The sonic hedgehog receptor patched associates with caveolin-1 in cholesterol-rich microdomains of the plasma membrane.

The Hedgehog signaling pathway is involved in early embryonic patterning as well as in cancer; however, little is known about the subcellular localization of the Hedgehog receptor complex of Patched and Smoothened. Since Hh has been found in lipid rafts in Drosophila, we hypothesized that Patched and Smoothened might also be found in these cholesterol-rich microdomains. In this study, we demonstrate that both Smoothened and Patched are in caveolin-1-enriched/raft microdomains. Immunoprecipitation studies show that Patched specifically interacts with caveolin-1, whereas Smoothened does not. Fractionation studies show that Patched and caveolin-1 can be co-isolated from buoyant density fractions that represent caveolae/raft microdomains and that Patched and caveolin-1 co-localize by confocal microscopy. Glutathione S-transferase fusion protein experiments show that the interaction between Patched and caveolin-1 involves the caveolin-1 scaffolding domain and a Patched consensus binding site. Immunocytochemistry data and fractionation studies also show that Patched seems to be required for transport of Smoothened to the membrane. Depletion of plasmalemmal cholesterol influences the distribution of the Hh receptor complex in the caveolin-enriched/raft microdomains. These data suggest that caveolin-1 may be integral for sequestering the Hh receptor complex in these caveolin-enriched microdomains, which act as a scaffold for the interactions with the Hh protein.

The Hedgehog signaling pathway, first described in Drosophila and conserved in vertebrates, is fundamental in early em-bryonic patterning of many structures, including the neural tube, axial skeleton, limbs, and lungs. Sonic Hedgehog (Shh), the most studied of the three vertebrate homologs of Drosophila Hedgehog, is a secreted protein that acts on target cells to increase transcription of several genes, including members of the Wnt and transforming growth factor-␤ families, and its receptor Patched. Patched (Ptc) is predicted to encode a large transmembrane protein that acts as a negative regulator of the pathway. It associates with a second transmembrane protein, Smoothened (Smo), which is a positive regulator of the pathway. Prior genetic and biochemical studies indicate that the two proteins form an unusual complex at the membrane that is inactive in the absence of the Shh ligand. Once the Shh protein binds to Ptc, this relieves the inhibition of Smo (by unknown mechanisms) and allows transduction of the Hedgehog signal.
Little is known about the structure or function of the Hedgehog receptor complex or the possible role of accessory proteins or lipids. The Ptc protein is predicted to have at least 12 transmembrane domains, and although it has little homology to other known receptors, it has been shown to directly bind Shh. The transmembrane domains of Ptc have a high degree of homology to the sterol-sensing domains of several proteins involved in cholesterol processing and trafficking, including NPC-1 (Niemann-Pick C protein-1) (1, 2), 3-hydroxy-3-methylglutaryl-CoA reductase, and SCAP (sterol regulatory elementbinding protein cleavage-activating protein). Smo encodes a serpentine, seven-transmembrane protein with characteristics of a G-protein-coupled receptor, including a glycosylated extracellular N terminus. Smo, however, does not directly bind Shh. The interaction with Ptc seems to be mediated through the N-terminal domain and/or the first two transmembrane domains of Smo, 1 with most of the signaling activity mediated through the third intracellular loop and the seventh transmembrane domain (3). The exact mechanism through which Smo transduces the Shh signal remains unclear, but it most probably involves a conformational change in the receptor complex, as the Ptc⅐Smo⅐Shh complex can be co-immunoprecipitated (4). The Shh protein undergoes autoproteolytic cleavage with covalent attachment of a cholesterol moiety to the N-terminal component of the protein. This modified N-terminal product is responsible for most of the apparent biological activity of the Shh protein. The cholesterol modification is not absolutely required for binding to Ptc or for limited biological activity, as several model systems have shown a response utilizing a bacterially derived Shh-N fusion protein (5). It has been proposed, however, that covalently linked cholesterol may modulate Shh activity, possibly through the sterol-sensing domains of Ptc, increasing the efficiency of signal transduction (6).
Caveolae are non-clathrin-coated invaginations of the plasma membrane that are important in endocytosis, cholesterol trafficking, and sequestering various lipid-modified signaling molecules in discrete microdomains. The caveolins, a family of three protein isoforms, are the major coat proteins of caveolae. Caveolin-1 directly binds to and transports cholesterol from the Golgi to the plasma membrane, and this association is required for caveolar formation (7)(8)(9)(10)(11). Caveolae are enriched in cholesterol and sphingolipids, are insoluble in nonionic detergents such as Triton X-100, and can be isolated as low density buoyant membranes in the absence of detergents. Associated with these complexes are various lipid-modified signaling molecules, including Ha-Ras, endothelial nitric-oxide synthase, serine/threonine kinases, several G-protein ␣-subunits, and Src tyrosine kinases (reviewed in Ref. 12). It has been postulated that caveolae may be signaling centers for multiple pathways and may regulate cross-talk between different pathways. Caveolin, per se, may directly influence signaling by serving as a molecular scaffold for signaling complexes (13)(14)(15) or indirectly modulate signaling by influencing cholesterol trafficking.
Since very little is known about the cellular localization of members of the Shh signaling pathway, we hypothesized that components of the pathway were in caveolae or lipid raft domains of cells. Recent data have shown that Hh is trafficked to lipid rafts in Drosophila, most probably through its association with cholesterol (16). Given these recent data and our preliminary observations of the Ptc trafficking pattern, we hypothesized that the Shh receptor complex is also targeted to these cholesterol-rich microdomains on the plasma membrane through an association between the Shh receptor Ptc and perhaps caveolin-1. In this report, we show that Ptc and caveolin associate with each other in the caveolar/lipid raft fraction of the plasma membrane, and we show data strongly implicating cholesterol as a key player in the transport and, most likely, the function of the Hedgehog receptor.

EXPERIMENTAL PROCEDURES
Constructs and Introduction of Mutations-A full-length human Ptc-GFP 2 construct (cDNA gift from Rune Toftgard) was made by creating an NheI/SalI cDNA fragment, ligating an 8-base pair linker to create an EcoRI site at the 5Ј-end, and then cloning this cDNA fragment into a mammalian GFP transfection vector (GFP vector C2, CLONTECH). This placed the GFP cDNA at the 5Ј-end of Ptc. The full-length human Smo-GFP construct was made by creating a HindIII/KpnI cDNA fragment (gift from Carol Wicking) and cloning it into a GFP transfection vector (GFP vector N3, CLONTECH), which placed the GFP cDNA at the 3Ј-end of Smo. The NheI/SalI Ptc cDNA fragment was also cloned into an expression vector (pCI-Neo mammalian expression vector, Promega), and the HindIII/KpnI Smo cDNA fragment was cloned into a FLAG mammalian transfection vector (pFLAG-CMV5a, Sigma), which placed the FLAG tag at the 3Ј-end of the Smoothened cDNA. All constructs were sequenced to confirm that the GFP or FLAG tags were in frame. The Myc-tagged caveolin-1 construct was a gift from Dr. Michael Lisanti (Albert Einstein College of Medicine).
The putative caveolin-binding motif found within the Patched protein sequence (amino acids 788 -798) was altered by polymerase chain reaction-based site-directed mutagenesis (Stratagene) using conditions recommended by the manufacturer. The peptide sequence YDFI-AAQFKYF was altered by replacing the underlined aromatic amino acids with alanine (ADAIAAQAKYA). Synthesis of an 82-base pair oligonucleotide (carried out at the Keck Facility, Yale University/ Howard Hughes Institute) that introduced eight nucleotide changes into the wild-type sequence resulted in a cDNA that would encode the altered protein. Sequencing of the entire PtcBS-GFP construct (where PtcBS is the caveolin-binding site on Ptc) confirmed the addition of the appropriate mutations. The altered construct was cloned into GFP vector C2 as described above (forward primer, 5Ј-GTACCTCGGGAAA-CCAGAGAA(T/G)(A/C)TGAC(T/G)(T/C)TATTGCTGCACAA(T/G)(T/C)-CAAATAC(T/G)(T/C)TTCTTTCTACAACATGTATATAGTCACCC).
Cell Culture and Antibodies-COS cells were maintained in Dulbecco's modified Eagle's medium with 10% fetal calf serum and antibiotics (penicillin and streptomycin). Chinese hamster ovary cells were maintained in Ham's F-12 medium with 10% fetal calf serum and antibiotics. An anti-Patched C terminus polyclonal antibody was a gift from Dr. Allen Bale (Yale University, New Haven, CT). The monoclonal antibodies were purchased as indicated: anti-Myc (Invitrogen), anti-FLAG (Sigma), anti-caveolin-1 (Transduction Laboratories, Lexington, KY), anti-␤-COP (clone maD, Sigma), and anti-GRP78 (Transduction Laboratories).
Co-immunoprecipitation and Western Blotting-COS cells were grown to 60% confluence and then transiently transfected (Effectene, QIAGEN Inc.) with the Patched-GFP and caveolin-Myc constructs under conditions recommended by the manufacturer. Cells were allowed to recover and express protein for 36 h and then lysed in lysis buffer on ice (50 mM Tris (pH 8.0), 140 mM NaCl, 1% Triton X-100, 0.4% deoxycholate, and protease inhibitors (Complete mini-protease inhibitor mixture tablet, Roche Molecular Biochemicals)). The lysate was incubated on ice for 30 min and then centrifuged at 12,000 ϫ g for 2 min at 4°C. The anti-Patched polyclonal antibody (5 l of serum) or the anti-caveolin-1 monoclonal antibody (2.5 g) was added to the lysate and rotated for 1.5 h at 4°C. Protein G-agarose (20 l of a 50% suspension in PBS; Sigma) was added and rotated for 1 h at 4°C. The beads were collected by centrifugation (brief pulse in a microcentrifuge) and then washed three times with lysis buffer. The beads were resuspended in 2ϫ Laemmli loading buffer and boiled for 5 min, and the supernatant was loaded onto an SDS-polyacrylamide gel.
Samples were separated by SDS-polyacrylamide gel electrophoresis using a 6% gel for the Patched and Smoothened proteins and a 15% gel for detection of the caveolin protein. The samples were transferred to nitrocellulose membrane (Biotrace TM , Gelman Instrument Co.), incubated in blocking solution (PBS, 0.1% Tween, and 5% nonfat dry milk) for 1 h at room temperature, and then washed twice with PBS/Tween. The membrane was incubated with the primary antibody (anti-caveolin-1 antibody or anti-Ptc polyclonal antibody) in blocking solution, rotated for 1 h at room temperature, and then washed three to four times with PBS/Tween. The membrane was then incubated with the appropriate secondary antibody tagged with horseradish peroxidase, and bands were detected by chemiluminescence using the ECL detection system and reagents supplied by Amersham Pharmacia Biotech. The membrane was exposed to film for up to 20 min and then developed in an Eastman Kodak X-Omat M43A processor.
Confocal Microscopy-COS cells were grown to 60% confluence on 35-mm coverslip plates (Mattek Corp., Ashland, MA), transiently transfected using Effectene transfection reagent, and allowed to recover for 24 -36 h. Cells were examined live under a confocal microscope (Zeiss, 60ϫ confocal objective; or Olympus, 60ϫ confocal objective) and assessed for localization and trafficking of the Ptc protein. Multiple images were obtained from the same cell, and a three-dimensional reconstruction of the cell was performed (NIH Image software). Time series and photobleaching experiments were also performed.
Immunocytochemistry-COS cells were grown to 60 -80% confluence in 6-well dishes containing untreated coverslips. They were transiently transfected and allowed to recover for 36 -48 h. The cells were fixed and permeabilized with methanol at Ϫ20°C for 2-4 min and then washed five times with PBS. Slides were blocked with goat serum; shaken for 1 h at room temperature; incubated with the anti-caveolin monoclonal antibody (1:100) diluted in PBS, 2% bovine serum albumin, and 10% goat serum; and shaken for 1 h at room temperature. The slides were washed four to five times with PBS and then incubated with Cy3-linked goat anti-mouse IgG secondary antibody (1:2000; Amersham Pharmacia Biotech) for 30 min at room temperature in the dark. Slides were washed five times with PBS and once with distilled water and sealed with Crystal Mount. Slides were then examined under the confocal microscope.
Subcellular Fractionation-Fractionation experiments were performed using a non-detergent method for isolation of caveolin-1-enriched buoyant membranes (20) and modified as described (39). 100-mm plates of COS cells were transiently transfected with Ptc-GFP or PtcB-SMut and allowed to recover and express protein for 48 h. To avoid interference from up-regulation of endogenous Ptc, cells transfected with Smo-GFP alone were processed after 40 h. The plates were then washed three times with PBS on ice. Cells were lysed in a solution containing 500 mM Na 2 CO 3 (pH 11.0) with protease inhibitors and incubated on ice for 10 min. The cells were scraped from the plate and homogenized in a Dounce homogenizer and then transferred to an Eppendorf tube and centrifuged at 1000 ϫ g for 10 min at 4°C. The supernatant was collected in a new Eppendorf tube and sonicated on ice. The supernatant was mixed with an equal volume of an 85% (w/v) sucrose/MB solution (MB ϭ 25 mM A-morpholine-ethanesulfonic acid (pH 6.5) and 0.15 M NaCl), allowed to equilibrate for 2 h at 4°C, and then placed at the bottom of a ultracentrifuge tube. An overlay of 6 ml of a 30% (w/v) sucrose/MB solution and then 3.5 ml of a 5% (w/v) sucrose/MB solution was added and centrifuged at 35,000 rpm for 18 h at 4°C in a Sorvall UltraPro-8 using a TH641 rotor. The gradient was fractionated into 1-ml fractions taken from the top, mixed with an equal volume of 2ϫ Laemmli loading buffer, and boiled for 5 min, and protein was separated by SDS-polyacrylamide gel electrophoresis. The gels were processed as described above, and the membranes were incubated with the anti-Ptc C terminus polyclonal antibody to detect Patched, with the anti-caveolin monoclonal antibody to detect caveolin-1, with the anti-FLAG M2 monoclonal antibody to detect Smo, with the anti-␤-COP monoclonal antibody to detect the Golgi protein, and with the anti-BIP/GRP78 antibody to detect the ER.
GST Fusion Proteins-GST fusion constructs of full-length caveolin-1 and the caveolin-binding site (amino acids 81-101) were made by creating two BamHI/EcoRI cDNA fragments by polymerase chain reaction and cloning these fragments into a GST vector (pGEX-2TK, Amersham Pharmacia Biotech). A GST fusion protein of the putative PtcBS (amino acids 772-810) was made by creating EcoRI/BamHI cDNA fragments by polymerase chain reaction and cloning this fragment into the GST vector. The constructs were sequenced to confirm that the GST cDNA was in frame and then transformed into Escherichia coli DH5␣ cells. A 1:100 dilution of an overnight culture of each construct and the GST vector alone was grown in 100 ml of 2ϫ yeast tryptone medium for 3 h at 37°C and then induced with isopropyl-␤-D-thiogalactopyranoside (0.5 mM) for 4 h at 37°C. The bacteria were pelleted at 7700 ϫ g for 10 min and washed with 3 ml of STE buffer (150 mM NaCl, 7.5 mM Tris (pH 8.0), and 3 mM EDTA). The bacteria were resuspended in 3 ml of STE buffer with lysozyme (100 g/ml) and placed on ice for 15 min. Dithiothreitol to a final concentration of 5 mM, phenylmethylsulfonyl fluoride to a final concentration of 100 M, N-lauroylsarcosine to a final concentration of 1.5%, and protease inhibitors were added, and the mixture was incubated on ice for 30 min. The lysate was homogenized with 15 strokes of a Dounce homogenizer and sonicated until clear. The lysate was centrifuged at 15,000 rpm in a Sorvall SS34 rotor for 15 min at 4°C. The supernatant was removed, and Triton X-100 was added to a final concentration of 2%. Glutathione beads (100 l of a 50% slurry in PBS) were added and incubated for 18 h at 4°C. The beads were collected by centrifugation at 500 ϫ g for 5 min and washed five times with STE buffer.
COS cells were transiently transfected with either the Patched or caveolin-1 construct and lysed 36 h later as described above. The GST beads (1-5 g of protein) were incubated with cell lysate (300 g of protein) for 2 h at 4°C. The complexes were washed four times with wash buffer (50 mM Tris (pH 8.0), 400 mM NaCl, and 1 mM EDTA); the beads were resuspended in 2ϫ Laemmli loading buffer, boiled for 5 min, and briefly centrifuged; and the supernatant was separated by SDS-

FIG. 1. Co-localization of Ptc and caveolin-1 by immunocytochemical staining.
In A, COS-1 cells were transfected with Ptc-GFP, fixed, and stained for native caveolin-1 (Cav) with the anticaveolin-1 monoclonal antibody. Single channel confocal imaging shows a vesicular pattern, which seems to be typical of both Ptc and caveolin-1. Dual channel confocal imaging shows co-localization of Ptc and caveolin-1 both within the cell and at the plasma membrane (yellow). In B, COS cells transfected with Ptc-GFP and a Myc-tagged form of caveolin-1 (Cav-Myc) again showed co-localization of the two proteins by confocal microscopy. polyacrylamide gel electrophoresis. The protein was transferred to a nitrocellulose membrane and processed as described above for detection of the Patched or caveolin-1 proteins.

RESULTS
Patched Associates with Caveolin-1-Ptc and Smo proteins are difficult to study in untransfected mammalian cells due to their low base-line levels of expression. Others who have studied endogenous Ptc in Drosophila (17,18) and epitope-tagged Ptc in mammalian cells (19) have shown that the majority of Ptc is found in intracellular vesicles, with a small proportion of the protein found at the membrane. To assess base-line localization and trafficking patterns for wild-type Ptc in our model system, we transiently transfected COS cells with a Ptc-GFP construct and examined living cells by confocal microscopy. As shown in Fig. 1A, wild-type Ptc-GFP was enriched in intracellular membranes, with the majority in the perinuclear region, reminiscent of the ER/Golgi (see also Supplemental Figs. 3 (lower panel) and 4 (upper panel)). Using time-lapsed microscopy, these small packets of protein trafficked from the Golgi to and along the cell membrane (Ptc.mov). The vesicular pattern displayed by Ptc-GFP was consistent across cell lines (Chinese hamster ovary, Madin-Darby canine kidney, HepG2, and the medulloblastoma cell line Daoy) and was consistent in cells transfected with untagged Ptc detected by immunocytochemistry using the anti-Ptc C terminus antibody (Supplemental Fig.  1A). A three-dimensional reconstruction of images obtained from a Z-series suggested that a portion of the Ptc protein localized at or just under the plasma membrane. The pattern was reminiscent of proteins associated with caveolae and their major structural protein, caveolin-1.
Next, we performed studies to determine if Ptc and caveolin-1 co-localize. Immunofluorescent microscopy of cells transfected with the Ptc-GFP cDNA (Fig. 1A, left panel) and immunolabeled for endogenous caveolin-1 (middle panel) demonstrated co-localization of the two proteins (right panel). In addition, cotransfection of a Myc-tagged version of caveolin-1 (caveolin-Myc) and Ptc-GFP constructs showed that the proteins co-localized both within the cell and at the plasma membrane (Fig. 1B). To examine if these proteins can interact biochemically, we performed co-immunoprecipitation in COS cells transiently transfected with Ptc-GFP and caveolin-Myc. As shown in Fig. 2 (A and B), immunoprecipitation of caveolin-1 (left panel) resulted in the co-association of Ptc, and immunoisolation of Ptc (middle panel) resulted in the association of caveolin-1. To investigate the potential interaction between Smo and caveolin-1, co-immunoprecipitation experiments were performed on lysates of COS-1 cells cotransfected with Smo-GFP and caveolin-Myc. Western blotting of the immunoprecipitates (Fig. 2C) showed that there was no association between these two proteins, and additional immunocytochemistry studies failed to show co-localization between the two proteins (Supplemental Fig. 2).
Smoothened Interacts with Patched, but Does Not Interact with Caveolin-1-It is known that several G-protein-coupled receptors are sequestered in a latent phase in caveolar or raft microdomains of the plasma membrane (9,14,15). Because onstrated that the Ptc and Smo proteins co-localized within discrete vesicles within the cytosol and at the membrane (third and fourth panels), suggesting that Ptc is necessary for trafficking Smo to the membrane. Moreover, Smo and Ptc were co-associated based on coprecipitation of the proteins from transfected cells (Fig. 3B). Collectively, these data suggest that the Ptc⅐Smo receptor complex forms prior to insertion in the plasma membrane and traffics as a heteromeric complex after synthesis in the ER.
Patched Associates with Caveolin-1 in Cholesterol-rich Microdomains of the Plasma Membrane-Given the above data showing a biochemical interaction between Ptc and caveolin-1 and co-localization of the proteins and the recent experiments showing Hh in lipid rafts (16), it seemed plausible that the Shh receptor complex might also be localized to these membrane microdomains. COS-1 cells were transiently transfected with the Ptc-GFP cDNA, and caveolin-enriched microdomains were isolated using a detergent-free purification method (20,21). Western blot analysis of sequential fractions from the sucrose gradient showed the majority of endogenous caveolin-1 in buoyant light membrane fractions 3 and 4 (Fig. 4A). Consistent with the confocal imaging studies, the majority of Ptc co-fractionated with Golgi and ER markers (lanes 8 -11); however, a significant amount of Ptc was also found in fractions 3 and 4, confirming that Ptc localizes to the same microdomains as caveolin-1. Similar results were obtained when these experiments were repeated on cells transfected with both Ptc and caveolin-Myc constructs.
Because our confocal microscopy data suggested that Ptc and GST alone served as a control for nonspecific binding. After extensive washing, GST fusion proteins were eluted and subjected to SDS-polyacrylamide gel electrophoresis analysis. Western blot analysis for the presence of Ptc was performed using the anti-Ptc C terminus antibody (Ptc CT Ab). Ptc bound specifically to both full-length caveolin-1 and the caveolin-1 scaffolding domain, whereas there was no binding of Ptc to GST alone. Equivalent amounts of cell lysates, GST fusion proteins, and GST were used in these experiments. In B, lysates from COS-1 cells expressing epitope-tagged full-length caveolin-1 were incubated with GST fusion proteins of PtcBS (amino acids 788 -798) or with GST alone. PtcBS was sufficient to specifically bind full-length caveolin-1 in these cell lysates, whereas there was no binding with GST alone. Endothelial cell lysate was used as a positive (Pos) control for the presence of caveolin-1.
pressing the Ptc-GFP construct. As shown in Fig. 5A, Ptc in cell lysates interacted with full-length caveolin-1 and the caveolin-1 scaffolding domain, but not with GST alone. Approximately 10 -30% of input Ptc interacted with the GST-caveolin fusion proteins.
Many proteins that can potentially bind to caveolin-1 contain a specific caveolin-binding sequence motif that may facilitate interaction with the caveolin-1 scaffolding domain (XXXXX, XXXXXX, or the composite XXXXXXX, where is the aromatic amino acid tryptophan, phenylalanine, or tyrosine, and X represents any other amino acid) (23). We searched the Ptc protein sequence for the presence of this motif and found a single sequence that matched the consensus binding motif: YDFIAAQFKYF (amino acid 788 -798). This sequence is highly conserved in Ptc between species as well as in the Ptc homolog Patched-2, but is not found in Smo. Thus, we expressed the putative caveolin-binding site on Ptc (PtcBS) as a GST fusion protein. As shown in Fig. 5B, PtcBS was able to isolate caveolin-1 from cell lysates expressing caveolin-1, whereas GST did not.
To assess the function of this putative binding motif on Ptc, we mutated the sequence by replacing the underlined aromatic amino acids with the amino acid alanine. Previous studies have shown that these substitutions inhibit the functional interaction between endothelial nitric-oxide synthase and caveolin-1 (12). The altered Ptc construct, PtcBSMut, was cloned into the GFP vector and transiently transfected into COS cells. Cells were examined live under the confocal microscope and compared with wild-type images. Unlike wild-type Ptc, which accumulated in the perinuclear region and moved to and from the membrane, the PtcBSMut protein accumulated in the same region, but failed to traffic throughout the cell (Fig. 6A, left  panel). Immunocytochemistry studies of PtcBSMut and caveolin-Myc also failed to show significant co-localization of the two proteins at the membrane, but did show some overlap within the perinuclear region (Fig. 6A, right panel). Co-immunoprecipitation studies on lysates derived from cells cotransfected with PtcBSMut and caveolin-1 showed that, despite altering this putative binding site, PtcBSMut and caveolin-1 continued to associate, but to a lesser degree (Fig. 6B). Fractionation of these cells revealed some PtcBSMut protein in the lipid raft component (Fig. 6C, middle panel), but this seemed decreased in comparison with wild-type Ptc-GFP (upper panel). This suggests that this binding site is important for the association between Ptc and caveolin-1, but that additional sites of interaction or accessory proteins may be necessary for the interaction of the proteins in vivo. (16). Prior studies in mammalian cells using methyl-␤-cyclodextrin (MBCD) have shown that cholesterol depletion abrogates the formation of caveolae, which is reversed upon cholesterol replacement (38). Due to the similarities between Ptc and several other proteins involved in cholesterol biosynthesis and transport (1,2) and the role of caveolin-1 in cholesterol trafficking (8 -11), we hypothesized that cholesterol might be involved in the transport of the Ptc⅐Smo⅐caveolin-1 complex to caveolae and/or insertion into the plasma membrane. We transfected COS cells with Ptc-GFP or PtcBSMut and Smo-FLAG and then treated them with serum depletion and/or MBCD. Confocal microscopy of cells transfected with wild-type Ptc-GFP and Smo-FLAG showed the receptor complexes localized to discrete vesicles, which remained intracellular and co-localized with Golgi markers (Fig. 7A, left and  middle panels). During live confocal imaging, there was little movement of these vesicles to the plasma membrane in the serum-depleted cell group and no movement of these vesicles in the MBCD-treated group. This abnormal trafficking pattern was reversed after serum repletion (Fig. 7A, right panel). In contrast, cells transfected with PtcBSMut and Smo-FLAG showed the formation of complexes in intracellular vesicles that were unable to traffic to the membrane (Fig. 7B, left and middle panels) even after cholesterol replacement (right panel). This indicates that cholesterol is most likely involved in the transport of the Ptc⅐Smo complex to the membrane. The failure of PtcBSMut to traffic correctly after replacement of cholesterol may be due to the fact that the mutated caveolin-1-binding site is also located in the sterol-sensing domain of Ptc.

Role of Cholesterol in Trafficking of the Hedgehog Receptor to Lipid Rafts on the Plasma Membrane-Cholesterol is a key component of lipid rafts in mammalian cells, and similar sterols seem to function as a necessary component in Drosophila raft formation
Detergent-free methods of caveolar isolation have shown that ϳ90% of caveolin-1 in the cell is associated with lipid rafts (24,25) and that cholesterol is necessary for the insertion of caveolin-1 into these rafts. We expected treatment with MBCD to cause a significant decrease in the amount of Ptc and caveolin-1 recovered from the lipid raft fraction of those cells. We performed detergent-free isolation and sucrose gradient frac- tionation on lysates of Ptc-GFP-expressing COS-1 cells treated with MBCD or control medium containing serum. Fig. 8 shows that, in control cells, the majority of caveolin-1 and a smaller proportion of Ptc-GFP were recovered in the raft fractions. In cells treated with MBCD, both caveolin-1 and Ptc were shifted from the raft fractions to the heavier membrane fractions. This further suggests that cholesterol is important for the correct trafficking of both Ptc and caveolin-1 to caveolae/lipid rafts. DISCUSSION In this report, we provide evidence supporting the concept that components of the Hh signaling pathway reside in caveolin-enriched microdomains. This assertion is supported by data demonstrating co-localization of Ptc with the caveolar coat protein caveolin-1, co-precipitation of Ptc (but not Smo) with caveolin-1, and co-fractionation of Ptc and Smo with caveolin-1 in buoyant membrane fractions containing caveolae/lipid raft microdomains. Furthermore, depletion of plasma membrane cholesterol results in the decreased amounts of both Ptc and caveolin-1 found in caveolae/lipid raft microdomains, suggesting that the local concentration of cholesterol influences the trafficking pattern of both proteins. Thus, our data, in conjunction with recent results demonstrating enrichment of Shh in lipid rafts, support the idea that Hh signaling resides in specialized, cholesterol-enriched microdomains of the cell.
The Hh receptor complex is an interesting and unusual heterodimer in that Ptc functions as the receptor domain for the Hh ligand, and Smo functions as the signaling domain for the complex. Little is known about the interactions between these two subunits, which compose the Hh receptor complex. Analysis of tumors suggests that Ptc functions not only at critical points in development, but also as a tumor suppressor where loss of both copies is required for tumor formation (29 -31). Activating mutations in Smo have also been described that implicate it as a potential oncogene (32,33).  (Fig. 3). Confocal microscopy of living cells transfected with Ptc alone shows that it is able to traffic effectively to the membrane from the Golgi without Smo. The results of our fractionation studies (Fig. 4) show that Ptc alone co-fractionates with caveolin-1 in the lipid raft compartment, whereas Smo alone does not. When Smo is cotransfected with Ptc, however, Smo is found to co-fractionate with Ptc and caveolin in the buoyant membrane fractions. Experiments with various Ptc mutants 1 also show that despite very abnormal trafficking patterns exhibited by some of the Ptc mutants, Smo consistently co-localizes with Ptc. Collectively, these data suggest that Ptc and Smo form a complex very early on, most probably in the Golgi, and are trafficked intact to lipid-rich microdomains on the plasma membrane (Fig. 9).
The raft hypothesis proposed by Simons and Ikonen (26) postulates that these lateral assemblies, composed of glycosphingolipids and cholesterol, function to aggregate certain proteins while excluding others. These rafts form a liquid order phase whose formation is driven by the interactions of the specific lipids involved. Caveolae, on the other hand, are specialized raft structures that are flask-shaped invaginations of the plasma membrane and are coated by caveolins. Indeed, caveolin-1 and -2 are necessary for the assembly of caveolae (36). In this study, we have used a well described detergent-free method for isolating caveolin-1-enriched membranes and lipid rafts. The strength of this method is that it can discern between buoyant membranes (caveolae/rafts) and heavy membranes The effects on trafficking seen with serum depletion or MBCD treatment were completely reversed after serum repletion (right panel). In B, COS-1 cells expressing PtcBSMut were subjected to serum depletion or MBCD treatment, followed by serum repletion. Similar to base-line studies using PtcBSMut constructs, the protein was not seen to traffic out of the Golgi to the plasma membrane (left and middle panels). This pattern was not reversed by serum repletion (right panel).
(ER, Golgi, and cytoskeleton); however, it cannot distinguish between proteins in caveolae versus lipid rafts. Here we show that Ptc co-segregates with caveolin-1 to these lipid-rich microdomains of the plasma membrane, consistent with the trafficking of other important signaling molecules such as Ha-Ras, endothelial nitric-oxide synthase, Src tyrosine kinases, and ␣-subunits of G-proteins (reviewed in Refs. 9 and 12).
The caveolin family consists of three distinct proteins that are differentially expressed in certain cell types. Caveolin-1 is the major structural protein in caveolae and is found in abundance in adipocytes, endothelial cells, and fibroblasts (10). The caveolin gene family has been conserved from Caenorhabditis elegans to humans, underscoring its evolutionary importance. Caveolin proteins form homo-and hetero-oligomers that directly bind and require cholesterol for insertion into the membrane. Caveolin-1 forms a hairpin-like structure, with the central portion of the protein inserted into the membrane and the N and C termini located in the cytoplasm (10). Work on heterotrimeric G-proteins as well as Ha-Ras, endothelial nitricoxide synthase, and others has shown that caveolin-1 interacts with these proteins through an area termed the caveolin-1 scaffolding domain (amino acids 82-101) (12)(13)(14). Our GST fusion protein studies have shown similar interactions between Ptc and caveolin-1, in which the full-length caveolin-1 protein and the caveolin-1 scaffolding domain were able to bind Ptc from transfected cell lysates, whereas GST alone did not. Analysis of proteins that interact with caveolin through its scaffolding domain have shown that these proteins contain a common sequence motif (XXXXX, XXXXXX, or XXXXXXX, where is Trp, Tyr, or Phe) (23). Sequence analysis of Ptc revealed that it contains such a motif (XXXXX) in the region of its seventh transmembrane domain, within the region of its sterol-sensing domain. As shown in Fig. 6, mutation of this binding motif by replacement of the aromatic amino acids with alanine produced a Ptc protein that was unable to traffic out of the Golgi complex to the membrane. However, mutation of these residues did not completely abrogate the association between Ptc and caveolin-1, indicating that other domains in Ptc, other proteins, and/or lipids may be involved in this interaction. It is known that caveolin-1 directly binds cholesterol and requires cholesterol for insertion into the plasma membrane and that its expression is regulated at the transcriptional level by cholesterol (8,11,34,35). Our initial studies using serum depletion and MBCD treatment suggest that it is quite likely that cholesterol is integral for the correct trafficking of the Hh receptor complex to the plasma membrane as well. This hypothesis is also supported by the similarities between Ptc and NPC-1 and the role of NPC-1 in cholesterol trafficking (1,2). The similarities between Ptc and NPC-1 raise interesting ideas about how Ptc traffics to the membrane and the role of its sterol-sensing domains in its interactions with Hh. Others have postulated that the Ptc sterol-sensing domain is probably important for binding of the Hh ligand, but it may also be critical for trafficking and localization of the receptor complex to caveolar domains of the plasma membrane.
Hh itself is an unusual morphogen in that it possesses an autocatalytic cleavage domain that cleaves the protein and covalently links cholesterol to the N terminus, forming the active portion. Recent data have shown that Hh is also modified by the addition of palmitate or other acyl residues to a Cys residue on the N terminus that seem to anchor Hh in the outer leaflet of the membrane bilayer, making it function like a glycosylphosphatidylinositol-anchored protein (16). This raises the conundrum of how a protein anchored in the lipid bilayer can function as a secreted morphogen. Sequestration of Ptc in these lipid-rich microdomains may be a method for the cell to separate this signaling pathway from others and to promote the interaction between Ptc and Shh.
Work on signaling molecules such as Ha-Ras, Src tyrosine kinases, endothelial nitric-oxide synthase, and G-protein-coupled receptors has shown that caveolin-1 plays an important role in holding these signaling molecules in a latent phase (13)(14)(15). Other studies have shown that this latent phase is also required for interaction of some of these molecules with caveolin-1 and that this interaction can be abolished by conversion of the signaling molecule to the activated form (20,37). Collectively, results from several labs suggest that Ptc sequesters Smo in an inactive state in the absence of the Hh ligand. An alternative interpretation, based on our studies, is that caveolin-1 may negatively regulate this heterodimeric receptor complex in a latent phase and that binding of Hh to Ptc removes this association, allowing Smo to transduce the signal. In addition, it is feasible that the Hh receptor complex is formed in the Golgi, associates with caveolin-1 and cholesterol, and is trafficked through an exocytic pathway to the plasma membrane. Additional studies examining the roles of caveolin-1 and cholesterol in signal transduction, stimulated by the Hh receptor complex and the biogenesis of these components, are clearly warranted.  The axis labels were omitted. Fig. 7A, the x axis should read Vc (mv) and the y axis normalized conductance. Fig. 7B, the x axis should read prepulse potential (mv), and the y axis should read normalized current. Fig. 7, C-E, the x axis should read Vc/prepulse potential (mV), and the y axis should read normalized conductance/current. Voltage-dependent activation and steady-state inactivation of ␣ 1I , ␣ 1G , and ␣ 1H calcium currents. A, activation curves. The current amplitude was converted to conductance by assuming a calcium reversal potential extrapolated from the linear, positive slope region of the I-V curve. The conductance at each potential was normalized to the maximum conductance and was averaged for each step potential. The symbols represent pooled data from ␣ 1I (filled triangles, n ϭ 5), ␣ 1G (filled circles, n ϭ 4), and ␣ 1H (filled squares, n ϭ 9). Solid lines represent the fitting with Boltzman equations with half-activation voltages (V 0.5a ) of Ϫ60.7, Ϫ51.73, and Ϫ43.15 mV and slope factors (k a ) of 8.39, 6.53, and 5.34 for ␣ 1I , ␣ 1G , and ␣ 1H , respectively. B, steady-state inactivation curves. The membrane potential was stepped to Ϫ30 mV from holding potentials ranging from Ϫ120 to Ϫ50 mV. The normalized peak amplitude of the currents elicited by the test pulse to Ϫ30 mV was plotted as a function of the holding potential. These data were fitted with a Boltzman equation (smooth curves). Half-inactivation voltage (V 0.5i ) and slope factor (k i ) were FIG. 8. Voltage dependence of deactivation kinetics. A, plot of mean deactivation time constants ( deac ) against repolarization potentials. Data represent mean and S.E. for the following number of cells: ␣ 1I (n ϭ 6), ␣ 1G (n ϭ 3), and ␣ 1H (n ϭ 4). Deactivation time constants were determined by fitting tail currents (B-D) with a single exponential. B-D, representative calcium current tail traces of ␣ 1I (B), ␣ 1G (C), and ␣ 1H (D). Currents were evoked using the following voltage protocols: a 9-ms step to Ϫ40 mV for ␣ 1G , a 20-ms step to Ϫ50 mV for ␣ 1I , and a 6-ms step to Ϫ30 mV for ␣ 1H followed by repolarization to potentials from Ϫ120 to Ϫ40 mV.