Functional interaction of Src family kinases with the acetylcholine receptor in C2 myotubes.

Tyrosine phosphorylation of the β subunit of the acetylcholine receptor (AChR) has been postulated to play a role in AChR clustering during development of the neuromuscular junction. We have investigated the mechanism of this phosphorylation in mammalian C2 myotubes and report that the tyrosine kinase Src binds and phosphorylates glutathione S-transferase fusion proteins containing the N-terminal half of the cytoplasmic loop of the β subunit. No binding occurs to the related kinases Fyn or Yes or to the corresponding regions from the γ and δ subunits. Furthermore, AChRs affinity-isolated from C2 myotubes using α-bungarotoxin-Sepharose were specifically associated with Src and Fyn and had tyrosine-phosphorylated β subunits. We suggest that AChRs are initially phosphorylated by Src and subsequently bind Fyn in a phosphotyrosine-dependent manner. These interactions are likely to play an important role in construction of the specialized postsynaptic membrane during synaptogenesis.

Protein-tyrosine phosphorylation is a widely used mechanism for regulating cellular functions, particularly those involving growth or differentiation factors. Several protein-tyrosine kinases are highly expressed in brain (1)(2)(3) and are associated with synaptic structures (4), suggesting that they play a general role in synaptic function. At the neuromuscular junction and at its homologous synapse in the electric organ of Torpedo, tyrosine phosphorylation appears to be important for regulating both the function and the distribution of the nicotinic acetylcholine receptor (AChR) 1 during development (5,6).
The AChR is a ligand-gated ion channel with a pseudosymmetric pentameric structure consisting of four homologous subunits in the ratio ␣ 2 ␤␥␦. Each subunit traverses the membrane four times, with a long, cytoplasmic loop between transmembrane domains 3 and 4 (7,8). In the Torpedo AChR, a single conserved tyrosine residue in the cytoplasmic loop of each of the ␤, ␥, and ␦ subunits is phosphorylated by a kinase activity in the postsynaptic membrane (9). In this tissue, two members of the Src family of tyrosine kinases, Fyn and Fyk, account for a substantial fraction of the total tyrosine kinase activity and have been shown in immunoprecipitation experiments to be associated with tyrosine-phosphorylated AChRs (10,11). Phosphorylation of the AChR subunits is accompanied by an increase in the rate of rapid desensitization of the receptor by cholinergic ligands, a change that is also produced by phosphorylation of the receptor on serine residues (12,13).
Tyrosine phosphorylation of the AChR appears to play an important role in synaptogenesis. At the mammalian neuromuscular junction, tyrosine phosphorylation in the postsynaptic membrane, possibly of the AChR, increases during the late, post-natal stage of synaptic maturation (14). Tyrosine phosphorylation of the AChR may also be related to one of the earliest steps in synapse formation, the clustering of AChRs in the postsynaptic membrane underlying the nerve terminal (5). Studies on the development of the chick neuromuscular synapse have shown that the earliest detectable AChR clusters in vivo contain phosphotyrosine and that in vitro AChR clustering and phosphotyrosine co-staining in chick myotubes depend on innervation by co-cultured neurons (15). Both in nerve-muscle cultures and in vivo, the aggregation of AChRs at the nascent neuromuscular junction is caused by agrin released from motor nerves (16,17). The addition of agrin to cultured myotubes induces both widespread AChR cluster formation and tyrosine phosphorylation of the AChR (15,18,19). In mammalian muscle, this phosphorylation, which is specific for the ␤ subunit of the receptor, reaches a peak in 1 h and precedes AChR clustering (20). The kinase inhibitors herbimycin A and staurosporine block both AChR phosphorylation and AChR clustering (20,21). A receptor tyrosine kinase, MuSK, which is localized to synapses in adult muscle, appears to be part of the signaling receptor for agrin (22). Activation of MuSK also results in tyrosine phosphorylation of the AChR (23,24).
The protein-tyrosine kinase or kinases that are responsible for phosphorylating the AChR in mammalian muscle cells are unknown. We report here experiments to identify the kinase(s) responsible for tyrosine phosphorylation of the AChR and to examine the interaction of Src family kinases with the AChR in mammalian muscle. Our results show that Src from C2 myotubes selectively binds to fusion proteins derived from the long, intracellular loop of the AChR ␤ subunit; by immunodepletion and in vitro phosphorylation we find that bound Src phosphorylates the fusion proteins. We also show that AChRs isolated from C2 myotubes are tyrosine-phosphorylated on their ␤ subunits and are associated with two members of the Src family, Fyn and Src. We describe a model for the interaction between the AChR and the Src family members and suggest that these kinases are likely to play a role in the construction of a postsyn-aptic complex and immobilization of the AChR at developing synapses.

EXPERIMENTAL PROCEDURES
Generation of GST Fusion Proteins-Glutathione S-transferase fusion proteins containing segments of the long cytoplasmic loop of the AChR subunits were generated using cDNAs encoding the ␣, ␤, and ␦ subunits of the mouse muscle nicotinic AChR in the vector pSM (25) as templates. AChR segments were amplified using the polymerase chain reaction in combination with oligonucleotides that bordered the domains of interest and were designed to create a 5Ј BamHI and a 3Ј EcoRI restriction site. Amplified DNA segments were isolated, digested with BamHI and EcoRI, and ligated into pGEX-2T vectors (Pharmacia Biotech Inc.) cleaved with the same enzymes. Polymerase chain reaction-amplified DNA inserts were control sequenced using a commercially available dideoxy sequencing kit from U.S. Biochemical Corp. In the resulting fusion proteins, the AChR segments are fused to the C terminus of GST. DNA constructs were used to transform E. coli HB101. Large scale cultures were grown, induced with isopropyl-␤-Dthiogalactopyranoside, sonicated on ice, and the GST fusion proteins purified on glutathione-Sepharose beads (26) and stored coupled to the beads at 4°C. The concentration of the fusion proteins was estimated by SDS-polyacrylamide electrophoresis (SDS-PAGE), protein staining with Coomassie Blue (Sigma), and comparison to a bovine serum albumin standard (Pierce). The following fusion proteins were prepared (see also Fig. 2; numbers indicate positions of amino acids within the fulllength subunits including the N-terminal signal sequence): ␤A,9,10,B (␤ subunit, amino acids 333-469); ␤A,9,10 (␤ subunit, amino acids 333-455); ␤A,9 (␤ subunit, amino acids 333-405); ␤9 (␤ subunit, amino acids 349 -405); ␤9,10 (␤ subunit, amino acids 349 -455); ␤10 (␤ subunit, amino acids 406 -455); ␥9 (␥ subunit, amino acids 346 -417); ␦9 (␦ subunit, amino acids 353-420). The amino acid sequence (N to C terminus) of the ␤ exon 9 region in the ␤9 fusion protein is IFIHKLPPYLGLKRPKPERDQLPEPHHSLSPRSGWGRGTDEYFIRK-PPSDFLFPKLN. The corresponding sequence of ␥9 is LFLRL-RPQLLRMHVRPLAPAAVQDARFRLQNGSSSGWPIMAREEGDLCLP-RSELLFRQRQRNGLVQAVLEKL, and the homologous sequence of ␦9 is FFLETLPKLLHMSRPAEEDPGPRALIRRSSSLGYICKAEEYFSLK-SRSDLMFEKQSERHGLARRLTTA.
Cell Culture-C2C12 mouse muscle cells were cultured on 10-or 15-cm plastic dishes (Nunc) as described previously (27). All cell culture reagents were purchased from Life Technologies, Inc. Cells were propagated at low density in growth medium consisting of Dulbecco's modified Eagle's medium with 4.5 g/liter D-glucose supplemented with 20% fetal bovine serum, 0.5% chick embryo extract, 2 mM glutamine, and penicillin/streptomycin. Cultures at about 90% confluence were induced to form myotubes by changing the medium to differentiation medium consisting of Dulbecco's modified Eagle's medium containing 5% horse serum and 2 mM glutamine. Fusion of myoblasts to generate myotubes was evident after 1 day, and contracting myotubes were usually observed after 2 days by which time the cultures were used for experiments.
Kinase Binding Assays and Kinase Expression-Myotubes grown on 10-cm dishes were rinsed at 4°C with phosphate-buffered saline supplemented with 1 mM sodium orthovanadate and 50 mM NaF and extracted in 1 ml of lysis buffer containing 1% Nonidet P-40, 150 mM NaCl, 10 mM triethanolamine, pH 7.6, 5 mM EGTA, 5 mM EDTA, 50 mM NaF, 1 mM sodium orthovanadate, and protease inhibitors (1 mM benzamidine, 1 mM N-ethylmaleimide, 1 mM sodium tetrathionate, 1 mM phenylmethylsulfonyl fluoride, 25 g/ml aprotinin, and 25 g/ml leupeptin). Insoluble material, such as nuclei and extracellular matrix, was removed by centrifugation at 18,000 ϫ g for 5 min. To analyze binding of fusion proteins to Src-related kinases, lysates were incubated for 2 h at 4°C with 2 or 5 g of fusion proteins adhering to glutathione-Sepharose beads. Beads were then washed twice in wash buffer 1 (0.4% Nonidet P-40, 500 mM NaCl, 10 mM triethanolamine, pH 7.6, 5 mM EGTA, 5 mM EDTA, 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride), twice in wash buffer 2 (same as 1, but containing 150 mM NaCl), and boiled in SDS-PAGE sample buffer. Samples were analyzed by reducing SDS-PAGE and immunoblotting using nitrocellulose (Micron Separations Inc.) and specific antibodies against Src (mouse monoclonal, Oncogene Science Inc.), Fyn (rabbit polyclonal, Santa Cruz Biotech), Yes (mouse monoclonal, Wako Chemicals), or with src-CT antiserum (rabbit polyclonal, Santa Cruz Biotech). Immunoreactive bands were visualized with horseradish peroxidase-conjugated secondary antibodies and enhanced chemiluminescence (ECL; Amersham Corp.). To analyze expression of Src, Fyn, and Yes in total ex-tracts of C2 myotubes, aliquots of cleared lysates were subjected to SDS-PAGE and immunoblotting with the antibodies mentioned above. Alternatively, Src-related kinases were first immunoprecipitated with 1 g of src-CT, Src, Fyn, and Yes antibodies followed by protein A-or protein G-Sepharose. Precipitates were then analyzed by nonreducing SDS-PAGE and immunoblotting with src-CT. To reprobe immunoblots, they were stripped as described (20) using 200 mM glycine, 0.1% Tween 20, pH 2.5, for 20 min and then reprobed with antibodies against GST (mouse monoclonal, Santa Cruz Biotech) to confirm that equal amounts of fusion proteins were present in all samples. For some experiments, a concentrated membrane fraction of C2 myotubes (18,28) was used for fusion protein adsorption. Briefly, myotubes grown on 15-cm dishes were homogenized in a lysis buffer without detergent, centrifuged at 6000 ϫ g to remove nuclei and unbroken cells, and then centrifuged at 100,000 ϫ g for 30 min. The pelleted membranes were extracted in 200 l of lysis buffer containing 1% Nonidet P-40 and centrifuged again at 100,000 ϫ g for 30 min. The supernatant, consisting of concentrated C2 myotube membranes, was then analyzed as described above. Quantitation of ECL immunoblotting data was performed by scanning the films with a computerized densitometer (Sierra Scientific) and using the NIH Image 1.54 software (National Institutes of Health).
In Vitro Phosphorylation and Kinase Depletion-To analyze the capability of Src to phosphorylate bound ␤9 fusion proteins, C2 myotube extracts were first adsorbed with fusion protein beads as mentioned above. Beads were then washed twice in kinase assay buffer (20 mM Tris, pH 7.4, 10 mM MgCl 2 , 2 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride) and incubated for 15 min at 37°C with kinase assay buffer containing 3 mM ATP. Beads were pelleted and boiled in SDS-PAGE sample buffer. To visualize tyrosine phosphorylation, samples were subjected to SDS-PAGE and immunoblotting using an anti-phosphotyrosine antibody, 4G10 (mouse monoclonal, UBI), and ECL. For immuno-depletion of Src-related kinases, C2 extracts were incubated for 2 h at 4°C with src-CT or antibodies specific for Src, Fyn, and Yes. Antibodies were removed by two rounds of precipitation with protein A-or protein G-Sepharose. Depleted lysates were then adsorbed with fusion proteins as described before. To analyze the ability of purified Src kinase to phosphorylate the ␤9 fusion protein, fusion proteins were first eluted from glutathione-Sepharose using 5 mM reduced glutathione in 50 mM Tris, pH 7.5, and 2 mM EDTA and dialyzed into the same buffer lacking glutathione. 5 g of fusion proteins were then incubated with 5 ng of baculovirus-expressed and purified Src kinase (a gift from Dr. D. Morgan, University of California, San Francisco, CA) in a volume of 40 l of phosphorylation buffer (0.5 Ci/l [␥-32 P]ATP, 10 M ATP, 20 mM Hepes, pH 7.4, 1 mM dithiothreitol, 10 mM MgCl 2 , 100 mM NaCl) for 1-24 h at 4, 25, or 37°C. Reactions were stopped by boiling in SDS-PAGE sample buffer and analyzed by SDS-polyacrylamide gel electrophoresis and autoradiography. For quantitation, the ␤9 protein bands (ϳ34 kDa) were excised from dried gels and analyzed by liquid scintillation counting. As background we measured pieces of the gels not containing any proteins. Comparison of the amount of incorporated 32 P with the amount of ␤9 fusion protein allowed us to calculate the percentage of fusion proteins phosphorylated in the reactions.
Isolation of AChRs -To assay association of endogenous Src-related kinases with the AChR in C2 myotubes, cells were grown and lysed as mentioned above with the exception of Nonidet P-40 which was replaced by digitonin in most experiments. The solubilized AChR was isolated as described (20) by incubating the extracts with ␣-bungarotoxin (Sigma) conjugated to Sepharose beads (Sigma). As a control for specificity, we added excess (10 M) free uncoupled ␣-bungarotoxin. Other control samples were precipitated with uncoupled Sepharose beads. Precipitates were washed as before and the proteins isolated on the beads eluted in SDS-PAGE sample buffer, either by boiling or at room temperature. Analysis by SDS-PAGE and immunoblotting with antibodies against Src-related kinases was performed as mentioned. Stripped immunoblots were reprobed with anti-phosphotyrosine antibodies (4G10, in a few experiments with PY20 (mouse monoclonal; Transduction Labs)). To reveal the AChR, the blots were stripped and reprobed a second time using mAB 124 which recognizes the AChR ␤ subunit (rat monoclonal; a kind gift from Dr. J. Lindstrom, University of Pennsylvania, Philadelphia, PA).

Expression of Src Family Kinases in C2
Myotubes-In our initial experiments we examined C2 myotubes for the expression of three prominent Src family kinase members, Src, Fyn, and Yes. In contrast to other Src-related kinases that are restricted to particular cell types, these kinases show a widespread tissue distribution and are thought to be involved in a variety of cellular signal transduction pathways (29,30). Extracts of C2 myotube cultures were immunoblotted with an antiserum, src-CT, that recognizes a highly conserved sequence at the extreme C terminus of all three kinases, and with antibodies specific for each of the kinases (Fig. 1A). All three kinases were found in the extracts. Each of the kinases was also immunoprecipitated from C2 extracts as shown by immunoprecipitation with the specific antisera followed by immunoblotting with src-CT (Fig. 1B). Thus each of the Src family members examined is present in C2 myotubes and is recognized by src-CT in immunoblot assays.
Fusion Proteins Containing the ␤ Subunit Cytoplasmic Loop-We then constructed GST fusion proteins containing all or part of the cytoplasmic loop connecting transmembrane domains 3 and 4 of the ␤ subunit of the mouse muscle AChR (Fig.  2). The gene encoding the ␤ subunit contains two exons, 9 and 10, that together encode most of the loop (31). The fusion protein ␤9 covers exon 9, whereas the fusion proteins ␤10 and ␤9,10 contain the products of exon 10 and both exon 9 and exon 10, respectively (Fig. 2). The region encoded by exon 9, which is N-terminal, contains two tyrosine residues. One of these (Tyr-390), which is conserved between species, is phosphorylated in the AChR isolated from the Torpedo electric organ (9). Exon 10 encodes most of the C-terminal half of the loop which contains a third tyrosine as well as the amphipathic helix (32). As controls we made fusion proteins with sequences from the cytoplasmic loops of the AChR ␥ and ␦ subunits that correspond to the exon 9 region of the ␤ subunit (␥9 and ␦9 proteins). Whereas the ␥ region lacks tyrosine residues, the ␦ region contains two, one of which is homologous to Tyr-390 of the ␤ subunit and is phosphorylated in the Torpedo AChR (9).
The N-terminal Half of the AChR ␤ Subunit Loop Specifically Interacts with Src-To determine whether the ␤ loop of the AChR interacts with members of the Src family, Nonidet P-40 extracts of C2 myotubes were incubated with purified GST ␤ fusion proteins adhering to glutathione-Sepharose beads. Bound kinases were identified by elution of the beads with SDS buffer, followed by SDS-gel electrophoresis and immunoblotting. Experiments with src-CT showed that one or more Src family member(s) bound to fusion proteins containing the ␤ subunit exon 9 region, but not to a fusion protein containing only ␤ exon 10 (␤10) nor to GST alone (Fig. 3A). Apparent differences in the extent of binding to Src kinase(s) between ␤ exon 9-containing fusion proteins (Fig. 3A) were not reproducible over the course of several experiments and may be related to variations in the folding efficiency of the different fusion proteins. Binding to the ␤ exon 9 region was specific, however, as interactions with ␤10 or GST were not observed, even when large amounts (20 g) of fusion proteins and long immunoblot exposure times were used. In control experiments, the ability of the AChR ␥ and ␦ subunits to bind to Src-related kinases was tested using the ␥9 and ␦9 fusion proteins (Fig. 3B). No binding was detected to the ␥9 region, consistent with the absence of tyrosine residues. Surprisingly, the ␦9 region, which contains FIG. 1. Expression of Src-related kinases in C2 myotubes and characterization of kinase-specific antibodies. A, C2 myotubes grown on 10-cm dishes were lysed in a buffer containing 1% Nonidet P-40, and the indicated fractions of the extracts were analyzed by reducing SDS-PAGE and immunoblotting as described under "Experimental Procedures." Nitrocellulose blots were incubated with antibodies specific for Src, Fyn, or Yes or with an antiserum, src-CT, that recognizes all three. Arrowheads indicate the position of Src-related kinases. Under reducing conditions, all three kinases show essentially the same mobility. Src-CT and Fyn antibodies also react with unknown proteins of ϳ90 and ϳ100 kDa, respectively. B, C2 myotube extracts were subjected to immunoprecipitation with 1 g of the indicated antibodies followed by protein A-or protein G-Sepharose. Immunoprecipitates (IP) were analyzed by nonreducing SDS-PAGE and immunoblotting with src-CT. A fraction of the initial lysate (L) was included for comparison. As src-CT and Fyn antisera are both from rabbit, whereas anti-Src and Yes are mouse monoclonal antibodies, immunoglobulin molecules (ϳ150 kDa) and Ig heavy chains (ϳ50 kDa) appear in the first two lanes (marked by asterisks). After immunoprecipitation and analysis with nonreducing SDS-PAGE, Yes displays a slightly slower mobility than Src or Fyn. All three kinases are recognized by src-CT antiserum on immunoblots.

FIG. 2. Sequences of AChR cytoplasmic portion constructs.
A, general transmembrane organization of the AChR ␤, ␥, and ␦ subunits. The long loop between transmembrane domains 3 and 4 accounts for most of the cytoplasmic portion of the subunits. B, GST constructs derived from the mouse ␤, ␥, and ␦ cytoplasmic loops. Intron/exon boundaries were previously determined for the mouse ␤, human, and chicken ␥, and chicken ␦ subunits (31,54,55). The scheme at the top represents the loop of the ␤ subunit and shows that most of this domain is encoded by two exons, exon 9 and exon 10. Two short flanking regions, termed A and B, account for the rest. Numbers refer to the position of amino acids within the uncleaved mouse ␤ subunit including the N-terminal signal sequence. As the amino acid sequences surrounding the exon/exon boundaries are highly conserved between species, we designed ␥9 and ␦9 constructs according to the human and chicken genomic structures. Tyrosine residues are highlighted and names of constructs are as indicated.
two tyrosine residues, also failed to bind Src family kinases. Together, these findings indicate that the exon 9 region of the AChR ␤ subunit specifically binds at least one kinase of the Src family.
To identify the specific kinase(s) involved, we analyzed the adsorptions with antibodies specific for Src, Fyn, or Yes. Each antibody was also tested against the original C2 myotube lysate to allow comparison of the results with different antisera. As shown in Fig. 4, ␤9 and ␤9,10 fusion proteins bound Src, but no binding was detected for Fyn or Yes. ␤10, ␥9, and ␦9 fusion proteins failed to bind Src, Fyn, or Yes, confirming the specificity of the ␤9-Src interaction. For each kinase the efficiency of binding to the ␤9 fusion protein was estimated by comparing the amount of kinase bound with the total kinase content of the initial cell lysates as measured by immunoblotting with the same antibody. With the Src-specific antibody we estimate that under the conditions of our experiments 5 g of ␤9 fusion protein bound ϳ1.4% of the Src initially present in the cell extracts. When more (15 g) fusion protein was incubated with a concentrated C2 membrane preparation instead of a total cell extract, this percentage was increased to about 15%. Using src-CT on immunoblots the comparable recovery values were 0.5 and 5% for 5 and 15 g of ␤9 fusion protein in total extracts and membrane preparations, respectively (data not shown). Together, these studies indicate that the ␤ exon 9 fusion protein specifically binds to Src, but not to Fyn or Yes, and that the homologous regions from the ␥ and ␦ subunits fail to bind to any of these three kinases.
The Exon 9 Region of the AChR ␤ Subunit Is Phosphorylated by Src-To determine whether associated Src is able to phosphorylate the ␤ exon 9 region, we adsorbed C2 extracts with ␤9 fusion protein beads and subsequently incubated the washed and immobilized Src-␤9 complex with ATP under phosphorylating conditions. Tyrosine phosphorylation was monitored by immunoblotting with antibodies against phosphotyrosine (Fig.  5). The ␤9 fusion protein showed a low degree of reactivity with this antiserum, even in the absence of incubation with ATP (Fig. 5A, lane 1). The same signal was seen with the ␥9 fusion protein, which lacks tyrosine residues in its AChR portion, and presumably represents nonspecific cross-reactivity of phosphotyrosine antibodies with both bacterially expressed proteins. Upon incubation of the ␤9 fusion protein with C2 extracts and subsequently with ATP, two phosphorylated bands were observed, a prominent band of ϳ34 kDa and a minor band of ϳ60 kDa (Fig. 5A, lane 2). Longer exposures of immunoblots revealed no other bands. Based on their molecular weights and the results of stripping and reprobing the blots with the appropriate antibodies (data not shown), we identified the 34-and 60-kDa bands as the ␤9 fusion protein and Src, respectively. When C2 extracts were adsorbed with ␥9 or with GST protein beads, followed by incubation with ATP, no specific phosphotyrosine staining was observed (Fig. 5A, lanes 16 and 18). Furthermore, no phosphorylation was detected using ␦9 or ␤10 fusion proteins, but strong phosphorylation did occur when the ␤9,10 fusion protein was used (data not shown). Thus tyrosine phosphorylation is specific for the ␤ exon 9 region of fusion proteins.
To analyze whether this phosphorylation is carried out specifically by bound Src, we immuno-depleted the C2 lysates with Src-specific antibodies prior to adsorption. Phosphorylation of the ␤9 fusion protein and Src was drastically reduced by depletion with either Src-specific antibodies or with src-CT but was unchanged when antibodies against Fyn or Yes were used FIG. 3. Fusion proteins containing the exon 9 region of the ␤-cytoplasmic loop bind to a Src-related kinase(s). Nonidet P-40 lysates from C2 myotubes were incubated for 2 h at 4°C with either 2 or 5 g of the indicated GST fusion proteins attached to glutathione-Sepharose. Adsorptions were analyzed by reducing SDS-PAGE followed by immunoblotting using src-CT antiserum (upper panels). Nitrocellulose blots were stripped and reprobed with an antibody against glutathione S-transferase to confirm that equal amounts of GST fusion proteins were present in the relevant adsorptions (lower panels). A, binding of constructs containing portions of the ␤-cytoplasmic loop to Src-related kinase(s). Apparent differences in kinase binding efficiencies varied between experiments. No binding was seen to ␤10 and GST. B, binding capabilities of ␥ and ␦ regions homologous to the exon 9 region in ␤. Binding of Src-related kinase(s) is only observed for ␤9 fusion proteins but not for the homologous regions of ␥ or ␦, even when 5 g of fusion protein is used. A fraction of the initial cell lysate was analyzed as a standard (L). Lower molecular weight products in the lower panels represent degradation products of the GST fusion proteins. The arrowhead indicates Src-related kinases.
FIG. 4. The exon 9 region of the ␤-cytoplasmic loop specifically binds to the tyrosine kinase Src. C2 myotubes grown on 10-cm dishes were lysed and adsorbed to 5 g of fusion proteins as described in the legend to Fig. 3. Kinases (arrowheads) binding to the various fusion proteins were analyzed by immunoblotting with src-CT and antibodies specific for Src, Fyn, and Yes. To normalize for different antibody affinities, a constant fraction (0.25%) of the initial lysate was included (L). ␤9 and ␤9,10 fusion proteins bound to Src, but no comparable binding was observed to Fyn and Yes. Densitometric quantitation of data revealed that 5 g of ␤9 bound to and immobilized ϳ1.4% of Src initially present in the cell extracts. Using src-CT to immunoblot, this value was ϳ0.5%. (Fig. 5A, lanes 5-12). Analysis of the depleted C2 lysates confirmed that each antibody removed a substantial fraction of its corresponding kinase without affecting the others (Fig. 5B). Src-CT removes all kinases, including Src, to an intermediate degree, yet depletes the phosphorylating activity slightly better than the Src-specific antibody. This discrepancy may be due to preferential depletion by src-CT of the enzymatically active form of Src in which its C terminus is not engaged in an intramolecular SH2 domain interaction but is exposed and therefore more accessible for antibody recognition (33). Together, these results show that the ␤ exon 9 fusion protein is phosphorylated by bound Src, and, as Src family kinases are known to phosphorylate themselves, they suggest that phosphorylation of bound Src is due to autophosphorylation. To confirm the ability of Src to phosphorylate the ␤9 fusion protein, we incubated the fusion protein with purified Src kinase and [␥-32 P]ATP. Phosphorylation of the ␤9 fusion protein, but not the parental GST protein, was observed (Fig. 5C). Phosphorylation was more efficient at 25 or 37°C than at 4°C (Fig. 5D); under optimal conditions, 24 h at 25°C, about 1% of the ␤9 fusion protein was phosphorylated.
Src and Fyn Are Bound to the AChR in C2 Myotubes-To determine whether Src interacts with the AChR in vivo as it does with the ␤ subunit fusion proteins in vitro, we examined the AChR purified from muscle cells for associated Src family kinases. C2 myotube cultures were extracted with a mild detergent (1% digitonin or Nonidet P-40) and the AChRs precipitated using ␣-bungarotoxin conjugated to Sepharose beads. The beads were then stripped under denaturing conditions and the proteins analyzed by SDS-PAGE and immunoblotting. When immunoblots were probed with src-CT (Fig. 6A), a Src family kinase was observed. The kinase was not observed when uncoupled Sepharose beads were used as a control nor when binding of the AChR to the beads was blocked by preincubation of the receptor in the cell lysates with free toxin. Thus the presence of the Src family kinase on the beads was dependent on its association with the AChR. Accordingly, the Src kinase was seen on immunoblots only when the AChR was also present, as observed by stripping the blots and reprobing them with an antibody to the ␤ subunit (Fig. 6C). In contrast, a contaminating src-CT reactive band of ϳ90 kDa and unknown identity was isolated under all conditions tested. Stripped immunoblots were also analyzed with antibodies to phosphotyrosine. Two bands were consistently seen (Fig. 6B), a 50-kDa band whose mobility was identical to the ␤ subunit of the AChR (Fig. 6C) and a 60-kDa band with the same mobility as the Src-related kinase seen with src-CT (Fig. 6A). A third band of ϳ110 kDa (Fig. 6B) was not consistently observed. Tyrosine phosphorylation of certain proteins, such as the ␦ subunit of the AChR in chick, is reported to be sensitive to degradation during boiling in SDS loading buffer (15). However, we did not see additional phosphotyrosine bands specifically precipitated by ␣-bungarotoxin-Sepharose beads, even when unboiled samples were used and when immunoblots were analyzed with an alternative phosphotyrosine antiserum (data not shown).
The identity of the Src-related kinase(s) associated with the AChR was determined on immunoblots using the kinase-specific antibodies. As shown in Fig. 7, two kinases were specifically co-isolated with the AChR, Fyn and Src; no detectable Yes was seen. Under these conditions, the fraction of each kinase associated with the AChR was small: approximately 0.04% of the total Fyn and 0.01% of the total Src in the extract were recovered with the AChR from ␣-bungarotoxin-Sepharose. By comparing the amount of isolated AChR to the receptor content of the initial lysate, on the other hand, we estimate that we recover only about 4% of the total AChR in the extract (data not FIG. 5. The exon 9 region of the ␤-cytoplasmic loop is tyrosinephosphorylated by Src. To determine whether bound Src phosphorylates the ␤ exon 9 region, in vitro tyrosine phosphorylation of ␤9 adsorbates and the effect of kinase depletion were analyzed. For kinase depletion, C2 myotube lysates were incubated with 1 or 5 g of src-CT, Src, Fyn, or Yes antisera (lanes [5][6][7][8][9][10][11][12]. Antibodies with bound kinases were removed by two rounds of precipitation with protein A-or protein G-Sepharose. In control experiments, incubation with antibodies was omitted (lanes 1-4, 13-18). Depleted and control lysates were then incubated with 5 g of ␤9, ␥9, or GST proteins as indicated. Adsorbates  (except lanes 1, 3, 15, and 17) were subjected to in vitro phosphorylation with ATP. A, tyrosine phosphorylation of fusion proteins and associated proteins was visualized by reducing SDS-PAGE and anti-phosphotyrosine immunoblotting. B, aliquots of control and depleted cell lysates were analyzed by immunoblotting with specific antibodies against Src, Fyn, and Yes (open arrowheads) to reveal the degree of kinase depletion prior to adsorption with fusion proteins. Phosphorylation and depletion assays show that ␤9 (but not ␥9 or GST) and an interacting protein of ϳ60 kDa (Src; solid arrowheads) become tyrosine-phosphorylated and that Src, but not Fyn or Yes, accounts for most of this activity. Residual activity in Src-depleted lysates is somewhat higher than after kinase removal with src-CT. This difference may originate from src-CT antiserum preferentially binding to the more exposed C terminus of kinase active Src. Lanes 1 and 3 and 2 and 4 are duplicates. PAS and PGS, protein A-and protein G-Sepharose, respectively. C, phosphorylation of ␤9 by purified Src. Isolated Src kinase was incubated for 3 h at 25°C with 5 g of ␤9, GST, or no substrate under phosphorylating conditions including [␥-32 P]ATP. Autoradiography reveals that ␤9 (ϳ34 kDa) and Src (ϳ60 kDa), but not GST, become phosphorylated efficiently. D, time and temperature dependence of Src-mediated ␤9 phosphorylation. Phosphorylated ␤9 protein bands were excised from SDS-polyacrylamide gels and quantitated by liquid scintillation counting. Phosphorylation reaches a plateau after about 5 h at 25 or 37°C.
shown). Therefore we calculate that in C2 myotubes about 1% of the total Fyn and about 0.25% of the total Src are associated with AChRs. Taken together these experiments show that the AChRs in C2 myotubes contain tyrosine-phosphorylated ␤ subunits and are specifically associated with two tyrosine kinases of the Src family, Src and Fyn. DISCUSSION We have investigated the mechanism by which the AChR becomes tyrosine-phosphorylated in mammalian C2 myotubes and find that the tyrosine kinase Src associates with, and phosphorylates, the region encoded by exon 9 in the N-terminal half of the cytoplasmic loop of the ␤ subunit of the AChR. In addition, AChRs isolated from C2 myotubes were specifically associated with two Src-related kinases, Src and Fyn, and the AChRs were found to have tyrosine-phosphorylated ␤ subunits.
As judged by several criteria, binding of Src to GST fusion proteins containing the ␤ exon 9 region was specific. First, binding of Src to this region occurred in a sequence-specific way. Binding was not observed to fusion proteins containing only the C-terminal half of the ␤ loop, encoded by exon 10, nor to GST protein lacking AChR segments. Second, binding of ␤ exon 9 fusion proteins specifically occurred to Src, but not to Fyn or Yes, two other members of the Src family. Third, the corresponding regions from the ␥ and ␦ subunits of the AChR, although highly homologous to the exon 9 region of ␤, did not show detectable binding to either Src, Fyn, or Yes. Fourth, and most importantly, the ␤ exon 9 region was a substrate for both purified Src and for bound Src derived from muscle cell ex-tracts. Moreover, as the phosphorylation activity was abolished by immunodepletion of C2 lysates with Src-specific antibodies, Src appears to be the major, and perhaps the only, tyrosine kinase in C2 myotubes that constitutively recognizes, binds, and phosphorylates initially unphosphorylated ␤ subunits of the AChR. Src-mediated phosphorylation of the ␤ subunit may involve Tyr-357 but is more likely to occur on Tyr-390; the corresponding tyrosine of the Torpedo AChR is phosphorylated and the flanking amino acids show features of consensus tyrosine phosphorylation sites (34). Detailed analysis reveals that the residues surrounding Tyr-390 resemble sites phosphorylated by receptor tyrosine kinases more closely than motifs preferred by cytosolic kinases (34). Our observation that Src phosphorylates the ␤9 region, however, still renders Tyr-390 a potential substrate for this kinase. Adsorption with fusion proteins followed by phosphorylation revealed that the ␤ exon 9-homologous regions of ␦ and ␥ fail to associate with tyrosinephosphorylating activity (data not shown), which correlates with the inability of Src family kinases to bind to these regions. This specificity is notable, because the AChR isolated from postsynaptic membranes of the Torpedo electric organ can be tyrosine-phosphorylated in vitro at a high stoichiometry on its ␤, ␥, and ␦ subunits (9) and because the critical ␤ and ␦ tyrosine residues are conserved in the mouse AChR subunits. The selective binding of the unphosphorylated ␤ subunit, but not the ␦ or ␥ subunits, to Src, suggests that this interaction may be relevant to AChR phosphorylation during synaptogenesis at the mammalian neuromuscular junction.
As binding of the ␤ exon 9 region to Src is not phosphotyrosinedependent, it does not appear to involve the SH2 domain of Src. Apart from the catalytic region, Src contains two other functional domains that could be responsible for binding to the ␤ exon 9 segment, its unique N-terminal domain and its SH3 domain (29,30,33). The exon 9 region of ␤ contains 10 prolines, whereas the corresponding ␥ and ␦ segments only show 5 and 4, respectively. However, since the proline residues in ␤ are distributed throughout the exon 9 region and do not show the high proline density found in typical polyproline-SH3 domain interactions (35,36), we assume that binding to ␤ does not occur via the SH3 domain of Src. Binding may rather involve the unique domain of Src, as the unique domains of Src-related kinases have been shown to mediate constitutive binding to a FIG. 7. AChRs from C2 myotubes are co-isolated with the Src family kinases Src and Fyn. AChRs were isolated from C2 myotube lysates as described in the legend to Fig. 6, with the exception of control isolations with nonconjugated Sepharose. Kinases (arrowhead) were analyzed by specific immunoblotting. A fraction (0.03%) of the initial lysate was included as a standard (L). Quantitation of immunoblots shows that under these conditions ϳ0.04% of Fyn and ϳ0.01% of Src present in the initial lysate are recovered with toxin-Sepharose. In src-CT immunoblots this value is ϳ0.02%. Replacing digitonin with Nonidet P-40 produced similar results.
FIG. 6. Specific co-isolation of a Src-related kinase(s) with the AChR from C2 myotubes. C2 myotubes were lysed in a buffer containing 1% digitonin. AChRs were isolated by an incubation for 2 h at 4°C with ␣-bungarotoxin conjugated to Sepharose beads. As controls, we either added 10 M excess free toxin to the lysate to compete for AChR binding to toxin-Sepharose, or we used nonconjugated Sepharose beads. A, analysis of co-isolated Src-related kinase(s) was performed by immunoblotting with src-CT. One or several kinases were isolated in a specific manner with toxin-Sepharose beads (solid arrowhead). The ϳ90-kDa src-CT-reactive, contaminating band represented nonspecific binding as it was found in both control and experimental samples. A fraction of the initial cell lysate was analyzed as a control (L). B, the nitrocellulose membrane was stripped and reprobed with an antibody recognizing phosphotyrosine. C, a second reprobing of the same blot was performed with antibody m124 reactive with the AChR ␤ subunit. The absence and presence of the ␤ subunit (open arrowhead) confirms that the AChR is specifically isolated with toxin-Sepharose. In this experiment, the amount of the control lysate loaded (L) was too small for the ␤ subunit to be visualized with m124 antiserum. Analysis shows that one or several Src-related kinases are co-isolated with the AChR and that the kinase(s) and the ␤ subunit of the AChR are tyrosine-phosphorylated. Similar results were obtained when using Nonidet P-40 instead of digitonin as the detergent. number of transmembrane receptors (29). For example, the unique region of Lck is responsible for its association with CD4 (37), and binding to components of the T-and B-cell receptors has been shown to involve the unique domains of Fyn and Lyn (38,39). Interestingly, to our knowledge, no constitutive interactions have been described between Src and transmembrane proteins lacking kinase activity, although Src is well known to bind via its SH2 domain to activated, phosphorylated receptor tyrosine kinases such as the platelet-derived growth factor receptor (40). Our observed binding of Src to the ␤ subunit of the AChR may thus represent a class of Src interactions that is specific for differentiated post-mitotic cells such as myotubes. One possibility that we cannot exclude is that a third, ancillary protein mediates the binding of Src to the ␤ exon 9 region. If such a protein is involved, however, it is not subject to tyrosine phosphorylation, as our assays did not identify any phosphorylated proteins other than Src and the fusion proteins.
None of the GST fusion proteins tested bound to Fyn or Yes, yet AChRs isolated from C2 myotubes by ␣-bungarotoxin-Sepharose are associated with Fyn as well as with Src. While we cannot exclude the possibility that association of the AChR with Fyn involves one or several domains of the receptor not covered by our fusion proteins, it seems most likely that this association involves an interaction between the tyrosine-phosphorylated AChR ␤ subunit and the Fyn SH2 domain. This inference is based on the observations that in C2 myotubes the ␤ subunit of the AChR is tyrosine-phosphorylated (Fig. 6) and that in the electric organ of Torpedo phosphorylated AChRs are associated with the SH2 domains of two kinases of the Src family, Fyn and Fyk (10,11). In co-immunoprecipitation experiments, Fyn interacted both with phosphorylated ␦ and ␤ subunits, whereas Fyk preferentially interacted with phosphorylated ␦ subunits. Expression of Fyk, a Src-related kinase with homologies to Fyn and Yes, has not been reported in organisms other than Torpedo. Furthermore, in our precipitates with ␣-bungarotoxin-Sepharose, we failed to detect tyrosine-phosphorylated proteins other than the AChR ␤ subunit, Src, or Fyn. Together, these findings suggest that abundant tyrosine phosphorylation of the AChR ␦ subunit, as well as binding of Fyk to phosphorylated ␦, may occur specifically in the electric organ of Torpedo but not in mammalian muscle. In mammalian C2 myotubes, on the other hand, our findings suggest a model in which newly synthesized, unphosphorylated AChRs are initially tyrosine-phosphorylated by Src on their ␤ subunits. The phosphorylated ␤ subunits may then interact with the SH2 domain of Fyn. Src may remain bound to the ␤ subunits to yield a Fyn⅐Src⅐AChR complex or dissociate from the AChR prior to its association with Fyn.
What is the functional significance of the association of the AChR with Src and Fyn in C2 myotubes? The most likely hypothesis is that AChR phosphorylation is related to clustering and immobilization of the AChR in the myotube membrane. Phosphotyrosine labeling, as observed by immunofluorescence, is concentrated both at the mature synapse and at AChR clusters in developing myotubes (14,15,19). In addition, a variety of physiological and nonphysiological stimuli that result in AChR aggregation, including agrin, neuregulin, basic fibroblast growth factor, polymer beads, and electric fields, appear to act through mechanisms that involve protein-tyrosine kinase activation (18,(41)(42)(43)(44). A number of postsynaptic proteins such as ␤-dystroglycan, syntrophin, paxillin, the 87-kDa protein, or dystrophin contain tyrosine phosphorylation consensus sites or have been shown to be tyrosine-phosphorylated in Torpedo electric organ (45)(46)(47)(48)(49). In our experiments with unstimulated C2 myotubes, small but significant fractions of the total kinases present in C2 cells are associated with the AChRs, ϳ1% of the total Fyn and ϳ0.25% of the total Src. Considering the high level of AChR expression in C2 myotubes, we thus expect only a small proportion of surface AChRs to be associated with Src and Fyn. C2 myotubes spontaneously form a small number of AChR clusters (27), and the phosphorylated AChRs and those associated with kinases could represent AChRs that are in the process of spontaneous cluster formation. Such a mechanism could involve phosphorylation of critical muscle proteins by AChR-associated Src and/or Fyn kinases and subsequent immobilization of phosphorylated AChRs by protein-protein recognition mediated by SH2 domain-phosphotyrosine interactions. In agreement with this idea, immunofluorescence studies have shown that C2 myotubes contain some domains composed of coextensive aggregates of phosphotyrosine and AChRs (19).
An important question raised by these experiments is whether Src and Fyn are part of the agrin signaling pathway. A number of experimental observations link tyrosine phosphorylation of the AChR with agrin-induced clustering. After treatment with agrin, tyrosine phosphorylation of the AChR occurs before AChR clustering, and inhibitors that block tyrosine phosphorylation of the AChR also block clustering (20,21,50). The agrin dose-response curve is the same for both events, and the two are correlated under several different conditions (20). Agrin has recently been shown to act via a receptor tyrosine kinase, MuSK, which apparently acts as part of a receptor complex involving other, unidentified proteins (23). Activation of MuSK has been shown to result in both autophosphorylation and tyrosine phosphorylation of the AChR (23,24). Thus, although the intracellular signaling pathway by which agrin induces AChR clustering is unknown, it appears to involve activation of protein-tyrosine kinases and tyrosine phosphorylation, perhaps of the AChR itself.
A possible interpretation of the association of Fyn and Src with the AChR in unstimulated cells and its relevance for agrin's signaling pathway is suggested by consideration of other receptors. Several signaling receptors that lack intrinsic tyrosine kinase activities, such as cytokine (e.g. interleukin-2) and lymphocyte (e.g. T-cell and B-cell) receptors, are heterooligomeric transmembrane proteins that constitutively associate with nonreceptor tyrosine kinases of the Jak and Src family, respectively, before onset of signaling (51)(52)(53). Upon extracellular stimulation with ligand, the cytoplasmic kinases associated with the receptors phosphorylate them, thereby beginning the signaling cascade. In a similar way, activation of the agrin receptor could, through as yet undefined steps, result in phosphorylation of the AChR by bound Src or Fyn and in increased association of the AChR with Fyn. This may then lead to phosphorylation of other postsynaptic proteins and/or their association with the AChR. Alternatively, tyrosine phosphorylation of the AChR may be unrelated to the initial clustering events but rather to downstream events in the pathway regulating AChR aggregation. Accordingly, the initial clustering of AChRs, leading to formation of microclusters, could itself result in activation of receptor-associated Src or Fyn. Tyrosine phosphorylation of the AChR might thus be related to enlargement and stabilization of the clusters or to the recruitment of other proteins to the aggregates. In any case, the elucidation of the signaling pathways and the protein-protein interactions related to AChR clustering is an important area of future research. Our experiments indicate that Src and Fyn, associated with the AChR, are likely to play important roles in these pathways. the preparation of ␣-bungarotoxin-Sepharose and Pam Schwartzberg for insights into signaling by Src-related kinases. Members of the Hall lab are acknowledged for critically reading the manuscript and Dr. Michael Ferns for helpful advice during the course of this study.
Note Added in Proof-Holmes et al. (Holmes, T. C., Fadool, D. A., Ren, R., and Levitan, I. B. (1996) Science, in press) have found that the potassium channel, hKv1.5, is also associated Src tyrosine kinase and that the channel is tyrosine-phosphorylated. Regulation of ionic channel function via interaction with Src may thus be a general phenomenon.