Protein Kinase CK2 Interacts at the Neuromuscular Synapse with Rapsyn, Rac1, 14-3-3γ, and Dok-7 Proteins and Phosphorylates the Latter Two*

Background: Knowledge about the role of the protein kinase CK2 at neuromuscular synapses is limited. Results: CK2 binds different postsynaptic proteins, Rapsyn, Rac1, 14-3-3γ, and Dok-7, and phosphorylates two of them. Phosphomimetic Dok-7 mutants improve neurotransmitter receptor aggregation. Conclusion: Phosphorylation of synaptic proteins by CK2 appears important for postsynaptic neurotransmitter receptor aggregation. Significance: Better understanding of signaling cascades involved in aggregation of neurotransmitter receptors offers possibilities for therapeutical intervention. Previously, we demonstrated that the protein kinase CK2 associates with and phosphorylates the receptor tyrosine kinase MuSK (muscle specific receptor tyrosine kinase) at the neuromuscular junction (NMJ), thereby preventing fragmentation of the NMJs (Cheusova, T., Khan, M. A., Schubert, S. W., Gavin, A. C., Buchou, T., Jacob, G., Sticht, H., Allende, J., Boldyreff, B., Brenner, H. R., and Hashemolhosseini, S. (2006) Genes Dev. 20, 1800–1816). Here, we asked whether CK2 interacts with other proteins involved in processes at the NMJ, which would be consistent with the previous observation that CK2 appears enriched at the NMJ. We identified the following proteins to interact with protein kinase CK2: (a) the α and β subunits of the nicotinic acetylcholine receptors with weak interaction, (b) dishevelled (Dsh), and (c) another four proteins, Rapsyn, Rac1, 14-3-3γ, and Dok-7, with strong interaction. CK2 phosphorylated 14-3-3γ at serine residue 235 and Dok-7 at several serine residues but does not phosphorylate Rapsyn or Rac1. Furthermore, phosphomimetic Dok-7 mutants aggregated nicotinic acetylcholine receptors in C2C12 myotubes with significantly higher frequency than wild type Dok-7. Additionally, we mapped the interacting epitopes of all four binding partners to CK2 and thereby gained insights into the potential role of the CK2/Rapsyn interaction.

Earlier, we reported that protein kinase CK2 (formerly called casein kinase 2) interacts and co-localizes with the receptor tyrosine kinase muscle specific receptor tyrosine kinase (MuSK) 3 at the neuromuscular junction (NMJ) (1). MuSK bind-ing partners are of fundamental interest at this synapse because MuSK activation is known to be responsible for the aggregation of AChRs, which represents a hallmark of postsynaptic apparatus formation (2). Protein kinase CK2 is a ubiquitously expressed serine/threonine kinase composed of four subunits: two catalytically active ␣ and two regulatory ␤ subunits (3). Earlier, we observed CK2-mediated phosphorylation of serine residues 680 and 697 within the cytosolic kinase domain of MuSK in vitro (1). MuSK activity was not changed, but AChR stability appeared to decrease in the presence of CK2 inhibitors (1). Moreover, muscle-specific CK2␤ knock-out mice develop a myasthenic phenotype likely due to impaired muscle end plate structure and function (1). Although CK2 is known to phosphorylate hundreds of proteins (3), up to now no more targets have been reported in a neuromuscular context.
Here, we asked whether protein kinase CK2, previously observed to be enriched at NMJs (1), might interact with other synaptic proteins and whether CK2 might be involved in their phosphorylation. We identified Rapsyn, Rac1, 14-3-3␥, and Dok-7 as binding partners of protein kinase CK2.
The cytosolic scaffold protein Rapsyn mediates the aggregation of AChRs in stoichiometric fashion (4 -6). It is known to be composed of seven amino-terminally localized tetratricopeptide repeats (TPRs) and a carboxyl-terminally localized coiled coil and RING domain (7), respectively. Rapsyn perfectly matches the localization of AChRs at NMJs (8,9).
It is known that AChR clustering in myotubes also depends on the small GTPases Rac1 and Cdc42 (10). Previous data show that for AChR clustering the activation of Rac1 and Cdc42 is required and that this activation is most likely controlled via reorganization of the actin cytoskeleton (11). The 14-3-3 proteins comprise several isoforms and are known to regulate cellular processes by binding signal transduction proteins. Forced expression of 14-3-3␥ in myotubes or muscle fibers induces both the specific repression of synaptic gene transcription and morphological perturbations of the NMJ (12). Dok-7 is the only member of the Dok family of cytoplasmic molecules that induces the aneural activation of MuSK and the subsequent clustering of AChRs in cultured myotubes (13). Like the other members of the Dok family, Dok-7 is composed of a pleckstrin homology (PH) domain and phosphotyrosine binding domain in its amino-terminal portion and Src homology 2 domain target motifs in its carboxyl-terminal region (13).
To further characterize the interaction of CK2 with these key players at the NMJ, we identified the epitopes within Rapsyn, Rac1, 14-3-3␥, and Dok-7 that interact with the protein kinase CK2. Additionally, CK2 even phosphorylated 14-3-3␥ and Dok-7 but not Rac1 and Rapsyn in vitro. For Rapsyn, we predicted serine 224 as a potential CK2 target in vivo, and we demonstrated that a phosphomimetic Rapsyn mutant was able to bind CK2 with higher affinity. Serine residue 235 of 14-3-3␥ was identified to be a CK2 phosphorylation target site. Tellingly, Dok-7 was strongly phosphorylated at several serine residues by CK2. Dok-7 phosphomimetic mutants improved AChR aggregation on C2C12 myotubes.

Experimental Procedures
Plasmid Constructs-For GST pulldown experiments and co-immunoprecipitation studies, the coding sequences of various mouse genes were cloned into pcDNA3-6HA to generate the respective expression plasmids. As template, cDNA of mouse hind limb muscle or C2C12 myotubes was used to PCRamplify the different DNA fragments with appropriate primer pairs. Expression plasmids for MuSK were described previously (1). The generation of full-length CK2␣ and CK2␤ fused to GST using the vector pGEX-KG and all other CK2 expression plasmids have been described (1).
For epitope mapping studies, different epitopes of mouse Dok-7, 14-3-3␥, Rapsyn, and Rac1 were amplified by PCR and cloned into pGEX-KG to generate GST fusions, and the recombinant GST fusion proteins were purified from Escherichia coli. For generating mutants (carboxyl-terminal truncations) of Dok-7, 14-3-3␥, Rapsyn, and Rac1, appropriate primers were used. For in vitro CK2 phosphorylation studies, different PCRamplified epitopes of mouse Dok-7 and 14-3-3␥ were fused to GST using pGEX-KG and the restriction sites BamHI and HindIII or BamHI and XhoI.
Substitutions of serine or threonine residues by alanines, aspartates, or glutamates were introduced into the full-length Dok-7, 14-3-3␥, or Rapsyn plasmids after cloning into pGEX-KG or pcDNA3-6HA and in the Dok-7 fragments aa238 -282 and aa283-347 after insertion into pGEX-KG using the QuikChange XL site-directed mutagenesis kit (Stratagene) or the Q5 site-directed mutagenesis kit (New England Biolabs) following the manufacturers' protocols. All mutations were verified by sequencing (LGC Genomics) after each cloning step. Phosphorylation site predictions were conducted using Kinase-Phos 2.0 (14).
For detection of proteins in Western blot analysis, monoclonal antibodies directed against the myc epitope (Cell Signaling Technology), the hemagglutinin epitope (Santa Cruz Biotechnology), CK2␤ (Olaf-Georg Issinger, Odense, Denmark), Dok-7 (Santa Cruz Biotechnology), and GAPDH (Santa Cruz Biotechnology) served as primary antibodies, and anti-mouse or -rabbit Ig-coupled horseradish peroxidase (Cell Signaling Technology) was used as secondary antibodies. For detection, the ECL detection system (Amersham Biosciences) was used.
Gastrocnemius and soleus muscles were dissected from wild type mice, dissolved in 3 ml of lysis buffer (1 M HEPES (pH 7.9), 1 M KCl, 0.5 M EDTA, 1 M DTT, and 10 g/ml aprotinin and leupeptin) supplemented with Nonidet P-40, and homogenized on ice. 250 l of 5 M NaCl were added, the solution was rotated for 15 min at 4°C and centrifuged, and the supernatant was frozen in liquid nitrogen and stored at Ϫ80°C.
Immunoprecipitation and GST Pulldowns-For each extract of transiently transfected HEK293 cells, the expression level of the desired protein was analyzed by Western blotting. Then, for co-immunoprecipitation studies, 1 l of a monoclonal antibody against the myc epitope was added, and the samples were incubated overnight at 4°C under constant rotation. Next, 20 l of PBS-equilibrated protein A-Sepharose CL-4B beads (Amersham Biosciences) were added, and incubation continued for another 2-4 h at 4°C. After washing the beads in 50 mM HEPES (pH 7.5), 150 mM NaCl, 1 mM EDTA, 10% glycerol, 1 mM PMSF, 10 g/ml leupeptin, and 10 g/ml aprotinin, the precipitated proteins were analyzed by SDS-PAGE and Western blotting.
For GST pulldowns, BL21 bacteria were transformed with the respective constructs, grown in 200-ml cultures at 37°C until A 600 was between 0.4 and 0.6, and induced by 1 mM isopropyl 1-thio-␤-D-galactopyranoside for 4 -6 h at 37°C to express the GST fusion proteins. Bacteria were collected by centrifugation, incubated in sonication buffer (50 mM NaH 2 PO 4 , 300 mM NaCl, 10 g/ml leupeptin, 10 g/ml aprotinin, 1 l/ml Triton X-100, 10 g/ml DNase I, and 15 units/l Lysozyme) at 4°C for 30 min, lysed by sonication for a total time of 1 min, and centrifuged. The supernatants containing the GST fusion proteins were supplemented with 30 l of equilibrated glutathione beads and incubated under constant rotation at 4°C for 2-12 h. After washing three times with washing buffer (4.3 mM Na 2 HPO 4 , 1.47 mM KH 2 PO 4 , 137 mM NaCl, and 2.7 mM KCl), an aliquot of the beads, now carrying the GST fusion protein, was incubated with the extract from HEK293 cells expressing the desired protein after transient transfection. After washing the beads another three times with washing buffer containing 137 mM NaCl or 20 mM NaCl, proteins bound to the beads were analyzed by SDS-PAGE and Western blotting.
For purification of recombinant Dok-7-and Rapsyn-GST fusion proteins, a different protocol was used. The bacterial pellet was resuspended in ice-cold STE buffer (10 mM Tris-HCl (pH 8), 1 mM EDTA, and 150 mM NaCl). Then, after adding 0.1 mg/ml lysozyme, the bacterial pellet was incubated on ice for 15 min, supplemented with 10 mM DTT and 1.2% Sarkosyl, and sonicated for a total time of 1 min. After centrifugation, the supernatant was diluted with STE buffer so that the concentration of Sarkosyl reached 0.7%, and Triton X-100 was added at a concentration of 2%. The lysate was incubated for 1 h on ice, then supplemented with PBS-equilibrated glutathione beads, and incubated with agitation at 4°C for 2-12 h. Finally, the beads were washed three times with cold PBS.
[␥-32 P]ATP Incorporation Assay-Recombinant purified full-length proteins or the indicated domains of CK2-interacting proteins were phosphorylated in vitro by CK2␣ (20 pmol; Kinase Logistics ApS, Denmark) alone or together with CK2␤ (20 pmol; Kinase Logistics ApS) in 50 mM Tris-HCl (pH 8), 10 mM MgCl2, and 1 mM DTT. After addition of [␥-32 P]ATP, samples were incubated for 30 min at 30°C; the reactions were stopped by addition of Laemmli buffer, and samples were analyzed by SDS-PAGE and autoradiography. As a positive control, CK2 activity was analyzed using a previously described synthetic peptide substrate (RRRDDDSDDD) (1). A 0.2 mM concentration of the peptide was incubated in 50 mM HEPES buffer (pH 7.5), 10 mM MgCl 2 , 0.5 mM DTT, 50 mM NaCl, and 50 M [␥-32 P]ATP (specific activity, 3500 -7800 cpm/pmol) with recombinant CK2␣ from Xenopus laevis for 30 min (verification by SDS-PAGE) at 30°C. The verification of the phosphorylation experiment by SDS-PAGE required [␥-32 P]ATP with a specific activity of 20,000 cpm/pmol. The reaction was spotted onto P81 phosphocellulose paper and washed three times in ice-cold 75 mM H 3 PO 4 . Paper-associated radioactivity was measured by scintillation counting. Control values obtained in assays without added enzyme were subtracted. For each peptide, the experiment was performed three times independently. For verification, reactions were analyzed by 14% SDS-PAGE. Radioactive bands were detected on a Molecular Imager FX apparatus (Bio-Rad).
Immunocytochemistry and Analysis of Dok-7-mediated Clustering-Electroporation with the Nucleofector TM kit V (Lonza) was used for transfection of C2C12 myoblasts. The cells were cultured on a 10-cm plate until confluence reached 60%, washed with sterile PBS, trypsinized, and harvested by centrifugation for 10 min at 90 ϫ g. The pellet was resuspended in 100 l of Nucleofector reagent, and the total amount of plasmid was added (4 g total; 3.5 g of expression construct ϩ 0.5 g of pCMV GFP-NLS). The cells were then electroporated in a Nucleofector cuvette, diluted in 15 ml of proliferation medium (DMEM, 20% FCS, and 1% penicillin/streptomycin), and distributed to 3.5-cm plates that had previously been coated with 25 l of Matrigel. After 24 h, the medium was changed to differentiation medium (DMEM, 1% penicillin/streptomycin, and 5% horse serum). Prior to staining, C2C12 cells were fixed with cold 2% paraformaldehyde for 30 min. To visualize AChR clusters, rhodamine-or Alexa Fluor 647-coupled ␣-bungarotoxin (␣-BTX) was applied at a 1:2,500 dilution in PBS, and cells were incubated at room temperature for 1 h. After three 5-min washes in PBS and 5 min of 1:5,000 DAPI staining, the cells were embedded with Mowiol. Analysis of clustering was performed as follows. Three independent transfections, each performed applying newly thawed and cultured C2C12 cells, were differentiated for 6 days with differentiation medium (DMEM, 5% horse serum, and 1% penicillin/streptomycin) and subjected to BTX staining. Myotubes showing BTX signals larger than 5 m in length in a circumferential area of 10 m around a transfected nucleus, as indicated by the green fluorescence of GFP-NLS, were counted in relation to the transfected tubes. Every tube was only counted once even if multiple nuclei were transfected to minimize confounding effects. This proportion for each condition was normalized to the respective proportion observed in pcDNA3-transfected controls from each transfection set, and the values were plotted and analyzed using GraphPad Prism.
Statistical Analysis-Data are presented as the mean values, and the error bars indicate ϮS.D. The number of biological replicates per experimental variable (n) is usually Ͼ5 or as indicated in the figure legends. The significance was calculated by unpaired two-tailed t test or as indicated by the figure legends and is provided as real p values that are categorized for different significance levels: ***, p Ͻ 0.001; **, p Ͻ 0.01; or *, p Ͻ 0.05.
Identification of the Interacting Epitope of Rapsyn with CK2: CK2 Does Not Phosphorylate Rapsyn in Vitro-To gain a better understanding of how CK2 interacts with Rapsyn, the epitope of Rapsyn that interacts with CK2␤ was mapped. Three mutants of Rapsyn were cloned, all of which carry differentially combined deletions of the RING domain, coiled coil motif, or only the amino-terminal domain until the end of TPR4 (Fig.   FIGURE 1. Searching for proteins interacting with protein kinase CK2 at the mammalian NMJ. a, proteins known to be localized at the NMJ were produced by HEK293 cells (HA-tagged) and used for GST pulldown assays together with purified recombinant GST-CK2␣ or GST-CK2␤ subunits of protein kinase CK2. Images of typical membranes after SDS-PAGE and Western blotting (WB) are presented. Note, for comparison, each time 1 ⁄10 of the input is shown. b, apart from AChR subunits, all other potential CK2 interactors were confirmed by co-immunoprecipitation (IP) studies. To this end, all protein-expressing plasmids were transfected into HEK293 cells, and lysates used for precipitation of the indicated CK2 subunits (myc tag-labeled) and analyzed for co-precipitations of CK2 interactors (HA tag-labeled). Images show membranes after SDS-PAGE and Western blotting. Note, the lower of the two bands visible for Rapsyn belongs to the antibody chain, which is detected by the secondary antibody. c-f, GST pulldown verified interaction of Rapsyn, Rac1, 14-3-3␥, and Dok-7 with endogenous protein kinase CK2␤ from wild type mouse hind limb muscles. Images show membranes after SDS-PAGE and Western blotting. cyto, cytosolic domain. Note, Rapsyn amino acid residues 1-160 are not sufficient to bind to CK2. d, both internal domains of Rapsyn, TPRs 5-7 and TPR6, are able to pull down protein kinase CK2␤. The image shows a Western blot membrane that is stained by anti-myc antibody. e, fusion proteins composed of GST and full-length Rapsyn were produced in bacteria, purified, and subjected to SDS-PAGE and Coomassie staining. The same amounts were used for in vitro phosphorylation assays as shown in f and g. f and g, [␥-32 P]ATP incorporation assays were performed with different amounts of GST-Rapsyn protein as shown in e. In f, only CK2␣ was used for kinase activity, whereas in g, CK2 holoenzyme was utilized. Note, CK2 holoenzyme phosphorylates CK2␣ and CK2␤ subunits but not Rapsyn. h, a phosphomimetic Rapsyn mutant pulls down more CK2␤ than wild type or alanine inactive Dok-7 mutants. The image shows a Western blot membrane after SDS-PAGE and staining. i, quantification of h as calculated from three independent GST pulldown assays. j, typical image of frozen hind limb cross-sections of wild type mice stained by CK2␤ and Rapsyn-specific antibodies. NMJs are labeled by Alexa Fluor 647-coupled BTX. The error bars indicate ϮS.D. *, p Ͻ 0.05; ns, not significant. 2a). The DNAs encoding these Rapsyn protein fragments were fused in-frame to GST, and recombinant fusion proteins were purified from bacteria and used for GST pulldown assays. The qualities of the purified GST-Rapsyn fusion proteins were confirmed by SDS-PAGE (Fig. 2b). Rapsyn variants were able to pull down CK2␤ as long as they contained TPRs 5-7 (Fig. 2c).
To further narrow the epitope of Rapsyn that interacts with CK2␤, we generated additional Rapsyn mutants composed only of TPRs 5-7 or TPR6, respectively. Both Rapsyn fragments were able to interact with CK2␤ (Fig. 2d). We conclude that TPR6 of Rapsyn is sufficient to mediate interaction with CK2␤. We next asked whether protein kinase CK2 phosphorylates Rapsyn. Increasing amounts of purified recombinant GST-Rapsyn fusion protein were incubated together with recombinant CK2 and subjected to a [␥-32 P]ATP incorporation assay (Fig. 2, e-g). These experiments were done by applying catalytically active CK2␣ subunit or CK2 holoenzyme. In none of these experiments was CK2 able to phosphorylate the GST part of the fusion protein (Fig. 2f). Neither CK2␣ nor CK2 holoenzyme was able to phosphorylate any Rapsyn protein, not even by using high amounts of Rapsyn (Fig. 2, f and g). When CK2 holoenzyme was used, phosphorylation of both CK2␣ and CK2␤ subunits by CK2 itself was detected (Fig. 2g). Interestingly, we identified in silico (KinasePhos 2.0) serine 224 of Rapsyn within a strongly suitable phosphorylation motif for CK2 (Fig. 2a). Although we did not detect any Rapsyn phosphorylation by CK2 in vitro, we cannot rule FIGURE 3. Rac1 interacts with protein kinase CK2 by its Switch I domain. a, schematic diagram showing the full-length Rac1 protein and its Switch regions, which change conformation on GDP/GTP exchange; its insert domain; and its prenylation motif at its carboxyl-terminal end. All Rac1 mutants truncated from their carboxyl-terminal end are shown underneath. b, image showing Coomassie-stained SDS-polyacrylamide gel of GST-Rac1 fusion proteins extracted from transfected HEK293 cells, purified, and separated by size. c, image showing a Western blot membrane of CK2␤, which was pulled down by GST-Rac1 fusion proteins as indicated. Note, a Rac1 mutant that lacks the Switch I domain does not pull down CK2␤. d, full-length Rac1 protein was fused to GST, expressed in bacteria, and purified for use in an in vitro phosphorylation assay. Increasing amounts of purified GST-Rac1 protein were verified by SDS-PAGE and Coomassie staining. e and f, the ability of CK2 to phosphorylate Rac1 was analyzed by [␥-32 P]ATP incorporation assays. Purified GST-Rac1 fusion protein was incubated with CK2␣ or CK2 holoenzyme, and phosphorylation events were detected by SDS-PAGE and autoradiography. g, typical image of frozen hind limb cross-sections of wild type mice stained by CK2␤-and Rac1-specific antibodies. NMJs are labeled by Alexa Fluor 647-coupled BTX.
out the possibility that CK2 phosphorylates Rapsyn in vivo. To this end, we investigated binding patterns of Rapsyn mutants with CK2␤. In fact, a phosphomimetic Rapsyn mutant (Ser-224 to Asp) was able to pull down more CK2␤ than wild type Rapsyn protein (Fig. 2, h and i). By immunofluorescence studies, we confirmed co-localization of Rapsyn and CK2␤ at NMJs of hind limb gastrocnemius/soleus muscle of wild type mice (Fig. 2j). Altogether, protein kinase CK2 interacts with TPR6 epitope of Rapsyn but does not phosphorylate it in vitro.
Identification of the Epitope of Rac1 That Interacts with CK2-Next, the nature of the interaction between Rac1 and CK2 was characterized. Four Rac1 mutants were generated by different truncations of wild type Rac1 at its carboxyl-terminal end (Fig. 3a). The full-size Rac1 and its mutants were fused to GST, recombinantly expressed, and purified from bacteria. The qualities of the purified proteins were confirmed by SDS-PAGE (Fig. 3b). Loss of the Switch I domain of Rac1 abolished its interaction with CK2␤ (Fig. 3c). Subsequently, we asked whether CK2 phosphorylates Rac1. To this end, increasing amounts of recombinant GST-Rac1 protein were incubated together with CK2␣ or the whole CK2 holoenzyme. The grade of purity of GST-Rac1 was verified by analyzing increasing protein amounts by SDS-PAGE (Fig. 3d). Using in vitro phosphorylation assays, we failed to detect CK2 phosphorylation of any Rac1 (Fig. 3, e and f). Only CK2 sub-units ␣ and ␤ were phosphorylated by CK2 itself (Fig. 3f). Finally, a perfect co-localization was detected between CK2␤ and Rac1 on frozen sections of wild type murine hind limb muscle by immunofluorescence staining (Fig. 3g).
Protein Kinase CK2 Phosphorylates 14-3-3␥ at Serine Residue 235-To identify serine and threonine residues of 14-3-3␥ that are phosphorylated by CK2, 14-3-3␥ was dissected into four different parts (Fig. 5a). Each protein fragment of 14-3-3␥ was purified as a GST fusion protein and in vitro phosphorylated by CK2 (Fig. 5, b and c). A deletion mutant of 14-3-3␥ consisting of FIGURE 4. Protein kinase CK2 interacts and phosphorylates 14-3-3␥. a, topography of full-length 14-3-3␥ and its carboxyl-terminally truncated variants. b, full-length 14-3-3␥ and its carboxyl-terminally truncated mutants were purified as GST fusion proteins, and their quality was verified by SDS-PAGE and Coomassie staining. c, GST-14-3-3␥ fusion proteins were utilized to pull down CK␤. Precipitates were analyzed by SDS-PAGE, and the image of a typical Western blot membrane is shown. d, recombinant GST-14-3-3␥ protein was purified from bacterial lysates, and its quality was verified by SDS-PAGE and Coomassie staining. e and f, in vitro phosphorylation assays were performed to determine whether CK2 is able to phosphorylate 14-3-3␥. CK2 was used either as the holoenzyme (f) or as its catalytically active ␣ subunit (e). Images were taken from autoradiographs. Note, 14-3-3␥ is significantly phosphorylated by CK2.
amino acid residues 226 -247 was very prominently phosphorylated as quantified by densitometric analysis of the autoradiograph (Fig. 5c). All potential CK2 target sites within the fragment of 14-3-3␥ ranging from amino acid residues 226 to 247 were identified by KinasePhos 2.0 algorithm or predicted by amino acid sequence (Table 1). 4 Serine residues at potential CK2 target sites of 14-3-3␥ were substituted by alanine residues. Only the 14-3-3␥ mutant carrying an alanine substitution at serine 235 was not phosphorylated by CK2 (Fig. 5d).
Protein Kinase CK2 Interacts with and Phosphorylates Dok-7-First, several deletion mutants of Dok-7 were generated to extract purified recombinant mutant Dok-7 proteins (Fig. 6, a and b). All Dok-7 mutant proteins were able to interact with CK2␤ as long as they contained the PH domain of Dok-7 (Fig. 6, c and d). In particular, a mutant Dok-7 protein lacking its amino-terminal end up to residue 108 (this part of the Dok-7 consists of the PH domain) failed to interact with protein kinase CK2 (Fig. 6d). Second, increasing amounts of GST-Dok-7 were incubated together with CK2␣ or CK2 holoenzyme and radioactive [␥-32 P]ATP to determine whether CK2 phosphorylates Dok-7 (Fig. 6, e-g). CK2-dependent phosphorylation of Dok-7 was easily detectable (Fig. 6, f and g). Finally, we identified Dok-7 co-localization with CK2␤ at the NMJs of frozen hind limb sections by immunofluorescence staining (Fig. 7d).

TABLE 1 Potential CK2-dependent phosphorylation sites of 14-3-3␥ and Dok-7 as predicted by KinasePhos 2.0
Prediction scores are given next to the amino acid residues. Band intensities, as extracted from in vitro phosphorylation assays using Image J, are presented for each analysed amino acid residue of 14-3-3␥ and Dok-7 phosphorylated by CK2. In some cases, several amino acid residues were mutated on the same protein fragment as indicated by the boxes within the table. np, not predicted. into six different protein fragments (Fig. 7a). Each of these Dok-7-encoding DNA fragments were fused in-frame to GST, expressed, purified as recombinant protein, and phosphorylated by CK2 in vitro. Intriguingly, except for Dok-7 aminoterminal amino acid residues 38 -157 encompassing the interaction site with CK2 and part of the phosphotyrosine binding domain, all other Dok-7 fragments were phosphorylated by CK2 (Fig. 7b). Quantitative scans of the autoradiographs revealed that Dok-7 fragments comprising amino acid residues 238 -282 and 283-347 were intensively phosphorylated by CK2 (Fig. 7c). The primary structure of Dok-7 was analyzed by KinasePhos 2.0 algorithm for CK2 target sites (Table 1), and numerous potential amino acid residues were identified (Table 1).

CK2 Interactors at the NMJ
Protein Kinase CK2 Phosphorylates Dok-7 at Many Different Serine Residues-CK2 seems to phosphorylate many amino acid residues of Dok-7 protein (Fig. 7b). We decided to identify only potential phosphorylation sites within the two Dok-7 fragments ranging from amino acid residues 238 to 347 as obviously these fragments featuring amino acid residue 238 -282 and 283-347, respectively, were the most strongly phosphorylated (Figs. 7c and 8a). To address this issue, several potential CK2 target sites within GST-Dok-7 fragment aa238 -282 were substituted by alanine residues. These purified mutant recombinant proteins were subjected to in vitro phosphorylation assays together with CK2 (Fig. 8, b-d). All band intensities of phosphorylated GST-Dok-7 fragment aa238 -282 alanine mutants were quantified (see also Table 1). It appeared likely that serines 249, 266, and 273 were all involved in CK2-dependent phosphorylation of the GST-Dok-7 fragment aa238 -282 (Fig. 8, b-d, and Table 1). We also analyzed potential CK2-dependent phosphorylation sites of the GST-Dok-7 fragment aa283-347 (Fig. 8a). A number of serine and threonine residues of the GST-Dok-7 fragment aa283-347 were substituted by alanines (Table 1). After purifying these recombinant mutant GST-Dok-7 fragments, they were used for in vitro phosphorylation assays with CK2 (Fig. 8, e-g), and the autoradiographs were scanned, normalized, and quantified (Table 1). After normal-FIGURE 6. Signaling adaptor protein Dok-7 binds by its pleckstrin homology domain to protein kinase CK2. a, Dok-7 protein presented schematically. Several deletion mutants of Dok-7 are depicted underneath. PTB, phosphotyrosine binding domain; SH2, Src homology 2 domain. b, full-length and different Dok-7 deletion mutants were purified (recombinant GST fusion proteins), and their quality was inspected by SDS-PAGE and Coomassie staining. c and d, GST-Dok-7 fusion proteins were used to pull down CK2␤. Reaction mixtures were monitored by SDS-PAGE and Western blotting. Note, the pleckstrin homology domain is necessary and sufficient for interaction with protein kinase CK2. e, full-length Dok-7 protein was fused to GST, purified, and quality-controlled by loading increasing amounts for SDS-PAGE, and analyzed by Coomassie staining. Note, the same amounts of protein were used for in vitro phosphorylation experiments as shown in f and g. f and g, [␥-32 P]ATP incorporation assays were performed by incubating CK2␣ or CK2 holoenzyme with full-length Dok-7 protein and radiolabeled [␥-32 P]ATP. After SDS-PAGE size separation, bands were detected by autoradiography. Note, CK2 is able to phosphorylate Dok-7.
ization, substitutions showed that a number of serine residues appear to be involved in CK2-dependent phosphorylation of the GST-Dok-7 fragment aa283-347, namely serine residues 289, 290, 291, 335, 336, 338, 346, and 347 (Fig. 8, e-g, and Table  1). Lack of phosphorylation was obvious by autoradiography of a Dok-7 mutant where serine residues 346 and 347 were substituted by alanines, but other substitutions also led to diminished phosphorylation that was recognized after normalization ( Fig.  8f and Table 1). Interestingly, Dok-7 proteins from different species have a highly conserved stretch of amino acid residues encompassing 327-355 (see Fig. 10).
Phosphorylation of Dok-7 by CK2 Influences the AChR Aggregation Ability of Dok-7 in Myotubes-Overexpression of Dok-7 is known to be sufficient for aggregation of AChRs in myotubes (13). We wondered whether CK2-dependent Dok-7 phosphorylation might influence Dok-7 function regarding aggregation of AChRs. To this end, we generated inactive alanine and phosphomimetic aspartate and glutamate Dok-7 mutants where we replaced several of the serine residues phosphorylated by CK2 (serine residues 249, 264, 265, 266, 273, 346, and 347). The expression of all Dok-7 mutants was confirmed by transfection into HEK293 cells and Western blotting (Fig. 9a). After transfection of the Dok-7 mutant-expressing plasmids into C2C12 myoblasts and differentiation to myotubes, we monitored AChR aggregation by BTX staining (Fig. 9b). Indeed, phosphomimetic Dok-7 mutants aggre-gated AChRs with significantly higher frequency than wild type or inactive alanine mutants (Fig. 9b). However, the morphology of Dok-7-induced AChR aggregation was not obviously altered (Fig. 9c).

Discussion
Here, we show for the first time that protein kinase CK2 strongly interacts with proteins Rapsyn, Rac1, 14-3-3␥, and Dok-7, which are all implicated in the postsynaptic apparatus of the NMJ (Fig. 1). A weak interaction with several AChR subunits was also detected but is beyond the scope of this study (Fig. 1a). Another interaction partner of protein kinase CK2 is Dsh (Fig. 1, a and b). Previously, protein kinase CK2 was reported to associate with and phosphorylate the central domain of Dsh (15). Moreover, Dsh was shown to regulate clustering of AChRs by interacting with MuSK (16). In accordance with those data, we performed a KinasePhos screen to predict CK2-dependent phosphorylated serine and threonine residues within the primary structure of Dsh. With Ͼ90% specificity, serine residues 202, 213, and 344 are likely phosphorylated by protein kinase CK2, with serine 213 having an Ͼ99% specificity. These serine residues are positioned within the central domain of Dsh and are in agreement with previously published data, which implicate this domain to be phosphorylated (15). Overexpression of Dfz2, a wingless receptor, was shown to stimulate FIGURE 7. Dok-7 is intensively phosphorylated by protein kinase CK2. a, to identify CK2 target sites within the primary structure of Dok-7, the protein Dok-7 was divided into six fragments. Full-length Dok-7 and the generated fragments are depicted. PTB, phosphotyrosine binding domain. b and c, all six Dok-7 fragments were utilized for in vitro phosphorylation assays and analyzed by SDS-PAGE and autoradiography (b). Band intensities were quantified to reflect phosphorylation degree (c). Note, almost all Dok-7 fragments were phosphorylated by CK2, and Dok-7 fragments aa238 -282 and aa283-347 exhibited the highest amounts of radioactivity (c). d, typical image of frozen hind limb cross-sections of wild type mice stained by CK2␤-and Dok-7-specific antibodies. NMJs are labeled by Alexa Fluor 647-coupled BTX.
phosphorylation of Dsh by CK2, suggesting a role for CK2 in Wnt signal cascades (15).
All protein kinase CK2 binding partners at the NMJ that are known up to now, are summarized in Table 2. Four of these were discovered by the present study.
Moreover, we provide evidence for CK2-dependent phosphorylation of 14-3-3␥ and Dok-7, and we even identified several of these phosphorylated serine residues for both proteins. Our assays failed to detect any CK2-dependent phosphorylation of Rapsyn and Rac1. However, our data do not exclude a potential phosphorylation of Rapsyn and Rac1 at the NMJ as in vivo phosphorylation of Rapsyn or Rac1 might not be detectable by an in vitro approach. In fact, KinasePhos screens did not detect any potential CK2-dependent phosphorylation site for Rac1, but two hits were revealed for Rapsyn (Ser-224, prediction score of 0.814; Ser-369, prediction score of 0.564). Moreover, potential phosphorylation of serine residues of Rapsyn was indicated by experiments with Torpedo electrocytes (17). Here, we showed that a phosphomimetic Rapsyn mutant fused to GST was able to pull down more CK2␤ protein (Fig. 2h). Conversely, for Rac1, it was reported that phosphorylation of serine residue 71 modulates downstream signaling in cultured cells (18). Whether phosphorylation of Rapsyn and Rac1 serine or threonine residues occurs at neuromuscular synapses in vivo has to be approached by future experiments.
Recently, protein kinase CK2 was identified as a MuSK-interacting protein (1). CK2 was shown to stabilize postsynaptic AChR aggregates by MuSK phosphorylation (1). In light of FIGURE 8. A systematic approach to identify amino acid residues of Dok-7 phosphorylated by CK2. a, scheme depicting the location of Dok-7 fragments aa238 -282 and aa283-347 in comparison with the topology of full-length Dok-7. PTB, phosphotyrosine binding domain. b-d, potential CK2 target sites within Dok-7 fragment aa238 -282 were predicted by KinasePhos 2.0, substituted by alanine residues, and analyzed by in vitro phosphorylation assays. Reaction mixtures were subjected to SDS-PAGE for size separation, and bands were detected by autoradiographs. Band intensities were quantified by ImageJ and are presented in Table 1. Note, several Dok-7 mutants contain more than one alanine substitution. e-g, similar to b-d but with the exception that all experiments were done with the Dok-7 fragment aa283-347. those data, it is very intriguing that CK2 appears not only to interact with other NMJ members but also phosphorylates them. For 14-3-3␥, only serine residue 235 was phosphorylated by CK2 in vitro (Fig. 5d). A comparison of the primary structure of 14-3-3␥ among different species (human, mouse, chick, frog, and zebrafish) revealed that they all have this serine 235 residue conserved (data not shown). Likewise, several serine residues of Dok-7 that were here identified to be phosphorylated by CK2 are mostly conserved within other species (Fig. 10). Evolutionary conservation of phosphorylatable amino acid residues of proteins between different species might reflect their biological importance (19).
It cannot be too strongly emphasized that this is the first description for potentially relevant serine phosphorylation of Dok-7 at the NMJ. Recently, an international phosphoproteomic effort provided evidence for phosphorylation of human Dok-7 serine residues in a different matter that might be indicative as a first hint for CK2-dependent Dok-7 phosphorylation at the NMJ (20).
To understand the biological role of the identified CK2-dependent phosphorylation events for both 14-3-3␥ and Dok-7, generation of inactive or phosphomimetic mutants of these proteins will be necessary. These mutants might be used to study their impact in cultured muscle cells on AChR clustering or morphology. As a start, in the present study, phosphomimetic mutants of Dok-7 mimicking the posttranslational phosphorylation by CK2 were generated and shown to aggregate AChRs more frequently than wild type Dok-7 in C2C12 myotubes (Fig. 9b). Of course, the concomitant expression of the endogenous 14-3-3␥ or Dok-7 proteins might influence the outcome of such experiments, but selective knockdown of both endogenous proteins by targeting UTRs of the respective mRNAs might be helpful to reduce their protein amounts in future studies. By this strategy, the biological effect of CK2-dependent phosphorylation of 14-3-3␥ and Dok-7 proteins might be elucidated.
Besides interaction and phosphorylation, the epitopes of each of the four proteins found here to interact with CK2 were mapped. The binding of CK2 to Rapsyn was abolished after Rapsyn was truncated by its TPRs 5-7 (Fig. 2c). This might be of interest because the predicted phosphorylated serine residue 224 of Rapsyn (see above) is located within TPR6. In fact, fur- Note, the graph indicates that GFP-NLS-positive myotubes aggregate AChRs much more frequently after introduction of Dok-7 phosphomimetic than with wild type or inactive Dok-7 protein. c, immunofluorescence staining of C2C12 myotubes. No obvious morphological change is detectable regardless of whether Dok-7 wild type or one of the mutants was transfected into the cells. The error bars indicate ϮS.D. ***, p Ͻ 0.001; *, p Ͻ 0.05; ns, not significant.

TABLE 2 Summary illustrating the function of individual CK2 binding partners at the NMJ
Note that this is the first report about interaction of CK2␤ with Rapsyn, Rac1, 14-3-3␥, and Dok-7. In the case of these four proteins, the citations refer to the first descriptions of these proteins at the NMJ. n.a., not applicable. ther experiments confirmed that TPR6 of Rapsyn is sufficient for interaction with CK2␤ (Fig. 2d). For Rapsyn self-clustering, only TPR1 and TPR2 are sufficient (7), whereas AChR clustering likely requires TPR8 itself and the coiled coil motif extending beyond the last TPR (7). Accordingly, it is attractive to assume that CK2-dependent phosphorylation regulates AChR clustering at several levels among others by acting on Rapsyn and as previously shown on MuSK (1) ( Table 2). In this context, it is remarkable that in patients with congenital myasthenic syndromes an insertion of a threonine amino-terminal to Ser-224 was reported (21) that might affect CK2-dependent phosphorylation of Ser-224 of Rapsyn. In fact, the prediction score for CK2-dependent Rapsyn Ser-224 phosphorylation changed from 0.81 to 0.69 by insertion of the additional threonine (KinasePhos 2.0). Binding of Rac1 to CK2 is determined by Rac1 amino acid residues 28 -47 (Fig. 3c). Only the first Switch domain of Rac1 is known to be located on this fragment between amino acid residues 30 and 40 (Fig. 3a). This and a second Switch domain flank the nucleotide binding site of Rac1. Previously, Rac1 was reported to be involved in Agrin-induced AChR cluster formation (10). Further investigation is required to understand the interaction between CK2 and the first Switch domain of Rac1 and its influence on AChR aggregation.

CK2 interactors
The importance of the phosphorylation of serine residues in consensus motifs was already reported in the regulation of 14-3-3 binding to target proteins (12). Consensus 14-3-3 recognition motifs such as RSXSXP and RX 1-2 SX 2-3 S in which at least one of the serine residues is phosphorylated were identified in many 14-3-3 ligands (22,23). Our pulldown data indicate binding of 14-3-3␥ to CK2 by its amino acid residues 166 -210 (Fig. 4c). Interestingly, in binding to CK2 it is serine residue 235 of 14-3-3␥ that is phosphorylated (Fig. 5d). Serine residue 235 of 14-3-3␥ is in close proximity to the epitope to which CK2 binds (Figs. 4a and 5a). It remains to be studied whether the interaction of CK2 is influenced by the phosphorylation status of serine residue 235 of 14-3-3␥.
Here, the most interesting finding might be the interaction between Dok-7 and CK2 because Dok-7 is also involved in human NMJ pathologies and is phosphorylated by CK2 at several serine residues ( Fig. 8 and Table 1). Recessive inheritance of mutations in Dok-7 were shown to result in a defective structure of the neuromuscular junction and to be the cause of congenital myasthenic syndromes with proximal muscle weakness (24); and later, the mutations were shown to impair Dok-7 function (25). Recruitment of two adapter proteins, Crk and Crk-like, to the carboxyl-terminal domain of Dok-7 occurs after Agrin stimulates phosphorylation of two tyrosine residues (26). Obviously, it is the PH domain within the amino-terminal area of Dok-7 that is necessary and sufficient to bind CK2 (Fig.  6, a and d). This PH domain is mandatory for Dok-7 to activate MuSK (13). Nevertheless, how the PH domain contributes to activation of MuSK is unclear (13). The PH domain of Dok-7 is farther away, and it seems questionable whether Dok-7 phosphorylation of amino acid residues 238 -347 interferes with Dok-7 PH/CK2 interaction. However, it is appealing to speculate that CK2 as a Dok-7 PH domain-interacting protein might also be important for activation of MuSK through Dok-7. MuSK itself was shown to be phosphorylated by CK2 (1). It might be challenging to differentiate the role of CK2 linked to MuSK or Dok-7 for AChR aggregation. A first step might be to overexpress the CK2-interacting epitope of Dok-7, its PH domain, in muscle cells to compete with the endogenous CK2/ full-length Dok-7 interaction. This might result in lower CK2dependent Dok-7 phosphorylation. It was reported that forced expression of Dok-7 induces aneural AChR clustering in C2 myotubes (13). Here, overexpression of phosphomimetic Dok-7 mutants was shown to induce aggregation of AChRs in C2C12 myotubes much more frequently than overexpression of wild type Dok-7 (Fig. 9b). The impact of these findings for FIGURE 10. Alignment of Dok-7 from different species. An asterisk indicates positions that have a single, fully conserved residue. A colon indicates conservation between groups of strongly similar properties, scoring Ͼ0.5 in the Gonnet PAM 250 matrix. A period indicates conservation between groups of weakly similar properties, scoring Յ0.5 in the Gonnet PAM 250 matrix. The alignment was made using the UniProt sequence alignment tool. Note, serine residues shown in red are likely to be phosphorylated by CK2.