Wnt/β-Catenin Signaling Suppresses Rapsyn Expression and Inhibits Acetylcholine Receptor Clustering at the Neuromuscular Junction*

The dynamic interaction between positive and negative signals is necessary for remodeling of postsynaptic structures at the neuromuscular junction. Here we report that Wnt3a negatively regulates acetylcholine receptor (AChR) clustering by repressing the expression of Rapsyn, an AChR-associated protein essential for AChR clustering. In cultured myotubes, treatment with Wnt3a or overexpression of β-catenin, the condition mimicking the activation of the Wnt canonical pathway, inhibited Agrin-induced formation of AChR clusters. Moreover, Wnt3a treatment promoted dispersion of AChR clusters, and this effect was prevented by DKK1, an antagonist of the Wnt canonical pathway. Next, we investigated possible mechanisms underlying Wnt3a regulation of AChR clustering in cultured muscle cells. Interestingly, we found that Wnt3a treatment caused a decrease in the protein level of Rapsyn. In addition, Rapsyn promoter activity in cultured muscle cells was inhibited by the treatment with Wnt3a or β-catenin overexpression. Forced expression of Rapsyn driven by a promoter that is not responsive to Wnt3a prevented the dispersing effect of Wnt3a on AChR clusters, suggesting that Wnt3a indeed acts to disperse AChR clusters by down-regulating the expression of Rapsyn. The role of Wnt/β-catenin signaling in dispersing AChR clusters was also investigated in vivo by electroporation of Wnt3a or β-catenin into mouse limb muscles, where ectopic Wnt3a or β-catenin caused disassembly of postsynaptic apparatus. Together, these results suggest that Wnt/β-catenin signaling plays a negative role for postsynaptic differentiation at the neuromuscular junction, probably by regulating the expression of synaptic proteins, such as Rapsyn.

The precisely patterned synaptic connections are crucial for normal information flow. At the vertebrate neuromuscular junction (NMJ), 2 AChR clusters underneath nerve terminals are stabilized by Agrin, a motor neuron-derived glycoprotein, and the AChR-associated protein Rapsyn (1)(2)(3). On the other hand, motor neurons release negative signals to disperse noninnervated clusters and refine clusters at the synapses (4,5). The dynamic interactions between positive and negative signals are thought to be necessary for AChR cluster remodeling to better match nerve terminals and to form perforated or pretzellike structures (2,3,6). Genetic studies suggest that ACh may serve as a negative signal, and Agrin acts by counteracting the inhibitory effect of ACh to stabilize AChR clusters (4,7). Recently, we show that Rapsyn stabilizes AChR clusters by inhibiting calpain, a calcium-dependent protease activated by the cholinergic stimulation and involved in the dispersion of AChR clusters (8). Nevertheless, it seems likely that there are other negative factors regulating neuromuscular synaptogenesis, although their identity has not been well defined (4, 8 -10). A recent study shows that Wnt signaling inhibits presynaptic differentiation of axonal branches and helps to establish the precisely patterned neuromuscular connectivity in C. elegans (11), suggesting a negative role of Wnt signaling in synapse formation. Thus, it is valuable to investigate the role of Wnt signaling during vertebrate neuromuscular synaptogenesis.
Wnt is a family of secreted proteins that are implicated in neural development (12,13), neurite outgrowth, navigation, and synaptogenesis (14 -18). Wnts bind to the seven-transmembrane receptor Frizzled (Frz) and, via Dishevelled (Dvl), activate distinct pathways. Activation of the canonical pathway inhibits GSK3␤, leading to accumulation of cytosolic ␤-catenin (19,20). Stabilized ␤-catenin is translocated into the nucleus to regulate gene transcription. Several members of Wnt proteins have been shown to regulate presynaptic differentiation (21)(22)(23). Drosophila Wnt, wingless, has been implicated in pre-and postsynaptic differentiation (24). Although several components of Wnt signaling pathway, including Dishevelled and adenomatous polyposis coli, have been shown to regulate AChR clustering at vertebrate neuromuscular junctions (25)(26)(27) by interacting with MuSK or AChR, respectively, the role of Wnt in vertebrate NMJ development remains largely unknown.
Here we show that Wnt canonical signaling plays a negative role for the assembly of the postsynaptic apparatus at the vertebrate NMJ. Treatment with Wnt3a, a canonical Wnt, or up-prepared using Trizol reagent (Invitrogen) and reverse transcribed into cDNA using oligo(dT) primers according to the manufacturer's instructions. Quantitative reverse transcription-PCR was performed with a SYBR Premix Ex Taq TM kit (TaKaRa) and carried out in an Mx3000P Real Time PCR instrument (Stratagene). The housekeeping gene GAPDH was used as the reference for quantification. Primer sequences are as follows: Rapsyn, 5Ј-ATATC GGGCC ATGAG CCAGT AC-3Ј (forward) and 5Ј-TCACA ACACT CCATG GCACT GC-3Ј (reverse); GAPDH, 5Ј-TGAAG CAGGC ATCTG AGGG-3Ј (forward) and 5Ј-CGAAG GTGGA AGAGT GGGAG-3Ј (reverse).
Immunohistochemistry-Mice muscles or muscle cells in culture were stained with the indicated antibodies, washed, and incubated with fluorescein isothiocyanate-conjugated goat anti-rabbit or anti-mouse antibody and together with R-BTX to label the AChR (29). Images were collected with a Zeiss confocal microscope or Nikon fluorescence microscope. The total length of all AChR clusters or number of clusters was quantitatively analyzed following the method described previously (8). Sometimes, AChR clusters were classified into different groups according to cluster sizes (1-5 m, 5-10 m, and Ͼ10 m), and the number of different sized clusters was quantified.
Electroporation of DNA into Mouse Skeletal Muscles-Mice operations in this study were approved by the Animal Administration Committee of the institute. 6 g of testing plasmids were mixed with fast green and injected into the tibialis anterior muscles of anesthetized 6-week-old male BALB/c mice. Plaque electrodes (1 cm 2 ) were then placed on each sides of the leg, and eight pulses (duration 20 ms with field strength 200 V/cm at 2 Hz) were applied to the injected muscle. The muscles were dissected out 6 weeks after electroporation and analyzed for AChR clusters by whole mount staining with R-BTX (see Ref. 29).

Wnt3a Inhibits Formation of AChR Clusters in Cultured
Muscle Cells-To investigate the role of Wnt in postsynaptic differentiation, we examined the expression of different members of Wnt proteins in the mouse limb muscle. There are 19 isoforms of Wnt in mice (19,20). Among them, Wnt1, -4, -5a, -7a, -8a, and -11 were barely detectable (data not shown), whereas Wnt3a, although barely detectable at embryonic day 12 (E12), gradually increased in developing skeletal muscles and peaked at embryonic day 17 (E17) (Fig. 1A), a pattern coinciding with the period of postsynaptic differentiation at the NMJ in vivo (30).
Next we investigated the role of Wnt3a in AChR clustering in cultured muscle cells. We treated C2C12 myotubes with Wnt3a-conditioned medium containing ϳ40 ng/ml Wnt3a, a concentration sufficient to increase cytosolic ␤-catenin (Fig. 3, A and B) for 30 h, followed by treatment with Agrin for another 6 h to induce the formation of AChR clusters. As shown in Fig.  1, B and C, compared with cultures treated with L cell-conditioned medium, cultures treated with Wnt3a showed marked reduction in the total length of AChR clusters induced by Agrin. We classified AChR clusters into three groups according to cluster size and found that the number of small clusters (1-5 m in length) was increased, whereas the number of large clusters (Ͼ10 m in length) was decreased by Wnt3a treatment (Fig. 1D). These results indicate that Wnt3a inhibits the formation of full-size AChR clusters.
Further, we determined the effect of other Wnts on AChR clustering. Due to the difficulty in the purification of most Wnt proteins, we chose to overexpress these proteins in C2C12 muscle cells (with co-transfected GFP to mark cell morphology) (supplemental Fig. S1A). After transfection, the fully differentiated C2C12 cells were treated with Agrin (5 ng/ml) for 6 h, and AChR clusters were analyzed. We found that overexpression of Wnt1 or Wnt3a, two canonical Wnts, but not other Wnts (Wnt5a, -7a, or -11), caused a reduction in total length of AChR clusters induced by Agrin (supplemental Fig. S1B). In contrast, transfection with DKK1, an antagonist of canonical Wnt signaling (31), caused an increase in the length of AChR clusters (supplemental Fig. S1B). Thus, canonical Wnt signaling may play a negative role in regulating the formation of AChR clusters.
Wnt3a Promotes Dispersion of AChR Clusters in Cultured Muscle Cells-Next, we determined the effect of Wnt3a on the stability of AChR clusters. C2C12 myotubes were first treated with Agrin for 6 h to induce the formation of AChR clusters and subsequently switched to Agrin-free medium containing either Wnt3a-conditioned medium or DKK1 or both of them for the next 12 h ( Fig. 2A). We found that Wnt3a significantly decreased the total length of AChR clusters (Fig. 2, B and C) and cluster numbers in all sizes (Fig.  2D), suggesting that Wnt3a treatment induced declustering of AChRs. The reduction in AChR cluster size and number by Wnt3a was ameliorated in DKK1-treated cells (Fig. 2, B-D) with the effect of DKK1 in a dose-dependent manner ( Fig. 2C), suggesting the specificity of the Wnt3a effect. Interestingly, under the condition without Wnt3a, DKK1 treatment caused a mild, but significant, increase in the size and number of AChR clusters (Fig. 2, B and D), suggesting a role of endogenous Wnt in AChR cluster dispersion. These observations are in line with the notion that Wnt3a acts as a negative factor for the stabilization of AChR clusters. Wnt Signaling Down-regulates the Level of Rapsyn-Given the finding that Wnt3a inhibits and destabilizes AChR clusters, we further investigated the mechanism of Wnt3a action. It is known that Wnt/ ␤-catenin signaling regulates target gene expression (32); we thus hypothesized that Wnt/␤-catenin signaling may regulate AChR clustering by modulating expression of synaptic genes at the NMJ. To test  this hypothesis, C2C12 myotubes were treated with Wnt3aconditioned medium (containing ϳ40 ng/ml Wnt3a) or LiCl (20 mM), an inhibitor of GSK3␤, both of which activate Wnt canonical pathway and stabilize ␤-catenin. As expected, treatment with either Wnt3a or LiCl caused an elevation of ␤-catenin in muscle cells (Fig. 3, A, B, and D). Interestingly, these treatments caused a reduction of Rapsyn (Fig. 3, A, C, and D) without altering the levels of other synaptic proteins, such as AChR or MuSK (Fig. 3A). The effect of Wnt3a or LiCl on Rapsyn levels was not due to the influence on muscle differentiation, since myosin heavy chain, which is a marker for differentiated myotubes, was not affected by Wnt3a or LiCl (Fig. 3, A  and D). That LiCl down-regulates Rapsyn suggests a mechanism by which LiCl prevents Agrin from inducing the forma-tion of AChR clusters, as shown in a previous report (33). Together, these results suggest that Wnt/␤catenin signaling negatively regulates AChR clustering, probably by down-regulating the expression of Rapsyn.
Up-regulation of ␤-Catenin Inhibits the Formation of AChR Clusters-Since Wnt3a treatment resulted in elevation of ␤-catenin (Fig. 3) and inhibited AChR clustering ( Fig. 1 and supplemental Fig.  S1), we investigated the role of ␤-catenin in AChR clustering. C2C12 myotubes were transfected with plasmid encoding constitutively stabilized ␤-catenin, a mutated form of ␤-catenin with four serine/threonine residues (Ser 33 , Ser 37 , Thr 41 , and Ser 45 ) changed to alanine, which is believed to be resistant to proteasome-mediated degradation (34), and then treated with Agrin to induce the formation of AChR clusters (Fig. 4A). In comparison with cells transfected with empty vector and EGFP, the cells overexpressing ␤-catenin exhibited reduced responsiveness to Agrin stimulation, since the total length of AChR clusters or cluster number was decreased in these cells (Fig. 4, B and C). Thus, up-regulation of ␤-catenin inhibits AChR clustering.
Wnt/␤-Catenin Signaling Inhibits Rapsyn Gene Expression-Given the finding that Wnt3a or LiCl treatment decreased the protein level of Rapsyn in muscle cells (Fig. 3), we further examined whether Wnt/␤-catenin signaling affects transcription of Rapsyn by using quantitative PCR or promoter reporter assays. We found that the level of Rapsyn mRNA was decreased in C2C12 myotubes treated with LiCl (Fig. 5A), a GSK3␤ inhibitor that increased ␤-catenin levels (as shown in Fig. 3D) and is believed to inhibit AChR clustering (33). To further investigate the negative role of Wnt/␤-catenin signaling in Rapsyn expression, we generated a reporter construct, R5k-Luc, that contains the 5605-nt 5Ј-flanking region of the Rapsyn gene and the downstream cDNA encoding firefly luciferase. The relative luciferase activity of the reporter over co-transfected Renilla luciferase encoded by pRL-TK was measured in C2C12 myotubes. Cultured myotubes transfected with R5k-Luc and pRL-TK (10:1) were treated with Wnt3a-conditioned medium for 36 h or LiCl for 12 h, respectively, and then lysed for reporter assay. We found that both Wnt3a and LiCl showed an inhibitory effect on Rapsyn promoter activity (Fig. 5, B and C). Similarly, transfection of C2C12 cells with active ␤-catenin also caused a reduction in Rapsyn promoter activity, with an effect in a dose-dependent manner (Fig. 5D). In contrast, transfection with ␤-catenin siRNAs, which were shown to knock down the expression of co-transfected ␤-catenin (Fig. 5E), caused an increase in Rapsyn promoter activity (Fig. 5F).
Normally, ␤-catenin functions as a co-transcription factor for T-cell-specific transcription factor (TCF)-directed tran-  scription (35). As expected, transfection of ␤-catenin caused a marked increase in the reporter activity of Top Flash, an indicator for Wnt canonical signaling (35). Co-transfection of the dominant negative form of TCF, DN-TCF, which loses responsiveness to ␤-catenin, inhibited ␤-catenin-induced reporter activity of Top Flash (Fig. 5H). However, under our experimental conditions, neither wild-type TCF nor DN-TCF appeared to have any effect on Rapsyn promoter activity (Fig. 5, D and G). Expression of DN-TCF had no effect on ␤-catenin-mediated suppression of Rapsyn promoter activity (Fig. 5G). To further determine whether the suppression effect of ␤-catenin is through the TCF-binding motifs in the Rapsyn promoter, we generated mutations on these sites. There are three consensus TCF-binding sites (TCAAAG) in the 5605-nt 5Ј-flanking region of the Rapsyn gene (supplemental Fig. S2A). We found that Rapsyn reporter constructs with the deletion of individual or all three TCF-binding sites had similar responsiveness to ␤-catenin up-regulation (supplemental Fig. S2B). Together, these results suggest the TCF-independent activity of ␤-catenin in inhibiting Rapsyn gene expression. In line with this notion, expressing DN-TCF in C2C12 myotubes had no effect on Agrin-induced formation of AChR clusters (supplemental Fig. S3).
Wnt3a Acts through Down-regulating Rapsyn Expression to Disperse AChR Clusters-The expression of Rapsyn is shown above to be inhibited by Wnt3a and ␤-catenin. To further examine whether the dispersing effect of Wnt3a is indeed caused by the reduction of Rapsyn expression, we reconstituted the expression of Rapsyn driven by a promoter that does not respond to Wnt regulation in cells with Rapsyn knockdown. We took advantage of Rapsyn-siRNA that has been shown to knock down the expression of Rapsyn (8) (Fig. 6A). As expected, transfection of Rapsyn-siRNA prevented Agrin from inducing the formation of AChR clusters (Fig. 6B).
To reconstitute the expression of Rapsyn in cells with Rapsyn knockdown, we generated an HA-tagged construct harboring a mutated DNA sequence of Rapsyn that was resistant to siRNA knockdown, hereafter referred as Rap-HA-res233 (Fig. 6A, lane 3). Overexpression of Rap-HA-res233 partially prevented the failure in the formation of AChR clusters caused by Rap-siRNA233 (Fig. 6C, compare rows 2 and 3 from the top). However, wild-type of Rapsyn had no rescue effect, presumably due to the effectiveness of siRNA in down-regulating Rapsyn expression. The apparent reduction in both total length and number of AChR clusters in cells co-transfected with Rap-HA-res and Rap-siRNA compared with cells transfected with vehicle plasmids (Fig. 6, D and  E) was probably due to the aberrant stoichiometry between exogenous Rapsyn and endogenous AChRs. It is believed that the 1:1 stoichiometry of AChR and Rapsyn in muscle cells is important for AChR cluster formation (36,37). Both decreased and notably increased expression of Rapsyn have been shown to disrupt AChR clustering in myotubes (37). The dispersing effect of Wnt3a was examined in these Rapsyn-reconstituted  1 g), either alone or together with TCF (0.2 g) or increasing amounts of active ␤-catenin (0.6, 1.2, or 1.8 g). Co-transfected pRL-TK was used as transfection control in the reporter assay. E, HEK293 cells were co-transfected with GFP-␤-catenin and pSUPER vectors that encode control siRNA or individual ␤-catenin siRNA (#1, #2, or #3). Cell lysates were subjected to immunoblotting with antibodies against GFP or GAPDH. F, C2C12 cells were co-transfected with R5k-Luc (0.1 g) and pSUPER-nonsilencing or pSUPER-␤catenin-siRNA (0.6 g). Rapsyn reporter activity was determined. G, Rapsyn reporter activity was determined in C2C12 myotubes expressing ␤-catenin (0.6 g), DN-TCF (0.2 g), or ␤-catenin plus DN-TCF. H, C2C12 cells were transfected with 0.1 g of Top Flash, either alone or together with ␤-catenin (0.6 g) or ␤-catenin (0.6 g) plus DN-TCF (0.2 g). The luciferase reporter assay in F-H was performed as shown in B-D, and the data represent relative promoter activity (mean Ϯ S.E., firefly/Renilla). The experiments were repeated at least three times with similar results, each time in triplicate. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001, one-way ANOVA with Tukey's honesty significant difference post hoc tests. Fig. 6C, treatment with Wnt3a promoted dispersion of AChR clusters in muscle cells transfected with vehicle plasmid pSUPER (Fig. 6, C (first row of control and Wnt3a panels), D, and E). However, this effect was not observed in cells with Rapsyn reconstitution (Fig. 6, C (the Rap-siRNA233 ϩ Rap-res233 row of control and Wnt3a panels), D, and E). To further exclude the off-target effect of siRNA, we performed the experiments above using another set of constructs encoding a different siRNA (Rap-siRNA616) and the corresponding resistant Rapsyn (Rap-HA-res616) (supplemental Fig. S4). Similarly, Wnt3a caused dispersion of AChR clusters in control muscle cells transfected with nonrelated siRNA, and reconstituting the Rapsyn expression in Rapsyn knock-down cells prevented the dispersing effect of Wnt3a (supplemental Fig.  S4B). Thus, the decreased expression of Rapsyn induced by Wnt3a is indeed responsible for the dispersing effect of Wnt3a.

cells. As shown in
The Disassembling Effect of Wnt3a or ␤-Catenin on Postsynaptic Apparatus in Vivo-To evaluate the role of Wnt3a and ␤-catenin in NMJ development in vivo, we expressed Wnt3a or ␤-catenin in the tibialis anterior muscle of 6-week-old male BALB/c mice by using the DNA electroporation method (38 -40). Wnt3a cDNA was subcloned in a pCAG-EYFP-CAG vector that had two CAG promoters driving enhanced yellow fluorescent protein and target gene, respectively (28). In agreement with the in vitro studies in cultured muscle cells, the expression of Wnt3a induced disorganized structures of postsynaptic apparatus at the transfected muscle fibers (Fig. 7A). We found that the shape of AChR clusters at control muscle fibers expressing empty vectors exhibited a typical "pretzellike" structure ( Fig. 7A, left), whereas fibers expressing Wnt3a showed a highly disrupted pattern (Fig. 7A, right). The NMJs of Wnt3a-expressing muscle fibers exhibited smaller areas (Fig. 7B) and fewer fragments of AChR clusters (Fig. 7C). Similarly, ectopic expression of ␤-catenin also caused a disassembly of the postsynaptic structures, with reduced area and fragment number of AChR clusters (supplemental Fig. S5). Together, these data further support the negative role of Wnt/␤catenin signaling in regulating postsynaptic apparatus at the NMJ.

DISCUSSION
The spatial correspondence between nerve terminals and the AChR-rich membrane becomes more precise as development proceeds, suggesting that the remodeling of synaptic connections requires both positive and negative factors. Most research has focused on molecules that promote synaptogenesis, whereas the contribution of signals that negatively regulate synaptic assembly is less understood. ACh elicits a negative pathway to disassemble the postsynaptic apparatus via a calpainand Cdk5-dependent mechanism (4,7,8,41). However, other ACh-independent dispersing factors may exist to regulate FIGURE 6. Forced expression of Rapsyn prevents the dispersing effect of Wnt3a on AChR clusters. A, C2C12 cells were transfected with empty vector or vectors encoding Rapsyn-HA (Rap-HA) or Rap-HA-res, together with pSUPER or pSUPER-Rapsyn-siRNA (Rap-siRNA233). Cell lysates were subjected to immunoblotting with antibodies against HA, Rapsyn, or GAPDH. B, C2C12 myoblasts were transfected with pSUPER or pSUPER-Rapsyn-siRNA. Fully differentiated myotubes were treated with Agrin for 6 h. AChR clusters were labeled with R-BTX. C, C2C12 myoblasts were transfected with the indicated plasmids, and fully differentiated myotubes were treated with Agrin to induce the formation of AChR clusters. After Agrin withdrawal, the myotubes were treated with Wnt3a-conditioned medium or control medium for 12 h, followed by labeling with R-BTX for AChR clusters. D and E, quantification of AChR cluster length (D) and number (E) in 100-m myotubes. Data are shown as means Ϯ S.E. from at least three experiments. *, p Ͻ 0.05; ***, p Ͻ 0.001, one-way ANOVA with Tukey's honesty significant difference post hoc tests. postsynaptic differentiation. A recent report shows that Wnt signaling determines precise neuromuscular connectivity by inhibiting synapse formation in C. elegans (11), although the mechanism remains unclear. Thus, it is important to test the role of Wnt signaling in vertebrate NMJ development. In this study, Wnt3a, a canonical Wnt, is identified as a negative factor for postsynaptic differentiation at the NMJ. We found that Wnt3a inhibits AChR clustering, probably by down-regulating the expression of synaptic protein Rapsyn.
Rapsyn is a synaptic protein essential for AChR clustering and neuromuscular junction formation (30,(42)(43)(44)(45). This protein is thought to slow the metabolic degradation of the AChR (46,47) and function downstream of the Agrin/MuSK signaling, leading to AChR clustering (48). Recently, we found that Rapsyn stabilizes AChR clusters by interacting with calpain, a calcium-dependent protease involved in ACh-induced dispersion of AChR clusters (8). Interestingly, it is believed that the 1:1 stoichiometry of AChR and Rapsyn in muscle cells is important for AChR cluster formation (36,37). Both decreased and notable increased expression of Rapsyn have been shown to disrupt AChR clustering in myotubes (37). Thus, the expression of Rapsyn must be tightly regulated at the neuromuscular junction. A recent report shows that the Rapsyn promoter is positively regulated by Kaiso and ␦-catenin (49). The present study reveals an unexpected role of Wnt/␤-catenin signaling in regulating Rapsyn expression at the neuromuscular junction, where Wnt-mediated up-regulation of ␤-catenin may inhibit Rapsyn expression. The negative regulation of target gene expression by ␤-catenin shown here is in line with a previous report show-ing that the TCF⅐␤-catenin complex represses E-cadherin expression by binding and blocking the E-cadherin promoter (50). In addition to its role in regulating gene expression, ␤-catenin is known to be crucial for the formation of cell adhesion complex by interacting with cadherins and ␣-catenin (51). Interestingly, during the course of our study, Zhang et al. (52) have reported that ␤-catenin regulates AChR clustering by physical interaction with Rapsyn. We also found that knockdown of the expression of ␤-catenin using siRNAs decreased the formation of AChR clusters (supplemental Fig. S6). This effect may be due to the disruption of Rapsyn association with cytoskeleton or enhanced expression of Rapsyn (see Fig. 5F), both of which are believed to inhibit AChR clustering (37,52). In contrast to the effect of ␤-catenin in cultured muscle cells, a recent report shows that muscle-specific deletion of the ␤-catenin gene resulted in an increase in the size of AChR clusters, probably due to the retrograde effect of muscle ␤-catenin on presynaptic transmitter release (53). It is known that ACh acts to disperse AChR clusters (4,5,8). The decrease of transmitter release may cause the enlargement of AChR clusters. It is also possible that there are adaptation or redundant compensatory mechanisms in the ␤-catenin Ϫ/Ϫ mice. Nevertheless, these observations suggest the multiple roles of ␤-catenin at the NMJ. The appropriate level and localization of ␤-catenin might be crucial for maintaining the normal structure and functions of the synapse.
It is believed that accumulated ␤-catenin in the nucleus forms a complex with TCF/lymphoid enhancer factor-1 to regulate target gene expression (19,20). However, neither wildtype nor dominant negative forms of TCF appeared to have any effect on AChR clustering and Rapsyn promoter activity, suggesting that ␤-catenin functions in NMJ development, probably in a TCF-independent manner. Activated ␤-catenin is found to inhibit the expression of NF-B target genes (54). Interestingly, the Rapsyn promoter contains several NF-B binding sites and E-box consensus sequence CANNTG that is responsive to a variety of myogenic determination factors, including MyoD, myogenin, Myf5, and Myf6 (55). E-box mutations in the Rapsyn promoter correlate with congenital myasthenic syndrome (55). It will be interesting to determine whether ␤-catenin regulates Rapsyn expression via these factors.
Wnt expression or release has been shown recently to be regulated by neuronal activity in cultured hippocampal neurons or slices (56,57); it shall be of interest to determine whether Wnt expression or release in the muscle or motor neuron is regulated by neurotransmitter ACh. It is also possible that nerve terminals may release some unknown factors to antagonize Wnt/␤-catenin to regulate Rapsyn expression. In addition, other members of the Wnt family may also participate in NMJ development by distinct mechanisms.
Acknowledgments-We are grateful to Dr. L. Mei for comments on the manuscript and thank Drs. C. Liu, X. He, and X. Yu for valuable reagents and Dr. Q. Hu of the ION Imaging Facility for microscope images.