Synaptic differentiation is defective in mice lacking acetylcholine receptor β-subunit tyrosine phosphorylation

Numerous ligand-gated channels, including receptors for glutamate,γ-amino butyric acid and acetylcholine, are tyrosine phosphorylated in vivo, but the role of this post-translational modification in neurotransmitter receptor function is poorly understood(Huganir et al., 1984; Qu et al., 1990; Moon et al., 1994; Wan et al., 1997; Swope et al., 1999). The nicotinic acetylcholine receptor (AChR) is a pentamer composed of four different subunits, and two of these subunits, β and δ, are tyrosine phosphorylated in vivo (Qu et al., 1990; Mittaud et al.,2001). Tyrosine phosphorylation of the AChR is stimulated by agrin, a ∼200 kDa protein that is synthesized by motor neurons and released from nerve terminals (Wallace et al., 1991). Agrin stimulates MuSK, a muscle-specific receptor tyrosine kinase that is concentrated in the postsynaptic membrane and critical for synapse formation (DeChiara et al.,1996; Glass et al.,1996). Activation of MuSK leads to the rapid tyrosine phosphorylation and subsequently to the clustering of AChRs. In mice lacking MuSK, AChRs are expressed, but they are neither tyrosine phosphorylated nor clustered (DeChiara et al.,1996; Glass et al.,1996; Smith et al.,2001). Taken together, these findings suggest that AChR tyrosine phosphorylation may have an important and possibly requisite role in clustering AChRs at synapses.

Consistent with this notion, prior studies, which analyzed transfected wild-type cultured myotubes reported that mutant AChR pentamers, which contain a β-subunit that cannot be tyrosine phosphorylated, cluster two-fold less efficiently than wild-type AChRs (Meyer and Wallace, 1998; Borges and Ferns, 2001). In addition, these mutant AChRs are extracted more readily than wild-type AChRs with a non-ionic detergent(Borges and Ferns, 2001). These data indicate that tyrosine phosphorylation of the AChR β-subunit has a role in clustering and stabilizing AChRs in cultured myotubes, but the role of AChR tyrosine phosphorylation in synaptic differentiation has not been studied.

To determine the role of AChR β tyrosine phosphorylation in clustering and anchoring AChRs at synapses, we generated mice with targeted mutations in the three tyrosines in the large intracellular loop of the AChRβ-subunit. Mice lacking AChR β-subunit tyrosine phosphorylation survive postnatally, but their neuromuscular synapses are simplified and contain a reduced number of AChRs, indicating that tyrosine phosphorylation of the AChR β-subunit has an important role in organizing and stabilizing AChRs at synapses.

The Chrnb1 gene was isolated from a mouse genomic BAC library(Research Genetics, Huntsville, AL) by screening for exon 10 using a PCR assay. A 8.5 kb HindIII fragment, extending from intron 7 to intron 11 of the Chrnb1 locus was subcloned, and a 7.8 kb XbaI-XhoI fragment was used to construct a targeting vector(Fig. 1A). PCR mutagenesis was used to convert codon 357 from TAC to TTC, codon 390 from TAT to TTT, and codon 442 from TAC to TTC (accession number M14537). Although substitution of these tyrosine residues with phenylalanine prevents phosphorylation, we cannot exclude the possibility that the tyrosine hydroxyl groups serve a different function and that removal of these hydroxyl groups leads to defects unrelated to phosphorylation. An frt-flanked PGK-neomycin resistance cassette was cloned into an EcoRV site, which we introduced in intron 10, and a β-actin-diphtheria toxin A cassette was introduced at the 3′-end of the genomic fragment to create the targeting construct(Fig. 1A). Nucleis (Lyon,France) performed the ES cell electroporation, ES cell screening and blastocyst injections. 384 G418-resistant clones were isolated; a PCR assay(one primer in neo, CCGGCAAGTTCCTATTCTC, and one primer 3′ to the targeting construct, TACACAACAAGCTCCAGG) identified 16 clones in which the targeting construct recombined into the AChR β locus. Gene targeting in 10 clones was verified by probing Southern blots of HindIII-digested genomic DNA with a HindIII-XhoI fragment(Fig. 1B). In addition, exons 10 and 11 were sequenced to ensure that recombination did not occur between the neo cassette and Y357; this analysis demonstrated that four clones contained the three desired tyrosine to phenylalanine mutations. Three targeted ES cell lines were injected into blastocysts. One of the lines (3B4)produced chimeric offspring and was subsequently outcrossed to C57BL6 mice. The frt-flanked neo cassette was excised by further mating to FLPE mice (Jackson Laboratory, Bar Harbor, ME, USA), and the mice were genotyped by PCR (CCCGGACCTGCGACGATTTA; ACTAGGGTCCCGACGCTTGT).

Whole leg muscles from P0 mice were homogenized in lysis buffer [50 mM sodium chloride, 30 mM triethanolamine pH 7.5, 50 mM sodium flouride, 5 mM EDTA, 5 mM EGTA, 2 mM sodium orthovanadate, 1 mM N-ethylmaleimide, 1 mM sodium tetrathionate, 1 μg/ml pepstatin plus Complete protease inhibitors (Roche, Basel, Switzerland)] with a PT 10/35 Polytron (Kinematica AG, Littau-Lucerne, Switzerland) at 4°C. NP-40 was added to a final concentration of 1%, and the extract was incubated with rocking for 30 minutes at 4°C. Insoluble proteins were removed by centrifugation, and the supernatant was pre-cleared for 1 hour at 4°C against streptavidin-agarose(Sigma, St Louis, MO). The supernatant was collected and incubated for 1 hour at 4°C with biotinylated-α-bungarotoxin (α-BGT; Invitrogen,Carlsbad, CA, USA). AChR complexes were precipitated overnight with streptavidin-agarose, followed by washing (three washes for 3 minutes each) in lysis buffer containing 1% NP-40. Western blots were probed with antibodies to phosphotyrosine (4G10, 1:1000; Upstate USA, Charlottesville, VA). The blots were subsequently stripped and reprobed with antibodies to the AChRβ-subunit (mAb124, 1:5000; a gift from J. Lindstrom, University of Pennsylvania, Philadelphia, PA).

P0 mouse diaphragms were dissected, and AChRs were labeled with 20 nM ¹²⁵I-α-BGT (Perkin Elmer, Waltham, MA) for 1 hour at 37°C in oxygenated L15 medium. Background binding, in the presence of 10 μM non-radioactive α-BGT, was ∼10% of the binding without competitor. Muscles were washed (five washes for 50 minutes) in oxygenated L15 medium. Bound ¹²⁵I-α-BGT was measured in a Wallac 1470 gamma counter(Perkin Elmer), and the muscle was weighed.

Tibialis anterior muscles from P30 mice were mildly fixed (in 0.1%paraformaldehyde in PBS) for 1 hour at 4°C, rinsed twice at 4°C in PBS, cryoprotected (in 30% sucrose-PBS) overnight at 4°C, and embedded in TissueTek (Sakura, Tokyo, Japan). Frozen sections (10 μm) were stained with the following antibodies: p-AchRβ1-Tyr 390 (1:200; sc-17087; Santa Cruz Biotechnology, Santa Cruz, CA), MuSK (1:1000; #83033), rapsyn (1:500; #232),APC (1:200; sc-895 Santa Cruz Biotechnology), Abl (1:200; sc-131 Santa Cruz Biotechnology), utrophin (1:10; Vector Labs, Burlingame, CA) and dystroglycan(1:100; gift from Kevin Campbell, University of Iowa, Iowa City, IA), and AChRs were labeled with Alexa Fluor 594-α-BGT (Invitrogen). The images were acquired with a 63× (1.4 NA) objective. AChRs, axons and nerve terminals from P0 and P30 diaphragms were stained, and synaptic AChR levels and size were measured as described previously(Jaworski and Burden, 2006). Briefly, we stained diaphragm muscle with Alexa Fluor 594-α-BGT,collected confocal image stacks at the same sub-saturating amplifier gain for both genotypes using a 40× (1.3 NA) objective, and measured the Alexa Fluor 594-α-BGT-stained area. We quantified actual levels from unprocessed images but adjusted image levels for display purposes. Quantification of the pattern (stripes and gaps) of AChR staining at individual synapses and assignment to one of three categories was done initially without knowledge of the genotype; after the data were acquired,using a 63× (1.4 NA) objective, the genotype was identified.

Levator auris muscles in terminally anesthetized mice were exposed and injected with 10 μg/ml tetrodotoxin in PBS to prevent muscle contraction,and subsequently fixed in vivo with 4% paraformaldehyde/1% glutaraldehyde in 110 mM sodium phosphate buffer, pH 7.3. After dissection, fixation was allowed to continue at room temperature for 90 minutes. The muscle was washed (five times for 25 minutes) in iso-osmotic phosphate buffer (150 mM sodium chloride,5 mM potassium chloride, 10 mM sodium phosphate, pH 7.3), and subsequently stained for acetylcholinesterase activity to locate the muscle area containing synapses (Karnovsky and Roots,1964). Following three washes (for 15 minutes) with 100 mM Tris,pH 7.2, 160 mM sucrose, the muscle was washed again (five washes for 25 minutes) in isosmotic phosphate buffer. Following treatment with osmium (4%osmium tetroxide in 140 mM sodium phosphate, pH 7.3) for 1 hour, the muscle was washed (twice for 1 hour) in water. The muscle was stained en bloc with saturated uranyl acetate for 1 hour, dehydrated in ethanol and embedded in Epon.

Diaphragm muscles from P30 mice were dissected in oxygenated high Mg²⁺-low Ca²⁺ Tyrode's solution (125 mM sodium chloride,5.37 mM potassium chloride, 24 mM sodium bicarbonate, 12 mM magnesium chloride, 0.5 mM calcium chloride and 5% dextrose). Glass microelectrodes were pulled with a Sutter P-2000 micropipette puller (Novato, CA) and filled with 3 M potassium chloride to give a resistance of between 40-60 MV. Recordings of miniature end-plate potential (mepp) amplitude were made with a Dagan IX-1(Minneapolis, MN) intracellular preamplifier connected to an ADInstruments Powerlab 8/30 data acquisition system (Colorado Springs, CO). Muscle fibers were impaled near the endplate, and the average rise-time (baseline to peak)was statistically similar and less than 3 mseconds for both genotypes. Mepp recordings were only made from fibers with resting membrane potentials between-70 and -85 mV and deviated no more than 15% during the duration of the recording. The data represent at least 30 mepps per muscle fiber.

We isolated two immortalized muscle cell lines from wild-type mice that carried the H-2Kb-tsA58 transgene and two muscle lines from the Chrnb1 mutant (hereafter referred to as AChR-β^3F/3F) mice that carried the H-2Kb-tsA58 transgene (Jat et al.,1991). Muscle cells were grown as described previously(Smith et al., 2001).

Myotubes were treated for 16 hours with 500 pM agrin (R&D, Minneapolis,MN), rinsed three times (in PBS) and fixed for 10 minutes (1% paraformaldehyde in PBS). After washing (twice for 10 minutes in PBS), the muscle was incubated in 0.1 M glycine (in PBS) for 10 minutes rinsed twice in PBS and incubated for 30 minutes in 1% BSA in PBS. AChRs were labeled with Alexa Fluor 594-α-BGT (Invitrogen; 1:1000, in 1% BSA-PBS) for 1 hour and washed(three washes for 30 minutes in PBS). Myotubes were permeabilized (in PBS containing 0.1% Triton X-100) for 5 minutes and stained with Alexa Fluor 488-phalloidin (1:250; Invitrogen) (in 1% BSA-PBS) for 20 minutes to label actin. Myotubes were washed (three washes for 15 minutes in PBS) and mounted in VectaShield (Vector Labs). Images of phalloidin-stained myofibers were collected using a Zeiss LSM 510 microscope (Oberkochen, Germany), with a 20× (0.7 NA) objective, and the number and size of AChR clusters in these myotubes was determined using the Volocity (Improvision, Lexington, MA)software package. The data represent the average from two cell lines for each genotype.

Differentiated myotubes were treated with 500 pM agrin in culture medium for 3 hours at 37°C. ¹²⁵I-α-BGT was added to fresh medium containing 500 pM agrin. Myotubes were incubated for a further 1 hour at 37°C and washed (three washes for 15 minutes in oxygenated L15 medium). AChRs were extracted by incubating myotubes in L15 medium containing 0.05%Triton X-100; the extraction solution was collected and replaced every 2 minutes for a total of 6 minutes. The remaining, unextracted AChRs were collected by incubating myotubes in L15 medium containing 1% SDS. The amount of AChR extracted with Triton X-100 and SDS was measured in a gamma counter. The data represent the average from two cell lines for each genotype. In some experiments, myotubes were pretreated with 20 nM staurosporine (Sigma) or vehicle (0.002% DMSO) for 16 hours prior to treating the myotubes with agrin.

Statistical analyses were performed using Minitab 15 (Minitab, State College, PA). All values are given as mean±standard error. Parametric tests were used only when the data were normally distributed as determined by the Kolmogorov-Smirnov test. Non-normal data were analyzed using the appropriate nonparametric test as indicated in the figure legend.

Previous studies showed that agrin stimulates phosphorylation of a single tyrosine residue, Y390, in the AChR β-subunit(Borges and Ferns, 2001). Because we were concerned that the two other tyrosine residues (Y357, Y442) in the main intracellular loop of the AChR β-subunit might be phosphorylated if phosphorylation of Y390 were prevented, we used site-directed mutagenesis and homologous recombination in ES cells to replace all three tyrosine residues in the main intracellular loop with phenylalanine(Fig. 1A,B; Materials and methods). We confirmed gene targeting by PCR, Southern blotting and DNA sequencing (Fig. 1B and data not shown). Mice lacking these tyrosine phosphorylation sites(AChR-β^3F/3F) are born alive, nurse normally, and are otherwise indistinguishable from wild-type littermates (see Fig. S1 in the supplementary material).

To determine whether mutation of these three tyrosine residues prevents AChR β-subunit tyrosine phosphorylation, we labeled AChRs in detergent lysates from P0 limb muscles with biotinylated-α-bungarotoxin(α-BGT), isolated AChR complexes with streptavidin-agarose and probed western blots with antibodies to phosphotyrosine. Fig. 1C shows that AChRs in AChR-β^3F/3F mutant mice, unlike AChRs in control mice, are not tyrosine phosphorylated. In addition, we stained frozen sections of muscle from wild-type and AChR-β^3F/3F mutant mice with antibodies against a phosphopeptide specific to AChR-β^Y390-P and found that these antibodies label synaptic sites in wild-type but not AChR-β ^3F/3F mutant mice(Fig. 1D). Taken together,these data indicate that tyrosine phosphorylation of the AChR β-subunit is eliminated in AChR-β^3F/3F mice.

AChRs at embryonic synapses are arranged in small ovoid plaques of uniform AChR density (Sanes and Lichtman,2001). To determine whether synapse formation is impaired in AChR-β^3F/3F mice, we stained whole mounts of diaphragm muscle from P0 mice with Alexa594-α-BGT to label postsynaptic AChRs and antibodies to Neurofilament and Synaptophysin to label presynaptic axons and nerve terminals. Fig. 2 shows that AChRs at synapses in wild-type and AChR-β^3F/3F mice are organized in small ovoid plaques, apposed by nerve terminals, indicating that this simple arrangement of AChRs at developing synapses is not dependent upon tyrosine phosphorylation of the AChR β-subunit(Fig. 2A).

In order to determine whether AChRs are clustered at synapses at their normal number and density, we measured the size of synaptic sites and the density of synaptic AChRs in wild-type and AChR-β^3F/3F mutant mice. We found that synaptic size is reduced by ∼30% (100±4.7%, n=4 for wild-type; 71.1±4.5%, n=6 for AChR-β^3F/3F mice), whereas AChR density is unchanged in AChR-β^3F/3F mice(Fig. 2B). These defects in AChR clustering are not due to a decrease in AChR expression, since the total number of surface AChRs, measured by ¹²⁵I-α-BGT binding, is comparable in AChR-β^3F/3F and wild-type mice(see Fig. S1D in the supplementary material).

The structure of neuromuscular synapses becomes more complex during postnatal development, as ovoid AChR plaques are transformed into complex,pretzel-like shapes, characteristic of adult synapses(Sanes and Lichtman, 2001). At P30, synapses in AChR- β^3F/3F mice are∼40% smaller than in wild-type mice (100±8.6%, n=3 for wild-type mice; 56.8±4.1%, n=5 for AChR-β^3F/3F mice), indicating that the decrease in synaptic size observed at P0 is not caused by a delay in synaptic differentiation (Fig. 2D). In addition, by P30 the density of synaptic AChRs is also reduced by ∼40%(100±6.3%, n=3 for wild-type mice; 61.2±5.1%, n=5 for AChR- β^3F/3F mice), leading to an approximately threefold reduction in the total number of AChRs at each synapse (Fig. 2D).

Since miniature end-plate potential (mepp) size is proportional to the density of synaptic AChRs, we measured the size and frequency of mepps in muscle from wild-type and AChR-β^3F/3F mutant mice in order to obtain a second, independent measure of AChR density. Fig. 2E-G shows that the frequency of mepps is comparable in mutant and control mice, whereas the amplitude of mepps is reduced by ∼30% in AChR-β^3F/3F mutant mice (0.95±0.05 mV, n=35 synapses from wild-type mice; 0.68±0.04 mV, n=35 synapses from AChR- β^3F/3Fmice). Thus, these data demonstrate that AChR β-subunit tyrosine phosphorylation is critical to cluster AChRs at their normal number and density at synaptic sites.

Concomitant with the formation of a complex, pretzel-shaped terminal arbor,the postsynaptic membrane becomes invaginated into deep and regularly spaced postjunctional folds (Salpeter,1987). AChRs are concentrated at the crests and along the upper portions of these postjunctional folds(Fertuck and Salpeter, 1976). When viewed en-face by light microscopy, this arrangement leads to a regular,striped appearance of AChRs at neuromuscular synapses(Anderson and Cohen, 1974). To determine whether AChR β-subunit tyrosine phosphorylation is necessary for this aspect of postsynaptic development, we collected confocal images of synapses, stained with Alexa Fluor 594-α-BGT and compared the organization of AChRs at wild-type and AChR-β^3F/3F mutant synapses in P30 mice(Fig. 3A-H). At synapses in wild-type mice, the synaptic area is organized into well-defined AChR stripes. Moreover, most synaptic branches contain few (two or less) gaps in AChR-rich areas (Fig. 3I,J). In contrast,the synaptic area in AChR- β^3F/3F mice is largely devoid of well-defined AChR stripes, and most terminal branches are interrupted by five or more gaps in AChR staining. These aberrations in synaptic morphology suggest that tyrosine phosphorylation of AChRβ-subunit is necessary for the normal development and/or maintenance of neuromuscular cytoarchitecture.

These alterations in AChR organization could be caused by defects in the formation of postjunctional folds. To investigate this possibility we examined the structure of neuromuscular synapses from P30 wild-type and AChR-β^3F/3F mice by electron microscopy. The organization of the neuromuscular junction in AChR-β^3F/3F mice appears largely normal(Fig. 3K-N): first, nerve terminals contain synaptic vesicles, some of which are focused across from the mouths of postjunctional folds; second, the synaptic basal lamina is interposed between presynaptic and postsynaptic membranes; third,postjunctional folds, which are lined by basal lamina, are readily apparent;fourth, the crests and upper portions of the postjunctional folds are often thick and darkly stained, probably because of the dense packing of AChRs(Fertuck and Salpeter, 1976)(Fig. 3M,N). We quantified the number of postjunctional folds in three ways: first, we calculated a fold-density by dividing the number of postjunctional folds, defined by invaginations lined with basal lamina, independent of whether these invaginations had visible mouths that open to the synaptic cleft, by the length of the synaptic cleft; second, we calculated a fold-index by dividing the total length of postjunctional fold membrane, independent of whether these folds had visible mouths that open to the synaptic cleft, by the length of the synaptic cleft; third, we calculated the density of postjunctional folds with openings, or mouths, to the synaptic cleft(Fig. 3O). According to the first and second methods, the number and length of postjunctional folds are normal in AChR-β^3F/3F mutant mice. According to the third method, however, the number of postjunctional folds is reduced by∼40% (0.38±0.02 mouths/μm, n=57 for wild-type mice;0.24±0.01 mouths/μm, n=75 for AChR-β ^3F/3F mice; Fig. 3O). Thus, in AChR-β^3F/3F mutant mice, postjunctional folds have fewer openings to the synaptic cleft than in wild-type mice, although the entire length of fold membrane appears normal(Fig. 3P,Q).

Agrin stimulates the clustering of multiple synaptic proteins, in addition to AChRs, in cultured myotubes, raising the possibility that the accumulation of these proteins at synapses may depend upon AChR β-subunit tyrosine phosphorylation. We stained frozen sections of muscle from P30 mice with antibodies to adenomatous polyposis coli (APC), rapsyn, MuSK, Abl1, utrophin and dystroglycan and found that each of these proteins is concentrated at synaptic sites in AChR-β^3F/3F mice(Fig. 4). Thus, the accumulation of these postsynaptic proteins at synapses is not dependent upon AChR β-subunit tyrosine phosphorylation.

To determine whether tyrosine phosphorylation of the AChR β-subunit has a role in clustering AChRs in response to agrin, we generated muscle cell lines from wild-type and AChR-β^3F/3F mice and examined their response to agrin. We stimulated wild-type and AChR-β^3F/3F myotubes with agrin for 16 hours, labeled AChRs with Alexa Fluor 594-α-BGT, and measured the number and size of AChR clusters (Fig. 5A). The number of spontaneous, agrin-independent AChR clusters is similar in mutant and wild-type myotubes(Fig. 5B). Agrin stimulation induces a ∼3.5-fold increase in the number of AChR clusters in wild-type myotubes but smaller, ∼1.4-fold increase in the number of AChR clusters in AChR-β^3F/3F myotubes(Fig. 5B). Wild-type and AChR-β^3F/3F myotubes contain large and small AChR clusters, and the number of AChR clusters of all sizes is reduced in AChR-β^3F/3F mutant myotubes(Fig. 5B). These results indicate that tyrosine phosphorylation of the AChR β-subunit is not required to cluster AChRs per se, but is necessary to fully respond to agrin and to maximally cluster AChRs.

Agrin treatment slows the rate of AChR extraction with non-ionic detergent,suggesting that agrin stimulation leads to a more stable association of AChRs with the cytoskeleton (Wallace,1992; Borges and Ferns,2001; Moransard et al.,2003). To determine whether tyrosine phosphorylation of the AChRβ-subunit is necessary to stabilize AChRs, we treated wild-type and AChR-β^3F/3F myotubes with agrin for 3 hours,labeled AChRs with ¹²⁵I-α-BGT for 1 hour in the presence of agrin, extracted labeled AChRs with non-ionic detergent for several minutes and measured the rate of AChR extraction. In wild-type myotubes, agrin stimulation causes a decrease in the rate of AChR extraction(Fig. 5C). By contrast, in AChR-β^3F/3F myotubes, agrin does not alter the rate of AChR extraction (Fig. 5C), indicating that tyrosine phosphorylation of the AChRβ-subunit is necessary to stabilize AChRs following agrin stimulation. Moreover, staurosporine treatment destabilizes AChRs in wild-type myotubes but fails to accelerate the rate of AChR extraction from AChR-β^3F/3F myotubes, indicating that staurosporine destabilizes AChRs by inhibiting tyrosine phosphorylation of the AChR β-subunit (Fig. 5C).

Agrin stimulation leads sequentially to MuSK tyrosine phosphorylation, AChR tyrosine phosphorylation and AChR clustering. Tyrosine phosphorylation of MuSK and clustering of AChRs are essential for the formation and function of neuromuscular synapses, but the role of AChR tyrosine phosphorylation in synapse formation had not been previously examined. Here, we show that tyrosine phosphorylation of the AChR β-subunit is necessary to form normal neuromuscular synapses. In the absence of AChR β-subunit tyrosine phosphorylation, developing synapses are smaller. Synaptic abnormalities become more profound as synapses mature, since synapses in postnatal mice are not only smaller, but also have a reduced density and an aberrant arrangement of synaptic AChRs and postjunctional folds. Nonetheless, mice lacking AChRβ tyrosine phosphorylation move and breathe normally, indicating that tyrosine phosphorylation of the AChR β-subunit is not an essential step in forming or maintaining functional neuromuscular synapses. The absence of behavioral defects in AChR-β^3F/3F mutant mice is probabbly the result of the large safety factor for synaptic transmission at neuromuscular synapses: although synapses in AChR-β^3F/3F mutant mice contain three-fold fewer AChRs than in wild-type mice, neuromuscular synapses contain four- to five-fold more AChRs than required to sustain synaptic transmission(Wood and Slater, 2001). Below, we discuss the implications of these results: first, the role of AChRβ-subunit tyrosine phosphorylation in clustering AChRs at synapses;second, the unexpected finding that synapses are smaller in mice lacking AChRβ tyrosine phosphorylation; third, the role of AChR β-subunit tyrosine phosphorylation in linking AChRs to the cytoskeleton and organizing postjunctional folds.

Previous studies showed that AChR pentamers that contain a mutantβ-subunit, which cannot be tyrosine phosphorylated, cluster two-fold less efficiently than wild-type AChRs (Borges and Ferns, 2001). Because the myotubes analyzed in these studies co-expressed wild-type as well as mutant AChR β-subunit, it remained possible that the wild-type AChR pentamers formed a scaffold that facilitated the clustering of mutant AChR pentamers and obscured a more dramatic role of AChR β tyrosine phosphorylation in clustering AChRs. We find that AChR clustering is reduced to a similar extent (2.5-fold reduction) in myotubes that express only mutant AChR pentamers. Thus, AChR β-subunit tyrosine phosphorylation contributes to, but is not essential for, AChR clustering.

We cannot exclude the possibility, however, that tyrosine phosphorylation of the β- and δ-subunits have redundant roles and that tyrosine phosphorylation of the AChR is essential for clustering AChRs. Nonetheless,the tyrosine phosphorylation sites in the β- and δ-subunits are embedded in different sequences, indicating that the phosphorylated subunits are unlikely to recruit the same adaptor protein(Colledge and Froehner, 1997). Notably, the tyrosine phosphorylation site in the δ-subunit conforms to a binding site for SH2 domains, whereas the tyrosine phosphorylation site in the β-subunit is not predicted to bind SH2 or PTB domains. For this reason, phosphorylation of the different subunits is unlikely to have a redundant role.

In wild-type myotubes, agrin induces the formation of AChR micro-clusters,in a Rac-dependent manner, and AChR macro-clusters, in a Rho-dependent manner(Weston et al., 2000; Weston et al., 2003). We find that the number of micro- and macro-clusters are reduced to the same extent in AChR- β^3F/3F mutant myotubes. Our findings are consistent with two possibilities: (1) Rac- and Rho-dependent pathways each depend upon AChR-β^3F/3F tyrosine phosphorylation or (2) micro- and macro-clusters are similarly unstable in the absence of AChR- β^3F/3F tyrosine phosphorylation.

Because AChRs containing a mutant AChR β-subunit cluster less efficiently in cultured myotubes (Borges and Ferns, 2001) (Fig. 5A,B), we were not surprised to find that the density of AChRs is reduced at synapses in AChR-β^3F/3F mutant mice. We did not anticipate, however, that synaptic size would be reduced in AChR- β^3F/3F mutant mice. Notably, the density of synaptic AChRs is reduced in utrophin mutant mice, yet the size of utrophin-deficient neuromuscular synapses is normal(Deconinck et al., 1997; Grady et al., 1997). Thus, a reduction in the density, or packing of synaptic AChRs, does not necessarily lead to a reduction in synaptic size. Neuromuscular synapses are smaller but contain a normal density of synaptic AChRs in humans carrying mutations in DOK7 (Beeson et al.,2006; Slater et al.,2006). DOK7 is an adaptor protein that is recruited to tyrosine phosphorylated MuSK and necessary for signaling downstream from MuSK,including AChR tyrosine phosphorylation(Okada et al., 2006). Thus, it is possible that a failure to phosphorylate AChRs in patients harboring DOK7 mutations is responsible for the decrease in synaptic size.

Our studies in AChR-β^3F/3F muscle cell lines suggest that the defects in AChR clustering at synapses may be caused,at least in part, by a failure to link mutant AChRs to an underlying cytoskeleton. Rapsyn, which is essential for clustering AChRs, binds the AChRβ-subunit and associates more efficiently with AChRs following agrin stimulation (Burden et al.,1983; Gautam et al.,1995; Moransard et al.,2003). APC is also reported to bind the AChR β-subunit, to associate more efficiently with AChRs following agrin stimulation, and to be required for clustering AChRs in cultured muscle cells(Wang et al., 2003). Although these data raise the possibility that alterations in the binding of rapsyn and/or APC to AChRs might underlie the synaptic defects observed in AChR-β^3F/3F mice, we find that APC and rapsyn are concentrated at synaptic sites in the absence of AChRβ-subunit tyrosine phosphorylation. Although we cannot exclude the possibility that reduced amounts of rapsyn and/or APC are present at synapses in AChR-β^3F/3F mutant mice, these data demonstrate that the accumulation of rapsyn and APC at synapses does not depend upon AChR β-subunit tyrosine phosphorylation.

The reductions in AChR density and postjunctional fold formation in AChR-β^3F/3F mice are strikingly similar to the synaptic defects observed in utrophin mutant mice(Deconinck et al., 1997; Grady et al., 1997). Utrophin and AChRs colocalize at the tops of postjunctional folds and form a complex in cultured muscle fibers (Bewick et al.,1992; Fuhrer et al.,1999). These data suggest that tyrosine phosphorylation of the AChR β-subunit and utrophin might function in the same pathway. Utrophin,however, remains concentrated at synapses in AChR-β^3F/3F mice, and the AChRβ-subunit is tyrosine phosphorylated in utrophin mutant mice (data not shown). Thus, these data do not support the idea that the defects in AChR-β^3F/3F mice are due to a failure to recruit utrophin or that the synaptic defects in utrophin mutant mice are due to a lack of AChR β-subunit tyrosine phosphorylation.

How might tyrosine phosphorylation of the AChR β-subunit regulate the formation of postjunctional folds? Our data are consistent with the possibility that tyrosine phosphorylation of the AChR β-subunit provides a docking site for a protein(s) that is involved in forming and/or stabilizing postjunctional folds. Alternatively, others have suggested that the dense packing of conically shaped AChRs may be sufficient to induce folding of the postsynaptic membrane (Unwin,2005) raising the possibility that the reduced density of synaptic AChRs in AChR-β^3F/3F mice might lead to a reduction in the number of postjunctional folds(Slater et al., 1997).

Interestingly, synapses in patients carrying mutations in DOK7have fewer postjunctional folds (Beeson et al., 2006; Slater et al.,2006). Since tyrosine phosphorylation of the AChR β-subunit depends upon DOK7, these findings raise the possibility that the reduced number of postjunctional folds in patients harboring mutations in DOK7 may be due, at least in part, to a failure to fully tyrosine phosphorylate the AChR β-subunit(Okada et al., 2006).

We are grateful to Jon Lindstrom for providing a monoclonal antibody(mAb124) to the AChR β-subunit, to Doju Yoshikami for helpful advice and to Kevin Campbell for kindly providing utrophin^-/- tissue and antibodies to dystroglycan. We thank Jihua Fan for excellent technical assistance, Peter Hallock for comments on the manuscript, and Anya Temple for the drawings of postjunctional folds. This work was supported by funds from the NIH (NS 36193) to S.J.B. and a NRSA predoctoral fellowship to M.B.F.