound proteins were subjected to SDS-PAGE, followed by immunoblot analysis with anti-MBP antibody for the detection of MBP-RSG. GST-NtCDPK1 proteins were visualized by CBB staining. This experiment was repeated three times with similar results. B, The quantification of band intensity of MBP-RSG in A. The intensity of MBP-RSG bands was normalized to the intensity of CBB-stained GST-NtCDPK1 bands in each lane. The value of nonautophosphorylated GST-NtCDPK1 wild type (WT) was set to 1. Each bar represents the mean ± se (n = 3). Different letters above the bars indicate significant difference in the binding affinity of GST-NtCDPK1 proteins for MBP-RSG (P < 0.01, one-way ANOVA with Tukey’s honestly significant difference test).

The reduction of the binding affinity of NtCDPK1 for RSG by autophosphorylation appeared to result from the negative charge of the phosphate groups on Ser-6 and Thr-21. Thus, we generated a S6D/T21D mutant version of NtCDPK1 by replacing Ser-6 and Thr-21 with Asp, which may mimic the phosphorylated state. The pull-down assay showed that the binding affinity of GST-NtCDPK1 S6D/T21D for MBP-RSG was comparable to that of autophosphorylated GST-NtCDPK1 wild type regardless of the autophosphorylation reaction (Fig. 3). This result suggested that S6D/T21D mutation mimics the autophosphorylated state of NtCDPK1 and that the negative charge of phosphorylated Ser-6 and Thr-21 causes the reduction of the binding affinity of NtCDPK1 for RSG.

Autophosphorylation Decreases Dimerization of NtCDPK1

Both inter- and intramolecular autophosphorylations have been reported in Arabidopsis receptor-like protein kinase RLK5 and Catharanthus roseus RLK1, respectively (Horn and Walker, 1994; Schulze-Muth et al., 1996). Furthermore, the homodimer formation has been reported in several kinases such as ERK2, p21-activated kinase1 (PAK1), and brassinosteroid-insensitive1 (BRI1; Khokhlatchev et al., 1998; Parrini et al., 2002; Wang et al., 2005). To examine whether NtCDPK1 forms a homodimer, we performed a pull-down assay using GST-NtCDPK1 and 10×His-tagged (His)-NtCDPK1. GST-NtCDPK1 bound to His-NtCDPK1 in the presence of Ca²⁺ but not in the presence of EGTA (Fig. 4A). This result suggested that NtCDPK1 forms a homodimer in a Ca²⁺-dependent manner. A bimolecular fluorescence complementation (BiFC) assay (Hu et al., 2002) was used to examine the dimerization of NtCDPK1 in plant cells. NtCDPK1 was translationally fused to the N- and C-terminal portion of yellow fluorescent protein, which generated NtCDPK1-YFP^N and NtCDPK1-YFP^C fusion proteins, respectively. These constructs were codelivered into the leaf cells of tobacco by particle bombardment. The reconstituted YFP signal, caused by the interaction between NtCDPK1-YFP^N and NtCDPK1-YFP^C, was observed in leaf epidermal cells (Fig. 4B, top row). Control experiments in which NtCDPK1-YFP^N was coexpressed with unfused YFP^C did not show any fluorescence (Fig. 4B, bottom row). These results indicated that NtCDPK1 forms a homodimer in living cells.

Figure 4.

The dimerization of NtCDPK1 was inhibited by autophosphorylation. A, NtCDPK1 forms a homodimer in a Ca²⁺-dependent manner. GST and GST-NtCDPK1 were immobilized on glutathione beads and incubated with His-NtCDPK1 in the presence of Ca²⁺ or EGTA (EG). NtCDPK1-bound proteins were subjected to SDS-PAGE, followed by immunoblot analysis with anti-NtCDPK1 antibody for the detection of His-NtCDPK1. GST and GST-NtCDPK1 were visualized by CBB staining. This experiment was repeated three times with similar results. B, BiFC analysis showed in vivo dimerization of NtCDPK1. BiFC constructs were delivered into leaf cells of tobacco by particle bombardment. After 24 h, the cells were observed by epifluorescence microscopy. Coexpression of NtCDPK1-YFP^N and NtCDPK1-YFP^C (top) and coexpression of NtCDPK1-YFP^N and YFP^C (bottom) are shown. Reconstituted YFP fluorescence (left), RFP fluorescence as a control for transfection efficiency (center), and bright-field images (right) are shown. This experiment was repeated three times with similar results. Bars = 100 µm. C, NtCDPK1 was autophosphorylated in an intermolecular manner. Phosphorylation of GST-NtCDPK1 by His-NtCDPK1 was examined. GST-NtCDPK1 wild type (WT), D219N, and S6A/T21A/D219N were reacted with His-NtCDPK1 WT in an autophosphorylation buffer. The autophosphorylation state of GST-NtCDPK1 proteins was examined using Phos-tag SDS-PAGE. GST-NtCDPK1 proteins were detected by immunoblot analysis with anti-GST antibody. His-NtCDPK1 WT were visualized by CBB staining. Letters represent phosphorylated GST-NtCDPK1 WT (a), unphosphorylated GST-NtCDPK1 WT and S6A/T21A/D219N (b), and phosphorylated GST-NtCDPK1 D219N (c). Phosphorylated GST-NtCDPK1 WT migrated slower than phosphorylated GST-NtCDPK1 D219N, which might reflect the effect of D219N mutation. This experiment was repeated three times with similar results. D, Autophosphorylation inhibits the dimerization of NtCDPK1. Autophosphorylated or nonautophosphorylated GST-NtCDPK1 were immobilized on glutathione beads and incubated with His-NtCDPK1 in the presence of Ca²⁺ or EGTA (EG). NtCDPK1-bound proteins were subjected to SDS-PAGE, followed by immunoblot analysis with anti-NtCDPK1 antibody for the detection of His-NtCDPK1. GST-NtCDPK1 proteins were visualized by CBB staining. This experiment was repeated three times with similar results.

To examine whether NtCDPK1 is autophosphorylated inter- or intramolecularly, we constructed GST-NtCDPK1 S6A/T21A/D219N in addition to the above-mentioned GST-NtCDPK1 D219N. Since these GST-NtCDPK1 proteins are catalytically inactivated, they are not intramolecularly autophosphorylated. The phosphorylation by His-NtCDPK1 was compared with GST-NtCDPK1 D219N and GST-NtCDPK1 S6A/T21A/D219N. Phos-tag SDS-PAGE showed that phosphorylated GST-NtCDPK1 D219N migrates slower than GST-NtCDPK1 S6A/T21A/D219N and dephosphorylated GST-NtCDPK1 wild type (Fig. 4C). This result suggested that NtCDPK1 can be autophosphorylated by an intermolecular mechanism.

To examine whether autophosphorylation also affects the dimerization of NtCDPK1, an in vitro pull-down assay was performed. When NtCDPK1 proteins were autophosphorylated, the binding of His-NtCDPK1 to GST-NtCDPK1 decreased (Fig. 4D). This result indicated that autophosphorylation decreases not only the binding of NtCDPK1 to RSG but also the dimerization of NtCDPK1.

The Kinase Activity of NtCDPK1 for RSG Is Decreased by Autophosphorylation

Autophosphorylation negatively regulates the binding of NtCDPK1 to RSG; therefore, autophosphorylation could affect the phosphorylation of RSG by NtCDPK1. To test this, we performed the in vitro kinase assay using MBP-RSG as a substrate and compared the kinase activity of NtCDPK1 wild type with S6A/T21A or S6D/T21D. As shown in Figure 5A, S6A/T21A mutation increased the phosphorylation of MBP-RSG, whereas S6D/T21D mutation did not. The 19S proteasome regulatory subunit, Rpn3, is also reported as a substrate of NtCDPK1 (Lee et al., 2003). The kinase assay using MBP-Rpn3 provided similar results, although MBP-Rpn3 was phosphorylated less efficiently than MBP-RSG (Fig. 5B). These results suggested that the autophosphorylation of NtCDPK1 negatively regulates the phosphorylation of substrates through the reduction of the binding affinity. Myelin basic protein (MyBP) is known as a conventional substrate, which is often used to measure the kinase activity. We performed the kinase assay using MyBP as a substrate to examine whether the mutations affect the phosphorylation of other substrates. The phosphorylation of MyBP decreased by S6A/T21A mutation but not by S6D/T21D mutation (Fig. 5C). Namely, the effect of autophosphorylation on the kinase activity of NtCDPK1 for MyBP seemed to be opposite to RSG and Rnp3. These results suggested that the autophosphorylation of NtCDPK1 does not affect the catalytic activity itself and that NtCDPK1 could alter the substrate preference before and after autophosphorylation.

Figure 5.

Autophosphorylation affects the substrate phosphorylation by NtCDPK1. A, C, and E, Phosphorylation of MBP-RSG (A), MBP-Rpn3 (C), and MyBP (E) by GST-NtCDPK1 wild type (WT), S6A/T21A, and S6D/T21D. Substrates were phosphorylated for the indicated time periods. The phosphorylation of substrates was detected by autoradiography using [γ-³²P]ATP. Substrates were visualized by CBB staining. GST-NtCDPK1 proteins were detected by immunoblot analysis with anti-GST antibody. W, A, and D represent NtCDPK1 WT, S6A/T21A, and S6D/T21D, respectively. This experiment was repeated three times with similar results. B, D, and F, The quantification of band intensity of MBP-RSG in A, C, and E, respectively. The values of GST-NtCDPK1 S6A/T21A at 6 min in A, S6A/T21A at 30 min in C, and WT at 6 min in E were set to 1, respectively. Each bar represents the mean ± se (n = 3). Different letters above the bars indicate significant difference in substrate phosphorylation by GST-NtCDPK1 proteins (P < 0.01, one-way ANOVA with Tukey’s honestly significant difference test). See also Supplemental Figure S3 for details.

Ser-6 and Thr-21 Are the in Vivo Phosphorylation Sites of NtCDPK1

Previous study revealed that GA treatment with tobacco plants induces the phosphorylation of NtCDPK1 (Ishida et al., 2008). To examine whether Ser-6 and Thr-21 are the phosphorylation sites of NtCDPK1 in response to GAs in plants, we generated the transgenic tobacco overexpressing S6A, T21A, or S6A/T21A mutant version of NtCDPK1 (abbreviated as S6A^OE, T21A^OE, and S6A/T21A^OE_, respectively). Tobacco seedlings were grown in the presence of GA₃ or Uniconazole P, a GA biosynthetic inhibitor. Proteins were extracted from tobacco plants and the phosphorylation status of NtCDPK1 proteins was determined by Phos-tag SDS-PAGE (Fig. 6). The mobility of NtCDPK1 S6A/T21A from GA-treated plants was comparable to Uniconazole P-treated plants. NtCDPK1 wild type, S6A, and T21A from GA-treated plants showed slower migration bands than S6A/T21A mutant version of NtCDPK1. These results suggested that both Ser-6 and Thr-21 are the autophosphorylation sites of NtCDPK1 in plant cells.

Figure 6.

Ser-6 and Thr-21 were phosphorylated in response to GA. Transgenic tobacco plants overexpressing NtCDPK1 wild type (WT), S6A, T21A, or S6A/T21A (shown as WT^OE, S6A^OE, T21A^OE, or S6A/T21A^OE, respectively) were grown on MS agar medium. One week after seed germination, seedlings were transferred onto the medium containing 50 µm GA₃, or 1 µm Uniconazole P (Uni). After GA₃ or Uniconazole P treatment for 3 d, proteins were extracted from the shoot of seedlings and subjected to Phos-tag SDS-PAGE. The possibly degraded NtCDPK1 was detected in extracts from plant cells (arrowhead). NtCDPK1 might be destabilized by T21A mutation in vivo. NtCDPK1 proteins were detected by immunoblot analysis with anti-NtCDPK1 antibody. CBB-stained Rubisco large subunit was used as a loading control. This experiment was repeated three times with similar results.

S6A/T21A Mutation Enhance the Sensitivity to a GA Biosynthetic Inhibitor

NtCDPK1 is involved in the feedback regulation of the expression of a GA 20-oxidase gene, NtGA20ox1, through the phosphorylation of RSG (Ishida et al., 2008; Fukazawa et al., 2010). NtCDPK1 negatively regulates RSG by promoting the 14-3-3 binding to RSG. Up-regulation of NtGA20ox1 in response to a GA biosynthetic inhibitor, Uniconazole P, was suppressed by the overexpression of the NtCDPK1 in the transgenic plants. Therefore, overexpressor of NtCDPK1 becomes sensitized to the GA biosynthetic inhibitor (Ishida et al., 2008). The substitution of the variable N-terminal domain of an Arabidopsis CPK9 with that of NtCDPK1 conferred RSG kinase activity and its overexpression also resulted in the sensitization to the GA biosynthetic inhibitor through the suppression of feedback regulation of NtGA20ox1 (Ito et al., 2010). In this study, we found that autophosphorylation of the variable N-terminal domain of NtCDPK1 decreases the phosphorylation efficiency of RSG. If the autophosphorylation of NtCDPK1 is functionally important in plants, the S6A/T21A mutation would enhance the sensitivity to Uniconazole P. The germination rate among wild-type tobacco plant SR1, NtCDPK1 WT^OE, S6A/T21A^OE, and S6D/T21D^OE was compared. S6A/T21A^OE was more sensitive to Uniconazole P than WT^OE (Fig. 7). No significant difference in the germination rate was detected between WT^OE and S6D/T21D^OE. The germination rate of these tobacco plants in Uniconazole P-containing medium was recovered by addition of GA₃. These results suggested that autophosphorylation of NtCDPK1 plays a role in the homeostasis of GAs.

Figure 7.

S6A/T21A mutation increased the sensitivity to Uniconazole P by the overexpression of NtCDPK1. A, Seeds of control wild-type SR1 tobacco and the transgenic plants overexpressing NtCDPK1 wild type (WT), S6A/T21A, and S6D/T21D were germinated in the MS agar medium containing 100 nm Uniconazole P (Uni) with or without 1 µm GA₃ for 7 d at 28°C. Transgenic tobacco overexpressing S6D/T21D mutant version of NtCDPK1 (S6D/T21D^OE) was generated for this experiment. B, The quantification of seed germination rate obtained from A. The bar graph represents means ± se (n = 3, three replicates of each of three transgenic lines). Different letters above the bars indicate significant differences in seed germination rate between wild-type SR1 and transgenic plants (P < 0.01, two-way ANOVA with Tukey’s honestly significant difference test). See also Supplemental Figure S4 for details.

DISCUSSION

Autophosphorylation is a common mechanism for the functional regulation of protein kinases in both prokaryotes and eukaryotes. One of the best-studied examples of functions of autophosphorylation is the self-activation by phosphorylation of activation loop in the catalytic domain (De Nicola et al., 2013). According to this report, approximately 45% of the human protein kinases can autophosphorylate their activation loop (Beenstock et al., 2016). Although autophosphorylation of other sites, outside the activation loop, is highly prevalent in protein kinases, molecular mechanisms, and physiological roles of autophosphorylations have been elucidated only in a handful of cases. EpsB is a Tyr kinase of Bacterium Bacillus subtilis that is involved in exopolysaccharide (EPS) biosynthesis required for biofilm formation. In the absence of EPS, EpsB is inactivated by autophosphorylation. The presence of EPS inhibits autophosphorylation and instead promotes the phosphorylation of the substrate glycosyltransferase (Elsholz et al., 2014). Budding yeast Ire1 (for inositol requiring enzyme1) is an endoplasmic reticulum-resident transmembrane sensor kinase involved in the unfolded protein response. Binding of misfolded proteins promotes Ire1 oligomerization and autophosphorylation that activates an endoribonuclease domain in the cytosolic domain of Ire1. This endoribonuclease activity of Ire1 excises the translation-inhibitory intron in Hac1 (for homologous to ATF and CREB) mRNA (Walter and Ron, 2011). Pom1 (for polarity misplaced1) is a fission yeast kinase that localizes to cell ends and couples cell size homeostasis with mitotic commitment. Local dephosphorylation of Pom1 permits direct association of Pom1 with the plasma membrane at cell tips. Autophosphorylation leads to Pom1 detachment from the membrane, which shapes Pom1 cortical gradients (Hachet et al., 2011).

In this study, we found an interesting role of autophosphorylation, namely, an influence on the affinity for the substrate. Although catalytic activity of CDPKs are mainly regulated by the intracellular concentration of Ca²⁺, autophosphorylation of NtCDPK1 can reduce the phosphorylation efficiency of RSG even in the presence of Ca²⁺ (Fig. 5). This mechanism could prevent the excessive phosphorylation of RSG, resulting in the transition from early repression of the transcription of NtGA20ox1 to moderate and continuous repression. The moderate repression state might be maintained until the concentration of GAs is decreased. Thus, autophosphorylation of NtCDPK1 may play a role in the homeostasis of GA by functioning as a compensation module in the GA feedback circuit.

Phosphorylation motifs are often recognized by the specific features with complementary characteristics of the active site of the catalytic domain, including its depth and charge or hydrophobicity (Ubersax and Ferrell, 2007). The amino acid sequence around Ser-114 of RSG is closely matched to the major consensus phosphorylation motif of CDPKs, basic-hydrophobic(Φ)-X-basic-X-X-Ser/Thr-X-X-X-Φ-basic (Harper and Harmon, 2005), whereas the amino acid sequences around Ser-6 and Thr-21 of NtCDPK1 is only partially related to the minor consensus phosphorylation motif of CDPKs, Ser-X-Basic (Neumann et al., 1996; Supplemental Fig. S5). This implies that autophosphorylation of NtCDPK1 cannot be explained by the standard recognition mode of the substrates. Thus, although both the autophosphorylation reaction and the substrate phosphorylation reaction use the single active site of the kinase domain, they nevertheless must have distinct structural and mechanistic properties. The network protein sequence analysis server predicted that the variable N-terminal domain of NtCDPK1 including Ser-6 and Thr-21 is mainly composed of random coils (NPS@, https://npsa-prabi.ibcp.fr/; Combet et al., 2000). The crystal structure of Toxoplasma gondii CDPK1 was determined; however, its variable N-terminal domain could not, suggesting a disordered structure (Wernimont et al., 2010, 2011). The flexible nature of the variable N-terminal domain of NtCDPK1 might help to orient the target residues correctly in the active site of the kinase domain.

Autophosphorylation can occur via several mechanisms. A simple mechanism would involve the formation of a homodimer, which can reduce the distance between the autophosphorylation sites of one monomer and the active site of the other monomer. However, this mechanism cannot explain how kinases recognize nonconsensus sequences on autophosphorylation. Rather, most of the experimentally supported models include dimerization, which allosterically leads the stabilization of a unique conformation that promotes autophosphorylation. Such examples include autophosphorylation of the activation loop of PAK1 that regulates cytoskeletal reorganization and cell motility (Wang et al., 2011). In the PAK1 dimer, one monomer serves as an enzyme and phosphorylates the other. Dimer formation induces conformational change resulting in the formation of a hydrogen bond between the catalytic Asp of the enzyme monomer with the target Thr in the activation-loop of the substrate monomer in a favorable orientation for phosphorylation. Thus, one monomer is considered to serve simultaneously as a substrate and as an allosteric activator of its partner. IL-1 receptor-associated kinase4 and epidermal growth factor receptor also employ autophosphorylation of the activation loop imposed by allosteric interactions between two monomers (Ferrao et al., 2014; Jura et al., 2011). We found that NtCDPK1 forms a homodimer (Fig. 4A). Dimerization of NtCDPK1 could similarly induce an allosteric effect that sets target residues in a correct position for autophosphorylation, supported by the flexible nature of the variable N-terminal domain, providing a model of how the catalytic domain can interact with nonconsensus sequences.

Another important finding of this study is that autophosphorylation alters the substrate preference of NtCDPK1. We observed the opposite effect of autophosphorylation on the phosphorylation activity of NtCDPK1 for RSG and MyBP. Autophosphorylation reduced the binding affinity for RSG and the efficiency of RSG phosphorylation (Figs. 3 and 5). On the other hand, autophosphorylation increased the efficiency of MyBP phosphorylation. This increased efficiency could result from the increased binding affinity of NtCDPK1 for MyBP. Although the effect of autophosphorylation on the kinase activity of BRI1 was reported (Oh et al., 2009), alteration of the substrate preference by autophosphorylation is unreported. Furthermore, we found that autophosphorylation of NtCDPK1 repressed its dimerization as well as the interaction with RSG (Fig. 4D), suggesting that dimeric NtCDPK1 and autophosphorylated monomeric NtCDPK1 prefers RSG and MyBP as a substrate, respectively. Protein phosphorylation often promotes protein-protein interactions through the creation of protein-binding sites for phosphobinding proteins in cellular signaling. We found that the variable N-terminal domain of NtCDPK1 containing autophosphorylation sites Ser-6 and Thr-21 is involved in substrate recognition (Ito et al., 2010). Although MyBP is merely a substrate for activity measurement, these findings raise the possibility that autophosphorylated NtCDPK1 might have distinct physiological substrates in plants. GAs evoke plant growth and feedback response. Since RSG, a preferable substrate for dimeric NtCDPK1, is involved in GA feedback regulation, an unknown substrate that is sustainably phosphorylated by autophosphorylated NtCDPK1 in the presence of GAs could be a regulator of plant growth. Identification of such a substrate will improve our understanding of the molecular mechanisms downstream of Ca²⁺ in GA signaling, which influences many aspects of plant growth and development.

MATERIALS AND METHODS

Preparation of Recombinant Proteins

Mutant versions of NtCDPK1 were generated by overlap extension PCR or whole-plasmid amplification method with plasmid harboring full-length NtCDPK1 as a template and appropriate primers (Supplemental Table S1). PCR products were cloned into pGEX-4T-1 (GE Healthcare). Some constructs were generated using seamless ligation cloning extract (SLiCE; Okegawa and Motohashi, 2015). GST-NtCDPK1s, MBP-RSG, His-14-3-3, and His-NtCDPK1 were expressed in Escherichia coli containing pGEX-4T-1-NtCDPK1, pMALc2-RSG, pET16b-14-3-3, and pET16b-NtCDPK1, and purified by glutathione-Sepharose 4B (GE Healthcare), amylose resin (New England Biolabs), and COSMOGEL His-Accept (Nacalai Tesque), respectively.

Autophosphorylation and Dephosphorylation Reactions

Purified GST-NtCDPK1 and His-NtCDPK1 were reacted in an autophosphorylation buffer containing 20 mm Tris-HCl, pH 7.5, 10 mm MgCl₂, 0.5 mm CaCl₂, 0.1% (v/v) Triton X-100, and 0.05% (v/v) β-mercaptoethanol with 1 mm ATP (cold assay) or 1 mm ATP supplemented with [γ-³²P]ATP (5,000 µCi/mmol, 10–20 µCi/reaction, hot assay) at 30°C for 30 to 60 min. GST-NtCDPK1 proteins were dephosphorylated by λ protein phosphatase (New England Biolabs) at 30°C for 30 min according to Ito and Takahashi (2015). Phosphorylation state was examined using Phos-tag SDS-PAGE (Kinoshita and Kinoshita- Kikuta, 2011) or Typhoon FLA 9500 biomolecular imager (GE Healthcare).

Mass Spectrometry Analysis

The autophosphorylation site of NtCDPK1 was identified using matrix-assisted laser desorption/ionization-time of flight mass spectrometry by Shimadzu Techno-Research. Autophosphorylated GST-NtCDPK1 was reduced with dithiothraitol, alkylated with iodeacetamide, and digested with Trypsin or V8 protease. The digested protein was dried by centrifugal evaporator and resuspended in a buffer containing 200 mm lactose, 0.1% trifluoroacetic acid (TFA), and 80% acetonitrile. Phosphorylated peptides of the resuspended sample were selectively enriched by Titansphere TiO (GL sciences). Sample was eluted with ammonium solution, neutralized with TFA, and purified with a µ-C18 ZipTip (Merck Millipore). Purified sample was spotted onto a matrix-assisted laser desorption/ionization plate, mixed with a matrix containing 10 mg/mL 2,5-dihydroxybenzoic acid, 0.1% TFA, and 50% acetonitrile and air-dried. The mass spectra were obtained using AXIMA Performance mass spectrometer (Shimadzu).

In Vitro Pull-Down Assay

GST-NtCDPK1 (2.0 μg) were incubated with 20 µL of glutathione-Sepharose 4B (GE Healthcare) and MBP-RSG (3.0 μg), His-NtCDPK1 (3.0 μg), or His-14-3-3 (3.0 μg) in 400 µL of a binding buffer containing 25 mm MOPS-NaOH, pH 7.0, 25 mm NaCl, 0.05% (v/v) β-mercaptoethanol, 0.1% (v/v) Triton X-100, 0.1 mm phenylmethylsulfonyl fluoride, and 0.5 mm CaCl₂ at 4°C for 30 min. Proteins bound to the beads were washed with the binding buffer, resolved by SDS-PAGE (Tris/Gly buffer) or Phos-tag SDS-PAGE and detected by Coomassie Brilliant Blue (CBB) staining or immunoblot analysis as described below. The intensities of bands were measured using ImageJ software (NIH, version 1.49).

Immunoblot Analysis

After electrophoresis, the separated proteins were transferred onto an Immobilon-P transfer membrane (Millipore). For Phos-tag SDS-PAGE gels, a transfer buffer containing 1 mm EDTA was used according to Kinoshita and Kinoshita-Kikuta (2011). The membranes were probed with anti-pS114 (antibody against the phosphorylated Ser-114 of RSG; Ishida et al., 2004), anti-NtCDPK1 (Ishida et al., 2008), anti-14-3-3 (Igarashi et al., 2001), anti-GST (GE Healthcare), or anti-MBP (MBL), followed by the horseradish peroxidase-conjugated secondary antibody. Chemiluminescence was detected using Immobilon western Chemiluminescent HRP Substrate (Millipore) and quantified using ImageQuant LAS 4000 with ImageQuant TL software (GE Healthcare).

BiFC Analysis

For BiFC analysis, the previously amplified DNA fragments encoding YFP^N and YFP^C were used (Ito et al., 2014b). We generated constructs expressing NtCDPK1 C-terminally fused with YFP^N or YFP^C and named them pJ4/NtCDPK1-YFP^N and pJ4/NtCDPK1-YFP^C, respectively. Primers used to generate these constructs are shown in Supplemental Table S1. pJ4/YFP^C (stop codon +) was used as a negative control. pJ4mRFP vector harboring mRFP was cotransfected as a control for transfection efficiency. Transfection was performed according to the previously described method (Ito et al., 2010, 2014b). Fluorescence images were acquired using ECLIPSE Ni-E (Nikon) equipped with DS-Ri1 digital camera (Nikon) and the YFP filter cube (500/20-nm excitation, 535/30-nm emission) and the RFP filter cube (540/25-nm excitation, 605/55-nm emission). The acquired images were analyzed using NIS elements D software (version 4.2; Nikon).

In Vitro Kinase Assay

The catalytic activity of GST-NtCDPK1 proteins was assayed in the autophosphorylation buffer as described above at 30°C. GST-NtCDPK1 (1.0 µg/mL), MBP-RSG (0.2 mg/mL), MBP-Rpn3 (0.2 mg/mL), and dephosphorylated myelin basic protein (0.2 mg/mL; Sigma-Aldrich) were used for the assay. After aliquots of reaction mixture were subjected to SDS-PAGE, the radioactivity of phosphorylated substrates was detected with an imaging system, Typhoon FLA 9500 (GE Healthcare). The intensities of bands were measured using ImageJ software (NIH, version 1.49).

Detection of in Vivo Phosphorylation of NtCDPK1

Transgenic tobacco plants overexpressing mutant versions of NtCDPK1 (S6A^OE, T21A^OE, or S6A/T21A^OE) were generated according to the previously described method (Ito et al., 2010), and previously generated NtCDPK1 WT^OE was used for this study (Ito et al., 2010). Plants were grown on Murashige and Skoog (MS) medium containing 0.8% agar. At 7 d after seed germination, seedlings were transferred onto MS agar medium containing 50 µm GA₃ (Wako Pure Chemical) or 1 µm Uniconazole P (Wako Pure Chemical). After GA₃ or Uniconazole P treatment for 3 d, proteins were extracted from the shoot of seedlings. Shoots were disrupted in liquid nitrogen by grinding with a mortar and pestle and then extracted in three volumes of an extraction buffer (40 mm MOPS-NaOH, pH 6.5, 20 mm glycerol 2-phosphate disodium, 25 mm NaCl, 10% glycerol, 0.1% Triton X-100, 0.05% β-mercaptoethanol, 1 mm phenylmethylsulfonyl fluoride, complete EDTA-free protease inhibitor cocktail [Roche], and a phosphatase inhibitor cocktail [Sigma-Aldrich]). The extract was centrifuged, and the supernatant was subjected to Phos-tag SDS-PAGE. NtCDPK1 proteins were detected by immunoblotting with anti-NtCDPK1 antibody.

Uniconazole P Tolerance Assay

Seeds of wild-type SR1 and transgenic tobacco plants were germinated on MS agar medium containing 100 nm Uniconazole P with or without 1 µm GA₃ for 7 d at 28°C.

Statistical Analysis

Statistical analysis was performed using R software (version 3.9.0). Comparisons were performed using two-way ANOVA with Tukey’s honestly significant difference test.

Accession Numbers

Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers AF072908 (NtCDPK1), AB040471 (RSG), AB071967 (14-3-3, D31), and NM_001324948 (Rpn3).

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Mass spectrometry analysis showing Thr-21 as the autophosphorylation site of NtCDPK1.

• Supplemental Figure S2. Kinase assay using [γ-³²P]ATP showing S6A/T21A mutant version of NtCDPK1 is not autophosphorylated.

• Supplemental Figure S3. Statistical analysis of the data shown in Figure 5.

• Supplemental Figure S4. Statistical analysis of the data shown in Figure 7.

• Supplemental Figure S5. Ser-6 and Thr-21 are located adjacent to the basic amino acid residues.

• Supplemental Table S1. Primer sequences used in this study.