Portal de Periódicos da CAPES

MuSK Glycosylation Restrains MuSK Activation and Acetylcholine Receptor Clustering

2002; Elsevier BV; Volume: 277; Issue: 52 Linguagem: Inglês

10.1074/jbc.m208664200

1083-351X

Anke Watty, Steven J. Burden,

Muscle Physiology and Disorders

MuSK, a muscle-specific receptor tyrosine kinase that is activated by agrin, has a critical role in neuromuscular synapse formation. In cultured myotubes, agrin stimulates the rapid phosphorylation of MuSK, leading to MuSK activation and tyrosine phosphorylation and clustering of acetylcholine receptors. Agrin, however, fails to stimulate tyrosine phosphorylation of MuSK that is force-expressed in myoblasts and fibroblasts, indicating that myotubes contain an additional activity that is required for agrin to stimulate MuSK. Certain glycosyltransferases are expressed selectively at synaptic sites in skeletal muscle, raising the possibility that carbohydrate modifications of MuSK, catalyzed by glycosyltransferases expressed selectively in myotubes, may be essential for agrin to bind and activate MuSK. We identifed two N-linked glycosylation sites in MuSK, and we expressed MuSK mutants lacking one or bothN-linked sites into MuSK mutant myotubes to determine whether N-linked carbohydrate modifications of MuSK have a role in MuSK activation. We found that N-linked glycosylation restrains ligand-independent tyrosine phosphorylation of MuSK and downstream signaling but is not necessary for agrin to stimulate MuSK. MuSK, a muscle-specific receptor tyrosine kinase that is activated by agrin, has a critical role in neuromuscular synapse formation. In cultured myotubes, agrin stimulates the rapid phosphorylation of MuSK, leading to MuSK activation and tyrosine phosphorylation and clustering of acetylcholine receptors. Agrin, however, fails to stimulate tyrosine phosphorylation of MuSK that is force-expressed in myoblasts and fibroblasts, indicating that myotubes contain an additional activity that is required for agrin to stimulate MuSK. Certain glycosyltransferases are expressed selectively at synaptic sites in skeletal muscle, raising the possibility that carbohydrate modifications of MuSK, catalyzed by glycosyltransferases expressed selectively in myotubes, may be essential for agrin to bind and activate MuSK. We identifed two N-linked glycosylation sites in MuSK, and we expressed MuSK mutants lacking one or bothN-linked sites into MuSK mutant myotubes to determine whether N-linked carbohydrate modifications of MuSK have a role in MuSK activation. We found that N-linked glycosylation restrains ligand-independent tyrosine phosphorylation of MuSK and downstream signaling but is not necessary for agrin to stimulate MuSK. receptor tyrosine kinase acetylcholine receptor digoxigenin neuraminidase horseradish peroxidase Vicia villosa agglutinin Galanthus nivalis agglutinin Datura stramonium agglutinin peanut agglutinin Maackia amurensis agglutinin Sambucus nigra agglutinin cytotoxic T cell carbohydrate The formation of neuromuscular synapses requires a complex series of interactions between developing motor neurons and muscle fibers (1Sanes J.R. Lichtman J.W. Nat. Rev. Neurosci. 2001; 2: 791-805Crossref PubMed Scopus (827) Google Scholar,2Arber S. Burden S.J. Harris A.J. Curr. Opin. Neurobiol. 2002; 12: 100-103Crossref PubMed Scopus (47) Google Scholar). Agrin, a heparansulfate proteoglycan that is synthesized by motor neurons and stably deposited into the synaptic basal lamina, plays a crucial role in the formation of neuromuscular synapses (3McMahan U.J. Cold Spring Harbor Symp. Quant. Biol. 1990; 55: 407-418Crossref PubMed Scopus (580) Google Scholar, 4Wallace B.G. Bioessays. 1996; 18: 777-780Crossref PubMed Scopus (8) Google Scholar, 5Ruegg M.A. Bixby J.L. Trends Neurosci. 1998; 21: 22-27Abstract Full Text Full Text PDF PubMed Scopus (160) Google Scholar, 6Glass D.J. Yancopoulos G.D. Curr. Opin. Neurobiol. 1997; 7: 379-384Crossref PubMed Scopus (60) Google Scholar). Agrin organizes postsynaptic differentiation by stimulating MuSK, a receptor tyrosine kinase (RTK)1 that is expressed selectively in skeletal muscle and in Torpedoelectric organ, a tissue that is homologous to skeletal muscle but more densely innervated (7Jennings C.G. Dyer S.M. Burden S.J. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 2895-2899Crossref PubMed Scopus (149) Google Scholar, 8Valenzuela D.M. Stitt T.N. DiStefano P.S. Rojas E. Mattsson K. Compton D.L. Nunez L. Park J.S. Stark J.L. Gies D.R. Thomas S. Copeland N.G. Jenkins N.A. Burden S.J. Glass D.J. Yancopoulos G.D. Neuron. 1995; 15: 573-584Abstract Full Text PDF PubMed Scopus (358) Google Scholar, 9Glass D.J. Bowen D.C. Stitt T.N. Radziejewski C. Bruno J. Ryan T.E. Gies D.R. Shah S. Mattsson K. Burden S.J. DiStefano P.S. Valenzuela D.M. DeChiara T.M. Yancopoulos G.D. Cell. 1996; 85: 513-523Abstract Full Text Full Text PDF PubMed Scopus (587) Google Scholar). Agrin and MuSK are essential for synapse formation, as mice lacking agrin or MuSK fail to form neuromuscular synapses and consequently die at birth because of a failure to move or breathe (10Gautam M. Noakes P.G. Moscoso L. Rupp F. Scheller R.H. Merlie J.P. Sanes J.R. Cell. 1996; 85: 525-535Abstract Full Text Full Text PDF PubMed Scopus (802) Google Scholar, 11DeChiara T.M. Bowen D.C. Valenzuela D.M. Simmons M.V. Poueymirou W.T. Thomas S. Kinetz E. Compton D.L. Rojas E. Park J.S. Smith C. DiStefano P.S. Glass D.J. Burden S.J. Yancopoulos G.D. Cell. 1996; 85: 501-512Abstract Full Text Full Text PDF PubMed Scopus (765) Google Scholar). Importantly, muscle-derived proteins, including acetylcholine receptors (AChRs), which are concentrated in the postsynaptic membrane of muscle in wild-type mice, are instead expressed uniformly in muscle of MuSK mutant mice (11DeChiara T.M. Bowen D.C. Valenzuela D.M. Simmons M.V. Poueymirou W.T. Thomas S. Kinetz E. Compton D.L. Rojas E. Park J.S. Smith C. DiStefano P.S. Glass D.J. Burden S.J. Yancopoulos G.D. Cell. 1996; 85: 501-512Abstract Full Text Full Text PDF PubMed Scopus (765) Google Scholar). Taken together with experiments showing that agrin can stimulate multiple aspects of postsynaptic differentiation in cultured myotubes (12Wallace B.G. J. Neurosci. 1989; 9: 1294-1302Crossref PubMed Google Scholar), including the clustering and tyrosine phosphorylation of AChRs (13Wallace B.G., Qu, Z. Huganir R.L. Neuron. 1991; 6: 869-878Abstract Full Text PDF PubMed Scopus (213) Google Scholar), these results indicate that agrin stimulation of MuSK leads to clustering of critical muscle-derived proteins, including AChRs, at synaptic sites (1Sanes J.R. Lichtman J.W. Nat. Rev. Neurosci. 2001; 2: 791-805Crossref PubMed Scopus (827) Google Scholar, 4Wallace B.G. Bioessays. 1996; 18: 777-780Crossref PubMed Scopus (8) Google Scholar, 5Ruegg M.A. Bixby J.L. Trends Neurosci. 1998; 21: 22-27Abstract Full Text Full Text PDF PubMed Scopus (160) Google Scholar, 6Glass D.J. Yancopoulos G.D. Curr. Opin. Neurobiol. 1997; 7: 379-384Crossref PubMed Scopus (60) Google Scholar). The mechanisms by which agrin activates MuSK, however, are poorly understood (1Sanes J.R. Lichtman J.W. Nat. Rev. Neurosci. 2001; 2: 791-805Crossref PubMed Scopus (827) Google Scholar, 6Glass D.J. Yancopoulos G.D. Curr. Opin. Neurobiol. 1997; 7: 379-384Crossref PubMed Scopus (60) Google Scholar). Agrin stimulates the rapid tyrosine phosphorylation of MuSK in myotubes, consistent with the idea that MuSK is a receptor, or a component of a receptor complex for agrin. Nonetheless, MuSK, forced-expressed in fibroblasts or myoblasts, is not phosphorylated by agrin (9Glass D.J. Bowen D.C. Stitt T.N. Radziejewski C. Bruno J. Ryan T.E. Gies D.R. Shah S. Mattsson K. Burden S.J. DiStefano P.S. Valenzuela D.M. DeChiara T.M. Yancopoulos G.D. Cell. 1996; 85: 513-523Abstract Full Text Full Text PDF PubMed Scopus (587) Google Scholar). These results, together with a failure to detect binding between recombinant agrin and recombinant MuSK in vitro, indicate that additional components, which are expressed in myotubes and not in myoblasts or fibroblasts, are essential for agrin to activate MuSK (9Glass D.J. Bowen D.C. Stitt T.N. Radziejewski C. Bruno J. Ryan T.E. Gies D.R. Shah S. Mattsson K. Burden S.J. DiStefano P.S. Valenzuela D.M. DeChiara T.M. Yancopoulos G.D. Cell. 1996; 85: 513-523Abstract Full Text Full Text PDF PubMed Scopus (587) Google Scholar). The additional myotube-specific component(s) are not known but could include: 1) a membrane protein that binds agrin and acts as a bona fide agrin receptor; 2) a co-ligand that acts with agrin to bind and activate MuSK; or 3) post-translational modifications of MuSK that allow agrin to bind directly to MuSK (14Burden S.J. Genes Dev. 1998; 12: 133-148Crossref PubMed Scopus (170) Google Scholar). Consistent with the latter possibility, certain glycosyltransferases are expressed selectively at synaptic sites in skeletal muscle (15Scott L.J. Balsamo J. Sanes J.R. Lilien J. J. Neurosci. 1990; 10: 346-350Crossref PubMed Google Scholar), resulting in the concentration of certain carbohydrate epitopes, attached to proteins, at neuromuscular synapses (17Scott L.J. Bacou F. Sanes J.R. J. Neurosci. 1988; 8: 932-944Crossref PubMed Google Scholar). We therefore sought to determine the sites of carbohydrate modification on MuSK and whether carbohydrate modification might be essential for agrin to activate MuSK. We identifed twoN-linked carbohydrate attachment sites on MuSK and found that these N-linked sites are dispensable for agrin to activate MuSK. Carbohydrate addition, however, restrains ligand-independent MuSK activity, because mutation of bothN-linked sites increases the level of agrin-independent MuSK phosphorylation, resulting in an increase in AChR tyrosine phosphorylation and clustering. The generation of MuSK mutant muscle cell lines has been described previously (18Herbst R. Burden S.J. EMBO J. 2000; 19: 67-77Crossref PubMed Scopus (145) Google Scholar). MuSKmutant myoblasts were grown in Dulbecco's modified Eagle's medium, 4.5 mg/ml glucose, 10% fetal calf serum, 10% horse serum, 0.5% chick embryo extract, 20 units/ml recombinant mouse interferon-γ (Invitrogen), and gentamycin at 33 °C and induced to differentiate as myotubes by growing the cells at 39 °C in medium lacking chick embryo extract and interferon-γ (18Herbst R. Burden S.J. EMBO J. 2000; 19: 67-77Crossref PubMed Scopus (145) Google Scholar). The antibodies used to immunoprecipitate rat MuSK and to immunoprecipitate and detectTorpedo MuSK by Western blotting have been described previously (19Watty A. Neubauer G. Dreger M. Zimmer M. Wilm M. Burden S.J. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 4585-4590Crossref PubMed Scopus (85) Google Scholar). Rat MuSK was detected by Western blotting using polyclonal antibodies to MuSK that were kindly provided by W. Hoch (20Hopf C. Hoch W. J. Biol. Chem. 1998; 273: 6467-6473Abstract Full Text Full Text PDF PubMed Scopus (78) Google Scholar). The monoclonal antibody, 4G10, to phosphotyrosine was purchased from Upstate Biotechnology (Lake Placid, NY). Digoxigenin (DIG)-labeled lectins and N-glycosidase F were purchased from Roche Molecular Biochemicals; biotin-labeled Vicia villosaagglutinin (VVA) was purchased from E-Y Laboratories; HRP-coupled antibodies to DIG, HRP-labeled steptavidin, and neuraminidase (NA) were purchased from Sigma. Single amino acid substitutions in the rat MuSK extracellular region were generated by PCR. The mutagenic primer (forward direction) to generate MuSK N222Q was CCTGAATCCCACCAGGTCACCTTTGGTTCC; the mutagenic primer to generate MuSK N338Q was CTTGTCTTCTTCCAGACCTCCTATCCC; the mutagenic primer to generate MuSK N459Q was GATTATAAAAAAGAACAGATAACAACATTC. The mutategenic primers for N222Q were used in combination with ATGGCATCTACTGCTTGCACAG (forward) and GAGTCTTGAGTCAATCACTCGG (reverse); the mutagenic primers for N338Q were used in combination with ATGGCATCTACTGCTTGCACAG (forward) and GAGCTCCTTTACTGCCAAGC (reverse); the mutagenic primers for N459Q were CGAGGCTCTGCTGTGTAATC (forward) and GTTATTCCTCGGATACTCCAG (reverse). The mutated PCR fragments were cloned into theNarI/PflMI (N222Q),PflMI/SacII (N338Q) andSacII/BlpI (N459Q) restriction sites in rat MuSK. The mutant MuSK constructs were subsequently subcloned into theEcoRI site of the retroviral vector pBabe/puro (21Morgenstern J.P. Land H. Nucleic Acids Res. 1990; 18: 3587-3596Crossref PubMed Scopus (1903) Google Scholar). Recombinant retrovirus was produced by transfecting Bosc 23 cells with the pBabe/puro plasmids encoding wild-type or mutant MuSK, as described previously (18Herbst R. Burden S.J. EMBO J. 2000; 19: 67-77Crossref PubMed Scopus (145) Google Scholar). Myoblasts, at ∼50% confluency, were infected with retrovirus and selected subsequently in medium containing 2 μg/ml puromycin. Myotubes were stimulated with 100 nm recombinant agrin (R&D Systems) or treated with 0.1 units/ml NA for 16 h. AChRs were labeled by staining with Texas Red-conjugated α-bungarotoxin (Molecular Probes) for one h. Myotubes were washed with phosphate-buffered saline, fixed with 1% formaldehyde for 20 min, washed, mounted in Vectashield (Vector Labs) and viewed at 630× with a Zeiss Axioskop. The number of AChR clusters was determined, and the number of AChR clusters per field was calculated. Myotubes were serum-starved for several hours before stimulation with agrin (30 min) or NA (2 h). Myotubes were rinsed with phosphate-buffered saline and lysed in ice-cold lysis buffer (50 mm Tris pH 7.4, 150 mm NaCl, 1 mm EDTA, 1% Triton X-100) containing 1 mm Na3VO4, 10 mm NaF, and a mixture of protease inhibitors (Roche Molecular Biochemicals). Lysates were pre-cleared by centrifugation, and antibodies to MuSK, together with Protein A-agarose, or biotin-coupled α-bungarotoxin and streptavidin-agarose were added and incubated overnight at 4 °C. The beads were centrifuged and washed four times with lysis buffer. Bound proteins were eluted from the beads with SDS-PAGE sample buffer, and the eluted proteins were separated by SDS-PAGE (7.5% polyacrylamide gels) and blotted onto nitrocellulose (Amersham Biosciences). Protein tyrosine phosphorylation, MuSK expression, and AChR β-subunit expression were detected by probing Western blots with monoclonal antibody 4G10, polyclonal antibodies to MuSK (20Hopf C. Hoch W. J. Biol. Chem. 1998; 273: 6467-6473Abstract Full Text Full Text PDF PubMed Scopus (78) Google Scholar), and a monoclonal antibody to the AChR β-subunit (mAb124, gift of J. Lindstrom, University of Pennsylvania), respectively. Antibody binding was visualized with HRP-coupled secondary antibodies (Jackson ImmunoResearch) and ECL (PerkinElmer Life Sciences), and x-ray films were scanned with a densitometer (BioRad). MuSK was immunoprecipitated from Torpedo californica electric organ (Winckler Enterprises, San Pedro, CA), as described previously (19Watty A. Neubauer G. Dreger M. Zimmer M. Wilm M. Burden S.J. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 4585-4590Crossref PubMed Scopus (85) Google Scholar). MuSK was immunoprecipitated from myotubes as described above. Two units of N-glycosidase F (in 20–30 μl of 50 mm sodium phosphate, pH 7.4) were added to the washed protein A-agarose beads; after 24 h at 37 °C, an additional two units of N-glycosidase F were added, and the reaction was allowed to proceed for a further 24 h. Immunoprecipitated MuSK was digested with NA (50 milliunits in 50 mm sodium acetate, 2 mm EDTA, pH 4.5) for 48 h at 37 °C. MuSK was immunoprecipitated fromTorpedo electric organ with antibodies to MuSK, and the single immunoprecipitate, solubilized in SDS-PAGE sample buffer, was divided and fractionated in different lanes of a SDS-polyacrylamide gel (7.5% polyacrylamide). Individual lanes from a Western blot were probed with DIG-labeled lectins, according to the manufacturer's instructions (Roche Molecular Biosciences), and the bound DIG-lectin was visualized with HRP-coupled antibodies to DIG (Roche Molecular Biosciences) and ECL (PerkinElmer Life Sciences). ImmunoprecipitatedTorpedo MuSK was treated with NA or N-glycosidase F, and the digested immunoprecipitates were divided and fractionated in different lanes of a SDS-polyacrylamide gel; Western blots were probed either with DIG-labeled lectins or antibodies to MuSK. In a first step to determine whether MuSK is glycosylated, we isolated membranes fromTorpedo electric organ, a rich source of synaptic proteins, including MuSK. MuSK was immunoprecipitated from nonionic detergent lysates, and proteins in the immunoprecipitate were separated by SDS-PAGE and transferred to nitrocellulose. Western blots were probed with DIG-labeled lectins and HRP-coupled antibodies to DIG (Fig. 1 A). This analysis reveals that Maackia amurensis agglutinin, a lectin that recognizes terminal sialic acid attached to galactose in an α2→3 linkage (22Wang W.C. Cummings R.D. J. Biol. Chem. 1988; 263: 4576-4585Abstract Full Text PDF PubMed Google Scholar), and Datura stramonium agglutinin (DSA), a lectin that recognizes terminal galactose attached to GlcNac in a β1→4 linkage (23Crowley J.F. Goldstein I.J. Arnarp J. Lonngren J. Arch. Biochem. Biophys. 1984; 231: 524-533Crossref PubMed Scopus (158) Google Scholar), bind to MuSK (Fig. 1 A). Other lectins, such as Galanthus nivalis agglutinin (GNA), which recognizes high-mannose structures inN-linked glycans, peanut agglutinin (PNA), which recognizes the core disaccharide in O-glycans after removal of terminal sugar residues, Sambucuc nigraagglutinin (SNA), which recognizes terminal sialic acid attached to galactose in an α2→6 linkage and VVA, which binds terminal GalNac, do not bind to MuSK (Fig. 1 A and data not shown). These results demonstrate that MuSK bears carbohydrate structures that are complex but not of the high-mannose type. The specificity of MAA binding was further demonstrated by treating immunoprecipitated MuSK with NA, a glycosidase that cleaves terminal sialic acid residues (Fig. 1 B). Treatment with NA abolishes MAA binding to MuSK, while leaving DSA binding unimpaired (Fig. 1 B). Moreover, NA treatment does not expose binding sites for GNA, PNA, SNA, or VVA (data not shown). Because PNA recognizes the core of O-linked carbohydrate structures after removal of terminal sugar residues, the lack of PNA binding in NA-digested MuSK does not support the presence of O-linked carbohydrates in MuSK. We next sought to determine whether the carbohydrates recognized by MAA and DSA are linked to asparagine (N-linked) or to serine/threonine (O-linked) residues. To distinguish between these possibilities, we treated immunoprecipitated MuSK withN-glycosidase F, which removes the entire carbohydrate structure from N-linked sites and converts the asparagine into an aspartate residue (Fig. 1 C).N-Glycosidase F treatment results both in a shift in the molecular weight of MuSK, revealed by probing Western blots with antibodies to MuSK (Fig. 1 C), and in a loss of MAA and DSA binding, demonstrating that the recognized carbohydrates are linked to MuSK via asparagine residues (Fig. 1 C). We next sought to determine whether the same carbohydrate structures are present on mammalian MuSK. The substantially lower abundance of MuSK in skeletal muscle, compared with that in Torpedo electric organ, made it difficult to analyze the carbohydrates attached to mammalian MuSK by probing Western blots with lectins. Therefore, we determined whether digestion of MuSK with N-glycosidase F or NA caused a shift in molecular weight, as observed in MuSK isolated from Torpedo electric organ. We immunoprecipitated MuSK from C2 myotubes, digested MuSK withN-glycosidase F or NA and probed Western blots with antibodies to MuSK (Fig. 1 D). Incubation withN-glycosidase F or NA caused a shift in the molecular weight of mammalian MuSK, similar to that observed with TorpedoMuSK, indicating that carbohydrate structures had been removed (Fig. 1 D). Moreover, these results demonstrate that mammalian MuSK bears N-linked carbohydrate, including sialic acid, and imply that glycosylation of mouse MuSK is not fundamentally different from glycosylation of Torpedo MuSK. Two consensusN-linked glycosylation sites, Asn-X-(Ser/Thr); (X = any amino acid, except Pro), are conserved in the extracellular domain of MuSK from fish, avian, amphibian, and mammals (Asn-222 and Asn-338 in rat MuSK) (7Jennings C.G. Dyer S.M. Burden S.J. Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 2895-2899Crossref PubMed Scopus (149) Google Scholar, 8Valenzuela D.M. Stitt T.N. DiStefano P.S. Rojas E. Mattsson K. Compton D.L. Nunez L. Park J.S. Stark J.L. Gies D.R. Thomas S. Copeland N.G. Jenkins N.A. Burden S.J. Glass D.J. Yancopoulos G.D. Neuron. 1995; 15: 573-584Abstract Full Text PDF PubMed Scopus (358) Google Scholar, 24Fu A.K. Smith F.D. Zhou H. Chu A.H. Tsim K.W. Peng B.H. Ip N.Y. Eur. J. Neurosci. 1999; 11: 373-382Crossref PubMed Scopus (38) Google Scholar, 25Ip F.C. Glass D.G. Gies D.R. Cheung J. Lai K.O., Fu, A.K. Yancopoulos G.D. Ip N.Y. Mol. Cell. Neurosci. 2000; 16: 661-673Crossref PubMed Scopus (56) Google Scholar). Asn-222 is located between the second and third Ig-like domains, and Asn-338 is within the C6 box, located between the third and fourth Ig-like domains (Fig. 2 A). A third consensusN-linked glycosylation site is present in avian, amphibian, and rat MuSK (Asn-459 in rat MuSK) but absent in fish and human MuSK; in rat MuSK, Asn-459 is located within an insert, produced by alternative splicing, between the fourth Ig-like domain and the transmembrane domain (Fig. 2 A). We mutated the three consensus N-linked glycosylation sites in rat MuSK, either individually or in combination, and stably expressed the MuSK mutants in MuSK-deficient muscle cells (18Herbst R. Burden S.J. EMBO J. 2000; 19: 67-77Crossref PubMed Scopus (145) Google Scholar). We immunoprecipitated MuSK from the muscle cell lines and probed Western blots of the immunoprecipitate with antibodies to MuSK (Fig. 2 B). A MuSK double mutant (N222Q/N338Q), lacking the two conserved N-linked glycosylation sites, migrates more rapidly than wild-type MuSK in SDS-PAGE (Fig. 2 B). These results indicate that one or both of the two conserved N-linked glycosylation sites, Asn-222 and Asn-338, are glycosylated in vivo. We were unable to distinguish between these two possibilities or to determine whether Asn-459 is a carbohydrate attachment site, because the three single mutants (N222Q, N338Q, or N459Q) lacking only one of the three consensus glycosylation sites, migrate similar to wild-type MuSK in SDS-PAGE (Fig. 2 B and data not shown). Moreover, a MuSK triple mutant (N222Q/N338Q/N459Q), lacking all three consensus sites, migrates similar to the MuSK double mutant (N222Q/N338Q) in SDS-PAGE (Fig. 2 B). We next sought to determine whether Asn-222, Asn-338, and Asn-459 are required for agrin to stimulate MuSK phosphorylation. We treated muscle cell lines, stably expressing wild-type or mutant MuSK, with agrin for 30 min, immunoprecipitated MuSK and probed Western blots with antibodies to phosphotyrosine. Following agrin stimulation, tyrosine phosphorylation of wild-type MuSK increases more than 10-fold (Fig. 3, A and B). Tyrosine phosphorylation of the MuSK double mutant (N222Q/N338Q) or the MuSK triple mutant (N222Q/N338Q/N459Q) is similarly stimulated by agrin (Fig. 3, A and B and data not shown). In the absence of agrin stimulation, however, tyrosine phosphorylation of the double and triple mutants is elevated (2- to 3-fold) relative to wild-type MuSK (Fig. 3, A and B). These results indicate that N-linked carbohydrate addition is not essential for agrin to stimulate tyrosine phosphorylation of MuSK but that glycosylation restrains MuSK activation, possibly by limiting ligand-independent dimerization. Agrin activation of MuSK leads to tyrosine phosphorylation of the AChR β- and δ-subunits (13Wallace B.G., Qu, Z. Huganir R.L. Neuron. 1991; 6: 869-878Abstract Full Text PDF PubMed Scopus (213) Google Scholar, 26Mittaud P. Marangi P.A. Erb-Vogtli S. Fuhrer C. J. Biol. Chem. 2001; 276: 14505-14513Abstract Full Text Full Text PDF PubMed Scopus (74) Google Scholar). The mechanisms that lead to tyrosine phosphorylation of the AChR are poorly understood and could, in principle, require glycosylation of MuSK. We treated myotubes expressing wild-type or mutant forms of MuSK with agrin for 30 min, isolated AChRs with α-bungarotoxin, and probed Western blots with antibodies to phosphotyrosine. We found that the AChR β-subunit is tyrosine-phosphorylated similarly in agrin-treated myotubes expressing either wild-type, double mutant (N222Q/N338Q), or triple mutant (N222Q/N338Q/N459Q) MuSK (Fig. 3, C andD and data not shown). Therefore, N-linked glycosylation of MuSK is not essential for agrin to stimulate tyrosine phosphorylation of AChRs. The extracellular domain of MuSK is thought to have an important role in the clustering of synaptic proteins, including AChRs (27Apel E.D. Glass D.J. Moscoso L.M. Yancopoulos G.D. Sanes J.R. Neuron. 1997; 18: 623-635Abstract Full Text Full Text PDF PubMed Scopus (202) Google Scholar). We therefore examined whether N-linked glycosylation of MuSK is required for agrin to stimulate clustering of AChRs. We treated myotubes expressing wild-type or mutant forms of MuSK with agrin, and we determined the number of AChR clusters that formed 8–16 h later. We found that the number of AChR clusters is similar in agrin-treated myotubes expressing either wild-type, double mutant (N222Q/N338Q), or triple mutant (N222Q/N338Q/N459Q) MuSK (Fig. 4). Thus, N-linked glycosylation of MuSK is not required for agrin to stimulate AChR clustering. We next asked whether the increase in agrin-independent MuSK phosphorylation, found in myotubes expressing the double (N222Q/N338Q) or triple (N222Q/N338Q/N459Q) MuSK mutant (Fig. 3, A andB) leads to an increase in the level of AChR tyrosine phosphorylation and clustering. The basal level of AChR tyrosine phosphorylation (Fig. 3, C and D) and clustering (Fig. 4) is indeed elevated in myotubes expressing MuSK N222Q/N338Q or MuSK N222Q/N338Q/N459Q. These data support the idea thatN-linked carbohydrate structures bearing terminal sialic acid minimize spontaneous MuSK activation and signal transduction. Because mutation of single asparagine residues in MuSK fails to increase MuSK phosphorylation (data not shown) and downstream signaling (Fig. 4), these data support the idea that both Asn-222 and Asn-338 function as N-linked carbohydrate-attachment sites but that glycosylation of either site is sufficient to restrain adventitious MuSK activation. Previous studies have shown that NA treatment of wild-type myotubes increases the level of MuSK and AChR tyrosine phosphorylation and AChR clustering, indicating that removal of sialic acid residues from MuSK or from other proteins expressed in skeletal muscle increases tyrosine phosphorylation of MuSK and MuSK signaling (28Martin P.T. Sanes J.R. Neuron. 1995; 14: 743-754Abstract Full Text PDF PubMed Scopus (74) Google Scholar, 29Grow W.A. Ferns M. Gordon H. Dev. Neurosci. 1999; 21: 436-443Crossref PubMed Google Scholar, 30Grow W.A. Ferns M. Gordon H. J. Neurobiol. 1999; 40: 356-365Crossref PubMed Scopus (32) Google Scholar, 31Grow W.A. Gordon H. Cell Tissue Res. 2000; 299: 273-279Crossref PubMed Scopus (18) Google Scholar). To distinguish between these two possibilities, we compared the level of MuSK tyrosine phosphorylation, AChR tyrosine phosphorylation, and clustering in 1) NA-treated myotubes expressing wild-type MuSK; 2) myotubes expressing glycosylation-deficient MuSK mutants; and 3) NA-treated myotubes expressing glycosylation-deficient MuSK mutants. We found that MuSK phosphorylation (Fig. 3, Aand B) and MuSK-dependent signaling, assessed by AChR tyrosine phosphorylation (Fig. 3, C and D) and AChR clustering (Fig. 4), were similarly elevated in untreated myotubes expressing the MuSK mutants (MuSK N222Q/N338Q and MuSK N222Q/N338Q/N459Q) and in NA-treated myotubes expressing wild-type MuSK. Moreover, NA treatment of myotubes expressing the MuSK double (N222Q/N338Q) or triple (N222Q/N338Q/N459Q) mutant failed to further increase the level of MuSK tyrosine phosphorylation, AChR tyrosine phosphorylation, or AChR clustering (Figs. 3 and 4). These results indicate that NA increases tyrosine phosphorylation of MuSK and MuSK-dependent signaling by removing sialic acid from from Asn-222 and Asn-338 in MuSK rather than by removing sialic acid from other muscle proteins. MuSK has three consensus N-linked glycosylation sites, and two of these sites are conserved in MuSK from all classes. Mutation of the two conserved glycosylation sites leads to an increase in MuSK tyrosine phosphorylation, AChR tyrosine phosphorylation, and AChR clustering. Because mutation of either site alone does not increase MuSK activation and signaling, our data are consistent with the idea that glycosylation at either site is sufficient to minimize adventitious MuSK activation. Although these data are compatible with the possibility that both sites serve as glycosylation attachment sites, it is also possible that one of the two N-linked sites serves as the primary glycosylation site, but that an alternative, second site can be used if the primary site is not available. This latter idea is consistent with experiments showing that the double, but not the single MuSK mutants migrate more rapidly than wild-type MuSK in SDS-PAGE. Because the double (N222Q/N338Q) and triple (N222Q/N338Q/N459Q) MuSK mutants migrate similarly in SDS-PAGE and induce similar numbers of AChR clusters, and because the Asn-459 glycosylation site is not conserved in human and Torpedo, our data favor the idea that Asn-459 does not function as aN-linked glycosylation site. The lectins tested in our experiments fail to bind to MuSK afterN-linked carbohydrates were removed byN-glycosidase F, indicating that the recognized carbohydrates are not O-linked to MuSK. Nonetheless, it is possible that other carbohydrate structures, which are not recognized by these lectins, are attached to MuSK through O-linked sites. Thus, although our experiments do not demonstrate the presence of O-linked glycans on MuSK, we cannot rule out this possibility. Attachment of carbohydrates to the consensus N-linked sites in MuSK is not required for agrin to activate MuSK and to stimulate tyrosine phosphorylation and clustering of AChRs. These data are inconsistent with the idea that cell-type differences in MuSK glycosylation account for the inability of agrin to activate MuSK in fibroblasts and myoblasts. The increase in MuSK activation and signaling found in myotubes expressing glycosylation-deficient MuSK mutants is comparable with the level of MuSK activation and signaling found in NA-treated myotubes expressing wild-type MuSK. Moreover, NA is ineffective in further increasing MuSK signaling in myotubes expressing glycosylation-deficient MuSK mutants. These results indicate that NA increases MuSK signaling in wild-type myotubes by removing sialic acid from MuSK. Our experiments, however, do not exclude the possibility that NA also acts on other proteins that exert an inhibitory effect on MuSK. Inhibition of N-linked glycosylation often causes defects in the trafficking of RTKs, including kinase insert receptor/flk-1, flt-1, and the insulin receptor (32Takahashi T. Shibuya M. Oncogene. 1997; 14: 2079-2089Crossref PubMed Scopus (273) Google Scholar, 33Collier E. Carpentier J.L. Beitz L. Carol H. Taylor S.I. Gorden P. Biochemistry. 1993; 32: 7818-7823Crossref PubMed Scopus (48) Google Scholar), resulting in a failure of these RTKs to accumulate at the cell surface. MuSK glycosylation mutants, however, are expressed at the cell surface, because mutants lacking individual or multiple N-linked glycosylation sites are tyrosine-phosphorylated by agrin stimulation. Moreover, in the absence of agrin, expression of double or triple mutants leads to an increase in AChR clustering at the cell surface. Thus,N-linked glycosylation of MuSK is neither required to stably express MuSK at the cell surface nor to activate MuSK by agrin. It remains possible, however, that MuSK glycosylation mutants are transported to the cell surface less efficiently than wild-type MuSK or that these mutants are less stable at the cell surface. The suppression of ligand-independent activation by N-linked glycosylation is a feature that MuSK shares with other RTKs.N-Glycosylation of TrkA (34Watson F.L. Porcionatto M.A. Bhattacharyya A. Stiles C.D. Segal R.A. J. Neurobiol. 1999; 39: 323-336Crossref PubMed Scopus (114) Google Scholar), insulin receptor (35Elleman T.C. Frenkel M.J. Hoyne P.A. McKern N.M. Cosgrove L. Hewish D.R. Jachno K.M. Bentley J.D. Sankovich S.E. Ward C.W. Biochem. J. 2000; 347: 771-779Crossref PubMed Scopus (43) Google Scholar), and epidermal growth factor receptor (36Tsuda T. Ikeda Y. Taniguchi N. J. Biol. Chem. 2000; 275: 21988-21994Abstract Full Text Full Text PDF PubMed Scopus (97) Google Scholar) suppresses ligand-independent activation of these RTKs. In TrkA, even partial deglycosylation is sufficient to cause ligand-independent TrkA phosphorylation, resulting in the recruitment of Shc and phospholipase C-γ (34Watson F.L. Porcionatto M.A. Bhattacharyya A. Stiles C.D. Segal R.A. J. Neurobiol. 1999; 39: 323-336Crossref PubMed Scopus (114) Google Scholar). N-Linked glycosylation, however, is not necessarily linked to inhibtion of RTKs, because N-linked glycosylation can induce conformational changes that lead to ligand-independent RTK dimerization and activation (37Fernandes H. Cohen S. Bishayee S. J. Biol. Chem. 2001; 276: 5375-5383Abstract Full Text Full Text PDF PubMed Scopus (126) Google Scholar). Although the mechanisms by which N-linked glycosylation restrains tyrosine kinase activity are poorly understood, glycosylation may minimize spontaneous dimerization simply by steric hindrance or by electrostatic repulsion and thereby inhibit trans-phosphorylation (38Hubbard S.R. Till J.H. Annu. Rev. Biochem. 2000; 69: 373-398Crossref PubMed Scopus (898) Google Scholar). In COS cells, however, the low level of MuSK tyrosine phosphorylation is not further increased by NA treatment (39Borges L.S. Ferns M. J. Cell Biol. 2001; 153: 1-12Crossref PubMed Scopus (91) Google Scholar). Thus,N-glycosylation may not inhibit MuSK activation by a simple steric or electrostatic mechanism. In the epidermal growth factor receptor (40Wang X.Q. Sun P. O'Gorman M. Tai T. Paller A.S. Glycobiology. 2001; 11: 515-522Crossref PubMed Scopus (75) Google Scholar), attached carbohydrate interacts with glycosphingolipids, and the epidermal growth factor receptor-ganglioside complex minimizes spontaneous epidermal growth factor receptor dimerization. By analogy, carbohydrate attached to MuSK may interact with glycolipids or with other glycoproteins either in the plasma membrane or in the extracellular matrix, and such interactions may minimize spontaneous MuSK dimerization. Because these interactions may be cell type-specific, it is possible that such interactions mediate inhibition of MuSK in muscle cells but not in COS cells. In this regard, other groups have suggested that NA and α-galactosidase induce AChR clusters by unmasking subterminal Galβ-1→3-GalNac and Galβ-1→4-GalNac (CT carbohydrate) in proteins within the extracellular matrix. Consistent with this idea, CT carbohydrate is concentrated at neuromuscular synapses, and addition of either disaccharide to cultured muscle cells induces tyrosine phosphorylation of MuSK and clustering of AChRs (41Parkhomovskiy N. Kammesheidt A. Martin P.T. Mol. Cell. Neurosci. 2000; 15: 380-397Crossref PubMed Scopus (39) Google Scholar, 42Martin P.T. Glycobiology. 2002; 12: 1R-7RCrossref PubMed Google Scholar). Moreover, agrin can stimulate MuSK tyrosine phosphorylation in non-muscle cells if these cells also express the CT carbohydrate, implicating the CT carbohydrate as a potential co-factor in agrin-induced MuSK activation, similar to that for heparin in fibroblast growth factor-induced fibroblast growth factor receptor activation (43Rapraeger A.C. Krufka A. Olwin B.B. Science. 1991; 252: 1705-1708Crossref PubMed Scopus (1292) Google Scholar). Taken together, these results favor a model in which the interaction between the N-linked carbohydrates on MuSK and other carbohydrates, possibly the Galβ-1→3-GalNac and Galβ-1→4-GalNac structures of the CT carbohydrate, regulate activation of MuSK. Such an interaction could mediate MuSK inhibition, partially relieved by removal of either the CT carbohydrate or MuSK glycosylation and fully relieved by agrin-binding, resulting in maximal MuSK activation. We thank Cindy Sadowski for constructing the MuSK glycosylation site mutants. We thank Matthew Friese for comments on the manuscript and Ruth Herbst for help with generating the muscle cell lines.