MuSK is a substrate for CaMK2β but this interaction is dispensable for MuSK activation in vivo

Abstract

The neuromuscular junction (NMJ) is the unique interface between lower motor neurons and skeletal muscle fibers and is indispensable for muscle function. Tight control of its localized formation at the center of every muscle fiber, and maintenance throughout lifetime are sustained by muscle-specific kinase (MuSK). MuSK acts as central regulator of acetylcholine receptor clustering at the postsynapse. Localized and temporally controlled signaling of MuSK is primarily achieved by tyrosine autophosphorylation and inhibition thereof. Previous investigations suggested serine phosphorylation of the activation domain as an additional modulator of MuSK activation. Here we identified calcium/calmodulin dependent protein kinase II (CaMK2) and in particular CaMK2β as novel catalyst of MuSK activation and confirmed its capability to phosphorylate MuSK in heterologous cells. However, whereas CaMK2β absence in muscle cells reduced AChR clustering, MuSK phosphorylation was unchanged. Accordingly, we ruled out MuSK phosphorylation as the cause of synapse fragmentation in a mouse model for myotonic dystrophy type 1, in which the muscle-specific splice-variant of CaMK2β is missing, or as the cause of ataxia or delayed muscle development in CaMK2β knockout animals. Histological characterization of muscles of CaMK2β knockout mice indicated specific roles of CaMK2β in fast glycolytic versus slow oxidative muscle. Taken together, our data shows that MuSK can be phosphorylated by CaMK2β, but loss of CaMK2β is likely compensated by other CaMK2 paralogs at the NMJ.

Fig. 1

CaMK2β phosphorylates MuSK S751 in vitro. A Radiometric kinase profiling identified CaMK2α, CaMK2β and CaMK2δ as S/T kinases capable of phosphorylating a 17 amino acid peptide of the MuSK activation domain with S751 (green) located at the center. S751 phosphorylation increased strongly when three tyrosines (grey) in the peptide were phosphorylated, while activity of 242 additional S/T kinases remained basal (Supplementary Table S1). Control reactions (blue) did not contain sample peptides. B The substrate specificity of CaMK2 proteins matches the sequence of the MuSK activation loop peptide displayed in Fig. 1A (adapted from Jones et al. 2023). C The activation domain of MuSK displays sequence similarity to multiple proteins that bind to CaMK2β (adapted from Özden et al. 2024). The DFG motif (orange) is involved in ATP binding and thus not available for protein-protein interaction. Phosphorylable serine residues are marked in green, conserved residues are marked in red.

Fig. 2

Constitutively-active CaMK2β boosts MuSK phosphorylation in heterologous cells. HEK293T cells were transiently co-transfected with MuSK and wildtype (WT), constitutively-active (T287D), or kinase-dead (K43R) CaMK2β. Co-expression of MuSK and vector plasmid (-) resulted in robust MuSK autoactivation, as expected and was used for normalization. Transfection of vector plasmid alone was used as technical negative control (- Ctrl). Transfection of constitutively-active MuSK served as technical positive control (+ Ctrl). MuSK was immunoprecipitated and subjected to immunoblotting. Membranes were probed with antibodies against tyrosine phosphorylation (pY) and MuSK (A) or activation-loop serine phosphorylation (pS751) and MuSK (B) using fluorescent immunoblotting (emission at 800 nm in green, 680 nm in red). Fluorescent signal overlaps (yellow) highlight matching band sizes. Fluorescent signal was normalized to the internal standard (-). A MuSK pY was increased when MuSK was co-transfected with CaMK2β T287D compared to WT (p = 0.0043) or K43R (p = 0.016). B MuSK pS751 was increased when MuSK was co-transfected with CaMK2β T287D compared to WT (p = 0.015) or K43R (p = 0.0421). C Total lysates of each experiment were subjected to immunoblotting and expression of transfected CaMK2β (Cb-beta-1) or MuSK was analyzed in absolute values (abs. val.). α-Tubulin served as loading control. Original blots are provided in Supplementary Fig. S4. All values are represented as mean ± standard deviation and were analyzed using one-way ANOVA and Tukey’s Honest Significant Difference test, n = 4.

Fig. 3

CaMK2β is localized at the NMJ and Camk2b-KO in muscle cells results in upregulation of AChRα and CaMK2 proteins. A Cryosections from M. tibialis anterior of WT and Camk2b^-/- mice were labelled with fluorescently conjugated α-BGT (red), and antibodies against MuSK (green) and CaMK2β (Cb-beta-1, blue). Confocal image stacks were acquired using a 40x oil immersion objective and displayed as maximum intensity projections, scale bar = 5 μm. Fluorescent signals co-localize at the NMJ. B CRISPR-Cas9 was used to generate muscle cell lines that lack all splice variants of CaMK2β by targeting exon 2 (KO1) or exon 8 (KO2) of the murine Camk2b gene. Total lysates were subjected to immunoblotting. Membranes were cut and probed with antibodies against proteins required for NMJ formation (Lrp4, MuSK, AChRα) or muscle cell differentiation (Cav3). Chemiluminescent signal was normalized to housekeeping protein (GAPDH). Resulting values were normalized to the WT condition to presented as change (Δ) in protein expression. Camk2b-KO cells displayed an increase in AChR expression (p = 0.0449, n = 11). C Total lysates from Camk2b -KO cells were assayed for expression of CaMK2 proteins (panCaMK2) using anti-panCaMK2 antibodies. Camk2b-KO cell lines exhibited an increase of CaMK2 proteins other than CaMK2β (p = 0.0346, n = 5). Original blots are provided in Supplementary Fig. S4. Values are represented as mean ± standard deviation and were analyzed using Welsh’s t-test, n ≥ 5. Dots, squares and triangles represent WT, KO1 and KO2 cell lines, respectively.

Fig. 4

Camk2b-KO in muscle cells does not affect MuSK phosphorylation but reduces AChR clustering. Camk2b-KO was achieved by separately targeting exon 2 (KO1) or exon 8 (KO2) of the murine Camk2b gene in wildtype (WT) muscle cells. Differentiated muscle cells were starved for 1 h and stimulated with 0.25 nM Agrin for 1 h, as indicated. A MuSK tyrosine (pY) and serine 751 (pS751) phosphorylation were quantitatively analyzed from immunoprecipitated protein (IP) by chemiluminescent immunoblotting. B Agrin stimulation resulted in robust increase of MuSK pY in both WT and (p < 0.0001) and KO cells (p < 0.0001). Similarly, pS751 in WT (p < 0.0001) and KO cells (p < 0.0001) was increased in response to Agrin. Agrin response did not differ between WT and KO. C AChRs were pulled down using biotin-conjugated α-BGT. AChRβ pY was quantitatively analyzed by chemiluminescent immunoblotting. Agrin stimulation resulted in robust increase of AChRβ pY in WT (p = 0.0102) and KO (p = 0.0102) cells, n = 4. There was no difference in Agrin response between WT and KO cells. Original blots are provided in Supplementary Fig. S4. D Differentiated muscle cells were stimulated with 10 ng/ml cAgrin 4.8 for 8 h. AChR clusters were labelled using fluorescently conjugated α-BGT and documented on an inverted widefield fluorescence microscope using a 40x objective. Scale = 20 μm. Cluster size did not differ between stimulated and unstimulated, or WT and KO cells. Agrin stimulation increased the number of AChR clusters in WT (p < 0.0001) and KO cells (p = 0.0013). AChR clustering was reduced in Camk2b-KO compared to WT cells (p = 0.0046). Chemiluminescent signal, cluster number and size were normalized to the WT untreated condition. Values are represented as mean ± standard deviation. Values derived from the Camk2b-KO cell lines KO1 and KO2 were pooled and data were analyzed using two-way ANOVA and Šidák’s multiple comparison test, n ≥ 3. Dots, squares and triangles represent WT, KO1 and KO2 cell lines, respectively.

Fig. 5

MuSK phosphorylation is not compromised at fragmented NMJs of Mbnl1^ΔE3/ΔE3 mice. A MuSK was immunoprecipitated from M. tibialis anterior of WT and Mbnl1^ΔE3/ΔE3 mice. MuSK pY and pS751 were quantitatively analyzed using chemiluminescent immunoblotting. Expression of CaMK2 proteins was analyzed from total lysates of corresponding animals. Original blots are provided in Supplementary Fig. S5. B Chemiluminescent signals were normalized as indicated. Mbnl1^ΔE3/ΔE3 animals exhibited decreased CaMK2β_M protein levels as expected (Falcetta et al. 2024). Furthermore, pS751 in MuSK was reduced (p = 0.0090) and MuSK protein level was increased (p = 0.0016), Data were analyzed using unpaired two-tailed Student’s t-test, n = 3. C AChR, MuSK and its phospho-tyrosines 754/55 (pY754/55) or phospho-serine 751 (pS751) residues were labelled on 8 μm cryosections from M. tibialis anterior using fluorescently-conjugated α-BGT or corresponding antibodies as indicated. Images were acquired at a widefield fluorescence microscope using a 20x objective. Scale = 20 μm. D AChR, CaMK2β proteins (panCaMK2) or their phospho-threonine 286/7 residues (pCaMK2) were labelled on 8 μm cryosections from M. tibialis anterior using fluorescently-conjugated α-BGT or corresponding antibodies as indicated. Images were acquired at a widefield fluorescence microscope using a 20x objective. Scale = 20 μm. E At least 15 NMJs of WT or Mbnl1^ΔE3/ΔE3 animals were manually segmented based on AChR labels. Mean fluorescent intensity of each fluorescent channel was quantified in Fiji v1.52 and normalized as indicated. Relative fluorescent intensities did not differ between WT or Mbnl1^ΔE3/ΔE3. Data were analyzed using unpaired two-tailed Student’s t-test, n = 4. F Mean AChR intensity was analyzed separately to assess comparability of the groups. Values are represented as mean ± standard deviation. Data were analyzed using paired two-tailed Student’s t-test, since background differed on sample slides, n = 8. Mbnl1^ΔE3/ΔE3 animals exhibited an increase in AChR levels at the NMJ (p = 0.0213).

Fig. 6

Camk2b^-/- mice exhibit reduced CaMK2 proteins at NMJs, yet MuSK phosphorylation and NMJ morphology are not affected. A MuSK was immunoprecipitated from TA of WT and Camk2b^-/- mice. MuSK pY and pS751 were quantitatively analyzed using chemiluminescent immunoblotting. Expression of CaMK2 proteins was analyzed from total lysates of corresponding animals. Original blots are provided in Supplementary Fig. S5. B Chemiluminescent signals were normalized as indicated. CaMK2β_M was reduced in Camk2b^-/- compared to WT animals (p = 0.0002). Data were analyzed using unpaired two-tailed Student’s t-test, n = 4. C AChR, MuSK and pY754/55 or pS751 in MuSK were labelled on 8 μm cryosections from TA using fluorescently-conjugated α-BGT or corresponding antibodies as indicated. Images were acquired at a widefield fluorescence microscope using a 20x objective. Scale = 20 μm. D AChR, panCaMK2 proteins or their phospho-residues (pCaMK) were labelled on 8 μm cryosections from TA using fluorescently-conjugated α-BGT or corresponding antibodies as indicated. Images were acquired at a widefield fluorescence microscope using a 20x objective. Scale = 20 μm. E At least 15 NMJs of WT or Camk2b^-/- animals were manually segmented based on AChR labels. Mean fluorescent intensity of each fluorescent channel was quantified in Fiji v1.52 and normalized as indicated. Relative fluorescence of panCamK2 (p = 0.0077) and phospho-CaMK2 proteins (p = 0.0003) was reduced in Camk2b^-/- animals. Data were analyzed using unpaired two-tailed Student’s t-test, n = 4. F Mean AChR intensity was analyzed separately to assess comparability of the groups. Data were analyzed using paired two-tailed Student’s t-test, since background differed on sample slides, n = 8. G NMJs from PFA-fixed diaphragm muscle were labelled using fluorescently-conjugated α-BGT. Images were acquired at a confocal laser scanning microscope using a 40x water immersion objective. Scale = 5 μm. H Selected variables of a standardized morphometric analysis of NMJs using a-NMJ morph (Minty et al. 2020). Male WT mice exhibited larger AChR area (p = 0.0198) and a tendency toward increased endplate size (p = 0.0641) compared to female WT mice. Endplate fragmentation was not different between the groups. Data were analyzed using two-way ANOVA and Šidák’s multiple comparison test, n ≥ 3). Values are represented as mean ± standard deviation.

Fig. 7

Fiber size and MyHC expression are altered in slow fibers of Camk2b^−/− mice. Muscle from two-month-old Camk2b^−/− mice was subjected to immunostaining against three isoforms of MyHC: type I (MYH1, blue), type IIA (MYH2, red) and type IIB (MYH4, green) and Laminin (LAMA1). MyHC, type IIX was determined by negative selection. Relative amounts of fiber types and size were determined according to the protocol from Ham et al. 2020. A, E Representative fluorescence images of 8 μm sections from TA (A) and soleus (E), illustrating fiber type distribution and size. Images were acquired using an upright widefield fluorescence microscope equipped with a 20x objective, and stitched. Scale overview = 500 μm, scale detail = 50 μm. B, F Relative amounts of abundant fiber types in TA (B) and soleus (F). The relative amount of type IIX fibers was higher in soleus of Camk2b^−/− animals (p = 0.0009). Data were analyzed using unpaired two-tailed Student’s t-test. C, G Quantitative analysis of type IIA and type IIX and type IIB fiber sizes in TA (C), or type I, type IIA and type IIX in soleus (G). Type IIA (p = 0.0097) and type IIX (p = 0.0252), but not type IIB fibers are smaller in TA of Camk2b^−/− animals. Type I (p = 0.0371), type IIA (p = 0.0688) and type IIX (p = 0.0304) fibers are smaller in soleus of Camk2b^−/− animals. Data were analyzed using unpaired two-tailed Student’s t-test. D, H Fiber size distribution of type IIA, type IIX and type IIB fibers in TA (D) or type I, type IIA and type IIX fibers in soleus (H) of WT and CaMK2b^−/− mice. TA: Mean size (Mean) of all Type IIA and type IIX fibers was shifted left, while mean size of type IIB fibers shifted right (p < 0.0001) in CaMK2b^−/− animals. Maximum fiber size (Amplitude, Amp) differed in type IIX fibers (p < 0.0001). Distribution width of type IIA and type IIX (p = 0.0001), but not type IIB fibers differed in CaMK2b^−/− animals. soleus: Mean size was shifted left in type I, type IIA and type IIX fibers (p < 0.0001). Maximum fiber size differed in all three fiber types (p < 0.0001, 0.0159, 0.0001, respectively). Size distribution width differed in all three fiber types (p < 0.0001, 0.0004, 0.0059, respectively). All values are represented as mean ± standard deviation. Frequency distributions were fit with gaussian distribution models and nonlinear regressions were calculated using the least squares method. Amplitude (Amp, maximum fiber size [%]), center of distribution (Mean, mean fiber size [µm]) and standard deviation (SD, width of distribution at the center of Y [µm] were compared using the extra sum of squares F test, n = 4.