CK2 interacts with MuSK. (A) Constructs used for baits in the yeast two-hybrid screen. They are composed of two MuSK wild-type or kinase-deficient intracellular domains fused by a flexible GE linker (MuSK2xwt, MuSK2xkd). Autophosphorylation-dependent interaction with candidates was investigated with the bait containing kinase-defective MuSK domains (MuSK2xkd). (B) Detection of MuSK2xwt and MuSKx2kd protein expression in yeast (left panel) and Cos7 cells (right panel) by Western blot. Note that only MuSK2xwt is autophosphorylated. (C–H) Interaction between MuSK and CK2 was investigated in detail by binding studies with extracts from transiently transfected HEK293 cells or cells expressing both proteins endogenously. Western blots are shown for coimmunoprecipitation of (1) MuSK2xwt by either CK2α or CK2β (C), (2) MuSK2xwt by CK2β in the presence of CK2α (D), (3) CK2α by MuSK2xwt in the presence of CK2β (E), (4) CK2α by CK2β in the presence of MuSK2xwt (F), and (5) CK2β by MuSK using extract either from nerve-derived Agrin-treated C2C12 myotubes (G), or from the synaptic area of hindlimb muscles isolated from adult mice (H). G and H show interactions between CK2β and MuSK with endogenous muscle proteins.

Mapping the interacting epitopes of MuSK and CK2β. (A) C-terminal deletion constructs for the intracellular domain of MuSK, generated and subcloned either into a bait plasmid for yeast two-hybrid studies or into a CMV-driven expression plasmid for coimmunoprecipitations. (B) C-terminal truncations of CK2β, subcloned either into a prey plasmid for yeast two-hybrid studies or fused to GST for GST pulldown. (C) Yeast two-hybrid interactions were performed to identify the epitope of MuSK interacting with full-length CK2β (left panel) or the region of CK2β that associates with MuSK2xwt (right panel). (D) Recombinant GST fusion proteins containing different CK2β epitopes were incubated with extracts of HEK293 cells transiently transfected with MuSK2xwt. The ability of any of the GST fusion proteins to pull down MuSK2xwt was detected by Western blot. At the left, one-fifth of the input is shown. (E, right panel) Coimmunoprecipitation of CK2β by different intracellular MuSK epitopes detected by Western blot. One-tenth of the extract used as input for immunoprecipitation is shown on the left panel. Note that coprecipitation of CK2β requires at least the juxtamembrane region and the kinase domain of MuSK.

CK2α and CK2β subunits colocalize with synaptic muscle membrane in vivo. (A) Typical images of cross-sections of mouse EDL, gastrocnemius, and unspecified early postnatal hindlimb muscle stained with BTX and CK2β-reactive antibody as indicated. (B,C) Images from innervated (B) or surgically denervated (C) contralateral mouse hindlimb muscle 5 d post-surgery, stained as indicated. Note the stability of synaptic CK2α and CK2β localization upon loss of nerve terminal. (D,E) CK2 is localized at least in part in subsynaptic muscle membrane. (D) CK2 is resolved at Agrin-induced, ectopic post-synaptic membranes from rat soleus, which are nerve cell- and Schwann cell-free (generated by injection of Agrin expression plasmids). (E) CK2 is resolved in P18 mouse hindlimb muscle denervated 5 d earlier. Such endplates are known to lose Schwann cells. Note that in A (only for P7), D, and E, images were obtained by confocal microscopy.

CK2 mediates phosphorylation of serine residues of MuSK. (A) Primary structure of synthetic peptides representing parts of the MuSK KI and containing potential CK2 phosphorylatable serine residues and the respective alanine substitutions. Numbering of peptides refers to the position of the amino acid residue of MuSK they start with. Peptides also containing alanine substitutions are specified. Potentially phosphorylatable serines are highlighted and shown in bold. (B) In vitro [³²P]γ-ATP incorporation into peptides shown in A. Phosphorylation of peptides was performed with the catalytic CK2 subunit (black bars) or the CK2 holoenzyme (open bars). Amount of incorporated isotope is given as picomoles (indicated above or within bars). (C) After CK2α-mediated in vitro [³²P]γ-ATP incorporation, radiolabeled peptides were verified by SDS-PAGE and autoradiography. (D) Kinetics of CK2-mediated in vitro phosphorylation of peptides presented in A. (E) Autoradiography of the purified recombinant wild-type intracellular part of MuSK and its S680/697A double mutant after in vitro phosphorylation by either CK2α or the CK2 holoenzyme. Note that only the wild-type variant of the intracellular part of MuSK is phosphorylatable by CK2. (F) Alignment of amino acids of the KI region of MuSK from different species. Areas representing KIs are underlined, serine residues 680 and 697 are boxed. Note that S697 is conserved in all shown species.

The physiological role of CK2-mediated serine phosphorylation of MuSK on Agrin-induced AChR cluster formation was examined in C2C12 myotubes. Treatment of these myotubes with nerve-derived Agrin induced AChR clusters (Fig. 5C). When CK2 activity was blocked by chemical inhibitors (10–60 μM)—like apigenin or DMAT (2-dimethylamino-4,5,6,7-tetrabromo-1H-benzimidazole), the latter of which is highly selective for CK2 over CK1 (Critchfield et al. 1997; Pagano et al. 2004)—the number of Agrin-induced AChR clusters increased in a dose-dependent fashion by up to ∼2.5-fold (Fig. 5A,C). However, the AChR clusters were smaller and highly dispersed (Fig. 5B,C), suggesting that serine phosphorylation in MuSK KI is important for normal AChR cluster formation and maintenance. Concentrations of CK2 inhibitors >80 μM were toxic.

MuSK mutants with substitutions of phosphorylatable serines by alanines, aspartates, or glutamates contribute differently to AChR clustering. (A) Exemplary images from BTX-labeled MuSK-deficient myotubes that were transiently transfected with pnlsGFP and wild-type MuSK or different mutants thereof. Clustering of AChRs was induced by nerve-derived Agrin-containing conditioned medium. In addition, myotubes were coincubated with the CK2 inhibitor DMAT as indicated. Note that regular AChR clustering was impaired (arrowheads) if a S680/697A MuSK mutant was transfected in Agrin-treated cells, while regular AChR clusters (arrows) occurred on Agrin-treated cells transfected with phosphorylation-imitating S680/697D or S680/697E MuSK mutants, even in the presence of the CK2 inhibitor. Same data were obtained using the CK2 inhibitor apigenin (data not shown). Bar, 50 μm. (B) Quantification of the effect shown in A by presenting the AChR cluster density as described in the Materials and Methods. (*) P values < 0.0001. (C) Expression of MuSK variants was confirmed by Western blot.

Knockdown of CK2 in cultured myotubes. Different siRNAs were examined regarding their ability to knockdown the mRNAs of different CK2 subunits by a luciferase-reporter assay. For that, bicistronic mRNAs bearing a full-length luciferase cDNA and the full-length cDNA encoding the respective CK2 subunit were created. (A,C,G) Transient transfections into HEK293 cells were carried out using these constructs together with specific siRNAs active against the genes encoding CK2 subunits. (B,D,H) The lack of the respective CK2 subunits at protein level was demonstrated after transient transfection in HEK293 cells by Western blot. Note that both CK2α and CK2α′ subunits are knocked down by the siRNA CK2α′-690. (E) Typical images of BTX-stained (shown in red) Agrin-treated MuSK-deficient myotubes after transient transfection of wild-type full-length MuSK together with pnlsGFP and respective siRNAs (bar, 50 μm). Note that formation of regular AChR clusters is significantly affected. Only very small and dispersed clusters form if the expression of the indicated CK2α subunit is inhibited. (F,I) Concomitant knockdown of both CK2α and CK2α′ or the CK2β subunit by siRNA corrupted cell survival countable by a strong decrease of GFP-positive MuSK-deficient cells.

Replacement of the KI epitope of MuSK by that of PDGFβ-R, ILGF-1-R, or IR. (A) Images represent typical patterns observed after transient transfection of MuSK-deficient myotubes with wild-type MuSK or with different MuSK KI mutants and subsequent immunostaining with BTX (shown in red). Note that regular formation of AChR clusters (arrows) occurred only after transfection of wild-type MuSK and of a mutant where the KI of MuSK was replaced by the KI of the PDGFβ-R. Affected AChR clustering (arrowheads) was observed if the KI of MuSK was replaced by the KI of either ILGF-1-R or IR. Transfection of a MuSK where the original KI was deleted was unable to promote any AChR clustering after treatment with Agrin (data not shown). Bar, 50 μm. (B) Quantification of the effect shown in A as described for Figure 6B. (*) P-values < 0.004. (C) Expression of MuSK KI mutants in HEK293 cells was confirmed by Western blot. (D) Amino acid residues of the KIs of PDGFβ-R, ILGF-1-R, and IR are depicted. All serine and threonine residues are shown in red. Potential CK2 phosphorylatable serine or threonine residues are also highlighted.

To examine whether CK2β is important for synaptic AChR clustering in vivo, we ablated the CK2β gene selectively in muscle precursors. To this end, we bred mouse mutants with floxed exons within the CK2β genomic locus (Buchou et al. 2003) and expressing Cre recombinase from E9 onward under the control of the human skeletal actin promoter (Schwander et al. 2003); i.e., before NMJs begin to form at E13. Mutant mice (CK2β^loxP/loxP, HSA-Cre) were born in the expected Mendelian ratio. Using the same transgene, previous experiments had demonstrated that Cre-dependent recombination is complete in all muscle nuclei (Escher et al. 2005). Consistent with muscle-specific deletion, CK2β was not detectable at synapses by immunofluorescence staining (data not shown). CK2β transcript levels in E18 and adult mutant muscles were significantly reduced but still detectable (Fig. 9F), as expected from the large (∼60%) fraction of nonmuscle nuclei present in adult muscle (Escher et al. 2005). The mutant mice did not display an obvious phenotype within the first 2 mo of age. Over the next 4 mo, however, their grip strength, as measured by the time for which they could cling upside-down to a wire grid, began to decline dramatically (grid test) (Fig. 9A), and body weight (∼5%–10%) was slightly reduced. The progressive weakness was accompanied by a significant decrease of miniature endplate currents (MEPC) of mutant muscles (Fig. 9B). In contrast, wild-type littermates retained their grip strength and MEPC amplitudes were unaltered. The vast majority of mutant NMJs at 6 mo revealed impaired morphological appearance of the synaptic AChR clusters. Specifically, the characteristic pretzel-like shape of age-matched wild-type endplates was dramatically fragmented in mutant muscles and began to further disintegrate into a spotty appearance (Fig. 9C). Interestingly, in spite of the fragmentation of the subsynaptic AChR clusters, the presynaptic terminal arborizations, as judged from immunostains for neurofilament, appeared largely intact even where postsynaptic AChR clusters had disintegrated into small spots (Fig. 9D). These changes were observed in all muscles examined; i.e., soleus, gastrocnemius, EDL, and diaphragm (data not shown). Surprisingly, measurement of CK2 kinase activity in wild-type and mutant mice of different age demonstrated a decline of kinase activity during aging in wild types (Fig. 9E). In CK2β-ablated mutant muscles, CK2 kinase activity was higher than in wild-type mice but still declined proportionally during aging (Fig. 9E).