Caveolin-3 directly interacts with the C-terminal tail of beta -dystroglycan. Identification of a central WW-like domain within caveolin family members.

Caveolin-3, the most recently recognized member of the caveolin gene family, is muscle-specific and is found in both cardiac and skeletal muscle, as well as smooth muscle cells. Several independent lines of evidence indicate that caveolin-3 is localized to the sarcolemma, where it associates with the dystrophin-glycoprotein complex. However, it remains unknown which component of the dystrophin complex interacts with caveolin-3. Here, we demonstrate that caveolin-3 directly interacts with beta-dystroglycan, an integral membrane component of the dystrophin complex. Our results indicate that caveolin-3 co-localizes, co-fractionates, and co-immunoprecipitates with a fusion protein containing the cytoplasmic tail of beta-dystroglycan. In addition, we show that a novel WW-like domain within caveolin-3 directly recognizes the extreme C terminus of beta-dystroglycan that contains a PPXY motif. As the WW domain of dystrophin recognizes the same site within beta-dystroglycan, we also demonstrate that caveolin-3 can effectively block the interaction of dystrophin with beta-dystroglycan. In this regard, interaction of caveolin-3 with beta-dystroglycan may competitively regulate the recruitment of dystrophin to the sarcolemma. We discuss the possible implications of our findings in the context of Duchenne muscular dystrophy.

Construction of cDNAs-The construction of the alkaline phosphatase-tagged ␤-dystroglycan (AP-␤-DG) was as described previously (15). Briefly, AP-␤-DG is a fusion protein carrying the transmembrane and cytoplasmic domain of ␤-dystroglycan fused to the ectodomain of alkaline phosphatase (AP). The cDNA for caveolin-3 was as previously described (10).
Immunofluorescence-NIH 3T3 cells were grown on coverslips coated with poly-L-lysine (Sigma), fixed in 2% paraformaldehyde, and permeabilized in 0.1% Triton X-100. Cells were then immunostained with monoclonal anti-caveolin-3 and polyclonal anti-placental alkaline phosphatase. Bound primary antibodies were visualized with a fluoresceinconjugated anti-mouse antibody and a rhodamine-conjugated anti-rabbit antibody (Jackson ImmunoResearch Inc.). As expected, omission of the primary antibodies prevented immunostaining.
Preparation of Caveolae-enriched Membrane Fractions-Transfected 293T cells were scraped into 0.7 ml of Mes-buffered saline (MBS, 25 mM Mes, pH 6.5, 0.15 M NaCl) containing 1% (v/v) Triton X-100 (19 -29). Homogenization was carried out with 10 strokes of a tight fitting Dounce homogenizer. The homogenates were adjusted to 40% sucrose by the addition of 0.7 ml of 80% sucrose prepared in MBS and placed at the bottom of an ultracentrifuge tube. A 5-30% linear sucrose gradient was formed above the homogenate and centrifuged at 44,000 rpm for 16 -20 h in a SW60 rotor (Beckman Instruments). A light scattering band confined to the 15-20% sucrose region was observed that contained caveolin-3, but excluded most other cellular proteins. From the top of each gradient, 0.375-ml gradient fractions were collected to yield a total of 12 fractions. An equal amount of protein from each gradient fraction was separated by SDS-PAGE and subjected to immunoblot analysis.
Immunoblotting-Samples were separated by SDS-PAGE under reducing conditions, and transferred to nitrocellulose membranes (Schleicher and Schuell). The protein bands were visualized with Ponceau S (Sigma). Membranes were blocked with 5% low-fat dried milk in TBST (20 mM Tris-HCl, 150 mM NaCl, 0.1% Tween 20). Blots were then incubated at room temperature for 1 h with either anti-␤-dystroglycan IgG or anti-caveolin-3 IgG, washed in TBST, and incubated with a secondary antibody conjugated with horseradish peroxidase (BD Transduction Laboratories). Bound IgG were detected using a chemiluminescent substrate (Pierce).
Co-immunoprecipitation Assay-Co-immunoprecipitation studies were performed essentially as previously described (11). Briefly, 293T cells stably expressing AP-␤-DG were transiently transfected with increasing amounts of the caveolin-3 cDNA and lysed in a buffer containing 50 mM Hepes, pH 7.5, 1 mM EDTA, 1 mM dithiothreitol, 0.5% CHAPS, protease inhibitor (Roche Molecular Biochemicals). Precleared lysates were then incubated overnight at 4°C with a specific mouse monoclonal IgG directed against caveolin-3, and immunoprecipitated using protein A-Sepharose CL-4B (Pharmacia Biotech Inc.). After extensive washing, samples were separated by SDS-PAGE (12% acrylamide) and transferred to nitrocellulose. Blots were then probed with a mAb directed against ␤-dystroglycan.
FIG. 3. Co-immunoprecipitation of caveolin-3 with ␤-dystroglycan. A 293T cell line stably expressing ␤-dystroglycan was transfected with increasing amounts of the caveolin-3 cDNA. Cells were lysed and subjected to immunoprecipitation with specific a mouse monoclonal IgG directed against caveolin-3. After extensive washing, samples were separated by SDS-PAGE and transferred to nitrocellulose. Blots were then probed with a mAb directed against ␤-dystroglycan. Note that caveolin-3 specifically co-immunoprecipitates with ␤-dystroglycan; no ␤-dystroglycan was found in the precipitate lacking caveolin-3.
GST-␤-dystroglycan "Pull-down" Assay-GST-␤-dystroglycan fulllength, a variety of deletion mutants and GST alone (bound to glutathione-agarose beads) were extensively washed first with TNET buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100) (3 times) and lysis buffer (10 mM Tris, pH 7.5, 50 mM NaCl, 1% Triton X-100, 60 mM octylglucoside), both containing protease inhibitors. SDS-PAGE followed by Coomassie staining was used to determine the approximate molar quantities of the fusion proteins per 100 l of packed bead volume. Pre-cleared lysates of 293T cells overexpressing caveolin-3 were diluted in buffer A (10 mM Tris, pH 8.0, 0.1% Tween 20) and added to approximately 100 l of equalized bead volume for overnight incubation at 4°C. After binding, the beads were extensively washed with phosphate-buffered saline (6 times). Finally, the beads were resuspended in 3 ϫ sample buffer and subjected to SDS-PAGE.
Overlay Assay-Overlay binding assays were performed essentially as described previously, with minor modifications (15). Peptide dot blots (SPOTs; Genosys Biotechnologies) were blocked overnight with blocking buffer supplied by the manufacturer (Genosys, Inc.) and then probed with 125 I-labeled GST-Cav-3-WW, GST-Cav-1-WW, or GST alone. Bound protein complexes were visualized by autoradiography.
hours post-transfection, cells were collected into 1 ml of lysis buffer (10 mM Tris, pH 7.5, 50 mM NaCl, 1% Triton X-100, 60 mM octylglucoside), containing protease inhibitors. After pre-clearing, biotinylated ␤-dystroglycan-derived peptides (either WT or the AA mutant) were added pre-bound to streptavidin-agarose beads. After incubation for 4 h at 4°C, the beads were extensively washed with phosphate-buffered saline (6 times). Finally, the beads were resuspended in 3 ϫ sample buffer, separated by SDS-PAGE (12% acrylamide), and transferred to nitrocellulose membranes. Blots were probed with a mouse mAb directed against the N terminus of caveolin-3 (11). To generate streptavidin beads containing pre-bound biotinylated peptides, the beads (50 l) were incubated overnight with a 1-ml solution containing ϳ15 g/ml of peptide dissolved in lysis buffer.
Dystrophin Competition Assay-A 10-cm plate of confluent cells overexpressing AP-␤-DG and four 10-cm plates overexpressing caveolin-3 were lysed in Buffer B (10 mM Tris-HCl, pH 7.4, 5 mM EDTA, 300 mM NaCl, 1% Triton X-100, 0.1% SDS, 1% sodium deoxycolate, 1% Nonidet P-40, and protease inhibitors). Clarified AP-␤-DG cell lysates were diluted in Tween buffer and incubated with glutathione-Sepharose immobilized GST-dystrophin fusion protein, glutathione-Sepharose immobilized GST, or with glutathione-Sepharose alone. Samples were then treated with 0.75 ml of cell lysate from 293T cells transfected with caveolin-3 cDNA, or with an equal volume of lysis buffer. Following an overnight incubation at 4°C, glutathione-Sepharose conjugates were pelleted by centrifugation and washed extensively in phosphate-buffered saline. Complexes were disrupted by boiling in 3 ϫ Laemmeli buffer. Equal volumes of each sample were subjected to SDS-PAGE as described above.
WW-ligand Peptide Competition-293T cells were transiently transfected with the cDNAs encoding full-length caveolin-3 and ␤-dystroglycan. Forty-eight hours post-transfection, cells were lysed in IP buffer containing protease inhibitors. After pre-clearing, antibodies directed against caveolin-3 and a given peptide competitor (200 g/ml) were added to the cell lysates. After incubation for 3 h at 4°C, the immunoprecipitates were washed 3 times with IP buffer, the samples were separated by SDS-PAGE, and transferred to nitrocellulose membranes. Blots were probed with antibodies directed against ␤-dystroglycan.

Caveolin-3 and ␤-Dystroglycan Interact in Vivo-Caveolin-3
and ␤-dystroglycan are known to localize to the sarcolemma and both coincide with the distribution of dystrophin (11,31). To examine a possible association between caveolin-3 and ␤-dystroglycan, we used a heterologous expression system. NIH 3T3 cells were co-transfected with the cDNAs encoding caveolin-3 and alkaline phosphatase-tagged ␤-dystroglycan (AP-␤-DG) and subjected to immunostaining. Caveolin-3 expression was detected with a specific mouse monoclonal Ab (clone 26); ␤-dystroglycan expression was detected with a rabbit polyclonal Ab that recognized the alkaline phosphatase epitope. Cells expressing both gene products were visualized with imaging by confocal laser fluorescence microscopy ( Fig. 1).
Our results indicate that caveolin-3 and ␤-dystroglycan demonstrate significant co-localization at the level of the plasma membrane.
We next examined the subcellular distribution of caveolin-3 and ␤-dystroglycan. To separate membranes enriched in caveolin-3 from the bulk of cellular membranes and cytosolic pro-  18 -⌽-X-⌽-(X) 8 -W-(X) 3 -P (h, human; m, mouse; r, rat, d, dog; c, chicken; f, fugu (puffer fish)). Note that the WW-like domain within caveolins overlaps with a domain previously suggested to function as a putative membrane spanning domain (based soley on its predicted hydrophobicity); however, experimental evidence has now shown that this region does not attach caveolins to the membrane (39,40). In addition, caveolin-2 lacks certain critical residues that are required to form a WW domain. B, other WW-like domains. WW-like sequences have also been identified in the PDGF-␤ receptor and related receptors (32) and in CD45-AP (33). An alignment of caveolin-3 with the WW-like regions of PDGF-␤ receptor, CD45-AP, and YAP is shown and highlights the conserved Trp, aromatic (⌽), and Pro residues.
Caveolin-3 Directly Interacts with ␤-Dystroglycan teins, an established equilibrium sucrose density gradient system was utilized (19 -29). In this fractionation scheme, immunoblotting with anti-caveolin-3 IgG can be used to track the position of caveolae-derived membranes within these bottom-loaded sucrose gradients. Fig. 2 shows that ϳ90 -95% of caveolin-3 (fractions 4 and 5) is separated from the bulk of cellular proteins. We also examined the distribution of ␤-dystroglycan in these sucrose density gradients. Fig. 2 demonstrates that ϳ20 -25% of ␤-dystroglycan co-fractionates with caveolin-3.
To further investigate whether caveolin-3 and ␤-dystroglycan are physically associated, co-immunoprecipitation experiments were performed. Lysates from 293T cells overexpressing both proteins were immunoprecipitated with anti-caveolin-3 IgG. As a negative control, a lysate from cells overexpressing only ␤-dystroglycan was used. Immunoprecipitates were then subjected to immunoblot analysis with a mAb directed against ␤-dystroglycan. Fig. 3 shows that caveolin-3 specifically coimmunoprecipitates with ␤-dystroglycan; no ␤-dystroglycan was found in the precipitate lacking caveolin-3. In addition, transfection experiments employing an increasing amount of the caveolin-3 cDNA showed that an increasing amount of ␤-dystroglycan co-immunoprecipitated.
Caveolin-3 Binds to the C-terminal Domain of ␤-Dystroglycan-To examine whether caveolin directly interacts with ␤-dystroglycan, we expressed and affinity purified a GST fusion protein carrying the full-length ␤-dystroglycan molecule GST-␤-DG FL-(467-895). GST-␤-DG FL, immobilized on glutathione-agarose beads, was incubated with lysates from 293T cells transiently overexpressing caveolin-3. After extensive washing, the samples were resuspended in 3 ϫ sample buffer and separated by SDS-PAGE. Blots were then immunostained with anti-caveolin-3 IgG. Fig. 4A shows that caveolin-3 bound specifically to full-length ␤-dystroglycan, as compared with GST alone or with beads alone.

Caveolin-3 Directly Interacts with ␤-Dystroglycan
Identification of a Central WW-like Domain in Caveolin-3-As the extreme C-terminal domain of ␤-dystroglycan serves as the dystrophin anchoring site (13,14), and as this interaction occurs via the WW domain of dystrophin (15,16), we searched for a putative WW domain within the caveolin-3 protein sequence. We found two tryptophan (Trp) residues separated by 29 amino acids; interestingly, these Trp residues are highly conserved among other caveolin family members. In addition, the second tryptophan (Trp) is followed by a highly conserved proline (Pro) residue. A similar sequence arrangement is found in the WW domains of other proteins. Fig. 5A shows an alignment of the caveolin family members and highlights the conserved Trp, Tyr, and Pro residues. Note that caveolin-2 lacks certain critical residues that are required to form a WW domain.
Other WW-like sequences have been identified in the PDGF-␤ receptor and related receptors (32), and in CD45-AP (33). Fig. 5B shows an alignment of caveolin-3 with the WWlike regions of the PDGF-␤ receptor, CD45-AP, and YAP and highlights the conserved Trp, aromatic (⌽), and Pro residues. In the case of the mPDGF-␤ receptor, the WW-like sequence was shown to bind proline-rich ligands. It is possible that the WW domain, like the SH3 module, could "tolerate" insertions and maintain the basic core fold that mediates interaction with proline-rich ligands (34,35).
Interestingly, the region of caveolins that contains the central WW-like domain has been previously proposed to function as a putative membrane-spanning domain. However, this proposal was based solely on primary sequence predictions (36 -38). Several independent lines of evidence now suggest that this region of caveolins does not function as a membrane anchor. For example, this region of caveolins shows little or no affinity for membranes in vitro or in vivo (39,40), while other adjacent regions experimentally confer membrane attachment (39,40). In addition, the putative membrane spanning domains of caveolins 1 and 2 interact with each other to form hetero-oligomeric complexes that contain both caveolins 1 and 2 (41). Thus, this provides evidence that this putative membrane spanning segment can function as a domain that mediates protein-protein interactions (41). In support of this notion, a GST fusion protein containing the putative membrane spanning domain of caveolin-1 is soluble, does not aggregate in FIG. 7. Fine mapping of the region of ␤-dystroglycan that interacts with the WW-like domain of caveolins 1 and 3. A, sequence and description of the 7 different peptides synthesized directly on SPOTs membranes. B, these SPOTs membranes were probed with 125 I-labeled GST-Cav-3-WW, GST-Cav-1-WW, or with GST alone. Note that the WW-like domains of both caveolin-3 and caveolin-1 recognize the C-terminal 12 amino acids of ␤-dystroglycan that corresponds to the ligand of the WW domain of dystrophin. No binding activity was observed with GST alone.

FIG. 8. Recognition of ␤-dystroglycan by the WW-like domain of caveolin-3 is dependent on the PPXY motif.
A, sequence and description of the 2 different biotinylated peptides used in the pulldown assay. B, 293T cells were transfected with the cDNA encoding caveolin-3. Lysates were then prepared and incubated with biotinylated peptides pre-bound to streptavidin beads. After washing, these precipitates were subjected to Western blot analysis with antibodies directed against caveolin-3. Note that the wild-type ␤-dystroglycan-peptide containing the PPXY motif (Biot-␤DG-WT) effectively pulls down caveolin-3. In contrast, a mutant of the same ␤-dystroglycan-peptide lacking the PPXY motif (Biot-␤DG-AA) shows little or no binding to caveolin-3. Similarly, streptavidin beads alone showed no caveolin-3 binding.

Caveolin-3 Directly Interacts with ␤-Dystroglycan
solution, and shows no affinity for membranes in vitro (5,39). In addition, this putative membrane spanning segment is 32-33 amino acids, which is much longer than most transmembrane domains (18 -22 amino acids) (3,4). Primary sequence prediction programs indicate that this region is predicted to assume a ␤-sheet conformation (not shown), rather than the typical ␣-helical conformation of traditional transmembrane domains. A predicted ␤-sheet conformation is more consistent with a WW domain-like structure.
To test the hypothesis that this caveolin-3 WW-like domain recognizes ␤-dystroglycan, we constructed a caveolin-3 GST fusion protein containing this WW-like domain (residues 34 -129). As a critical control for these studies, we compared the binding of caveolin-3 to GST alone and agarose beads alone to rule out any possible nonspecific binding. As substrate for these binding experiments, lysates from cells overexpressing ␤-dystroglycan were utilized. Binding of AP-␤-DG was detected by Western blotting using a mAb against ␤-dystroglycan. Fig. 6, A and B, shows that the GST-caveolin-3 containing the WW-like domain specifically bound ␤-dystroglycan. In contrast, the homologous region of caveolin-2, which does not conform to the WW consensus (Fig. 5A), fails to bind ␤-dystroglycan (Fig. 6C).
Fine Mapping of the ␤-Dystroglycan Region which Interacts with the WW-like Domain of Caveolin-3-The dystrophin-binding site on ␤-dystroglycan has been localized to the C-terminal 15 amino acids of ␤-dystroglycan (13,14). Thus, we next investigated whether the caveolin-3 WW-like domain is able to recognize the same C-terminal 15 amino acids of ␤-dystroglycan.
A series of 12 amino acid peptides were directly synthesized on SPOTs membranes. These peptides included a number of known WW domain ligands and the ␤-dystroglycan ligand for the WW domain of dystrophin. Fig. 7A shows the sequence and description of these peptides. The SPOTs membranes were then probed with 125 I-labeled GST-Cav-3-WW, GST-Cav-1-WW, or GST alone. Fig. 7B shows that the WW-like domains of caveolin-1 and caveolin-3 are both capable of binding only SPOT number 5 (TPYRSPPPYVPP) which is the ␤-dystroglycan ligand (residues 884 -895) for the WW domain of dystrophin. Importantly, no nonspecific binding was observed with GST alone. These results suggest that caveolin-3 and dystrophin are able to bind the same site on ␤-dystroglycan. cell lysates were subjected to immunoblot analysis with a pAb directed against the AP-epitope. C, lysates from 293T cells transiently co-expressing AP-␤-DG (WT or ⌬PPXY) and caveolin-3 were immunoprecipitated with anti-caveolin-3 IgG. As negative controls, lysates from cells overexpressing AP-␤-DG alone were also utlilized. Immunoprecipitates were then subjected to immunoblot analysis with a mAb directed against the AP epitope. Note that caveolin-3 specifically co-immunoprecipitates with AP-␤-DG-WT, while little or no binding is observed with AP-␤-DG-⌬PPXY.

FIG. 10. Caveolin-3 and dystrophin compete for binding to the C-terminal tail of ␤-dystroglycan.
Lysates from a stable line overexpressing ␤-dystroglycan were incubated overnight with affinity purified GST-dystrophin (residues 3046 -3447) or GST alone immobilized on glutathione-agarose beads. In a parallel experiment, a cell lysate from 293T cells overexpressing caveolin-3 was mixed with the lysate from the stable line overexpressing ␤-dystroglycan. This mixed lysate was then incubated overnight with affinity purified GST-dystrophin or GST-alone immobilized on glutathione-agarose beads. Note that the presence of caveolin-3 disrupts the binding of dystrophin to ␤-dystroglycan. These results are as predicted and suggest that caveolin-3 competes with dystrophin for the same or an overlapping binding site on ␤-dystroglycan.
Recognition of ␤-Dystroglycan by the WW-like Domain of Caveolin-3 Is Dependent on the PPXY Motif-WW domains are known to bind to PPXY containing protein ligands; mutation of these critical residues is sufficient to prevent binding. To determine whether recognition of ␤-dystroglycan by the caveolin-3 WW-like domain is dependent on the PPXY motif, we next used a peptide-based pull-down assay. 293T cells were transfected with the cDNA encoding caveolin-3. Lysates were then prepared and incubated with biotinylated peptides pre-bound to streptavidin beads. After washing, these precipitates were subjected to Western blot analysis with antibodies directed against caveolin-3. Fig. 8 shows that the wild-type ␤-dystroglycan-peptide containing the PPXY motif (TPYRSPPPYVPP) effectively pulls down caveolin-3. In contrast, a mutant of the same ␤-dystroglycan-peptide lacking the PPXY motif (TPYRSAAAAVPP) shows little or no binding to caveolin-3. Thus, the PPXY motif and its surrounding sequence is necessary for caveolin-3 binding in vitro.
To further investigate the requirement for the PPXY motif, we next generated a C-terminal truncation mutant of ␤-dystroglycan (AP-␤-DG-⌬PPXY) that lacks the PPXY motif (Fig. 9, A  and B). To assess its possible interaction with caveolin-3, coimmunoprecipitation experiments were performed. Lysates from 293T cells transiently co-expressing AP-␤-DG (WT or ⌬PPXY) and caveolin-3 were immunoprecipitated with anticaveolin-3 IgG. As negative controls, lysates from cells overexpressing AP-␤-DG alone were also utlilized. Immunoprecipitates were then subjected to immunoblot analysis with a pAb directed against the AP-epitope. Fig. 9C shows that caveolin-3 specifically co-immunoprecipitates with AP-␤-DG-WT, while little or no binding is observed with AP-␤-DG-⌬PPXY. Taken together, these results suggest that the PPXY motif is critical for the interaction of ␤-dystroglycan with caveolin-3.
Competition of Caveolin-3 and Dystrophin for Binding to ␤-Dystroglycan-Since caveolin-3 and dystrophin are able to bind the same site on ␤-dystroglycan, we next wanted to test the hypothesis that they interact competitively with ␤-dystroglycan. Lysates from cells overexpressing AP-␤-DG were prepared and incubated with the GST-dystrophin-WW domain fusion protein. In a parallel experiment, cell lysates from cells overexpressing ␤-dystroglycan were premixed with lysates from cells overexpressing caveolin-3. The mixture was then added to beads containing the GST-dystrophin-WW domain fusion protein. As negative controls, reactions with GST alone and beads alone were performed. Fig. 10 shows that in the absence of caveolin-3, the GSTdystrophin-WW domain is able to pull-down ␤-dystroglycan. In contrast, the presence of caveolin-3 interferes with the interaction between ␤-dystroglycan and dystrophin, such that ␤-dystroglycan binding to GST-dystrophin-WW is no longer detected.
Is Binding of the Caveolin-3 WW-like Domain to ␤-Dystroglycan Affected by Phosphorylation of the PPXY Motif?-It has been recently shown that phosphorylation of the tyrosine in the PPXY motif of ␤-dystroglycan blocks its recognition by the WW domains of both utrophin (42,43) and dystrophin (16). Thus, we next examined if tyrosine phosphorylation of the motif might also block the recognition of this motif by the WW-like domain of caveolin-3.
Interestingly, Fig. 11 shows that the WW domain of caveolin-3 recognizes the wild-type unphosphorylated peptide (TPYRSPPPYVPP) and two different phosphorylated forms of the same peptide (TP(pY)RSPPPYVPP or TPYRSPPP(pY)VPP) equally well. Thus, unlike the binding of utrophin and dystro-phin to ␤-dystroglycan, recognition by caveolin-3 is tyrosine phosphorylation independent.
Deletion Mutagenesis of the Cav-3 WW-like Domain-In order to define a minimal functional region of the caveolin-3 WW domain, we used a mutagenesis approach. A series of eight GST-Cav-3 WW deletion mutants were generated and are shown schematically in Fig. 12A. These mutants were then used as the substrate for binding to ␤-dystroglycan.
To further investigate a possible role for the caveolin-3 scaffolding domain in ␤-dystroglycan binding, we next used a second independent approach, i.e. peptide competition. 293T cells were doubly transfected with cDNAs encoding full-length caveolin-3 and ␤-dystroglycan. Lysates from these cells were prepared and subjected to immunoprecipitation with anticaveolin-3 IgG. Co-immunoprecipitation of ␤-dystroglycan was then detected by Western blotting.
Prior to immunoprecipitation, peptides were added to these lysates. Fig. 13A shows that when a peptide encoding a known ligand (THFTFKDLHFKMFDV; from G␣ i2 ) to the caveolin scaffolding domain (44) was added, little or no effect was observed. In contrast, when a peptide encoding the WW domain ligand from ␤-dystroglycan (TPYRSPPPYVPP) was added, coimmunoprecipitation of ␤-dystroglycan with caveolin-3 was blocked (Fig. 13B). Importantly, a mutant of the same ␤-dystroglycan-peptide lacking the PPXY motif (TPYRSAAAAVPP) FIG. 11. Binding of the caveolin-3 WW-like domain to ␤-dystroglycan is not affected by phosphorylation of the PPXY motif. A, sequence and description of the 3 different peptides synthesized directly on SPOTs membranes. Arrows point at potential sites of tyrosine phosphorylation and the PPXY motif is underlined. B, these SPOTs membranes were probed with 125 I-labeled GST-Cav-3-WW. Note that the WW domain of caveolin-3 recognizes the wild-type unphosphorylated peptide (lane 1) and two different phosphorylated forms of the same peptide (lanes 2 and 3) equally well. Thus, unlike the binding of utrophin and dystrophin to ␤-dystroglycan, recognition by caveolin-3 is tyrosine phosphorylation independent. No binding activity was observed with GST alone (not shown).
These results are consistent with the idea that the caveolin-3 scaffolding domain does not participate in ␤-dystroglycan binding in vivo. In addition, they suggest that the WW domain within full-length caveolin-3 can functionally recognize the Cterminal tail of ␤-dystroglycan in vivo.

DISCUSSION
Dystrophin is the protein product of the Duchenne muscular dystrophy (DMD) gene (45,46) and is tightly associated with the sarcolemmal membrane (31). Dystrophin forms a complex with a series of specific dystrophin-associated glycoproteins, termed DAGs. One major component of this complex is dystroglycan (47).
Dystroglycan provides a continuos link between laminin-2 in the extracellular matrix and dystrophin that is attached to the intracellular cytoskeleton (48). Dystroglycan begins as a pre-cursor protein that is proteolytically cleaved into two interacting subunits, ␣and ␤-dystroglycan (49). ␣-Dystroglycan is a heavily glycosylated extrinsic membrane protein that interacts directly with laminin-2; in contrast, ␤-dystroglycan is an integral membrane glycoprotein that binds tightly to dystrophin. The dystrophin-anchoring site on ␤-dystroglycan is localized to the extreme C terminus of ␤-dystroglycan at amino acids 880 -895 (13,14).
Recent studies have shown that ␣-dystroglycan can function as an agrin receptor, suggesting that it may play a role in neuromuscular synapse formation (50,51). In addition, several lines of evidence suggest that ␤-dystroglycan is part of a membrane-anchored signal transduction complex that interacts with the Src homology 3 domain of Grb-2. Grb-2 is an adaptor protein that helps to initiate the Ras-MAP kinase signal transduction cascade and is involved in controlling cytoskeletal organization (52). Disruption of the dystrophin-glycoprotein com- FIG. 12. Deletion mutagenesis of the caveolin-3 WW-like domain. A, to define a minimal functional region of the caveolin-3 WW domain, we used a mutagenesis approach. A series of eight GST-Cav-3 WW deletion mutants were generated and are shown schematically. These mutants were then used as the substrate for binding to ␤-dystroglycan, as detailed in the legend of Fig. 6. B, note that, as predicted, a central region from residues 64 to 114 appears most critical for binding. C, note that GST-Cav-3 fusion proteins (residues 34 -74 and 55-74) containing the full-length caveolin scaffolding domain (residues 55-74) show no binding activity. plex underlies the molecular pathogenesis of a variety of forms of muscular dystrophy. This suggests that the extracellular matrix-cytoskeletal linkage is critical for maintaining the structural integrity of the sarcolemma (53).
DMD is one of the most common and severe muscle disorders caused by a deficiency of dystophin. Several morphological observations seemingly implicate muscle cell caveolae in the pathogenesis of DMD: (i) dystrophin has been localized to plasma membrane caveolae in smooth muscle cells using immunoelectron microscopy techniques (54) and (ii) another electron microscopy study demonstrates that skeletal muscle caveolae undergo characteristic changes in size, number, and their distribution in patients with DMD, but not in other forms of neuronally based muscular dystrophies examined (55). In accordance with an increased number of caveolae in DMD patients, recent studies have shown that caveolin-3 protein levels are dramatically up-regulated in mdx mice and in patients with DMD (56,57). These results suggest that up-regulation of caveolin-3 may contribute to the pathogenesis of DMD.
Previous co-localization, co-immunoprecipitation, and cofractionation studies have suggested that caveolin-3 is associated with the dystrophin-glycoprotein complex (11). However, under certain conditions caveolin-3 can be physically separated from the dystrophin-glycoprotein complex (12). Here, we have studied the relationship between caveolin-3 and the dystrophin-glycoprotein complex. We examined the direct interaction of caveolin-3 with an integral membrane component of the dystophin-glycoprotein complex, namely ␤-dystroglycan. Using co-localization, co-fractionation, and co-immunoprecipitation experiments, we demonstrate that caveolin-3 may interact with the dystrophin-glycoprotein complex through the integral membrane protein ␤-dystroglycan.
We have mapped the caveolin-binding domain on ␤-dystroglycan to the extreme C terminus of ␤-dystroglycan. Surprisingly, dystrophin is known to bind ␤-dystroglycan through 15 residues at the C-terminal of ␤-dystroglycan (13,14). Interestingly, the dystrophin/␤-dystroglycan interaction occurs primarily through the WW domain of dystrophin (15,16). The WW domain is a small domain of 38 -40 semiconserved amino acids that is widely distributed among various structural, regulatory, and signaling proteins (58,59). The WW domain is named after two highly conserved tryptophan (W) residues spaced ϳ20 -22 amino acids apart. Various WW domains have been implicated in mediating protein-protein interactions by binding to peptide sequences containing Pro-rich motifs, such as PPXY (60,61).
Thus, we searched for the presence of a conserved domain which is similar to the WW domain in the caveolin-3 protein. We found two highly conserved tryptophan residues (Trp) separated by 29 residues and followed by a highly conserved proline (Pro). We next constructed a caveolin GST fusion protein carrying this WW-like domain. We demonstrated that this FIG. 13. The ␤-dystroglycan WW domain ligand prevents the interaction of full-length caveolin-3 with ␤-dystroglycan. To further investigate the mechanism of the interaction of caveolin-3 with ␤-dystroglycan, we used a peptide competition approach. 293T cells were doubly transfected with cDNAs encoding full-length caveolin-3 and ␤-dystroglycan. Lysates from these cells were prepared and subjected to immunoprecipitation with anti-caveolin-3 IgG. Co-immunoprecipitation of ␤-dystroglycan was then detected by Western blotting. Prior to immunoprecipitation, peptides were added to these lysates. A, caveolin-scaffolding domain ligand. Note that when a peptide encoding a known ligand (THFTFKDLHFKMFDV; from G␣ i2 ) for the caveolin scaffolding domain is added, little or no effect is observed. B, WW domain ligand. Note that when a peptide encoding the WW domain ligand from ␤-dystroglycan (TPYRSPPPYVPP) is added, co-immunoprecipitation of ␤-dystroglycan with caveolin-3 is blocked. Importantly, a mutant of the same ␤-dystroglycan-peptide lacking the PPXY motif (TPYRSAAAAVPP) does not prevent co-immunoprecipitation of ␤-dystroglycan with caveolin-3.
FIG. 14. Schematic diagram summarizing the interaction of caveolin-3 with ␤-dystroglycan. In this report, we show that a WW-like domain within caveolin-3 recognizes the extreme C terminus of ␤-dystroglycan that contains a PPXY motif. Note that this WW-like domain is located between two distinct domains that have been experimentally shown to attach caveolins to the membrane (40). We have previously termed these functional domains N-MAD (N-terminal membrane attachment domain) and C-MAD (C-terminal membrane attachment domain) (40). central portion of the caveolin-3 protein is sufficient to mediate ␤-dystroglycan binding and that this binding is dependent on the PPXY motif in ␤-dystroglycan. We show that the WW-like domain within caveolin-3 competitively recognizes the same site as dystrophin on ␤-dystroglycan. We also demonstrate that the presence of caveolin-3 is able to disrupt the interaction between ␤-dystroglycan and dystrophin. These results are summarized schematically in Fig. 14.
In this regard, the interaction of caveolin-3 with ␤-dystroglycan may competitively regulate the interaction and recruitment of dystrophin to the sarcolemma. As such, the interaction between caveolin-3 and ␤-dystroglycan may play a key role in stabilizing and regulating the activity of the sarcolemmal membrane. Thus, understanding the competitive interaction of caveolin-3 and dystrophin with ␤-dystroglycan provides further insight into the structure and the molecular organization of the dystrophin-glycoprotein complex and may be of vital importance in elucidating the pathogenesis of a number of different forms of muscular dystrophy.
Recently, we have created transgenic mice that overexpress caveolin-3 (62). Analysis of skeletal muscle tissue from these mice reveals DMD-like myopathic changes and down-regulation of dystrophin and ␤-dystroglycan protein expression (62). Our current results now provide a possible mechanism to explain this phenotype. As we show here that caveolin-3 competes with dystrophin for binding to ␤-dystroglycan, transgenic overexpression of caveolin-3 would then be expected to prevent the interaction of dystrophin with ␤-dystroglycan. Inhibition of the dystrophin/␤-dystroglycan interaction is known to result in degradation of the dystrophin complex and is the molecular basis for Duchenne muscular dystrophy. Thus, these results are consistent with the idea that caveolin-3 competes for binding to ␤-dystroglycan in vivo.