A Critical Role of Cavin (Polymerase I and Transcript Release Factor) in Caveolae Formation and Organization^*

Caveolae were first defined from their appearance as “little caves” that were visible in electron micrographs of cell surfaces (1, 2). These flask-like invaginations of the plasma membrane are expressed in many but not all eukaryotic cell types, and they are most abundant in vascular endothelial cells and in primary and cultured adipocytes (3-9) (reviewed in Ref. 10). Caveolae have been suggested to play a role in a variety of pathologies, including cancer, diabetes, and cardiovascular disease (10). Postulated mechanistic functions of caveolae include a role in vesicular transport via their endocytosis/transcytosis in endothelium and other tissues (11-14) and the regulation of cellular cholesterol (15, 16). Caveolae have also been suggested to play a role in signal transduction by serving as organizing centers for a variety of receptors and signal molecules, thus regulating their function (17).

In certain cells, morphologically defined caveolae can be formed by the expression of their structural protein caveolin-1 (18, 19), which is a small (178 amino acids) integral membrane protein whose 34 hydrophobic amino acids form a “U,” which is inserted into the inner leaflet of the membrane bilayer and never reaches the outside of the cell. The rest of the protein is cytosolic, and the C terminus aligns along the inner leaflet of the bilayer by virtue of multiple palmitoylations (20). There are three caveolin isoforms, and they are expressed in a tissue-specific manner to a certain extent (13, 18, 21, 22). Caveolin-3 is expressed primarily in cardiac and skeletal muscle (22, 23). The tissue distribution of caveolin-1 and -2 is broad but not universal. Their expression is virtually absent in liver and brain and is particularly high in vascular endothelial cells and adipocytes (10). Caveolin-2 appears to require the expression of caveolin-1 to target it to plasma membrane caveolae (24), and it does not seem capable of forming caveolae when expressed on its own. However, although the expression of caveolin-1 has been described to be necessary and sufficient for the formation of morphological defined caveolae (18, 19), recent results suggests that both caveolin-1 and -2 may be necessary for formation of caveolae in some cells (25).

Numerous proteins have been found in caveolae using a variety of protocols for their isolation (see Ref. 26 for a discussion of the relevant methodology). We employed an immunoisolation method to purify caveolae after these structures are pinched off from the adipocyte plasma membrane, either during cell disruption or following sonication (8, 26). The major protein components of immunoisolated caveolae were identified by mass spectrometry-based sequencing to be caveolin-1 and -2, CD36, semicarbazide-sensitive amine oxidase, and polymerase I and transcript release factor (PTRF)² (8, 26), also known as cavin (27).

Vinten and colleagues (28, 29) identified PTRF as a caveolae component by immunological and morphological procedures and called it p60 cavin. They produced a monoclonal anti-p60 cavin antibody and showed its specific localization to the cytoplasmic face of adipocyte plasma membrane caveolae by electron microscopy. They also showed by Western blot that the expression of cavin closely mirrors that of caveolin in tissue distribution (27). Cavin was also found by mass spectrometry to be a component of human adipocyte caveolae (30). PTRF was originally identified by two independent groups using yeast two-hybrid screens (31, 32). The subsequent analysis of its putative eponymous function in the nucleus was performed in vitro, and no localization studies of PTRF in cells was performed by these workers (31, 32). However, it is evident that cavin is exceptionally abundant in plasma membrane caveolae (reviewed in Ref. 26). There are no obvious structural motifs in the cavin sequence, e.g. a transmembrane domain, that would obviously link it to caveolae or to membranes, but there are three leucine zippers, phosphorylation sites, and a putative nuclear localization sequence (30).

The physiological role of cavin in caveolae remains unknown. Here, our data confirm that cavin co-localizes with caveolin-1 in adipocytes by confocal microscopy and co-distributes with caveolin-1 in lipid raft fractions in sucrose gradients. We show by several techniques that altering the level of cavin expression results in concomitant changes in caveolin-1 levels. Moreover, cytoskeletal perturbation and cholesterol depletion also result in diminished cavin and caveolin-1 levels. We postulate that cavin plays a critical role in the stabilization and/or formation of caveolae.

EXPERIMENTAL PROCEDURES

Reagents—Dexamethasone, 3-isobutylmethylxanthine, insulin, sodium fluoride, sodium orthovanadate, fetal bovine serum (Australian origin), benzamidine, β-methylcyclodextrin (β-CD), and mouse immunoglobulin G (IgG) were purchased from Sigma. Nocodazole (NCZ) and latrunculin B (LatB) were purchased from BIOMOL Research laboratories (Plymouth Meeting, PA). LB base, ampicillin, kanamycin, aprotinin, leupeptin, and pepstatin A were obtained from American Bioanalytical (Natick, MA). Calf serum was purchased from Invitrogen, and Dulbecco's modified Eagle's medium (DMEM) was from Mediatech (Herndon, VA). Lipofectamine 2000 reagent and the pcDNA 3.1 expression vector were purchased from Invitrogen. A BCA protein assay kit was from Pierce. Protein A-agarose was from Santa Cruz Biotechnology (Santa Cruz, CA). Penicillin, streptomycin, and trypsin were purchased from Invitrogen. Transfection vectors (pEGFP-N3 and pEGFP-C3) were purchased from BD Bioscience Clontech (Palo Alto, CA).

Antibodies and Western Blotting—Monoclonal antibodies recognizing cavin (2F11) (28), caveolin-1 (7C8) (8), have been previously described. The following antibodies were commercially acquired: anti-caveolin-1 was from BD Transduction Laboratories (San Jose, CA), anti-actin was from Sigma; anti-transferrin receptor was from Zymed Laboratories Inc. Invitrogen. Additional anti-cavin/PTRF antibodies were purchased from BD Transduction. Polyclonal rabbit anti-cavin antibody was also produced against a peptide sequence at the C terminus of the protein (21st Century Biochemicals, Hopkinton, MA). Primary antibodies were detected in Western blots using secondary antibodies conjugated to horseradish peroxidase (Sigma) diluted 1:3000 and chemiluminescent substrate (PerkinElmer Life Sciences).

Cell Culture—3T3-L1 fibroblasts were maintained in DMEM containing 4.5 g/liter glucose and l-glutamine supplemented with 10% calf serum and 100 units/ml penicillin, and 100 μg/ml streptomycin. Two days after confluence, cells were induced to differentiate by changing media to DMEM with 10% fetal bovine serum, 0.5 mm 3-isobutylmethylxanthine, 1 μm dexamethasone, and 1.7 μm insulin. After 48 h, the induction medium was removed and cells were maintained in DMEM with 10% fetal bovine serum as described previously (33). CHO cells were maintained in DMEM with 10% fetal bovine serum in 5% CO₂. When they reached 90% confluence, they were transfected with 2 μg of the cDNA of interest by means of the Lipofectamine 2000 reagent.

Subcellular Fractionation of Adipocytes—This procedure was performed on differentiated 3T3-L1 adipocytes (6) essentially as described for rat adipocytes (34). Cells were incubated with appropriate effectors as described in the figure legends, and then washed by cold PBS for three times, cold HES buffer once, and homogenized with a Teflon-glass tissue grinder in HES buffer. Subcellular fractions (plasma membrane (PM), heavy microsomes (HM), and light microsomes (LM)) were obtained by differential centrifugation and resuspended in HES. Buffers used with subcellular fractionation contained a mix of protease inhibitors consisting of 1 μm aprotinin, 10 μm leupeptin, 1 μm pepstatin, and 5 mm benzamidine.

Triton X-100 Solubilization and Sucrose Gradient Centrifugation—This protocol was performed as described in previous studies (35, 36). Briefly, 3T3-L1 adipocytes were incubated with appropriate effectors as described in the figure legends, then 40 × 10⁶/ml cells were lysed in 2 ml of MBS (25 mm MES and 150 mm NaCl, pH 6.5) containing 1% Triton X-100 and supplemented with a protease inhibitor mix (Roche Molecular Biochemicals). The samples were then incubated at 4 °C for 20 min with end-over-end rotation. The solubilized cells were homogenized with 10 strokes of a Dounce homogenizer, and 1 ml of the homogenate was added to an equal volume of 80% (w/v) sucrose in MBS. The solubilized cells (in 40% sucrose) were placed at the bottom of a centrifuge tube and overlaid successively with 2 ml of 30% sucrose and 1 ml of 5% sucrose (in MBS). After centrifugation at 240,000 × g in a Beckman SW55 rotor for 18 h, 0.3- to 0.4-ml fractions were collected from the bottom of the gradient (designated fractions number 1 (top) through 14 (bottom)) and immediately supplemented with Laemmli sample buffer.

Immunoprecipitation—The whole cell lysates or PM fraction was solubilized with 1% Triton X-100 or 60 mm octylglucoside for 2 h at 4 °C with constant agitation. Insoluble material was removed by pelletting for 10 min in a microcentrifuge. Monoclonal or polyclonal anti-caveolin-1 antibodies, monoclonal anti-cavin antibodies, and nonspecific mouse and rabbit IgGs (5 μg) were incubated with the supernatant 1 h at 4 °C, then 20 μl of protein A beads (Santa Cruz Biotechnology) was added for 1 h. The supernatant with unbound proteins was collected, and the beads were washed four times and eluted with SDS-PAGE loading buffer containing 2% SDS.

Confocal Microscopy—This protocol was performed as described by (37, 38). Briefly, cells were washed three times with PBS and fixed with 3% (w/v) paraformaldehyde in PBS for 20 min at room temperature. The cells were permeabilized, and nonspecific binding sites were blocked in PBS containing 0.1% saponin, 0.4% bovine serum albumin, and 5% normal goat serum for 10 min at room temperature. The cells were then incubated with primary antibodies for 2 h at room temperature, and washed three times with PBS containing 0.1% saponin. Next, the cells were incubated with Cy3-conjugated anti-rabbit IgG and Cy2-conjugated anti-mouse IgG (1:200 dilution) for 1 h at room temperature. Finally, the cells were washed with PBS containing 0.1% saponin, mounted in 50% glycerol saturated with n-propyl gallate as an anti-bleaching reagent, and observed with an fluorescence microscope equipped with a laser confocal system (Zeiss). Captured images were processed with LSM 5 Image Browser software.

Small Double-stranded RNA-mediated Interference—The RNA-meditated interference experiments were carried out using retroviral vector (pSUPER) expressing shRNA fragments. Three sequences for RNA interference were selected from the cavin cDNA-coding region (5′-AAGGAGAACCTGGAGAAGA-3′, 5′-TCCGACGAGCTGATCAAGT-3′, and 5′-AAGTTGCTGGAGAAGGTGC-3′) and analyzed by a BLAST search to ensure that they did not have significant sequence homology with other genes. Three pairs of oligonucleotides were synthesized and annealed to construct the RNA interference retroviral vector: The annealed double-stranded oligonucleotides were inserted into the BglII/HindIII site of pSUPER retrovirus vector. Three stable cell lines were obtained after retroviral transduction and puromycin selection. The most effectively cavin knockdown cell line was obtained with probe 1 and the resultant cell line was used in all experiments.

Retrovirus Production and Transduction—Cells (HEK 293) were transfected with viral DNA at 50% confluence by the use of FuGENE 6 reagent. Forty-eight hours after transfection, virus supernatant was harvested, centrifuged, and filtered. Forty percent confluent 3T3-L1 cells were transduced with virus supernatant diluted with one volume of fresh standard medium. On the following day, the cells were subjected to puromycin selection. Stable cell populations were obtained after 10-14 days of selection.

Fatty Acid Uptake—The fatty acid uptake assays were performed as described previously (39). Briefly, a 100 mm nonradioactive oleic acid stock solution was prepared in 0.1 mm NaOH by heating at 70 °C. The appropriate amount of 100 mm oleic acid stock solution was added to 173 μm fatty acid-free bovine serum albumin solution at a ratio of 1:1 and incubated for 30 min at 55 °C. Just prior to use, the radioactively labeled oleic acid ([9,10-³H]oleic acid) were conjugated to the fatty acid bovine serum albumin stock solution. The oleate/bovine serum albumin solution (2 ml) was incubated with each 3T3-L1 adipocytes in a 3.5-cm Ø culture dish at 37 °C. The uptake was stopped by removal of the solution followed by addition of 2 ml of an ice-cold stop solution containing 0.5% (w/v) albumin and 200 μm phloretin. The stop solution was discharged after 2 min, and the culture dishes were washed by dipping them four times in ice-cold incubation buffer. NaOH (0.5 m) was added to lyse the cells, and aliquots of the lysate were used for protein and radioactivity determination. Radioactivity was determined by a liquid scintillation counter.

Statistics and Data Analysis—Results are expressed as the mean ± S.E. for at least three independent experiments in all figures. Student's t test was performed to determine statistical significance.

RESULTS

Cavin Co-localizes with Caveolin-1 in Lipid Rafts in Adipocytes—Previously we and others found cavin to be a major caveolae protein in primary adipocytes by biochemical and immunological procedures (26-30). Here we investigated the localization of cavin and caveolin-1 by confocal microscopy and sucrose gradient flotation. As shown in Fig. 1A, caveolin-1 and cavin co-localize to a high extent on the plasma membrane of 3T3-L1 cultured adipocytes, whereas the transferrin receptor serves as a negative control and does not co-localize at all with caveolin-1. Next, we used a well established protocol based on the detergent resistance and low buoyant density of lipid rafts to separate caveolae fractions from bulk cellular proteins (35, 36). 3T3-L1 adipocytes were solubilized in 1% Triton X-100 for 30 min, and then were centrifuged in a discontinuous sucrose gradient. As shown in Fig. 1 (B and C), this procedure separates solubilized proteins from those in detergent-resistant lipid rafts. Solubilized proteins remain in the 40% sucrose layer, whereas proteins present in lipid rafts accumulate near the interface of the 5 and 30% sucrose layers. Western blot analysis of recovered fractions revealed that 50% of cavin is present in raft fractions and co-distributes with caveolin-1 (Fig. 1C), whereas no transferrin receptor is found there as expected. These data further confirm that cavin is a lipid raft constituent that co-localizes with caveolin-1 in caveolae. We also found 22% of cellular actin to be present in lipid raft fractions, indicating a possible link between caveolae and the cytoskeleton.

Association of Cavin with Caveolin-1—The localization of cavin in lipid raft fractions requires that it must interact with protein components of caveolae, cytoskeleton components, or lipids. To test possible protein-protein interactions between cavin and caveolin-1 we performed co-immunoprecipitation protocols. As shown in Fig. 2, IP with anti-caveolin-1 antibody, from plasma membrane solubilized in lysis buffer (1% Triton X-100), results in the co-immunoprecipitation of cavin and the converse IP of caveolin-1 by cavin gives a similar result. However, because of the detergent-resistant nature of lipid rafts, these solubilization conditions do not lead their complete disruption (26, 40) raising the possibility that the observed co-immunoprecipitation results from an indirect interaction. To test this possibility, we used 60 mm octylglucoside, a relatively gentle detergent that generally does not affect protein-protein interactions, but efficiently solubilizes lipid rafts (40). As shown in Fig. 2, in octylglucoside buffer the interaction of caveolin-1 and cavin was almost completely eliminated, indicating that their association is most likely indirect and dependent on intact raft or caveolae structure (see “Discussion”).

β-CD Treatment Dramatically Decreases Cavin Expression Levels in Lipid Rafts—Cholesterol is a critical and major component of caveolae, and it is known that β-CD-mediated cholesterol extraction causes a rapid loss of morphologically identifiable caveolae (41, 42). Accordingly we performed separation of plasma membranes into soluble and lipid raft fractions from 3T3-L1 adipocytes treated with β-CD or not. As shown in Fig. 3A, after β-CD treatment the amount of cavin and caveolin-1 was significantly reduced in both PM and raft fractions, the portion from the lipid raft (insoluble) fractions being almost totally lost (Fig. 3, B and C). Cavin shows a more extensive loss than caveolin-1, 88 and 49%, respectively, in lipid raft fractions (Fig. 3C). Interestingly, the total amount of actin as well as that in the lipid raft was significantly reduced after β-CD treatment (71%), in confirmation of a potential link between caveolae and the cytoskeleton, possibly mediated via cavin.

The Localization of Cavin in Lipid Raft Fractions Is Dependent on Microtubules and Actin Filament Structures—Given the localization of actin in the lipid raft fraction, we considered that there might be an interaction between caveolae and the cytoskeleton, although a direct interaction of caveolin-1 or cavin with actin or tubulin could not be demonstrated by IP (data not shown, also see “Discussion”). To further evaluate the relationship between caveolae components and the cytoskeleton, we treated 3T3-L1 adipocytes with NCZ and LatB, which disrupt microtubules and actin filaments, respectively, then performed sucrose gradient fractionation. As shown in Fig. 4, NCZ and LatB treatment reduced the distribution of cavin from lipid raft fractions by 52 and 55%, respectively, with no effect on raft-associated caveolin-1. Theses data further support a link between caveolae/lipid rafts and the cytoskeleton via cavin.

Cavin Can Promote Caveolin-1 Incorporation into Lipid Raft Fractions in HEK293 Cells—Structural studies suggested that the expression of caveolin-1 oligomers can form rafts, but that expression of one or more additional proteins is likely to be required for the formation and maintenance of their cave-like morphology (43). Previously we created cell lines expressing high levels of caveolin-1 (44). The expression of caveolin-1 increased the level of surface cholesterol by 70%, but electron microscopy revealed minimal caveolae in either parental or caveolin-1-transfected HEK293 cells (data not shown). To determine whether cavin expression affects the raft partitioning of caveolin-1 in the plasma membrane, we transfected cavin cDNA into a HEK293 cell line highly expressing caveolin-1, and then performed sucrose gradient fractionation. As show in Fig. 5, 33% of caveolin-1 was localized to lipid rafts in untransfected HEK293 cells. After cavin transfection into this cell line, the total amount of caveolin-1 and that distributed in lipid raft faction increased by 2.5-fold, indicating cavin plays a role in forming or stabilizing caveolin-1 expression in lipid raft fractions and presumably caveolae.

Knockdown of Cavin in 3T3-L1 Cells Reduces Caveolin-1 Expression—To test the hypothesis that cavin is a critical component of caveolae, we specifically suppressed its expression by retrovirus-driven shRNA in a stably transfected 3T3-L1 cell line. We then evaluated the expression levels of cavin, caveolin, and other caveolae-localized proteins in total cell lysates of 3T3-L1 adipocytes. As shown in Fig. 6 (A and B), cavin protein expression was reduced by 75% and caveolin-1 was comparably reduced by 66%, whereas total cellular actin was unchanged. Moreover, the amount of the non-caveolar raft protein, flotillin (8), was unchanged indicating the caveolae specificity of cavin reduction. Next, we examined the distribution of cavin, caveolin-1, and actin in lipid rafts as in previous figures. As shown in Fig. 6 (C and D), cavin and caveolin-1 were correspondingly decreased in the lipid raft fractions (box). Importantly, actin protein was also reduced in lipid raft fractions from cavin shRNA knockdown cells, yet another piece of data supporting a link between cavin/caveolae and the cytoskeleton network. These data immediately demonstrate an important partnership between caveolin and cavin in lipid rafts and suggest that cavin plays a functional role to stabilize caveolae morphology or mediate their formation. Finally, neither transferrin receptor nor tubulin shows any significant changes due to cavin knockdown, which serves as a control for non-raft associated proteins and microtubules, respectively.

A Cavin C-terminal Truncation Accumulates with Microtubules When Overexpressed in CHO Cells—We generated cavin-eGFP, then overexpressed this construct in CHO cells. As show in Fig. 7, wild-type cavin-eGFP gives punctate signals throughout the cell, likely indicative of caveolae cluster structures (45) and overlaps substantially, though not completely, with caveolin-1, probably due to relative expression levels. Interestingly when we generated a truncated cavin form, which lacks 76 amino acids in its C terminus (cavin 1-322-eGFP), it showed a clear cytoskeleton-like structure as shown in Fig. 7A. To further confirm this localization, we co-stained with Alexa-phalloidin anti-tubulin and anti-cavin antibodies. The signals of truncated cavin-1-322-eGFP and tubulin merged almost completely, indicative of cavin-1-322-eGFP accumulation at microtubule bundles (Fig. 7B).

Cavin/Caveolin-1 Play a Functional Role on Fatty Acid Uptake—Previously we reported that transport of fatty acid across the plasma membrane is modulated by caveolin-1 (16). To further evaluate the functional role of cavin/caveolin-1, we performed a fatty acid uptake assay in our cavin-shRNA knockdown cells and compared them to control cells. As show in Fig.8, fatty acid uptake was decreased by 16% in cavin shRNA knockdown cells, consistent with our previous proposal for a possible physiological functional of caveolae.

DISCUSSION

We identified cavin as a major protein component of caveolae (26) in confirmation of results from two other laboratories (27-30). We show here that cavin co-localizes with caveolin-1 in primary rat and 3T3-L1 cultured adipocytes by confocal microscopy (Fig. 1A), and by a lipid rafts sucrose gradient flotation protocol (Fig. 1B). Cavin levels, like caveolin-1, are sensitive to cholesterol depletion, because β-CD treatment caused a dramatic down-regulation of cavin from plasma membrane and lipid raft fractions (Fig. 3). Increasing cavin expression by transfection (Fig. 5) or decreasing it by means of shRNA (Fig. 6) resulted in comparable changes in caveolin-1 levels, suggesting that cavin is a critical protein component of caveolae.

The essential role of caveolins in the formation of caveolae has been well established. The most compelling evidence for the structural role of caveolin in the formation of caveolae comes from knockout mice (46-51). It has been shown that, upon caveolin-1 gene loss, endothelial cells and adipocytes lack caveolae while they are still present in other cell types that express caveolin-3 (46-49). Conversely, in caveolin-3 null mice, the complete lack of caveolae invaginations is readily seen in muscle cells (50, 51). Although caveolin expression is a necessary condition for caveolae formation, whether or not caveolin is sufficient for caveolae structure is not clear. Richard Anderson's group reported, based on structural studies, that caveolin-1 alone could not account for the membrane curvature of caveolae (43). Several electron microscopy studies of caveolae suggest the localization of caveolin occurs only in the necks and not in the bodies of caveolae (13, 42, 52, 53). Thus it is likely that caveolae formation requires proteins other than caveolins, and we document here that cavin is one such protein. We created two independent cell lines, HEK293-caveolin-1-cavin overexpressing cells, and 3T3-L1 cavin-shRNA knockdown cells. Upon the overexpression of cavin, caveolin-1 recruitment into lipid raft fractions was increased by about 2.5-fold compared with the control cells (Fig. 5). On the other hand in 3T3-L1 cavin knockdown cells, cavin levels were reduced by 70% and caveolin-1 was comparably down-regulated from whole cell lysates as well as from lipid raft fractions (Fig. 6). These data immediately indicate an important functional role of cavin in caveolae formation and/or stabilization. We have not determined by electron microscopy the number of caveolae structures in our knockdown cell line. However, preliminary data from a cavin knockout mouse model indicate a dramatic disruption of caveolae morphology in endothelial cells (data not shown). Manipulation of cavin levels will therefore likely provide another useful way to study the functional role of caveolae.

Caveolae can be anchored to the plasma membrane by cytoskeleton components (45, 54). An actin cross-linking protein, filamin, is one of the proteins identified as a ligand for caveolin-1 (55). However, the molecular mechanism for binding of caveolae to the cytoskeleton has not been established. Here we show that a truncated form of cavin, eGFP-cavin-322, accumulates in a manner similar or identical to the microtubular filament network (Fig. 7), suggesting cavin may serve as a direct connection between the caveolae component and the cytoskeleton. It is reported that caveolae can be decorated with myosin fragments making them more prominent, suggesting the participation of a myosin interacting molecule in their structure (56). A sequence homology search on www.ncbi.nlm.nih.gov/ shows that cavin has a 20% sequence identity and 44% sequence similarity in the region from amino acid residues 60-296 with myosin heavy chain (27). Our data also show depolymerization of microtubules with NCZ or actin filaments with LatB dramatically reduced the amount of cavin from lipid raft fractions (Fig. 4A, top), whereas there was no change in caveolin-1 levels (Fig. 4A, bottom). All these data are consistent with the role of cavin in stabilizing caveolae formation through the cytoskeleton network. It is yet to be determined how cavin participates in the formation of the caveolae structure and what the roles of tubulin and actin are in this regard. Actin is present and tubulin is absent from caveolin-1/cavin-containing lipid rafts (Fig. 6), yet C-terminal truncated cavin aligns with tubulin and microtubules. The 74 amino acids from the C-terminal of cavin may be important for the localization of cavin to caveolae, but we have been unable to express this sequence, possibly due to its susceptibility to proteolysis.

How is cavin localized to caveolae? We failed to see a direct interaction between caveolin-1 and cavin by IP in octylglucoside (Fig. 2), suggesting that these proteins do not directly interact or that the interaction is weak. Its possible that there are additional proteins in caveolae that are involved in their structure and/or function, and these could be serum deprivation response protein and protein kinase Cδ-binding protein. These proteins are structurally related to cavin (30), and both proteins were identified by mass spectrometry as being present in human adipocyte caveolae (30). A possible functional role for protein kinase Cδ-binding protein in binding protein kinase C is implied in its name, and a similar activity was verified experimentally for serum deprivation response protein by Mineo et al. (57). We have also verified the presence of these two proteins in purified adipocyte caveolae (data not shown). Thus, further study of the possible interaction of these proteins is underway.

Recent data implicate a connection between cytoskeleton components and caveolae-regulated endocytosis (54, 58). A few studies also suggest microtubules may play a role in caveolae-dependent endocytosis (54, 59-61). Cholera toxin is one of the best studied caveolae ligands, which binds to a ganglioside GM1 (monosialoganglioside) and is transported by caveolae endocytosis to an intracellular compartment. However, we did not see any difference in cholera toxin uptake in cavin-shRNA knockdown cells as compared with control cells (data not shown). Indeed, caveolae are relatively stable structures in adipocytes (8, 26) and in other cells (62), with only a small percentage undergoing internalization at a slow rate.

The physiological one or more functions of cavin in caveolae remain unknown. Previous studies suggested that caveolae play an important role in lipid raft-dependent fatty acid uptake in adipocytes (63) and that the transport of fatty acid across the plasma membrane is modulated by caveolin-1 (16). Although it is not clear if cavin is directly involved in fatty acid transport, here our data show the fatty acid uptake is down-regulated in cavin-shRNA cells (Fig. 8), supporting the hypothesis that caveolae may play a functional role in fatty acid transport. The relatively small effect of diminishing cavin/caveolae expression on fatty acid uptake can possibly be explained by the non-linear relationship between caveolin expression and uptake (16).

Taken together, our data support a model in which cavin serves as an important caveolae organizer, stabilizing their morphology and linking them to the cytoskeleton. Manipulation of cavin expression is mirrored by essentially identical changes in caveolin-1 expression in confirmation of its critical role in caveolae formation and/or stabilization. As discussed above, cavin may be the structural protein of caveolae that mediates the curvature of the “little cave” structures and may be the rope-like protein seen in cryoelectron microscopy and previously thought to be due to caveolins (13). Further work is in progress to test this and other possible functions of cavin.