Regulation of cAMP-mediated Signal Transduction via Interaction of Caveolins with the Catalytic Subunit of Protein Kinase A*

cAMP-dependent processes are essential for cell growth, differentiation, and homeostasis. The classic components of this system include the serpentine receptors, heterotrimeric G-proteins, adenylyl cyclase, protein kinase A (PKA), and numerous downstream target substrates. Evidence is accumulating that some members of this cascade are concentrated within membrane microdomains, termed caveolae and caveolae-related domains. In addition, the caveolin-1 protein has been shown to interact with some of these components, and this interaction inhibits their enzymatic activity. However, the functional effects of caveolins on cAMP-mediated signaling at the most pivotal step, PKA activation, remain unknown. Here, we show that caveolin-1 can dramatically inhibit cAMP-dependent signaling in vivo. We provide evidence for a direct interaction between caveolin-1 and the catalytic subunit of PKA both in vitro and in vivo. Caveolin-1 binding appears to be mediated both by the caveolin scaffolding domain (residues 82–101) and a portion of the C-terminal domain (residues 135–156). Further functional analysis indicates that caveolin-based peptides derived from these binding regions can inhibit the catalytic activity of purified PKA in vitro. Mutational analysis of the caveolin scaffolding domain reveals that a series of aromatic residues within the caveolin scaffolding domain are critical for mediating inhibition of PKA. In addition, co-expression of caveolin-1 and PKA in cultured cells results in their co-localization as seen by immunofluorescence microscopy. In cells co-expressing caveolin-1 and PKA, PKA assumed a punctate distribution that coincided with the distribution of caveolin-1. In contrast, in cells expressing PKA alone, PKA was localized throughout the cytoplasm and yielded a diffuse staining pattern. Taken together, our results suggest that the direct inhibition of PKA by caveolin-1 is an important and previously unrecognized mechanism for modulating cAMP-mediated signaling.

Caveolae are vesicular invaginations of the plasma mem-brane with a characteristic ⍀-shaped morphology and a diameter of ϳ50 -100 nm (1,2). Although they are present in many cell types, caveolae are most abundant in terminally differentiated cells, such as adipocytes, endothelial cells, fibroblasts, and muscle cells (3).
Caveolae membranes are rich in cholesterol, glycosphingolipids, and a 21-24-kDa protein, caveolin (also known as caveolin-1). It appears that all three components are important for caveolae formation (4 -8). Recently, two other members of the caveolin gene family have been described, termed caveolin-2 and -3 (9,10). Caveolin-2 has the same tissue distribution as and co-localizes with caveolin-1, whereas caveolin-3 is found only in cardiac and skeletal muscle cells (11,12). Caveolin-1 contains a 41-amino acid region that self-associates and induces the formation of caveolin-1 homo-oligomers that contain ϳ14 -16 individual caveolin monomers. It is thought that these caveolin oligomers are the assembly units that drive the formation of caveolae in intact cells (13,14).
Caveolae are known to participate in vesicular trafficking (i.e. endocytosis and transcytosis), as well as signal transduction processes. Biochemical and morphological experiments have shown that a variety of signaling molecules are concentrated within these plasma membrane microdomains. This is particularly true of lipid-modified signaling molecules, such as Src family tyrosine kinases, Ha-Ras, endothelial nitric-oxide synthase, and heterotrimeric G-proteins (15)(16)(17)(18)(19)(20). The clustering of these proteins in microdomains of the plasma membrane is thought to facilitate signaling events and expedite cross-talk between distinct signaling pathways (reviewed in Ref. 1).
A role for the caveolin gene family in regulating such signaling cascades is emerging. Recent evidence points to a functional interaction between the caveolins and certain classes of signaling molecules. More specifically, caveolin-1 binding can functionally suppress the GTPase activity of heterotrimeric G-proteins and inhibit the kinase activity of Src family tyrosine kinases, the epidermal growth factor receptor kinase, and protein kinase C through a common caveolin domain, termed the caveolin scaffolding domain (15, 20 -22).
Signaling via cAMP is one of the first and most rigorously studied signal transduction pathways. A wealth of information is known about the events beginning with ligand binding and the subsequent intracellular effects. Briefly, the activation of serpentine receptors by their cognate ligands causes activation of the heterotrimeric G-protein, G s , in turn activating adenylyl cyclase and leading to the generation of cAMP. cAMP is bound by PKA 1 holoenzymes, which are composed of two regulatory and two catalytic subunits. PKA catalytic subunits, the main effectors of cAMP, dissociate from cAMP-saturated regulatory subunits and phosphorylate numerous downstream targets, thereby enhancing or repressing cellular processes and gene expression (23,24).
Much of this cascade initially occurs at or near the plasma membrane, given that G-protein-coupled receptors and adenylyl cyclase are integral membrane proteins. Diversification and subcellular regionalization of the pathway then occurs at the level of PKA, because various protein kinase A anchoring proteins (AKAPs) can sequester PKA molecules until the arrival of liberating cAMP signals (25,26). Recent studies suggest that many components of the cAMP cascade are concentrated within caveolae microdomains, and the caveolins have been shown to mitigate their function in vivo and in vitro. Several G-protein-coupled receptors have been morphologically and biochemically localized to caveolae (27,28). Also, heterotrimeric G-proteins and adenylyl cyclase are concentrated within these microdomains (20, 29 -34), and caveolin acts to inhibit their functions (15,20,32). Although several kinases have been shown to be either negatively or positively affected by the caveolins (15,21,22,(35)(36)(37), virtually nothing is known about the relationship of caveolins to PKA.
Here, we have used both in vivo and in vitro approaches to investigate the effects of caveolin-1 expression on PKA signaling.

EXPERIMENTAL PROCEDURES
Materials-The PKA␣ catalytic rabbit polyclonal IgG was purchased from Santa Cruz Biotechnology. Caveolin-1 mouse monoclonal antibody 2297 (used for immunoblotting; see Ref. 38) and monoclonal antibody 2234 (used for immunofluorescence; see Ref. 38) were the gifts of Dr. Roberto Campos-Gonzalez, Transduction Laboratories, Inc. The AV12-664 cell line was obtained from ATCC (CRL-9595). Purified bovine PKA␣ catalytic and truncated RII␤ subunits were as we described previously (39,40). A variety of other reagents were purchased commercially as follows: cell culture reagents were from Life Technologies, Inc., the PathDetect CRE/cis-acting reporter system was from Stratagene, and Effectene liposomal transfection reagent was from Qiagen.
Immunoblotting-Forty-eight hours post-transfection, AV12 cells were washed with PBS and incubated with lysis buffer (10 mM Tris, pH 7.5; 50 mM NaCl; 1% Triton X-100, 60 mM octyl glucoside) containing protease inhibitors (Roche Molecular Biochemicals). Where indicated, protein concentrations were quantified using the BCA reagent (Pierce). Samples were separated by SDS-PAGE (12.5% acrylamide) and transferred to nitrocellulose. The nitrocellulose membranes were stained with Ponceau S (to visualize protein bands) followed by immunoblot analysis. All subsequent wash buffers contained 10 mM Tris, pH 8.0, 150 mM NaCl, 0.05% Tween 20, which was supplemented with 1% bovine serum albumin and 2% nonfat dry milk (Carnation) for the blocking solution and 1% bovine serum albumin for the antibody diluent. Primary antibodies (either polyclonal or monoclonal) were used at a 1:500 dilution. Horseradish peroxidase-conjugated secondary antibodies (1:5000 dilution, Transduction Laboratory) were used to visualize bound primary antibodies with the Supersignal chemiluminescence substrate (Pierce).
In Vivo Reporter Assay for Signal Transduction-To assess PKAmediated signal transduction in vivo, we employed the PathDetect CRE cis-reporting system (Stratagene). In this assay, the luciferase reporter is driven by a promoter containing four cAMP response element (CRE) consensus sequences. Upon PKA catalytic subunit activation, endogenous CRE binding (CREB) protein is phosphorylated thereby allowing its interaction with CRE sequences and enhancing the transcription of luciferase. Transient transfections were performed using the Effectene liposome-mediated method (Qiagen) as described previously (35,36) with minor modifications. Briefly, 300,000 AV12 cells were seeded in 6-well plates 12-24 h prior to transfection. For stimulation experiments, each plate of cells was transfected with 0.5 g of the caveolin-1 (pCB7-Cav-1) or empty vector (pCB7) and 0.5 g of the pFR-Luc reporter construct. Twelve hours post-transfection, the cells were rinsed with PBS, incubated in 1% fetal bovine serum and, where indicated, supplemented with either 10 M forskolin, 10 M dideoxy-forskolin, or 500 M IBMX. Twenty-four hours post-stimulation, the cells were lysed in 200 l of extraction buffer, 50 l of which was used to measure luciferase activity as described (42). For the analysis of the effects of caveolin-1, caveolin-1 deletions, and caveolins-2 and -3 directly on PKA, all points were additionally transfected with 100 ng of pFC-PKA, a vector encoding the PKA catalytic subunit (Stratagene), and the cells were not serum-starved post-transfection. Each experimental value represents the average of two separate transfections performed in parallel; error bars represent the observed standard deviation. All experiments were performed at least three times independently and yielded virtually identical results. By using this CRE assay system, we have determined that transient expression of recombinant PKA increases total PKA activity by ϳ4 -6-fold; thus, the ratio between exogenous and endogenous PKA activity is ϳ5:1. These results are consistent with Western blot analyses showing that recombinant expression of PKA increases total PKA expression by ϳ5-fold (see Fig. 1, upper panel).
Interaction of the PKA␣ Catalytic Subunit with GST-Cav-1 Fusion Proteins-GST alone or GST-Cav-1 fusion proteins 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) (3ϫ) and lysis buffer (1ϫ), 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. Approximately 100 l of equalized bead volume was incubated with precleared lysates of AV12 cells transfected with PKA␣ catalytic subunit by rotating overnight at 4°C. After binding, the beads were extensively washed with TNET buffer and protease inhibitors (6ϫ). Finally, the associated proteins were eluted with TNET buffer supplemented with 20 mM glutathione and an additional 100 mM Tris-Cl, pH 8.0. The eluate was mixed 2:1 with 3ϫ sample buffer and subjected to SDS-PAGE. The immunoblotting procedure was as described above using anti-PKA␣ catalytic rabbit polyclonal antibody IgG (Santa Cruz Biotechnology) as the primary antibody.
Caveolin-derived Synthetic Peptides-The caveolin-based peptides were synthesized using standard methodology and subjected to amino acid analysis and mass spectroscopy (Massachusetts Institute of Technology Biopolymers Laboratory/Research Genetics) to confirm their purity and composition, as we described previously (15,21,22,35,36).
In Vitro Kinase Assays-Purified bovine PKA␣ catalytic subunit (5 g/ml) was incubated with caveolin-derived synthetic peptides and the truncated (residues 1-158) PKA RII␤ subunit in 100 l of kinase reaction buffer (30 mM KHPO 4 , 12.5 M cAMP, 625 M dithiothreitol, 12.5 mM MgCl 2 ) at 4°C. The reaction was initiated by the addition of 15 Ci of [␥-32 P]ATP. After incubation for 15 min at 4°C, the reaction was stopped by the addition of 3ϫ sample buffer and boiling for 5 min. Samples were subjected to SDS-PAGE and transferred to nitrocellulose membranes. Bands corresponding to phosphorylated RII␤ were visualized by autoradiography using an intensifying screen.
Immunofluorescence-The procedure was performed as we previously described (15). AV12 cells transfected with the PKA␣ catalytic subunit and/or caveolin-1 were fixed for 30 min in PBS containing 2% paraformaldehyde, rinsed with PBS, and quenched with 50 mM NH 4 Cl for 10 min. The cells were then incubated in permeabilization buffer (PBS; 0.2% bovine serum albumin; 0.1% Triton X-100) for 10 min, washed with PBS, and double-labeled with a 1:400 dilution of anticaveolin-1 mouse monoclonal antibody 2234 (Transduction Laboratories, Inc.) and PKA␣ catalytic rabbit polyclonal antibody (Santa Cruz Biotechnology) for 60 min. After rinsing with PBS (3ϫ), secondary antibodies (7.5 g/ml) ((lissamine-rhodamine (LRSC)-conjugated goat anti-rabbit and fluorescein isothiocyanate-conjugated goat anti-mouse) antibodies) were added for a period of 60 min. Cells were washed with PBS (3ϫ), and slides were mounted with Slow-Fade anti-fade reagent (Molecular Probes). A Bio-Rad MR600 confocal fluorescence microscope was used for visualization of bound secondary antibodies.

Functional Effect of Caveolins on PKA-mediated Signaling in
Vivo-Given the ability of caveolin to inactivate a variety of signal transduction pathways (15,21,22,35,36), we sought to investigate its possible role in PKA signaling. It is well known that one of the important downstream effectors of PKA is cAMP response element binding (CREB) protein. CREB binds CRE sequences in promoters of cAMP-responsive genes (24). PKAcatalyzed phosphorylation of Ser 133 of CREB elicits a conformational change that enables CREB to promote transcription of downstream target genes.
Thus, we employed a well established CRE reporter system (Stratagene, Inc.) to evaluate the in vivo effects of caveolin-1 expression on PKA signaling. In this assay, a series of consensus CRE sequences drive expression of a luciferase reporter. PKA directly activates CREB which up-regulates the transcription of the luciferase reporter. AV12-664, a tumor-derived cell line from hamsters, was chosen for this and subsequent assays because of its failure to endogenously express detectable levels of caveolin-1 ( Fig. 1). Therefore, overexpression of recombinant caveolin-1 in this system should allow us to assess the effect of caveolin-1 on PKA signaling.
As shown in Fig. 2A, caveolin-1 expression (driven by the cytomegalovirus promoter; pCB7-Cav-1) dramatically inhibits endogenous basal PKA signaling as compared with the empty vector (pCB7) or ␤-galactosidase. Most likely, this reflects the ability of caveolin-1 expression to inhibit endogenous PKA activity. However, this may still reflect the ability of caveolin-1 to inhibit signaling at the level of G s ␣ subunits or adenylyl cyclase, as well.
To assess the effects of caveolin-1 in the presence of PKA pathway activators, 10 M forskolin (a potent stimulator of adenylyl cyclase) and 500 M IBMX (an inhibitor of cyclic nucleotide phosphodiesterase) were used in the same assay system. Treatment with either pharmacological agent enhanced PKA-mediated signaling nearly 3-fold; in striking contrast, caveolin-1 expression completely abolished this induction (Fig. 2B).
In contrast to the above experiments, we co-expressed the PKA␣ catalytic subunit (PKA␣-cat) driven by the cytomegalovirus promoter with wild-type caveolin-1 and the corresponding deletion mutants to study the modulation of this system in the presence of PKA overexpression. The constructs used in these reporter assays are schematically shown in Fig. 3A and have all been previously well characterized (41). Full-length caveolin-1 (also known as the ␣-isoform of Cav-1) contains residues 1-178; ⌬C is a C-terminal deletion mutant containing residues 1-140; ⌬N is a N-terminal deletion mutant containing residues 96 -178, and the ␤-isoform is a naturally occurring alternatively translated form of caveolin-1 lacking residues 1-31. As shown in Fig. 3B, PKA␣-cat expression dramatically induces luciferase activity over pCB7 alone, whereas caveolin-1 co-transfection efficiently inhibits this induction. Most caveolin-1 deletions contained inhibitory activity comparable to caveolin-1. However, the ⌬N construct was ϳ2-fold less potent than the others, possibly implicating caveolin-1 residues 32-95 in this inhibition (Fig. 3B); this region contains the caveolin-1 scaffolding domain (residues 82-101). In addition to caveolin-1, PKA activity was similarly abrogated by co-transfection with the cDNAs encoding caveolins-2 and -3.
Direct Interaction of the PKA␣ Catalytic Subunit with the Note that transfection of the PKA␣-cat cDNA produces a distinct band that co-migrates with endogenous PKA␣ (lane 3); this antibody also recognizes the PKA␤-catalytic isoform. Lower panel, immunoblot analysis using anti-caveolin-1 IgG. Note that caveolin-1 expression is undetectable in AV12 cells but is easily detected in NIH 3T3 and Chinese hamster ovary cells.

FIG. 2. Recombinant expression of caveolin-1 inhibits forskolin-and IBMX-mediated activation of a PKA-responsive reporter in vivo.
A, AV12 cells were transiently transfected with the CREB-responsive luciferase reporter plasmid. These cells were also co-transfected with either empty vector alone (pCB7), caveolin-1 (pCB7-Cav-1), or ␤-galactosidase (␤-Gal). Note that caveolin-1 inhibits PKA signaling by greater than 10-fold. B, these experiments were performed as detailed in A, except AV12 cells were treated in the presence or absence of pharmacological agents that activate endogenous PKA signaling. Note that forskolin (an activator of adenylyl cyclase) or IBMX (an inhibitor of phosphodiesterase) activated the PKA pathway approximately 2-3-fold, whereas dideoxy-forskolin (an inactive form of forskolin) had no effect. In addition, caveolin-1 expression dramatically inhibited forskolin-and IBMX-mediated activation of PKA signaling. A and B, each experimental value represented graphically is the average of two separate transfections performed in parallel; error bars represent the observed standard deviation. These experiments were performed several times independently and yielded virtually identical results.

Caveolin scaffolding Domain and the C-terminal Domain of
Caveolin-1 in Vitro-Caveolin has been shown to interact with and inhibit the function of a series of signal-transducing molecules, including Ha-Ras, c-Src, G-protein ␣-subunits, adenylyl cyclase, and PKC (15,20,22,32). The dramatic reduction of PKA activity by caveolin-1 expression in vivo (as assessed by the above reporter assays) led us to evaluate the possibility of a direct interaction between caveolin-1 and PKA␣-cat.
To test this hypothesis, we expressed and affinity purified a GST fusion protein carrying the full-length caveolin-1 molecule (GST-Cav-1-FL). GST-Cav-1-FL, attached to glutathione-agarose beads, was incubated with detergent extracts of AV12 cells transiently overexpressing PKA␣-cat. After extensive washing and elution with reduced glutathione, the samples were separated by SDS-PAGE and subjected to immunoblot analysis with anti-PKA␣-cat IgG (Santa Cruz Biotechnology). Fig. 4A shows that PKA␣-cat bound specifically to full-length caveo-lin-1, as compared with GST alone.
These experiments implicate the caveolin scaffolding domain and a region of the caveolin-1 C-terminal domain as interaction ⌬C (lacking residues 141-178), ⌬N (lacking residues 1-95), and ␤-isoform (alternatively translated form of caveolin-1; lacking residues 1-31). B, AV12 cells were transiently co-transfected with PKA␣-cat and a variety of forms of caveolin-1 or vector alone (pCB7). Note that transfection of the PKA␣-cat cDNA induces PKA signaling as compared with vector alone (pCB7), whereas co-transfection with full-length caveolin-1 dramatically inhibits this PKA-induced signaling. The various caveolin-1 deletion mutants were equally effective as wild-type, except that ⌬N displayed ϳ2-fold less inhibition. In addition to caveolin-1, PKA activity was similarly abrogated by co-transfection with the cDNAs encoding caveolins-2 and -3. Each experimental value represented graphically is the average of two separate transfections performed in parallel; error bars represent the observed standard deviation. These experiments were performed several times independently and yielded virtually identical results.

FIG. 4. Binding of PKA to GST-caveolin-1 fusion proteins in vitro.
A, lysates from AV12 cells transfected with the PKA␣-cat cDNA were incubated with affinity purified GST alone or GST-Cav-1-FL (fulllength; residues 1-178) immobilized on glutathione-agarose beads. After extensive washing and elution with reduced glutathione, the eluates were analyzed by SDS-PAGE and subjected to immunoblot analysis with anti-PKA-specific IgG. Note that only GST-Cav-1-FL bound PKA␣cat; no binding was observed with GST alone. Untransfected AV12 cell lysates were used as a positive control for the migration of PKA isoforms (not shown). B, schematic diagram summarizing the construction of a panel of GST-Cav-1 fusion proteins. They are as follows: GST-Cav-1-(1-101), GST-Cav-1-(1-81), GST-Cav 1-(61-101), and GST-Cav-1-(135-178). C, lysates from AV12 cells transfected with the PKA␣-cat cDNA were incubated with a variety of affinity purified GST-Cav-1 fusion proteins immobilized on glutathione-agarose beads and treated as described in A. Note that GST-caveolin-1 fusion proteins containing the caveolin scaffolding domain (residues 82-101), i.e. GST-Cav-1-(1-101) and GST-Cav-1-(61-101), retain binding activity. In contrast, a GST-caveolin-1 fusion lacking the scaffolding domain (GST-Cav-1-(1-81)) does not retain any binding activity. In addition, a weaker interaction is observed with a GST fusion encoding the C-terminal domain of caveolin-1 (GST- Cav-1-(135-178)). sites for PKA␣-cat. A second interacting region in caveolins (other than the scaffolding domain) has been described for the interaction of caveolin-3 and -1 with neuronal nitric-oxide synthase (44,45). In light of these findings, the importance of this second interaction site may in part explain caveolin-1-mediated inhibition of PKA signaling (as determined by luciferase assays) despite deletion of either the N-terminal or C-terminal caveolin-1 domains. In support of this notion, a peptide derived from the C-terminal domain of caveolin-1 (residues 135-156) inhibits PKA activity in vitro (see below; Fig. 6B, lane 8).
Functional Inhibition of PKA␣ Catalytic Activity by Caveolin-derived Peptides-The demonstration of a direct interaction between specific caveolin-1 domains and PKA␣-cat led us to determine the potential functional consequences of this binding via in vitro kinase assays. Since PKA␣-cat can phosphorylate Ser 95 on its own regulatory subunit (RII␤) in vivo (46), we used a purified portion of the RII␤ subunit (residues 1-158) as a physiologically relevant substrate for PKA␣-cat in these kinase assays. PKI, a potent peptide inhibitor of PKA, was used as a positive control in these experiments.
We have previously described a series of caveolin-derived peptides spanning various regions of the caveolin molecule (15,20,22,32,35). In a preliminary analysis, peptides derived from the N-terminal region of caveolin-1 (including the caveolin-1 scaffolding domain) and the scaffolding domains of caveolin-2 and -3 were utilized (Fig. 5A). As shown in Fig. 5B, peptides derived from the scaffolding domain of caveolin-1 and -3 demonstrated potent inhibition of PKA activity, whereas an irrelevant caveolin-1 peptide and the scaffolding domain of caveo-lin-2 had little or no effect. This caveolin-mediated PKA inhibition is sequence-specific as demonstrated by the following observations: (i) such inhibition is abrogated when the caveolin-1 scaffolding domain is divided into two halves (residues 84 -92 and 93-101) and (ii) the caveolin-3 scaffolding domain (with the highest homology to caveolin-1) yielded similar results, in contrast to the more divergent caveolin-2 scaffolding domain (Fig. 5B).
To dissect the region of inhibitory activity within the caveolin-1 scaffolding domain, we successively truncated the original 20-mer caveolin-1 scaffolding domain (Fig. 6A). The inhibitory activity of the caveolin-1 scaffolding domain seemed to be unaffected until the 12-mer peptide, possibly implicating either Ser 88 or Phe 89 in direct binding (Fig. 6B). A region of the C-terminal domain of caveolin-1 (residues 135-156) was also inhibitory in these assays (Fig. 6B). This is intriguing since the above GST fusion experiments demonstrated the binding of this region to PKA␣-cat (Fig. 4C), and other investigators have proposed the existence of a second C-terminal inhibitory region within caveolin-1 (44).
The above assays were conducted with caveolin peptides at a concentration of 10 M. To determine the minimal inhibitory concentration, we serially diluted the caveolin-1 scaffolding domain from 10 to 0.5 M. As shown in Fig. 7B, potent inhibition was first observed at 3 M, yielding an IC 50 between 1 and 3 M. To investigate further the sequence specificity of this caveolin-1-mediated inhibition, we generated a peptide that encodes a partially scrambled caveolin-1 scaffolding domain. This partially scrambled peptide is 100% compositionally iden- Note the similarity between the scaffolding domains (SD) of caveolins-1 and -3 and that they are significantly divergent from the caveolin-2 scaffolding domain. B, the purified bovine PKA␣-catalytic subunit was used to phosphorylate a fragment of its own regulatory subunit (RII␤, residues 1-158). The RII subunit and either PKI (10 M; a potent peptide inhibitor of PKA) or caveolin-derived peptides (10 M) were incubated with PKA␣-cat in the presence of [␥-32 P]ATP and subjected to SDS-PAGE. Note that only the caveolin-1 peptides including the scaffolding domain (residues 82-101) are inhibitory and that the effect is reversed if the scaffolding domain is divided into two peptides. Also, note that the caveolin-3 scaffolding domain is inhibitory, whereas the caveolin-2 scaffolding domain shows no inhibitory activity. These observations are consistent with the sequence divergence of caveolin-2 from caveolins-1 and -3.  (157-178)). B, the purified bovine PKA␣-catalytic subunit was used to phosphorylate a fragment of its own regulatory subunit (RII␤, residues 1-158). The RII subunit and either PKI (10 M) or caveolin-derived peptides (10 M) were incubated with PKA␣-cat in the presence of [␥-32 P]ATP and subjected to SDS-PAGE. Note that all of the caveolin-1 scaffolding domain N-terminal deletions remain inhibitory, with the exception the 12-mer peptide which shows a partial reversal of this effect. This may demarcate a region of importance within the caveolin-1 scaffolding domain. Also note the inhibitory activity of the C-terminal Cav-1-(135-156) peptide in contrast to that of Cav-1-(157-178), demonstrating the presence of a second inhibitory domain within caveolin-1.
tical to the wild-type caveolin scaffolding domain, with 60% sequence identity (Fig. 7A). At a peptide concentration of 3 M, complete inhibition was observed with the wild-type caveolin-1 scaffolding domain, whereas no inhibition was observed with the partially scrambled caveolin-1 scaffolding domain (Fig. 7C). This experiment clearly demonstrates the importance of sequence specificity in the direct binding and inhibition of PKA␣cat by caveolin-1.
We used alanine scanning mutagenesis of a 16-mer caveolin-1 scaffolding domain peptide to determine which residues within the caveolin-1 scaffolding domain are required for inhibition of PKA catalytic activity. Note that significant inhibition of PKA activity was observed with the wild-type 16-mer caveolin-1 scaffolding domain (Fig. 6B). As shown in Fig. 8A, Phe 89 , Phe 92 , Thr 93 , Trp 98 , and Phe 99 seem to be pivotal for the inhibition of PKA activity. Note that the importance of Phe 89 is corroborated by the peptide deletion analysis described above.
Interestingly, via binding studies, we have previously shown that the FTVT sequence (residues 92-95) is an essential component of the caveolin-1 scaffolding domain (21), and the analogous caveolin-3 scaffolding domain TFT sequence is deleted in patients with an autosomal dominant form of limb-girdle muscular dystrophy (LGMD-1C (47)) (Fig. 10B).
Co-localization of Caveolin-1 and PKA␣-cat in Vivo-As we have shown above that caveolin-1 interacts with PKA␣-cat in vitro and caveolin-1-derived peptides can functionally regulate its catalytic activity, we next examined the co-localization of these proteins in vivo. AV12 cells were transiently transfected with the PKA␣ catalytic subunit with or without co-transfection with caveolin-1. These cells were then doubly-immunostained and examined by confocal microscopy. Fig. 9B shows that co-expression of caveolin-1 and PKA in cultured cells results in their dramatic co-localization as seen by immunofluorescence microscopy. In cells co-expressing caveolin-1 and PKA, PKA assumed a punctate distribution that strictly coincided with the distribution of caveolin-1. In contrast, in cells expressing PKA alone, PKA was localized throughout the cytoplasm and yielded a diffuse staining pattern (Fig. 9A).

DISCUSSION
By using several independent and complementary approaches, we have examined the functional relationship between caveolin-1 and the catalytic subunit of PKA. Caveolin-1 dramatically inhibited PKA signaling in vivo as measured by the transcriptional activation of CREB using a luciferase-based reporter system. Support for a direct interaction between caveolin-1 and PKA␣-cat was provided by employing a series of GST-caveolin-1 fusion proteins. This binding activity within caveolin-1 was localized to residues 82-101 (i.e. the caveolin scaffolding domain). In addition, weaker binding was observed with the C-terminal domain of caveolin-1 (residues 135-178), identifying a possible secondary binding site for PKA␣-cat. In vitro kinase assays employing caveolin-derived peptides identified the caveolin-1 scaffolding domain (residues 82-101) and a peptide derived from the C-terminal domain of caveolin-1 (residues 135-156) as potent inhibitors of PKA catalytic activity. Mutational analysis of the caveolin-1 scaffolding domain identified five critical residues that are important for maintaining this inhibitory activity: Phe 89 , Phe 92 , Thr 93 , Trp 98 , and Phe 99 . Interestingly, four of these five residues are aromatic amino acids. Finally, co-expression of caveolin-1 and PKA␣-cat in cultured cells results in their co-localization as seen by immunofluorescence microscopy.
We find that although co-expression of caveolin-2 with PKA blocks transcriptional activation of CREB in vivo, a peptide derived from the caveolin-2 scaffolding domain did not block the activity of PKA in vitro. One possible explanation for this difference between in vivo and in vitro results is that there may be another region within caveolin-2, other than the scaffolding domain, that contains inhibitory activity. In support of this idea, recent experimental evidence suggests that caveolins may FIG. 7. Inhibition of PKA by the caveolin-1 scaffolding domain occurs in the low micromolar range and is sequence-specific. A, the sequences of the caveolin-derived peptides used in these in vitro kinase experiments are shown. Note that the partially scrambled caveolin-1 scaffolding domain peptide is 100% compositionally identical to the wild-type caveolin-1 scaffolding domain, with 60% sequence identity. Identical residues are boxed. B, the purified bovine PKA␣-catalytic subunit was used to phosphorylate a fragment of its own regulatory subunit (RII␤, residues 1-158). The RII subunit and either PKI (10 M) or the caveolin-1 scaffolding domain (0 -10 M) were incubated with PKA␣-cat in the presence of [␥-32 P]ATP and subjected to SDS-PAGE. Note that the 1/3 M boundary marks the minimal effective concentration for the wild-type caveolin-1 scaffolding domain. C, comparison of the inhibitory activity of the wild-type and the partially scrambled caveolin-1 scaffolding domain at a concentration of 3 M. In vitro kinase reactions with the PKA␣-catalytic subunit were performed as described in B. Note that the partially scrambled caveolin-1 scaffolding domain has no inhibitory activity, as compared with the wild-type peptide.
FIG. 8. Alanine-scanning mutagenesis of the caveolin-1 scaffolding domain identifies five residues that are critical for mediating PKA inhibition. The wild-type sequence of the caveolin-1 scaffolding domain (16-mer) is shown. Each residue within this sequence was changed one at a time to alanine, as indicated. In vitro kinase reactions with the PKA␣-catalytic subunit were performed as described in legends of Figs. 5-7. Note that this analysis of the caveolin-1 scaffolding domain identified five critical residues that are important for maintaining inhibitory activity (Phe 89 , Phe 92 , Thr 93 , Trp 98 , and Phe 99 ) in the context of the 16-mer peptide.
also contain a second inhibitory domain within the C terminus (44). Alternatively, as caveolins-1 and -2 can form a stable hetero-oligomeric complex in vivo (9), the in vivo effect of caveolin-2 may be mediated by the caveolin-1/-2 hetero-oligomer.
To date, a variety of physiological regulators of PKA have been identified, such as the regulatory subunits (RI and RII) and PKI. A common motif in these known PKA inhibitors is a region resembling the consensus phosphorylation site for PKA (i.e. the RRX(S/T) sequence) (46,48). RII␣ and RII␤ contain this motif and are in fact phosphorylated by PKA␣-cat, whereas RI␣, RI␤, and PKI have an altered residue at the S/T site, thereby acting as pseudo-substrates of PKA (Fig. 10A (46)). Here, we showed the caveolin-1 scaffolding domain is an inhibitor of PKA␣-cat. However, the critical residues within the caveolin scaffolding domain and the overall sequence itself do not resemble the consensus phosphorylation site for PKA (Fig.  10A). This may imply that caveolin-1 allosterically binds to a distinct region of the catalytic subunit to mediate such inhibition.
By using phage display libraries, we have previously identified peptide ligands for the caveolin scaffolding domain. These peptide ligands or "caveolin-binding motifs" are as follows: ⌽X⌽XXXX⌽, ⌽XXXX⌽XX⌽, and ⌽X⌽XXXX⌽XX⌽, where ⌽ indicates an aromatic residue, Trp, Phe, or Tyr (49). Interestingly, the residues within the caveolin-1 scaffolding domain that are important for mediating inhibition of PKA catalytic activity are also primarily aromatic residues, with the sequence ⌽XX⌽TXXXX⌽⌽ (Fig. 10C). The similarity between these residues within the caveolin scaffolding domain and the known consensus for caveolin-binding motifs is quite striking and may suggest that the interaction between the scaffolding domain and binding motifs is mediated through the side-byside stacking of aromatic amino acid side chains.
The cellular compartmentalization of PKA␣-cat is diverse and largely depends on its binding via RII regulatory subunits to AKAPs (25,26). It is thought that AKAPs allow a serine/ threonine kinase such as PKA to remain active in distinct cellular regions, thereby exerting control over its target substrates. AKAP79 and AKAP18 are associated with the plasma FIG. 9. Co-localization of caveolin-1 and PKA in vivo. AV12-664 cells were co-transfected with the cDNAs encoding caveolin-1 and PKA␣-cat or with PKA␣-cat alone. Cells were then doubly immunostained with antibodies that specifically recognize either caveolin-1 or PKA␣-cat. Bound primary antibodies were visualized with distinctly tagged secondary antibody probes (see "Experimental Procedures"). A, cells transfected with PKA␣-cat alone and stained with the appropriate antibody showed a diffuse cytoplasmic distribution for PKA␣-cat. B, cells co-transfected with caveolin-1 and PKA␣-cat and doubly immunostained with their respective antibodies revealed significant co-localization between PKA␣-cat and caveolin-1. Areas of co-localization are as indicated (see white arrowheads). Positive and negative controls omitting a given primary antibody and with cells singly transfected with a given cDNA were also performed and yielded the expected results (data not shown).
FIG. 10. The caveolin scaffolding domain does not share sequence homology with known PKA peptide inhibitors or PKA substrates. A, comparison of the sequence of the caveolin-1 scaffolding domain with the sequences of classic PKA substrates and pseudosubstrate inhibitors (i.e. RI␣, RI␤, RII␣, RII␤, and PKI). Note that PKA substrates contain the RRX(S/T) motif, whereas PKA pseudo-substrate inhibitors contain RRX(A/G). However, the caveolin-1 scaffolding domain does not show any homology with either PKA substrates or pseudo-substrate inhibitors. B, caveolin-1 scaffolding domain residues that are important for mediating PKA inhibition, as determined experimentally in Fig. 8. These include Phe 89 , Phe 92 , Thr 93 , Trp 98 , and Phe 99 . Interestingly, four of these residues are aromatic amino acids and are designated here as ⌽ in panel C. Residues within this caveolin domain that have been shown to be important for limb-girdle muscular dystrophy (47), G␣/epidermal growth factor receptor binding activity (21,49), and inhibition of adenylyl cyclase (AC) (32) are indicated. C, comparison with the previously defined caveolin binding motif (49) reveals interesting similarities, suggesting a potential mechanism for caveolin's recognition of caveolin binding motifs. See text for details. membrane via electrostatic interactions or undergo lipid modifications (i.e. myristoylation and palmitoylation), respectively (50 -52). Furthermore, a glycine at the extreme N terminus of PKA␣-cat undergoes myristoylation (53). We and others have previously shown that (i) caveolin-1 is a fatty acid-binding protein (7,54), and (ii) these lipid modifications can function as signals to target other cytoplasmic proteins to caveolae and caveolae-related microdomains of the plasma membrane (17)(18)(19)55). Therefore, it is perhaps not surprising that a fraction of total cellular PKA is localized at or near caveolae membrane domains. The apparent ability of caveolins to interact with and functionally inhibit PKA␣-cat provides an additional mechanism for modulating cAMP-dependent signaling at the level of the PKA catalytic subunit.
Here, we have determined that a peptide encoding the caveolin-1 scaffolding domain inhibits PKA activity in vitro, yielding an IC 50 between 1 and 3 M. However, it is likely that the affinity of this interaction is higher in vivo because (i) caveolin-1 forms high molecular mass homo-oligomers of ϳ350 kDa that contain ϳ14 -16 caveolin-1 monomers per oligomer (13,14), and (ii) these caveolin homo-oligomers then interact with each other via their C-terminal domains, forming a caveolin-1rich scaffold or network on the cytoplasmic side of the plasma membrane (41). These two stages of oligomerization greatly increase the local effective concentration of caveolin-1 at the membrane. In support of this notion, we show here by immunofluorescence microscopy that recombinant expression of caveolin-1 is sufficient to recruit PKA onto caveolin-1-rich areas of the membrane (see Fig. 9B). In certain cell types, such as smooth muscle cells, endothelial cells, and adipocytes, it has been reported that up to ϳ50% of the total plasma membrane surface area is represented by caveolae-derived membranes (31, 56 -58). Thus, caveolin-1-mediated regulation of PKA signaling events may be particularly important in cell types that have abundant caveolae.