A Role for the Caveolin Scaffolding Domain in Mediating the Membrane Attachment of Caveolin-1

Here, we have created a series of caveolin-1 (Cav-1) deletion mutants to examine whether the membrane spanning segment is required for membrane attachment of caveolin-1 in vivo. One mutant, Cav-1-(1–101), contains only the cytoplasmic N-terminal domain and lacks the membrane spanning domain and the C-terminal domain. Interestingly, Cav-1-(1–101) still behaves as an integral membrane protein but lacks any known signals for lipid modification. In striking contrast, another deletion mutant, Cav-1-(1–81), behaved as a soluble protein. These results implicate caveolin-1 residues 82–101 (also known as the caveolin scaffolding domain) in membrane attachment. In accordance with the postulated role of the caveolin-1 scaffolding domain as an inhibitor of signal transduction, Cav-1-(1–101) retained the ability to functionally inhibit signaling along the p42/44 mitogen-activated protein kinase cascade, whereas Cav-1-(1–81) was completely ineffective. To rule out the possibility that membrane attachment mediated by the caveolin scaffolding domain was indirect, we reconstituted the membrane binding of caveolin-1 in vitro. By using purified glutathioneS-transferase-caveolin-1 fusion proteins and reconstituted lipid vesicles, we show that the caveolin-1 scaffolding domain and the C-terminal domain (residues 135–178) are both sufficient for membrane attachment in vitro. However, the putative membrane spanning domain (residues 102–134) did not show any physical association with membranes in this in vitro system. Taken together, our results provide strong evidence that the caveolin scaffolding domain contributes to the membrane attachment of caveolin-1.

The protein domains responsible for some of the biological properties of caveolins have recently been defined. For example, homo-oligomerization is mediated by a 41-amino acid domain that is located at the membrane-proximal region of the N-terminal domain (residues 61-101 in Cav-1) (10); deletion of this oligomerization domain (and all sequences N-terminal to it) prevents multimerization in vivo (13). Furthermore, fusion of the oligomerization domain of Cav-1 to GST results in GST multimerization in vitro (10). Thus, the oligomerization domain is necessary and sufficient for caveolin self-association.
However, it remains untested whether the putative membrane spanning segment actually functions in membrane attachment. Here, we find that the putative membrane spanning domain is not required for membrane attachment of caveolin-1 in vivo or in vitro. Instead, we provide evidence that the caveo-lin scaffolding domain and the C-terminal domain participate in membrane attachment.
Deletion Mutagenesis-Standard polymerase chain reaction strategies were used in constructing the various deletions shown in Fig. 1A. The canine caveolin-1 cDNA (29) was used as the template for these reactions. Various caveolin-1 deletion mutants were subcloned into the pCB7 vector using the HindIII/BamHI sites within the polylinker (13). The correctness of the intended mutations was confirmed by DNA sequencing.
Cells, Media, and Transfection Methods-Except where indicated, all experiments were performed using HEK 293T cells. These cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 2 mM glutamine, 100 units/ml penicillin, 100 g/ml streptomycin. Cells were transfected using the calcium phosphate precipitation method. All experiments involving HEK 293T cells were performed 36 h post-transfection.
Immunoblotting-Samples were subjected to SDS-PAGE under reducing conditions and transferred to nitrocellulose membranes. The protein bands were visualized with Ponceau S (Sigma). Membranes were then incubated with primary antibody, washed, and incubated with a secondary antibody conjugated with horseradish peroxidase (Transduction Laboratories). Bound IgG were detected using a chemiluminescent substrate (Pierce).
Hypotonic Lysis and High Salt Extraction-Post-nuclear lysates were separated into soluble and insoluble components, as described previously (30). Cells grown to confluence in a 60-mm diameter dish were scraped into cold PBS, pelleted by centrifugation, and resuspended in 0.5 ml of hypotonic buffer (5 mM Tris, pH 7.5, 1 mM MgCl 2 , 1 mM EGTA, 0.1 mM EDTA) containing protease inhibitors. Following a 30-min incubation on ice, the lysate was passed through a 21-gauge needle 10 times and centrifuged at 1000 ϫ g to remove nuclear debris. The post-nuclear lysate was transferred to an 11 ϫ 34-mm polycarbonate tube and centrifuged at 56,000 rpm for 30 min in a TLA-100.2 rotor (Beckman). The supernatant was collected, and the pellet was resuspended in an equal volume of 1% SDS. Equal volumes of both fractions were subjected to SDS-PAGE and immunoblot analysis, as described above. In the salt extraction experiments, the lysate was treated with NaCl (to the indicated final concentration) after removal of nuclear debris and then subjected to centrifugation as described above.
Alkaline Carbonate Extraction-Carbonate extraction was performed as we described previously (27). Briefly, cells were grown to confluence in a 60-mm diameter dish and were washed twice in cold PBS and once in 150 mM NaCl. After aspiration of the NaCl solution, 1 ml of 100 mM NaCO 3 , pH 11.3, containing protease inhibitors was used to scrape the cells off the dish. The sample was transferred to a loose fitting 1-ml Dounce homogenizer and homogenized with 5 strokes. Following a 30-min incubation on ice, the sample was transferred to an 11 ϫ 34 mm polycarbonate tube and centrifuged at 100,000 rpm in a TLA-100.2 rotor. The pellet was resuspended in an equal volume of 1% SDS. Both fractions were passed through a 26-gauge needle 10 times, and 50 l of each were subjected to SDS-PAGE and immunoblot analysis, as described above.
Triton Solubility-Extraction of Triton-soluble proteins was performed as we described previously (29). Briefly, cells were grown to confluence in a 35-mm diameter dish and washed twice with cold PBS. Three hundred microliters of cold MBS (25 mM Mes, pH 6.5, 150 mM NaCl), 1% Triton X-100 containing protease inhibitors was then added. Following a 30-min incubation on ice without agitation, the soluble fraction was collected. An equal volume of 1% SDS was added to the plate to dissolve the Triton X-100-insoluble material. The latter fraction was passed through a 26-gauge needle 10 times in order to lower its viscosity. Fifty microliters of the Triton X-100-soluble and -insoluble fractions were separated by SDS-PAGE and subjected to immunoblot analysis as described above.
Purification of Caveolae-enriched Membrane Fractions-Caveolaeenriched membrane fractions were purified essentially as described previously (13,29). Cells grown to confluence in a 150-mm diameter plate were washed twice in cold PBS, scraped into 2 ml of MBS containing 1% Triton X-100, passed 5 times through a loose fitting Dounce homogenizer, and mixed with an equal volume of 80% sucrose (prepared in MBS lacking Triton X-100). The sample was then transferred to a 12-ml ultracentrifuge tube and overlaid with a discontinuous sucrose gradient (4 ml of 30% sucrose, 4 ml of 5% sucrose, both prepared in MBS, lacking detergent). The samples were subjected to centrifugation at 200,000 ϫ g (39,000 rpm in Sorval rotor TH-641) for 16 h. A light scattering band was observed at the 5/30% sucrose interface. Twelve 1-ml fractions were collected, and 50-l aliquots of each fraction were subjected to SDS-PAGE and immunoblot analysis.
Incorporation of Recombinant GST-Cav-1 into Lipid Vesicles-GST-Cav-1 fusion proteins were constructed and affinity purified as we described previously (10,21,26). The incorporation of recombinant GST-Cav-1 into membranes was monitored by flotation in sucrose density gradients, essentially as we described previously for bacterially expressed full-length His-tagged caveolin-1 (12). Briefly, a given purified GST-Cav-1 fusion protein was mixed with purified lipid components in 2 ml of MBS containing 60 mM octyl glucoside and dialyzed overnight against MBS lacking detergent to allow the association of caveolin with lipids. More specifically, ϳ85 g of a given GST-Cav-1 fusion protein was added to 1.8 mg of the purified lipid extract dissolved in octyl glucoside. The dialysate was then adjusted to 40% sucrose by the addition of 2 ml of 80% sucrose prepared in MBS and placed at the bottom of an ultracentrifuge tube. A 5-30% discontinuous sucrose gradient was formed above (4 ml of 5% sucrose, 4 ml of 30% sucrose; both in MBS lacking detergent) and centrifuged at 200,000 ϫ g (39,000 rpm in Sorval rotor TH-641). A lightscattering band at the 5-30% sucrose interface contained the associated lipids and formed regardless of whether or not caveolin was added. The light-scattering band at the 5-30% sucrose interface was collected by centrifugation and subjected to immunoblot analysis.

Construction and Expression of Caveolin-1 Deletion Mutants Lacking the Putative Membrane Spanning Domain
It has been postulated that caveolin-1 is anchored to the membrane by a central 33-amino acid hydrophobic segment (32)(33)(34). To examine its requirement for membrane anchoring, we created a panel of caveolin-1 deletion mutants lacking this protein domain. Briefly, Cav-1 (⌬TM) lacks only the putative membrane spanning domain but contains both the complete Nand C-terminal domains of caveolin-1; Cav-1-(1-101) lacks both the putative membrane spanning domain and the C-terminal domain but retains the complete cytoplasmic N-terminal domain; and Cav-1-(1-81) consists solely of the N-terminal domain but lacks the caveolin scaffolding domain (residues 82-101). We created the latter two mutants lacking the C terminus, as the C-terminal domain of caveolin-1 undergoes palmitoylation, and palmitoylation is known to mediate membrane an-choring. Fig. 1A diagrammatically illustrates the construction of these "anchor-minus" caveolin-1 mutants.
We chose 293T cells as our expression system as they do not express significant levels of endogenous caveolin-1. Fig. 1B shows that all of these caveolin-1 deletion mutants were expressed to equivalent levels in 293T cells and were easily detected with mono-specific antibodies directed against the extreme N-terminal domain of caveolin-1 (residues 2-21).

Phenotypic Behavior of Caveolin-1 Mutants Lacking a Putative Membrane Spanning Domain
Membrane Attachment-We evaluated the membrane association of these caveolin-1 deletion mutants using a variety of established assay systems. First, 293T cells transiently expressing wild-type caveolin-1 (FL) or a variety of deletion mutants (⌬TM, 1-101, and 1-81) were subjected to hypotonic lysis and separated into pellet and supernatant fractions. Fig. 2A shows that wild-type caveolin-1 and two of the three deletion mutants (⌬TM and 1-101) partitioned with the pellet fraction, suggesting an association with membranes. In contrast, Cav-1-(1-81) remained completely soluble under these condition. These results immediately suggested a role for residues 82-101 in mediating membrane attachment.
Proteins that resist alkaline carbonate extraction are operationally considered to be "integral membrane proteins" (35). Fig. 3, A and B, demonstrates that wild-type caveolin-1 (FL) and two of the deletion mutants (⌬TM and 1-101) completely resisted extraction with sodium carbonate at pH 11.3. In contrast, Cav-1-(1-81) and guanine nucleotide dissociation inhibitor (a known cytosolic protein) quantitatively partitioned with the supernatant fraction under these conditions, behaving as soluble proteins.
To evaluate the possibility that Cav-1-(1-101) was associated with the membrane via an ionic interaction, we employed increasing concentrations of salt in the extraction buffer (0 -1.0 M NaCl). Fig. 3C shows that even under conditions of 1.0 M NaCl, Cav-1-(1-101) remained completely membrane-associated, behaving as expected for an integral membrane protein.
To separate membranes enriched in caveolae from the bulk of cellular membranes and cytosolic proteins, an established equilibrium sucrose density gradient system was utilized (21,26,29,37,38,(41)(42)(43)(44)(45)(46)(47). In this fractionation scheme, immunoblotting with anti-caveolin IgG can be used to track the position of caveolae-derived membranes within these bottom-loaded sucrose gradients. By using this procedure, caveolin-1 is purified ϳ2000-fold relative to total cell lysates as ϳ4 -6 g of caveolinrich domains (containing ϳ90 -95% of total cellular caveolin-1) are obtained from 10 mg of total cellular proteins (21,44). We and others (29,37,38) have shown that these caveolae-enriched fractions exclude Ͼ99.95% of total cellular proteins (Fig. 5, A and  B) and also markers for non-caveolar plasma membrane, Golgi,  1-101, and 1-81). Cells were also transfected with vector alone (pCB7) as a negative control. Caveolin-1 expression was detected by immunoblot analysis with a mono-specific anti-peptide antibody probe that recognizes the unique N terminus of caveolin-1 (residues 2-21). Note that wild-type full-length caveolin-1 and the caveolin-1 deletion mutants were all expressed to equivalent levels in 293T cells.
Signal Transduction-To evaluate the functional effects of deletion of the putative membrane spanning domain on the ability of caveolin-1 to inhibit signal transduction, we employed a MAP kinase reporter system (from Stratagene, Inc.) to measure EGFR-mediated signal transduction in vivo. By using this assay system, we have previously shown that co-transfection of wild-type caveolin-1 with EGFR is sufficient to inhibit dramatically EGFR-mediated signaling via the p42/44 MAP kinase cascade to the nuclear transcription factor Elk (15,17). Fig. 7A shows that wild-type caveolin-1 (FL) and two of the deletion mutants (⌬TM and 1-101) all inhibited EGFR-mediated signaling. In addition, they were capable of inhibiting signaling initiated by co-transfection with downstream elements of the pathway, such as Ha-Ras (G12V), Raf-1ca, and MEKca. These results indicate that caveolar localization is not required for the ability of caveolin-1 to inhibit signaling, as Cav-1 (⌬TM) retains inhibitory activity and is quantitatively excluded from caveolae domains (Fig. 5). However, both Cav-1 (⌬TM) and Cav-1-(1-101) were significantly less potent than wild-type caveolin-1 in inhibiting signaling.

The Caveolin-1 Scaffolding Domain Mediates Membrane Attachment in Vitro
As the above experiments suggested a role for the caveolin-1 scaffolding domain in mediating membrane attachment, we decided to test this hypothesis in vitro. For this purpose, we used a panel of purified recombinant GST-Cav-1 fusion proteins and a purified lipid extract to reconstitute the membrane attachment of caveolin-1. The construction and expression of these GST-Cav-1 fusion proteins is shown in Fig. 8, A and B. We first examined the incorporation of GST-Cav-1-(1-178) that encodes the full-length caveolin-1 molecule, as compared with GST alone. Fig. 8C shows that the full-length caveolin-1 molecule (1-178) was able to confer membrane attachment of GST, whereas GST alone did not show any affinity for these reconstituted membranes. These observations are consistent with our previous data showing that a bacterially expressed His-tagged form of full-length caveolin-1 was efficiently incorporated into membranes in vitro (12).
In addition, GST-Cav-1 (135-178), encoding the entire caveolin-1 C-terminal domain, showed membrane binding activity (Fig. 8D). This suggests that the C-terminal domain of caveolin-1 may play a role in its membrane attachment. In support of these in vitro observations, we find that a GFP (green fluorescent protein)-fusion containing only the C-terminal domain of caveolin-1 (residues 135-178) is sufficient to target this reporter to Golgi membranes in vivo, suggesting that C-terminal domain contains a signal for Golgi membrane attachment.
Perhaps surprisingly, GST-Cav-1 (102-134) encoding the putative membrane spanning domain showed no affinity for membranes (Fig. 8D). However, we cannot rule out the possibility that this may be due to steric hindrance. But this seems unlikely, as other short protein segments of caveolin-1 fused to GST (61-101 and 134 -178) efficiently associated with membranes. In addition, we recently showed that GST-Cav-1 (102-134) specifically interacts with caveolin-2, mediating the formation of caveolin-1/-2 hetero-oligomers (48). These results indicate that this GST fusion protein carrying the putative membrane spanning domain is clearly functional in other assay systems, both in vitro and in vivo (48) (see "Discussion").

FIG. 5. Caveolar targeting of caveolin-1 deletion mutants.
Transiently transfected 293T cells were used to separate membranes enriched in caveolae from the bulk of cellular membranes and cytosolic proteins, by employing an established equilibrium sucrose density gradient system (21,26,29,37,38,(41)(42)(43)(44)(45)(46). In this fractionation scheme, immunoblotting with anti-caveolin IgG can be used to track the position of caveolae-derived membranes within these bottom-loaded sucrose gradients. A and B, the distribution of total cellular protein is shown, as demonstrated by Ponceau S staining (A) and protein quantitation (B). C, the distribution of wild-type caveolin-1 and the deletion mutants is shown. Note that wild-type caveolin-1 (FL) is correctly targeted to these low density Triton-insoluble membranes (fractions 4 or 5) that are enriched in caveolae membranes. In contrast, two of the caveolin-1 mutants (⌬TM and 1-81) were quantitatively excluded from these caveolae-enriched fractions. Interestingly, Cav-1-(1-101) was still efficiently targeted to these caveolae-enriched domains (fractions 4 or 5), although less efficiently than wild-type caveolin-1.

DISCUSSION
Several independent lines of evidence indicate that caveolin-1 is an integral membrane protein with a cytoplasmic Nterminal domain and a cytoplasmic C-terminal domain. First, wild-type caveolin-1 contains a central 33-amino acid hydrophobic stretch of amino acids that encodes a putative membrane spanning segment, and biochemically caveolin-1 is carbonate-inextractable (27,29,49), behaving as expected of an integral membrane protein. Second, both the N-and C-terminal domains face the cytoplasm, as shown using anti-peptide antibodies and by employing epitope tags (9,26,47,49). In addition, both the N-and C-terminal domains of caveolin-1 undergo cytoplasmic modifications as follows: the extreme N terminus is tyrosine-phosphorylated by v-Src on Tyr-14 and several cysteines within the C-terminal domain are S-acylated by palmitoylation (18,47,50). Third, caveolin-1 is inaccessible to cell surface biotinylation, indicating no detectable portion of the protein is extracellular (10). Fourth, consistent with the idea that caveolin-1 is an integral membrane protein, caveo-lin-1 in native plasma membranes can be photoaffinity labeled with a variety of lipids, including a glycosphingolipid (GM 1 ) and a free-fatty acid (51,52); additional evidence has been presented that caveolin-1 is a cholesterol-binding protein (11,12,53). Taken together, these data have been used to construct a model of caveolin-1 in which the putative membrane spanning domain forms a hairpin loop within the membrane, allowing both the N-and C-terminal domains of caveolin-1 to remain entirely cytoplasmic (the "hairpin loop" model). However, no experimental evidence has yet been presented demonstrating that this central hydrophobic stretch (residues 102-134) is required for membrane attachment of caveolin-1.
Here, we have created a series of caveolin-1 deletion mutants to examine whether the membrane spanning segment is required for membrane attachment of caveolin-1 in vivo. One mutant, Cav-1-(1-101), contained only the cytoplasmic N-terminal domain and lacked the membrane spanning domain and the C-terminal domain. However, Cav-1-(1-101) still behaved as an integral membrane protein after transient expression in FIG. 7. Regulation of signal transduction by caveolin-1 deletion mutants. A, note that wild-type caveolin-1 (FL) and two of the deletion mutants (⌬TM and 1-101) all inhibited EGFR-mediated signaling. In addition, they were capable of inhibiting signaling initiated by co-transfection with downstream elements of the pathway, such as Ha-Ras (G12V), Raf-1ca, and MEKca. However, both Cav-1 (⌬TM) and Cav-1-(1-101) were significantly less potent than wildtype caveolin-1 in inhibiting signaling. In addition, co-transfection with an irrelevant soluble protein (␤-galactosidase (␤gal)) had no effect on signaling (approximately constitutively active). B, note that Cav-1-(1-81) did not possess any inhibitory activity in this assay system. These results are consistent with the hypothesis that the caveolin-1 scaffolding domain (residues 82-101) possesses inhibitory activity, as Cav-1-(1-101) retains significant inhibitory activity, whereas Cav-1-(1-81) is completely ineffective. A and B, the mean Ϯ 1 S.D. of three data points is shown. In addition, these results were reproduced three times independently.
To rule out the possibility that membrane attachment mediated by the caveolin scaffolding domain was indirect, we reconstituted the membrane binding of caveolin-1 in vitro. By using purified GST-caveolin-1 fusion proteins and reconstituted lipid vesicles, we showed that the caveolin-1 scaffolding domain is both necessary and sufficient for membrane attachment in vitro (Table II). Interestingly, we also observed that the Cterminal domain (residues 135-178) may participate in membrane attachment. However, the putative membrane spanning domain (residues 102-134) did not show any physical association with membranes in this in vitro system. These data provide strong evidence that the caveolin scaffolding domain and the C-terminal domain contribute to the membrane attachment of caveolin-1.  1-81, 1-101, and 61-101), the putative membrane spanning region (102-134), and the C-terminal domain (135-178). C and D, the incorporation of recombinant GST-Cav-1 fusion proteins into membranes was monitored by flotation in sucrose density gradients (12). GST alone or a given purified GST-Cav-1 fusion protein was mixed with purified lipid components in a buffer containing 60 mM octyl glucoside and dialyzed overnight against a buffer lacking detergent to allow the association of caveolin with lipids. The dialysate was then adjusted to 40% sucrose by the addition of 2 ml of 80% sucrose prepared in MBS and placed at the bottom of an ultracentrifuge tube. A 5-30% discontinuous sucrose gradient (lacking detergent) was formed above and centrifuged. A light-scattering band confined to the 5-30% sucrose interface contained the associated lipids and formed regardless of whether or not caveolin was added. The light-scattering band at the 5-30% sucrose interface was collected by centrifugation and subjected to immunoblot analysis with anti-GST IgG. C, note that the full-length caveolin-1 molecule was able to confer membrane attachment of GST, whereas GST alone did not show any affinity for these reconstituted membranes. D, note that GST-Cav-1-(1-101) and GST-Cav-1-(61-101) efficiently bind to membranes, whereas GST-Cav-1-(1-81) shows no affinity for membranes. These results implicate residues 82-101 (the caveolin-1 scaffolding domain) in membrane attachment. In addition, GST-Cav-1-(135-178), encoding the entire caveolin-1 C-terminal domain, showed membrane binding activity, whereas GST-Cav-1-(102-134) encoding the putative membrane spanning domain showed no affinity for membranes. What is the function of the putative membrane spanning domain of caveolin-1? One possibility is that it still functions as a membrane anchor in the context of the full-length caveolin-1 molecule. Alternatively, it may subsume another unexpected functional role in mediating specific protein-protein interactions. For example, in a recent report we showed that the putative membrane spanning domains of caveolins-1 and -2 mediate the formation of caveolin hetero-oligomers, i.e. the interaction between caveolins 1 and 2 (48). By using a panel of GST-caveolin-1 fusion proteins, we localized the caveolin-2binding site to the caveolin-1 membrane spanning domain. In addition, by using an overlay assay employing radiolabeled GST-caveolin-1, we identified two regions within caveolin-2 that bind to caveolin-1 as follows: a membrane proximal region of the caveolin-2 scaffolding domain and a region of the caveolin-2 membrane spanning domain. Finally, we demonstrated that recombinant expression of a GST fusion protein carrying the membrane spanning domain of caveolin-1 in mammalian cells is sufficient to stabilize the formation of these caveolin hetero-oligomers in vivo. Taken together, our results indicated that the membrane spanning domains of both caveolins-1 and -2 play a critical role in mediating their ability to interact with each other. However, these findings do not exclude the possibility that these protein-protein interactions occur within lipid bilayer of the plasma membrane (48).