Members of the Plant CRK Superfamily Are Capable of Trans- and Autophosphorylation of Tyrosine Residues*

Background: Protein kinases that catalyze Tyr phosphorylation in plants in vivo are largely unknown. Results: CDPK/CPK-related protein kinases (CRKs) that auto/trans-phosphorylate Tyr residues and six substrates of these were identified. CRK knock-out mutants show reduced Tyr phosphorylation of β-tubulin proteins. Conclusion: CRKs can phosphorylate Tyr residues of β-tubulin and certain transcription factors. Significance: CRKs might be responsible for much of the protein Tyr phosphorylation in vivo. Protein phosphorylation on Tyr residues is a key post-translational modification in mammals. In plants, recent studies have identified Tyr-specific protein phosphatase and Tyr-phosphorylated proteins in Arabidopsis by phosphoproteomic screenings, implying that plants have a Tyr phosphorylation signal pathway. However, little is known about the protein kinases (PKs) involved in Tyr phosphorylation in plants. Here, we demonstrate that Arabidopsis calcium-dependent protein kinase (CDPK/CPK)-related PKs (CRKs) have high Tyr-autophosphorylation activity and that they can phosphorylate Tyr residue(s) on substrate proteins in Arabidopsis. To identify PKs for Tyr phosphorylation, we examined the autophosphorylation activity of 759 PKs using an Arabidopsis protein array based on a wheat cell-free system. In total, we identified 38 PKs with Tyr-autophosphorylation activity. The CRK family was a major protein family identified. A cell-free substrate screening revealed that these CRKs phosphorylate β-tubulin (TBB) 2, TBB7, and certain transcription factors (TFs) such as ethylene response factor 13 (ERF13). All five CRKs tested showed Tyr-auto/trans-phosphorylation activity and especially two CRKs, CRK2 and CRK3, showed a high ERF13 Tyr-phosphorylation activity. A cell-based transient expression assay revealed that Tyr16/Tyr207 sites in ERF13 were phosphorylated by CRK3 and that Tyr phosphorylation of endogenous TBBs occurs in CRK2 overexpressing cells. Furthermore, crk2 and crk3 mutants showed a decrease in the Tyr phosphorylation level of TBBs. These results suggest that CRKs have Tyr kinase activity, and these might be one of the major PKs responsible for protein Tyr phosphorylation in Arabidopsis plants.


Protein phosphorylation on Tyr residues is a key post-translational modification in mammals. In plants, recent studies have identified Tyr-specific protein phosphatase and Tyr-phosphorylated proteins in Arabidopsis by phosphoproteomic screenings, implying that plants have a Tyr phosphorylation signal pathway. However, little is known about the protein kinases (PKs) involved in Tyr phosphorylation in plants. Here, we demonstrate that Arabidopsis calcium-dependent protein kinase (CDPK/CPK)-related PKs (CRKs) have high Tyr-autophosphorylation activity and that they can phosphorylate Tyr residue(s) on substrate proteins in Arabidopsis.
To identify PKs for Tyr phosphorylation, we examined the autophosphorylation activity of 759 PKs using an Arabidopsis protein array based on a wheat cell-free system. In total, we identified 38 PKs with Tyrautophosphorylation activity. The CRK family was a major protein family identified. A cell-free substrate screening revealed that these CRKs phosphorylate ␤-tubulin (TBB) 2, TBB7, and certain transcription factors (TFs) such as ethylene response factor 13 (ERF13). All five CRKs tested showed Tyr-auto/trans-phosphorylation activity and especially two CRKs, CRK2 and CRK3, showed a high ERF13 Tyr-phosphorylation activity. A cell-based transient expression assay revealed that Tyr 16/ Tyr 207 sites in ERF13 were phosphorylated by CRK3 and that Tyr phosphorylation of endogenous TBBs occurs in CRK2 overexpressing cells. Furthermore, crk2 and crk3 mutants showed a decrease in the Tyr phosphorylation level of TBBs. These results suggest that CRKs have Tyr kinase activity, and these might be one of the major PKs responsible for protein Tyr phosphorylation in Arabidopsis plants.
Protein phosphorylation of serine (Ser), threonine (Thr), and tyrosine (Tyr) residues is a key post-translational modification required for signal transduction in eukaryotes. In animals, protein-tyrosine kinases (PTKs) 2 play a central role in many signaling pathways, including hormone response, differentiation, development, and cancer formation (1). In contrast, it is unclear whether Tyr phosphorylation signaling cascades exist in plants, because no PTK homologous genes have been reported in Arabidopsis and rice genomes (2,3).
Recently, Tyr-specific protein phosphatase (PTP1) (4) and more than 1000 Tyr-phosphorylated proteins have been identified by proteomics analysis (5,6) in plants. A previous report suggests that not only PTP1 but also dual-specific (pSer/pThr and pTyr) phosphatases involve in abiotic stress and hormone signaling are present in plants (7). On the other hand, a phosphoproteomic approach revealed that ϳ4% of phosphopeptides are Tyr-phosphorylated peptides in plants and the proportion of Tyr phosphorylation is equivalent to that found in human cells (5). In addition, meta-analysis of phosphoproteomic data revealed that Tyr-phosphorylated peptides are overrepresented in mitochondrial and characterized two Tyr phosphorylation motifs (6). Other studies have demonstrated that PTK and PTP inhibitor application to plants alter abscisic acid (8), gibberellin (9), cold stress (10), and sugar responses (11), as well as cytoskeleton organization (12) and cell division (13). These findings strongly suggest that plants possess Tyr phosphorylation signaling pathways. According to recent studies, Tyr residues (as well Ser and Thr) of several dual-specific (Ser/Thr/Tyr) type or Ser/Thr-specific type PKs, e.g. receptorlike/receptor-like cytoplasmic PKs (RLK/RLCKs), CDPKs, glycogen synthase kinase 3 (GSK3)/Shaggy-like PKs, and MAPK have been observed to autophosphorylate (14 -16). These PKs probably participate in Tyr phosphorylation signaling in plants.
However, the molecular mechanisms of signaling pathways that are controlled by Tyr phosphorylation remain to be clarified because the substrate proteins for Tyr phosphorylation have not been identified.
To understand the Tyr phosphorylation signaling, it is important to identify PKs and their substrate protein for Tyr phosphorylation. Although PKs are one of the largest gene families, representing ϳ4% (more than 1,000) of all the genes in Arabidopsis and rice, the biochemical characteristics of most plant PKs are unclear. In our previous study (17), we demonstrated the Ser/Thr autophosphorylation activity using a highthroughput profiling method combining the Arabidopsis 759 PKs array that was produced using a wheat cell-free system, and a luminescent method "AlphaScreen." Here, by modifying this approach, we have identified and characterized 38 Tyr autophosphorylation PKs. Among them, we focused on angiosperm-specific CRK2, CRK3, and CRK8, and screened for the substrate protein for Tyr phosphorylation using a TFs protein array (18) and a pulldown assay using cellular extracts. We identified 6 substrate proteins in total. Cell-based transient expression assay and analysis of crk2 and crk3 mutants revealed that CRK2 and CRK3 were able to phosphorylate Tyr residue(s) of substrate proteins such as TBBs or ERF13 in cells or plants. In addition, the five proteins tested from the CRK family showed Tyr-auto/trans-phosphorylation activity in vitro. These findings suggest that CRK proteins possibly function as PKs for Tyr phosphorylation in plants.

Experimental Procedures
General-The following procedures were previously described (19 -22): wheat cell-free protein production, splitprimer PCR for construction of the DNA templates, parallel syntheses of mRNAs and their translated proteins, protein biotinylation, purification of synthesized proteins, hydrolysis of radioisotope-labeled protein, and quantification of proteins synthesized using densitometer scans of Coomassie Brilliant Blue-stained proteins or radiolabeled proteins.
Analysis of Protein Kinase Autophosphorylation Using a Luminescent Method-In vitro autophosphorylation assays were carried out as previously described with slight modifications (17). Ser/Thr or Tyr autophosphorylation was detected by anti-phospho-Ser/Thr (Upstate Biotechnology, Lake Placid, NY) or anti-phospho-Tyr antibody (4G10) (Millipore), respectively. All data represent the average of two independent experiments and the background for each experiment was controlled using the relevant non-biotinylated PK. For in vitro dephosphorylation assays, autophosphorylated biotinylated PKs were incubated with crude FLAG-PTP1 at 26°C for 60 min. Tyr autophosphorylation was detected by anti-phospho-Tyr antibody (4G10). All data represent the average of three independent experiments, and dephosphorylation efficiency was defined as the ratio of PTP1-treated kinases to untreated kinases.
Phosphoamino Acid Analysis-The phosphoamino acid analysis was performed according to a previous method (23). Biotinylated PKs were purified using streptavidin Magne-Sphere Paramagnetic Particles (Promega), and incubated at 30°C for 30 min in a total volume of 50 l consisting of 50 mM Tris-HCl (pH 7.6), 37 kBq of [␥-32 P]ATP, 100 mM potassium acetate, 10 mM MgCl 2 , and 1 mM DTT. Then, biotinylated PKs were hydrolyzed using 6 N HCl at 110°C for 4 h. After drying the product, each amino acid was separated by thin-layer chromatography (TLC) using ethanol:ammonium hydroxide:water at a ratio of 105:42:6 (v/v) and each 32 P-labeled amino acid was detected by autoradiography.
In Vitro Phosphorylation and Dephosphorylation Assays-For in vitro kinase assays, biotinylated proteins and FLAGtagged proteins were purified using streptavidin Magne-Sphere Paramagnetic Particles and anti-FLAG M2-agarose (Sigma), respectively. Then, in vitro kinase assays were carried out in a total volume of 50 l consisting of 50 mM Tris-HCl (pH 7.6), 100 mM potassium acetate, 10 mM MgCl 2 , 1 mM DTT, and 100 M ATP or 37 kBq of [␥-32 P]ATP at 26°C for 30 -60 min. Dephosphorylation by AtPTP1 was carried out in a total volume of 50 l consisting of 50 mM Tris-HCl (pH 7.5), 10 mM MgCl 2 , 2 mM DTT, 0.01% Brij 35, and 50 ng of purified FLAG-AtPTP1 at 26°C for 30 -60 min. For immunoblotting, antiphospho-Tyr antibody (4G10) was used to detect phosphotyrosine. Biotinylated proteins were detected with Alexa Fluor 647 streptavidin conjugate (Invitrogen). The chemiluminescent signal, fluorescent signals, and filmless autoradiography were detected using an ImageQuant LAS-4000 mini biomolecular imager (GE Healthcare), a Typhoon 9400 imager (GE Healthcare) with a 633-nm laser and a 670-nm BP30 emission filter, and a Typhoon FLA 9000 phosphorimager (GE Healthcare), respectively.
In-Gel Protein Digestion-Coomassie Brilliant Blue-stained protein bands were excised from the SDS-PAGE gels and further destained using 50% (v/v) acetonitrile in 100 mM ammonium bicarbonate (pH 8.9). Destained gel pieces in 30 l of 100 mM ammonium bicarbonate (pH 8.9) were reduced by adding 10 l of 40 mM dithiothreitol for 2 h at 37°C and alkylated by adding 10 l of 250 mM acrylamide for 30 min at room temperature. In-gel digestion of each protein was performed with 0.1 g of sequencing grade modified trypsin or chymotrypsin (Promega, Madison, WI) at 37°C for 12 h. After repeated extractions of tryptic digests from the gel with 50% (v/v) acetonitrile, 5% (v/v) trifluoroacetic acid, the solution containing the extracted peptides was concentrated using a vacuum microcentrifuge. The peptide sample was reconstituted with 0.1% (v/v) trifluoroacetic acid for mass spectrometry (MS) analysis.
Liquid Chromatography-Mass Spectrometry Analysis-Tandem MS (MS/MS) analysis was performed using the LTQ XL linear ion trap mass spectrometer (Thermo Fisher Scientific) coupled with DiNa nano LC system (KYA Technologies). Peptide separations were performed at a constant flow rate of 300 nl/min with a fused silica capillary column packed with C18 resin (75 m ϫ 15 cm). Mobile phases used for separation were 0.1% formic acid (A) and 80% acetonitrile with 0.1% formic acid (B). A gradient (2-50% mobile phase B) was applied for 25 min, followed by a 10-min wash at 100% mobile phase B, and a equilibration for 15 min with 2% mobile phase B. For the identification of autophosphorylation sites, MS/MS spectra were processed using the Proteome Discoverer software version 1.1 (Thermo Fisher Scientific). Peptide identification was performed using the SEQUEST search algorithm with the follow-ing parameters: two missed cleavages were allowed; precursor mass tolerance, 2 Da; fragment mass tolerance, 0.8 Da; static modification, propionamide (cysteine); dynamic modifications, phosphorylation (serine, threonine, and tyrosine), methionine oxidation, and pyroglutamic acid. For the identification of substrate proteins, the acquired MS/MS spectra were searched against the IPI Arabidopsis database (ipi.ARATH version 3.85) using the SEQUEST software and we required at least one highquality peptide for positive identification (peptide probability score Ն10).
Construction of Transient Expression Plasmids for Cultured Cells-Full-length cDNAs of CRK2, CRK3, and ERF13 were cloned into pDONR221 vectors via Gateway reactions. After the clone sequence was confirmed, the C-terminal HA-tagged form and kinase dead (K D ) form mutants, CRK2 (Lys 176 to Arg) and CRK3 (Lys 175 to Arg), were generated using a PrimeSTAR Mutagenesis Basal kit (Takara Bio) according to the manufacturer's instructions. Fragments containing the gene coding sequence and the HA tag were subcloned into p35S⍀-GW-NOST vectors (which we generated from the 35S⍀-sGFP vector (24)), 3 using LR clonase reaction (Invitrogen). We produced expression vectors for YFP or GFP fusion proteins by LR clonase recombination of CRK2 or CRK3 and p35S⍀-GW-YFP-NOST vectors, and ERF13 and p35S⍀-GW-GFP-NOST vectors.
Tyr Phosphorylation Analysis of Substrate Proteins and Subcellular Localization Analysis by Transient Expression-Isolation of Arabidopsis suspension-cultured cell protoplasts and polyethylene glycol-mediated DNA transfection were performed as previously described (25). For Tyr phosphorylation analysis of ERF13-GFP, 100 g of plasmid DNA of ERF13-GFP and 40 g of plasmid DNA of CRK2-HA or CRK3-HA were transfected into 2 ϫ 10 6 protoplasts. After overnight incubation in the dark, crude extracts were obtained by homogenizing protoplasts in radioimmunoprecipitation assay buffer supplemented with phosphatase inhibitor mixture (PhosSTOP, Roche) and protease inhibitor mixture (Sigma). Immunoprecipitation of ERF13-GFP was performed with 50 l of Protein G-Sepharose (GE Healthcare) and 1 l of rabbit polyclonal anti-GFP antibody (MBL), and aliquots were analyzed by immunoblot analysis using rabbit polyclonal anti-GFP antibody or anti-phospho-Tyr antibody (4G10). For Tyr phosphorylation analysis of endogenous TBBs, 40 g of plasmid DNA of CRK2-HA or CRK3-HA was used for transfection, and crude extracts were obtained in cell lysis buffer (20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, phosphatase inhibitor mixture, and protease inhibitor mixture). Immunoprecipitation of endogenous TBBs was performed with 50 l of Protein G-Sepharose and 5 l of mouse polyclonal anti-TBB antibody (TUB2.1) (Sigma), and aliquots were analyzed by Phos-Tag SDS-PAGE (100 M Phos-Tag and 100 M MnCl 2 ) and immunoblot analysis using mouse polyclonal anti-TBB antibody or anti-phospho-Tyr antibody (4G10). Expression of CRK2-HA and CRK3-HA was detected by anti-HA-HRP antibody (3F10) (Roche Applied Science). For subcellular localization analysis, 10 g of plasmid DNA was transfected to 100 l of 2 ϫ 10 4 protoplasts. YFP and GFP fluorescence was observed with a confocal laser-scanning microscope LSM5 PASCAL (Zeiss).
Identification of Tyr-phosphorylated Proteins from Extracts of Arabidopsis Cultured Cells-For substrate screening by co-immunoprecipitation, biotinylated CRK3 was attached to streptavidin MagneSphere Paramagnetic Particles (Promega). After washing, CRK3 was incubated with Arabidopsis suspensioncultured cells lysate in immunoprecipitation buffer (25 mM Tris-HCl, 150 mM NaCl, 0.5% Triton X-100, phosphatase inhibitor mixture, and protease inhibitor mixture) at 26°C for 60 min. Following the above reaction, the beads were washed 4 times with immunoprecipitation buffer, and then boiled in the sample buffer. The precipitated interacting proteins were analyzed by immunoblotting with anti-phospho-Tyr antibody, Coomassie Brilliant Blue stain, and MS analysis.
Identification of Tyr-phosphorylated Proteins Using Protein Library of Arabidopsis TFs-For substrate screening using AlphaScreen, 188 TFs were selected from the Arabidopsis TF library previously described (18). In vitro protein-protein interaction (PPI) assays were carried out in a total volume of 15 l consisting of 100 mM Tris-HCl (pH 8.0), 0.1% Tween 20, 1 mg/ml of BSA, 1 l of biotinylated TFs, and FLAG-PKs at 25°C for 1 h in a 384-well Optiplate (PerkinElmer Life Sciences). In accordance with the AlphaScreen IgG (Protein A) detection kit (PerkinElmer) instruction manual, 10 l of detection mixture containing 100 mM Tris-HCl (pH 8.0), 0.1% Tween 20, 1 mg/ml of BSA, 5 g/ml of anti-FLAG M2 antibody (Sigma), 0.1 l of streptavidin-coated donor beads, and 0.1 l of Protein A-coated acceptor beads was added to each well of the 384-well Optiplate, followed by incubation at 25°C for 1 h. Luminescence was analyzed using the AlphaScreen detection program. All data represent the average of two independent experiments and the background was controlled using a dihydrofolate reductase from Escherichia coli.
Mutational Analysis-Mutagenesis was carried out using a PrimeSTAR Mutagenesis Basal kit (Takara Bio) according to the manufacturer's instructions. The mutated genes were sequenced using an ABI PRISM 310 DNA sequencer (Applied Biosystems).
For Tyr phosphorylation analysis of endogenous TBBs, 14-day-old seedlings were homogenized in the cell lysis buffer and cleared by centrifugation at 12,000 ϫ g for 10 min. Immunoprecipitation, Phos-Tag SDS-PAGE, and immunoblot analysis were carried out as described above.

Identification of Tyr Autophosphorylation Protein Kinases
Using a Wheat Cell-free Protein Array-Autophosphorylation of PKs is an important aspect of regulatory systems such as activation or inactivation by the autophosphorylation-dependent conformational change (26). Many biochemical analyses demonstrate that almost all PKs catalyze intra/inter-molecular autophosphorylation reactions (27). Actually, in our previous report, many Arabidopsis PKs showed the Ser/Thr autophosphorylation activity (17). These findings prompt us to propose that Tyr-autophosphorylation activity might be a useful clue to identify plant PK(s) for Tyr phosphorylation. To search for Arabidopsis PKs for Tyr phosphorylation, we therefore profiled Tyr-autophosphorylation activity using a wheat cell-free based luminescent method with anti-phospho-Tyr (Tyr(P)) antibody in this study (Fig. 1A). In our previous study, more than half of the PKs that showed four times stronger luminescent signal than background had indeed the high Ser/Thr autophosphorylation activity (17). Thus, we concluded that these PKs are more likely to have autophosphorylation activity. Based on this threshold, 89 of 759 PKs were considered candidates for Tyrautophosphorylation activity by autophosphorylation profiling (supplemental Table S1). Furthermore, to further enrich PKs having the high Tyr-autophosphorylation activity, profiling data of the Tyr-autophosphorylation activity was compared with our previous Ser/Thr-autophosphorylation data (Fig. 1B). Finally, we selected 38 Tyr-autophosphorylation PK candidates that showed 4-fold stronger luminescence with anti-Tyr(P) antibody than with anti-Ser(P)/Thr(P) antibody (Tyr(P)/ Ser(P)-Thr(P)) ( Table 1). However, we cannot conclude that these PKs have high Tyr-autophosphorylation activity based on antibody detection alone as the signal value is correlated to the number of phosphorylated molecules and phosphorylation site. Furthermore, the effects of the three-dimensional structure of the PK on the accessibility and specificity of the antibodies cannot be discounted. Unfortunately, the affinity and specificity of currently available anti-Ser(P)/Thr(P) antibodies are lower than those of anti-Tyr(P) antibodies. Thus, currently available anti-Ser(P)/Thr(P) antibodies are almost certainly unable to solve these problems because of their low specificity. Nevertheless, at present, an approach to compare the luminescent signal would be probably effective as an indicator for high-throughput autophosphorylation profiling of PKs.
To confirm Tyr autophosphorylation, we performed phosphoamino acid analysis and dephosphorylation assays. By TLC, 22 of the 25 tested PKs showed phosphorylation of the Tyr residue (Fig. 1C). A phospho-Tyr residue was difficult to detect in the other 3 PKs (PK-B11, -C05, and -C06 in Fig. 1C). This might be due to the differences in PK kinetic properties. Another explanation for this might be that they are not able to autophosphorylate further because the autophosphorylation levels of these PKs had been saturated during the in vitro translation. However, the Tyr-autophosphorylation signal of all 38 PKs as observed decrease by incubation with AtPTP1 (Fig. 1D). These results suggest that these 38 PKs have Tyr-autophosphorylation activity (Table 1). Next, we randomly selected PK-A09, -B01, -B08, -B12, -C02, and -D01, and tested whether these PKs have re-autophosphorylation activity. Expectedly, all tested PKs showed activity after dephosphorylation by treatment with -protein phosphatase (Fig. 1E). In addition, we randomly selected PK-A01, -A11, -B01, -B07, -B10, and -D01, and generated K D proteins by mutating conserved Lys residues in the ATP binding sites to Arg. All the mutated PKs failed to react with anti-Tyr(P) antibody (Fig. 1F). Taken together, these results suggest that Tyr phosphorylation in PKs is due to the kinase activity itself. According to the PlantsP classification, these 38 PKs are mainly classified into five groups: GSK3/Shaggy-like PK (8 clones), MAPK (6 clones), RLK/RLCK (8 clones), casein kinase I/casein kinase I-like (CK1/CKL) (4 clones), and CRK (3 clones). Because the Tyr-autophosphorylation activity of some of the PKs classified within the GSK3/Shaggy-like PK, MAPK, RLK/RLCK, and CK1/CKL families has been identified in previous studies (14,28), suggesting that this approach is able to find PKs having the Tyr-autophosphorylation activity.
On the other hand, the CRK family, which was classified as part of the Ser/Thr-type CDPK-sucrose non-fermenting-1-related PK superfamily (29), has not been previously characterized. In the profiling data, three CRKs showed Tyr-autophosphorylation activity. This finding supports the suggestion that CRKs probably has Tyr-autophosphorylation activity and we focused on the analysis of the biochemical properties of CRKs. To confirm the Tyr-autophosphorylation sites of CRK3, we predicted phosphorylation sites using a web page based pro-gram (PhosPhAt, NetPhos 2.0). Following this, the predicted Tyr 559 residue sites were changed to Phe and the Tyr 559 -Phe (Y559F) mutant was used for in vitro kinase assays. Immunoblot analysis showed that autophosphorylation was decreased in CRK3-Y559F (Fig. 1G). Similarly to immunoblotting, luminescent analysis showed that the Tyr-autophosphorylation signal of CRK3-Y559F decreased by 80% in comparison with that of the wild-type (Fig. 1H). The CRK3-autophosphorylation site was also analyzed by MS analysis and phospho-Tyr 559 was detected (supplemental Table S2).
Substrate Screening of CRK3 by Pulldown Assay Based on a Wheat Cell-free System-To better understand the biochemical properties of CRKs, we selected CRK3 and explored the sub-strate proteins that Tyr residue(s) are phosphorylated by CRK3. Protein extracts from Arabidopsis cultured cells were used for pulldown assays with recombinant biotin-labeled CRK3 and anti-Tyr(P) antibody was used to detect Tyr phosphorylation. Some Tyr-phosphorylated proteins interacting with CRK3 were detected ( Fig. 2A) and, subsequently, Coomassie Brilliant Blue-stained bands of ϳ50 kDa molecular mass showing Tyr phosphorylation were analyzed using MS. These proteins identified TBB2, TBB7, and TBB9 (supplemental Table S3).
To confirm whether the TBBs are substrate proteins of CRK3 for Tyr phosphorylation, these three TBB proteins were used for in vitro kinase assays. We also tested whether CRK2 and CRK8 could phosphorylate these three TBB proteins because FIGURE 1. Screening of Tyr-autophosphorylation PKs using a wheat cell-free based protein array. A, autophosphorylation analysis of 759 biotinylated PKs using a luminescent system with anti-Ser(P)/Thr(P) or anti-Tyr(P) antibodies. All data represent the average of two independent experiments and the background for each experiment was controlled using the relevant non-biotinylated PK. B, relative luminescent signals between anti-Ser(P)/Thr(P) (horizontal axis) and anti-Tyr(P) antibody (4G10) (vertical axis). Black circles represent PKs that are more reactive with anti-Tyr(P) antibody than with anti-Ser(P)/Thr(P) antibody. C, phosphoamino acid analysis. Phospho-Tyr (pTyr), phospho-Ser (pSer), and phospho-Thr (pThr) in PKs were confirmed by one-dimensional TLC with [␥-32 P]ATP. The ninhydrin panel shows the location of standard Tyr(P), Ser(P), and Thr(P) (upper). The autoradiography panel shows 32 P-labeled Tyr(P), Ser(P), and Thr(P) (middle: short exposure and bottom long exposure). D, the dephosphorylation assay of 38 PKs by AtPTP1 using a luminescent system with anti-Tyr(P) antibody. All data represent the average of three independent experiments and dephosphorylation efficiency was defined as the ratio between PTP1-treated kinases and untreated ones. E, Tyr autophosphorylation of PKs was confirmed by immunoblotting with anti-Tyr(P) antibody (4G10) with or without ATP. F, in vitro autophosphorylation assay using WT or K D mutants. Biotinylated PKs were detected using Fluor 647 strepavidin. G, in vitro autophosphorylation assay of autophosphorylation site mutants. Wild-type biotinylated CRK3 and CRK3 Y559F mutants were used. H, autophosphorylation analysis of CRK3 by a luminescent system. The data presented are the average of three independent experiments. Black and gray bars represent wild-type CRK3 and autophosphorylation site mutants, respectively. CRK2 and CRK8 also showed Tyr-autophosphorylation activity. Our results show that the three CRK proteins tested can phosphorylate TBB2 and TBB7 at the Tyr residue but not TBB9 (Fig. 2B). Comparison of amino acid sequences showed that TBB2 and TBB7 have an extra C-terminal region including Tyr residue(s) (Fig. 2C), whereas TBB9 is missing this region. By mutation of the Tyr residue to Phe (Y443F or Y449F), Tyr phosphorylation of TBB7 was decreased (Fig. 2D), suggesting that the two Tyr residues are major phosphorylation sites for CRK2, CRK3, and CRK8. Interestingly, Tyr phosphorylation of ␤-tubulins in Arabidopsis and tobacco plants has already been reported (12,30,31). Our finding suggests that the three CRKs are the responsible PKs for Tyr phosphorylation of ␤-tubulins.
Substrate Screening of CRK2 and CRK3 by PPI Assay Using TFs Array-We found only two TBB substrate proteins by pulldown assays using cell extracts. Thus, we used a protein array of Arabidopsis TFs as previously described (18) to explore more substrate proteins phosphorylated at the Tyr residue(s) by CRK2 and CRK3. From the 647 TFs in the protein array, we selected 188 TFs characterized in previously published papers (supplemental Table S4). To identify substrate proteins for Tyr phosphorylation, we adopted a two-step screening procedure: 1) PPI between CRK2 or CRK3, and TFs using a luminescent method (Fig. 3A and supplemental Table S4), and 2) in vitro Tyr-phosphorylation assay using the top 20 proteins that showed the strongest PPI assay signals. After this double screening, four TFs, ethylene response factor 13 (ERF13) (At2g44840), WRKY DNA-binding protein 14 (WRKY14) (At1g30650), ERF subfamily B-4 of ERF/AP2 transcription factor 2.6 (RAP2.6) (at1g43160), and cryptochrome-interacting basic helix-loop-helix 5 (CIB5) (At1g26260), were found to be Tyr phosphorylation substrates of CRK2 or CRK3 (Fig. 3B and Table 2). Because ERF13 and WRKY14 were phosphorylated by both CRK2 and CRK3, they were used for further analysis. Tyr phosphorylation of the two substrates was not detected when K D forms of CRK2 and CRK3 were used, indicating that ERF13 and WRKY14 were phosphorylated by CRK2 and CRK3 in an activity-dependent manner (Fig. 3C). In summary, we identified a total of six substrate Tyr-phosphorylated proteins in vitro by CRK2, CRK3, and CRK8 using pulldown and PPI assays ( Table 2).
CRK Family Has Tyr Phosphorylation Activity-Next, we analyzed the biochemical properties of CRK2 and CRK3 by using ERF13 as a substrate. First, we analyzed whether CRK2 and CRK3 phosphorylate not only Tyr but also Ser/Thr residues. AtPTP1 is known to be a Tyr-specific protein phospha-  a Ser(P)-Thr(P) and Tyr(P) are the relative values of the non-biotinylated PKs. All data represent the average of two independent experiments and the background was controlled for each experiment using the relevant non-biotinylated PK. b Results of phosphoamino acid analyses (PPA). ϩ, Tyr(P) residue was detected; ND, not detected; NT, not tested. c The classification is based on the PlantP database. Ϫ, not annotated. tase (4,23). A dephosphorylation assay revealed that both CRK2-dependent and CRK3-dependent phosphorylations of ERF13 dramatically were decreased upon treatment with AtPTP1 (Fig. 3D). Furthermore, AtPTP1 significantly dephosphorylated even radioisotopically CRK-phosphorylated ERF13 proteins (Fig. 3E). In another approach, Tyr residue(s) were mutated to Phe (Fig. 3F). A significant decrease in Tyr phosphorylation was observed in Tyr 16 -Phe (Y16F) and Tyr 207 -Phe (Y207F) mutants (Fig. 3F), whereas the Tyr 16/207 -Phe (Y16F/ Y207F) double mutation completely lost Tyr phosphorylation (Fig. 3G). These mutation analyses revealed that Tyr 16 and Tyr 207 residues are the main phosphorylation sites in ERF13.

A01
The Arabidopsis genome contains eight CRK genes (Fig. 3H). To test the activity of other CRKs, we synthesized five CRKs and analyzed their Tyr-auto-and trans-phosphorylation activity. All CRKs exhibited Tyr-autophosphorylation activity, and particularly CRK2, CRK3, and CRK4 showed high activity (Fig. 3I). In addition, five CRKs could phosphorylate ERF13 (Fig. 3I), even though the degree of Tyr ERF13 phosphorylation differed. CRK2 and CRK3 showed the highest kinase activity, using ERF13 protein as a substrate, when compared with the other CRKs. Furthermore, we searched for the CRK3 orthologous gene in soybean and found Glyma07g05750 (74% similarity to AtCRK3), which we named GmCRK3 (Fig. 3H). The cDNA was cloned from soybean seedlings and GmCRK3 protein was synthesized using the cell-free system. In vitro kinase assay revealed that GmCRK3 also phosphorylated Arabidopsis ERF13 (Fig. 3J), indicating that GmCRK3 has PTK activity.

Overexpression of CRK2 and CRK3 Could Phosphorylate Exogenous ERF13 and Endogenous ␤-Tubulin in Cultured
Cells-To confirm the Tyr phosphorylation activity of CRK2 and CRK3 in cells, we used cultured Arabidopsis cells. First, we investigated the cellular localization of these proteins. Transient expression analyses revealed that CRK3-YFP and ERF13-GFP are partially localized in the nucleus of cultured cells. CRK2-YFP is likely to localize to the plasma membrane as well as to the nucleus to a lesser degree (Fig. 4A). Similar results were observed in onion epidermal cells transiently expressing CRK3-GFP (32). Next, we expressed CRK2 or CRK3 together with ERF13 in cells. Immunoprecipitation with anti-GFP followed by immunoblot analysis with antibody to phosphorylated Tyr showed that CRK3WT phosphorylated ERF13-GFP, whereas CRK2WT did not (Fig. 4B), and ERF13Y16F/Y207F-GFP was not phosphorylated by CRK3WT. Because expression of K D forms of CRK2 and CRK3 was very low in cells compared with wild-type forms, we failed to analyze K D forms (Fig. 4B). Taken together,theseresultssuggestthatCRK3,insteadofCRK2,phosphorylates the Tyr 16 and Tyr 207 sites of ERF13 in cells. In addition, we tried to investigate the Tyr phosphorylation of endogenous TBBs. Because Tyr phosphorylation of TBBs was very weak, we used an immunoprecipitation method with anti-TBB antibody to increase the detection sensitivity. A phosphoprotein mobility shift assay with Phos-tag acrylamide (33) was used because the immunoprecipitated TBBs could not be detected on normal SDS-PAGE gel incubated with heavy chain anti-TBB antibody. Due to the covalent modification of the phosphory-  JULY 3, 2015 • VOLUME 290 • NUMBER 27

JOURNAL OF BIOLOGICAL CHEMISTRY 16671
lated TBB by the Phos-tag reagent, a band of low mobility was detected (phospho-TBB in Fig. 4C). Following this, Tyr phosphorylation of the TBBs was detected by anti-Tyr antibody 4G10 (shown as TBB(pY) in Fig. 4C). A low level of Tyr phos-phorylation of endogenous TBBs was observed in control cells (Mock lane), indicating that the Tyr residue of endogenous TBBs is slightly phosphorylated as shown by a previous report (31). Overexpression of CRK2WT induced Tyr phosphoryla- FIGURE 3. Substrate identification of CRKs, and biochemical characterization of CRK2 and CRK3. A, heat map of the PPI relative AlphaScreen signals between 188 biotinylated TFs and FLAG-CRK2, FLAG-CRK3, or FLAG-cyclin-dependent kinase F;1 (CDKF;1). All data represent the average of two independent experiments and the background was controlled using dihydrofolate reductase (DHFR) from E. coli. B, in vitro Tyr phosphorylation assay using the top 20 proteins of interacting clones. After the incubation of FLAG-CRK2 or FLAG-CRK3, and biotinylated TF, Tyr phosphorylation was detected by immunoblot analysis with anti-Tyr(P) antibody (4G10) (upper panel). Biotinylated proteins were detected by Fluor 647 streptavidin (lower panel). Black arrowhead indicates the Tyr phosphorylation signal. C, in vitro kinase assay of WT or K D form of biotinylated CRK2 or CRK3 with biotinylated substrates. D and E, in vitro AtPTP1 dephosphorylation assay of biotinylated ERF13 phosphorylated by FLAG-CRK2 and FLAG-CRK3. Phosphorylation was detected by anti-Tyr(P) antibody (4G10) (D) or autoradiography using [␥-32 P]ATP (E). F and G, in vitro kinase assay of CRK2 and CRK3 with wild-type ERF13 and its Phe mutants. Eight Tyr residues were mutated to Phe in ERF13 (Y16F, Y47F, Y92F, Y122F, Y132F, Y157F, Y207F, and Y221F) and used for in vitro kinase assays (F). In vitro CRK2 and CRK3 phosphorylation site(s) of ERF13 were analyzed using its mutants (Y16F, Y207F, and Y16F/Y207F) (G). H, phylogenetic tree showing the sequence relationships between CRK proteins from Arabidopsis and Glycine max (soybean). The amino acid sequences of the Arabidopsis CRK family were retrieved from the TAIR and GreenPhyl databases. The amino acid sequences for the soybean homolog of Arabidopsis CRK were retrieved by BLAST searches against the Phytozome and GreenPhyl databases. A phylogenetic tree was constructed using ClustalW and the Neighbor joining (NJ) method using the tools of the DDBJ program. The scale bar indicates 0.   tion of endogenous TBBs (CRK2 WT lane), whereas that of CRK3WT did not (CRK3 WT lane). These results suggest that CRK2, but not CRK3, phosphorylates endogenous TBB in cultured cells.
Tyr Phosphorylation of TBBs Was Decreased in CRK2-and CRK3-deficient Mutants-To further clarify Tyr phosphorylation of endogenous CRK2 or CRK3 in plants, the Tyr phosphorylation level of TBBs in CRK2-and CRK3-deficient mutants (crk2 and crk3 respectively) was analyzed. CRK2 gene expression dramatically decreased in the crk2 mutant (Fig. 5A). The CRK3 gene expression level in the crk3 mutant was 50% lower than that in the Col-0 wild-type. Protein levels of the endogenous TBBs were almost the same as in wild-type plants (right panel in Fig. 5B). Using the phosphoprotein mobility shift assay, we observed that Tyr phosphorylation levels of endogenous TBBs were lower in crk2 and crk3 mutants than in wild-type plants (left panel in Fig. 5B). This result suggests that in Arabidopsis plants, Tyr phosphorylation of TBBs is mainly catalyzed by CRK2 and CRK3. Taken together, these data strongly suggest that Arabidopsis CRKs function as PKs for Tyr phosphorylation in plants.

Discussion
In this study, we profiled the autophosphorylation activity of 759 PKs by combining a wheat cell-free system and a high-throughput AlphaScreen method, and identified 38 PKs with Tyr-autophosphorylation activity. The 38 PKs were mainly classified into five groups: GSK3/Shaggy, MAPK, RLK/RLCK, CK1/CKL, and CRK. Like in mammals, several plant MAPKs and GSK3/Shaggy-like kinases have already been reported to show Tyr-autophosphorylation activity (14,34,35). Indeed, Tyr phosphorylation sites of most of these PKs have been identified already by MS analysis (6,36). In mammals, CK1 is a well conserved Ser/Thr-specific PK (37). In our study, four CK1/CLKs showed Tyr-autophosphorylation activity and Arabidopsis CKL9 has previously been reported to have a dual-specific kinase activity (28). Therefore, the biochemical properties of CK1/CKLs might be different in plants from those in mammals. Many previous studies have suggested that several RLK/RLCKs such as BRASSINOSTEROID-INSENSITIVE 1 (BRI1) and BRI1-associated kinase 1 (BAK1) have Ser-, Thr-, and Tyrautophosphorylation activity (38,39). Our results revealed that the novel eight RLK/RLCKs identified have Tyr-autophosphorylation activity. However, BRI1 and BAK1 did not show a strong Tyr-autophosphorylationsignal.ThelackofTyr-autophosphorylation signal in our study might be because full-length BRI1 and BAK1 may require a brassinosteroid ligand and/or BRI1/ BAK1-heterodimerization for the activation (40 -42) or, alternatively, the three-dimensional structure of the two PKs might disturb the accessibility of the anti-Tyr(P) antibody.
By autophosphorylation profiling and in vitro kinase assay, we showed that all tested CRKs, five Arabidopsis CRKs, and one soybean GmCRK3 have Tyr-autophosphorylation and transphosphorylation activities. A previous study using MS analysis revealed that Tyr phosphorylation was observed at the Tyr 52 of CRK4 (6). Furthermore, we identified a new Tyr-autophosphorylation site (Tyr 559 ) of CRK3 in addition to the previously reported Ser/Thr-autophosphorylation sites (43). These findings suggest that the CRK family is a novel angiosperm-specific PK family for Tyr phosphorylation. CRK family members have a Ser/Thr kinase domain and a degenerate calcium-binding EFhand motif, and these structures resemble those found in Ser/ Thr-specific CDPKs (29). Previous studies have shown that the kinase activity of Arabidopsis CRK1, and rice and tobacco CRKs is enhanced by calcium/calmodulin but that of CRK3 is not (44 -49). In this study, CRK2 and CRK3 showed the highest auto/trans-phosphorylation activity among the Arabidopsis CRK family members. Similarly to CRK3, CRK2 might not require calcium/calmodulin for activation. The biochemical and biological functions of CDPKs have been well characterized, and these PKs have been shown to play an important role in the signaling pathways of the abscisic acid, stress and metabolic responses (50,51). Recent studies have shown that CPK16, CPK18, and CPK28, phylogenetically closely related to CRKs, phosphorylated Tyr residues in plants, and CPK28 had Tyr-autophosphorylation activity (5,15,16,36,52). In addition, Tyr phosphorylation was also observed in other CDPKs. Based on these findings, we conclude that CRKs and several CDPKs might have Tyr-phosphorylation activity.
Previous studies have already shown that the Tyr residues of ␣and ␤-tubulin are phosphorylated in Arabidopsis and tobacco, but the phosphorylation site(s) and the kinases responsible for phosphorylation remain unknown (30, 31). By pulldown assays using protein lysate from cultured cells and in vitro kinase assays, we identified the Tyr residues of TBB7 that are phosphorylated by CRK2, CRK3, and CRK8. A cell-based transient assay revealed that CRK2 could phosphorylate Tyr residues of TBBs. Furthermore, Tyr phosphorylation of endogenous TBBs decreased in crk2 and crk3 mutants. Our results strongly suggest that CRKs are the responsible kinases for the Tyr phosphorylation of TBBs. Pharmacological approaches showed that PTK and PTP inhibitors can alter the stability and orientation of the microtubules (MTs) (12,13). In mammals, membrane-associated TBBs were phosphorylated by membrane-associated pp60c-src in nerve growth cones membranes (53). In addition, pp60c-src-dependent Tyr phosphorylation of TBB inhibited the polymerization of MTs (54). However, no significant difference in growth was observed between crk2 or crk3, and wild-type Col-0 plants under normal conditions. This might be because other CRKs might play complementary roles. The role of Tyr phosphorylation of TBBs by CRKs remains unclear. The AtGenExpress global stress expression data set revealed that expression of CRK2 increased under osmotic stress (55). Thus, CRK2-dependent Tyr phosphorylation might be involved in MT depolymerization or MT orientation maintenance under osmotic stress (56).
In some cases, phosphorylation of TFs can provide a key link between cell signaling and the control of gene expression. By PPI and in vitro kinase assays, we identified four Tyr-phosphorylated TFs that are substrate proteins of CRK2 and CRK3. Our approach, which combines a wheat cell-free system and a highthroughput AlphaScreen method, is able to analyze the direct interactions between the PK and the substrate protein with a high sensitivity level, as shown by the previous reports (57,58). We hypothesized that this approach might be applicable to identify substrate proteins or kinases involved in phosphorylation. By dephosphorylation analysis and mutational analysis of the identified substrate proteins, we showed that CRK2 and CRK3 could phosphorylate ERF13 Tyr residues (Tyr 16 and Tyr 207 ). In contrast, CRKs-dependent phosphorylation of Ser or Thr residues was hardly observed. These results suggest that CRKs catalyze mainly the trans-phosphorylation of ERF13 Tyr residues. To uncover whether CRK2 and CRK3 can preferentially trans-phosphorylate the Tyr residue of a substrate protein, other substrate proteins would need to be found because it is not possible to clarify the essential biochemical properties of PKs by using pseudosubstrates such as myelin basic protein or peptide. We found only two substrates, ERF13 and TBB7, which Tyr residues were phosphorylated. Thus, we could not confirm whether CRK2 and CRK3 preferentially phosphorylate Ser/Thr or Tyr residues because our data from these two substrate proteins is very limited.
Using a cell-based transient assay, we showed that CRK3 could phosphorylate the Tyr residues (Tyr 16 and Tyr 207 ) of ERF13-GFP in cells. However, CRK2-dependent Tyr phosphorylation was below the detection limit in this study. Using subcellular localization analysis, we observed that CRK2 was mainly localized to the plasma membrane and slightly to the nucleus. However, cytoplasmic localization might also be possible because we did not compare its localization with a plasma membrane-localized marker protein. Among the CRK family members, six CRKs (1, 2, 3, 5, 7, and 8) were predicted to have a myristoylation/palmitoylation motif in the N terminus (29) and a predicted nuclear localization signal motif was found in all CRKs (59). In fact, tomato CRK1 and CPK/CRK-related tobacco CPK5 that have a myristoylation/palmitoylation motif in the N-terminal are localized to the plasma membrane, whereas motif mutants of these are localized to the nucleus and cytoplasm (60,61). These findings suggest that CRKs have the potential for nuclear and membrane localization and are able to phosphorylate nuclear and/or membrane proteins.
Previous studies (62) have shown that overexpression of ERF13 causes growth retardation, and increases abscisic acid and glucose sensitivities in Arabidopsis. In addition, expression of CRK3 is induced by abscisic acid treatment (53). These findings suggest that Tyr phosphorylation of ERF13 by CRK3 might be involved in the regulation of abscisic acid signaling. However, unfortunately, direct downstream target genes of ERF13 remain unknown. Both CRK3 and CRK2 could phosphorylate WRKY14, and CRK3 phosphorylates CIB5 in vitro. Thus, it is possible that both CRK3 and CRK2 play a role in the phosphorylation of these TFs in plants. On the other hand, previous studies have also suggested that some CRKs are involved in the signal transduction of different extracellular stimuli such as salt, low/high-temperature stress, wounding, and phytohormones in Arabidopsis, tobacco, tomato, and pea (61,(63)(64)(65)(66)(67). Phenome analysis based on genetic techniques might provide significant information regarding the role of CRK-dependent Tyr phosphorylation in plants.
Although many previous studies have strongly suggested the existence of plant PTKs, evidence linking a particular PK to Tyr phosphorylation of substrates in vivo has been missing. In this study, we demonstrate that Arabidopsis has an angiosperm-specific CRK family involved in Tyr phosphorylation. Furthermore, we have identified six substrate proteins for CRKs. Our results provide the first direct evidence that plant PKs are able to phosphorylate substrate proteins on Tyr residues, and that typical TFs are the target of CRKs as observed in mammals. Furthermore, our data supports the identification of CRKs as the PKs that phosphorylate tubulin in vivo. In future studies, the functions of CRKs will be investigated to further clarify the role of Tyr phosphorylation in plants.