A role for caveolin in transport of cholesterol from endoplasmic reticulum to plasma membrane.

Caveolin is a 22-kDa membrane protein found associated with a coat material decorating the inner membrane surface of caveolae. A remarkable feature of this protein is its ability to migrate from caveolae directly to the endoplasmic reticulum (ER) when membrane cholesterol is oxidized. We now present evidence caveolin is involved in transporting newly synthesized cholesterol from the ER directly to caveolae. MA104 cells and normal human fibroblasts transported new cholesterol to caveolae with a half-time of ∼10 min. The cholesterol then rapidly flowed from caveolae to non-caveolae membrane. Cholesterol moved out of caveolae even when the supply of fresh cholesterol from the ER was interrupted. Treatment of cells with 10 μg/ml progesterone blocked cholesterol movement from ER to caveolae. Simultaneously, caveolin accumulated in the lumen of the ER, suggesting cholesterol transport is linked to caveolin movement. Caveolae fractions from cells expressing caveolin were enriched in cholesterol 3-4-fold, while the same fractions from cells lacking caveolin were not enriched. Cholesterol transport to the cell surface was nearly 4 times more rapid in cells expressing caveolin than in matched cells lacking caveolin.

on day 3 and day 5. The day 5 medium contained 10% (v/v) human lipoprotein-deficient serum. The cholesterol pool was radiolabeled by changing the medium on day 6 or day 7 to DMEM plus 20 mM Hepes, pH 7.4, adding [ 3 H]acetate (50 Ci/dish) to the dish, and was incubated for the indicated times. All experiments were carried out on day 7. MA104 cells were also grown according to a standard format (24). On day zero, 3.0 ϫ 10 5 cells were seeded into a T-75 culture flask and cultured for 5 days in folate-free medium 199 supplemented with 5% (v/v) fetal calf serum and 100 units/ml penicillin/streptomycin. The cholesterol pool was labeled as described for human fibroblasts. For [ 3 H}folate uptake assays, the cells were used directly on day 5 as described previously (25). L1210-JF cells (kindly provided by Dr. Bart Kamen) are a murine lymphocyte cell line expressing the folate receptor (26). On day zero, 5 ϫ 10 6 cells were seeded into T-75 flasks in RPMI medium 1640 plus glutaMax I (Life Technologies, Inc.) and 25 mM Hepes with 10% bovine serum. Transfected cell medium also contained 300 g/ml Geneticin. Cells were grown for 5 days. On day 5 the medium was changed to the same medium containing 10% (v/v) human lipoprotein-deficient serum, and on day 6 or day 7 the cholesterol pool was labeled as for the human fibroblasts.
Radiolabeled Cholesterol Determination-Thin-layer chromatography was used, as described previously (27), to measure the amount of [ 3 H]cholesterol in membrane fractions from cells incubated in the presence of [ 3 H]acetate. Fractions (0.65 ml) were adjusted to 1 ml (final volume) by adding 0.33 ml of 30% taurodeoxycholate and 0.02 ml Buffer A before mixing with 2 ml of Dole reagent (78:20:2; 2-propanol:heptane: water) and 1 ml of heptane. The samples were vortexed and spun in a table-top centrifuge for 10 min (3000 ϫ g). The heptane phase (upper) containing lipids was saved for the thin-layer chromatography plates. The heptane phase was dried under N 2 and suspended in 50 l of the solvent system (80:20:1; petroleum ether:ethyl ether:acetic acid). Pure cholesterol and pure cholestenone were dissolved in the solvent system and used as standards (5 g/spot). Lipids were visualized by charring with sulfuric acid-dichromate and heating at 180°C for 10 min. Unlabeled cholesterol and cholestenone were added to each fraction to facilitate visualization. The appropriate spots were scraped and the amount of radiation quantified by liquid scintillation counting.
Purification of Caveolae-The caveolae were isolated as described previously (28) and modified (29). All steps were carried out at 4°C. A plasma membrane fraction was first prepared from 10 100-mm dishes or T-75 flasks of confluent tissue culture cells (8 -10 mg of total protein). A post-nuclear supernatant fraction (ϳ4 mg of total protein) was layered on the top of 23 ml of 30% Percoll in Buffer A, and centrifuged at 84,000 ϫ g for 30 min in a Beckman Ti 60 rotor. The cytosol corresponded to the top two fractions of the Percoll gradient, the plasma membrane was a 1-ml fraction (a visible band) taken ϳ5.7 cm from the bottom of the centrifuge bottle, and the internal membrane fraction was the bottom 21 1-ml fractions in the gradient. The plasma membrane, which contained ϳ0.6 mg of protein, was collected with a Pasteur pipette, adjusted to 2.0 ml with Buffer A and placed in a Sorvall TH641 centrifuge tube on ice. The membrane was sonicated before mixing with 1.84 ml of Buffer C and 0.16 ml of Buffer A (final OptiPrep concentration, 23%) in the bottom of the same TH641 tube. A linear 20 -10% OptiPrep gradient (prepared by diluting Buffer C with Buffer A) was poured on top of the sample and then centrifuged at 52,000 ϫ g, 90 min in a Sorvall TH641 swinging bucket rotor. Fourteen fractions were collected from this first OptiPrep gradient. The top seven fractions contain the bulk of the caveolae membrane markers but very little protein, while the bottom seven fractions contain low concentrations of caveolae markers but the bulk of the membrane protein (29). Caveolae were prepared from the top seven fractions of the first OptiPrep gradient as described previously (29).
Protease Protection-Cells that had been preincubated for 3 h in the presence of 50 g/ml cycloheximide were washed extensively in PBS and subjected to the indicated treatments, all in the presence of cycloheximide. At the end of the treatments, cells were fractionated as described above. The appropriate samples were incubated with 300 g of trypsin or trypsin plus 0.5% SDS for 30 min on ice. Soybean trypsin inhibitor (300 g) was then added before caveolin was immunoprecipitated.
Immunoprecipitation of Caveolin-Protein A-Sepharose beads were first blocked by incubating them for 4 h at 4°C with human fibroblast cell lysate (200 g/ml) plus 30 mg/ml BSA in Buffer E. Blocked beads were used to preclear each experimental fraction after it had been adjusted to 1% Triton X-100 and 60 mM octylglucoside. Precleared fractions were then incubated for 19 h at 4°C with a 1:400 dilution of anti-caveolin monoclonal antibody before adding blocked, Protein A-Sepharose beads and incubating an additional 2 h at 4°C. Beads were removed by centrifugation, dissolved in Buffer F (30), and proteins separated by electrophoresis. The immunoprecipitated caveolin was detected by immunoblots with a polyclonal anti-caveolin IgG.
Folate Internalization Assay-The [ 3 H]folic acid internalization assay was carried out as described previously (31). MA104 cells grown for 5 days in low folate medium were washed with PBS and then subjected to the indicated experimental treatments. All L1210-JF cells were grown without serum for 18 h in folate-free medium before incubation in the presence of 5 nM [ 3 H]folic acid for 1 h at 37°C. After treatment the cells were chilled on ice for 20 min and washed twice with cold PBS. Surface-bound folate corresponded to the amount of [ 3 H]folic acid released when cells were incubated on ice for 30 s in the presence of acid saline (0.15 M NaCl, adjusted to pH 3.0 with glacial acetic acid). Internalized folate was the amount of [ 3 H]folic acid remaining associated with the acid-saline-treated cells. The latter was collected by adding 0.1 N NaOH to the flask to dissolve the cells. Radioactivity was measured by liquid scintillation counting using a Tri-carb 1900A liquid scintillation analyzer (Packard Instruments Co., Downers Grove, IL). Nonspecific binding, which was measured by adding 100-fold excess unlabeled folic acid, was less than 5% of specific binding.
Transfection with Caveolin cDNA-A full-length human caveolin cDNA was subcloned into pJB20 vector using EcoRI sites. This vector contains an SV40 promoter and a neomycin-selectable marker. Twentyfour hours before transfection, ϳ10 6 cells were seeded per 100-mm dish. On the day of transfection, 5 g of plasmid DNA (either vector alone or vector with full-length insert) was diluted in 200 l of serum-free Ham's F-12 medium. In a separate tube, 20 l of LipofectAMINE (Life Technologies, Inc.) was diluted in 200 l of serum-free Ham's F-12 medium. The diluted DNA and the LipofectAMINE were then gently mixed and incubated at 25°C for 30 min. After the incubation, 6 ml of serum-free Ham's F-12 medium was added to the DNA/LipofectAMINE mixture, mixed, then placed onto cells that had been rinsed with serum-free Ham's F-12 medium. The cells were incubated for 5 h at 37°C. Without washing, 3.6 ml of Ham's F-12 medium containing 20% serum was added. The cells were grown for 24 h. The medium was removed, and Ham's F-12 medium containing 10% serum and 1.5 mg/ml Geneticin was added. Selected cells were grown under constant selection in medium containing 300 g/ml Geneticin.
Electrophoresis and Immunoblots-Samples were concentrated by trichloroacetic acid precipitation and washed in acetone. Pellets were suspended in Buffer F containing 1.2% ␤-mercaptoethanol and heated at 95°C for 3 min before being loaded onto gels. Samples that were immunoblotted with anti-folate receptor IgG did not contain ␤-mercaptoethanol. Proteins were separated in a 12.5% SDS-polyacrylamide gel using the method of Laemmli (30). The separated proteins were then transferred to nylon. The nylon was blocked in Buffer D that contained 5% dry milk for 1 h at room temperature. Primary antibodies were diluted in Buffer D that contained 1% dry milk and incubated with the nylon samples for 1 h at room temperature. The nylon was washed four times, 10 min each in Buffer D ϩ 1% dry milk. The second antibody (goat anti-mouse IgG, goat anti-rabbit IgG, or rabbit anti-goat IgG all conjugated to horseradish peroxidase) was diluted 1/30,000 in Buffer D ϩ 1% dry milk and incubated with the nylon for 1 h at room temperature. The nylon was then washed and the bands visualized using enhanced chemiluminescence.
Other Methods-Each experiment was conducted at least three times. Although the absolute numbers varied between experiments, the results were the same. Representative experiments are shown. For the quantification of radiation in the OptiPrep gradients, 50 l from each fraction was measured. Each fractionation experiment was conducted at least four times with similar results.
Protein concentrations were determined by the Bradford method (32). A 100-l sample of each gradient fraction was mixed with 5 ml of diluted (1:4) Bradford reagent and the absorbance measured at 595 nm. OptiPrep interfered with other standard protein determination methods including the micro-Bradford assay.
For ultrastructural analysis, cells were rinsed in PBS, fixed with 2% glutaraldehyde for 30 min at room temperature, dehydrated, and embedded in Epon 812. Sections were stained with 4% uranyl acetate and lead citrate before viewing with a JEOL 100CX electron microscope. Quantitative analysis of progesterone-treated cells was carried out by direct quantification of randomly photographed EM negatives. Caveolae in L1210-JF cells were quantified directly in the electron microscope by counting the number of cells that had caveolae. Each "cell" corresponds to the portion of the cell captured within individual squares of the EM grid.

RESULTS
Previous measurements of caveolae cholesterol were made on detergent-resistant caveolae separated from soluble plasma membrane by sucrose gradient centrifugation (12). In the current study we used a new detergent-free method of purification that retains resident molecules removed by detergents (28). In addition we made measurements on both MA104 cells and normal human fibroblasts (Table I). Cells were incubated overnight in the presence of [ 3 H]acetate before the [ 3 H]cholesterol content of different fractions was measured. Most of the labeled cholesterol was in the plasma membrane (ϳ80% for fibroblasts and ϳ92% for MA104 cells). A substantial portion of this cholesterol was in the caveolae fraction (16% for fibroblasts and 21% for MA104 cells). When normalized for the amount of protein, caveolae were significantly enriched in cholesterol relative to the remainder of the plasma membrane (3.18-fold for fibroblasts and 4.16-fold for MA104 cells).
Transport of Cholesterol to Caveolae-Cholesterol synthesized at 14°C is largely retained in the ER (16). Shifting the temperature to 37°C allows this cholesterol to migrate to the cell surface. To see if caveolae in MA104 cells were involved in transport of newly synthesized cholesterol, we incubated cells in the presence of [ 3 H]acetate for 1 h at 14°C before adding excess unlabeled acetate and shifting the temperature to 37°C for various times. We measured the amount of radiolabeled cholesterol in the plasma membrane, non-caveolae, and caveolae fractions (Fig. 1A). The pattern of transport was similar in human fibroblasts (data not shown). A small amount of labeled cholesterol was in the plasma membrane (Ç) at the end of the pulse but all of this cholesterol was in the caveolae fraction (Ⅺ). During the first 10 min at 37°C, most of the labeled membrane cholesterol was in the caveolae fraction (compare Ç with Ⅺ). The level of label in the caveolae fraction peaked at 10 -20 min and then declined while the amount in the non-caveolae fraction steadily increased after 10 min (E). Membrane and noncaveolae membrane cholesterol reached a plateau as the amount in the caveolae fraction declined to zero. Therefore, newly synthesized cholesterol appears at the cell surface first in caveolae and then in non-caveolae membrane.
Another way of looking at the dynamics of caveolae cholesterol is to uniformly label the plasma membrane cholesterol pool before removing the label and measuring the amount of labeled cholesterol in the various fractions during a chase period (Fig. 1B). MA104 cells were incubated for 24 h in the presence of [ 3 H]acetate at 37°C, washed, and then chased for the indicated times. Initially we found ϳ400 ϫ 10 5 dpm of [ 3 H]cholesterol in the caveolae fraction (Ⅺ). During the chase period, the amount of label declined to zero and a corresponding amount disappeared from the plasma membrane (Ç). The labeled cholesterol pool in non-caveolae membrane remained unchanged. The loss of [ 3 H]cholesterol from the caveolae fraction, either during a pulse-chase (A) or a chase (B), suggests cholesterol flows from the ER through caveolae on its way to the surrounding membrane.
Progesterone interrupts two-way traffic of cholesterol between the plasma membrane and internal membranes (17). It also inhibits cholesterol synthesis. If cholesterol movement to the plasma membrane involves caveolae, then progesterone should block the appearance of radiolabeled cholesterol in this fraction. We labeled the cholesterol pool by culturing MA104 cells for 24 h in the presence of [ 3 H]acetate ( Fig. 2A). Progesterone was added to the dish without removing the [ 3 H]acetate and the cells further incubated for various times before preparing cell fractions. The presence of progesterone caused a rapid decline in the level of [ 3 H]cholesterol in the caveolae fraction (Ⅺ). The plasma membrane fraction (Ç) also lost [ 3 H]cholesterol during the incubation, while the non-caveolae membrane (E) did not change. The effect of progesterone further indicates cholesterol first appears in caveolae after leaving the ER.
Cholesterol returned to caveolae after progesterone was removed (Fig. 2B). Uniformly labeled MA104 cells were incubated in the presence of progesterone for 1 h. The progesterone was removed, and either caveolae fractions were prepared immediately (0 min) or the cells were incubated an additional 1 h (60 min) in the presence (striped bar) or absence (solid bar) of 100 M compactin before caveolae were isolated. The amount of [ 3 H]cholesterol in the caveolae fraction (solid bar) rose from 0 to 311 ϫ 10 5 dpm within 60 min after progesterone was removed. The presence of compactin completely blocked the appearance of radiolabeled cholesterol (60 min, striped bar), indicating a requirement for new cholesterol biosynthesis. This suggests that [ 3 H]cholesterol does not accumulate in internal membranes during the exposure to progesterone, nor does noncaveolae [ 3 H]cholesterol from contiguous plasma membrane migrate back into caveolae when transport to the cell surface is blocked.
Progesterone Inhibits Caveolae Internalization-Previously we found that internalization of folate by caveolae is reduced in cells starved of cholesterol (7). This treatment, however, has too many side effects to ever be a useful experimental tool for studying caveolae function. Progesterone should have a similar effect to cholesterol depletion with the advantage of being rapid and reversible (Fig. 3). MA104 cells were incubated in the presents of different concentrations of progesterone for 1 h at 37°C before [ 3 H]folic acid was added to the dish and the cells further incubated for 1 h (A). At the end of the incubation, the amount of internal (Ⅺ) and external (E) bound [ 3 H]folic acid was measured (24). In the absence of any progesterone, the cells had equal amounts of internal and external bound [ 3 H]folic acid, indicating normal caveolae function. As little as 10 g/ml progesterone caused a dramatic relocation of the internal receptor pool (Ⅺ) to the external receptor pool (E). Higher concentrations of progesterone had similar effects. Progesterone, therefore, appears to prevent folate receptor internalization.
Inhibition by progesterone occurred coordinately with the loss of cholesterol from caveolae (Fig. 3B). Cells were incubated in the presence of [ 3 H]folic acid at 37°C for 1 h to label both internal and external receptors (B). Progesterone (10 g/ml) was then added to the dish and the cells further incubated for the indicated time. Within 20 min, most of the internal receptors (Ⅺ) became exposed at the cell surface (E), which matches closely the time it takes for cholesterol to leave the caveolae fraction (Fig. 1B). In other experiments (data not shown), we found that it took 60 -90 min for the receptor ratio to returned to normal after progesterone was removed.
Lowering the cholesterol content of caveolae can have two effects: (a) prevention of caveolae internalization along with reducing the number of invaginated caveolae (7) and (b) unclustering of folate receptors (6,29). We used fibroblasts to determine if progesterone changed the number of invaginated caveolae because these cells respond to progesterone exactly the same as MA104 cells (data not shown). Cells were prepared for electron microscopy after they had been incubated in the presence or absence of progesterone for 1 h (Table II). Progesterone caused a 32-36% decline in the number of invaginated caveolae. We then used the caveolae isolation procedure to assess how progesterone affected the surface distribution of the folate receptor (Fig. 4). Plasma membranes were isolated from untreated cells, sonicated, and separated on the first OptiPrep gradient (Control). Immunoblots of total protein in each fraction showed most of the folate receptor to be in the light membrane fractions (1-7) used to purify caveolae on the second OptiPrep gradient (see "Experimental Procedures"). Little receptor was present in the heavier membrane fractions (8 -14) containing the bulk of the membrane protein. By contrast, a large portion of the receptor was shifted to the bulk membrane fractions in progesterone-treated cells. This suggests the receptor became dispersed in the plane of the membrane (29). The distribution of the folate receptor returned to normal after the progesterone was removed and the cells incubated for 1 h (Wash-out). Therefore, the acute loss of cholesterol from caveolae has the same effect as cholesterol starvation. Progesterone may be a useful tool for studying those caveolae functions that depend on cholesterol.
A Role for Caveolin in Cholesterol Transport-In human fibroblasts caveolin can move directly between the plasma membrane and the ER (12, 13). Caveolin also binds both cholesterol (19) and fatty acids (21). 2 This raises the possibility that caveolin might shuttle newly made cholesterol to caveolae. FIG. 4. Progesterone unclusters the folate receptor. MA104 cells were incubated in the presence (Progesterone and Wash-out) or absence (Control) of progesterone plus 0.5 mg/ml fatty acid-free BSA for 1 h at 37°C. One set of cells treated with progesterone was incubated an additional hour at 37°C in the absence of progesterone (Wash-out). Cells were processed to isolate caveolae up to the first OptiPrep gradient step. Total amount of protein in each fraction from this gradient was separated on polyacrylamide and immunoblotted with anti-folate receptor IgG. For the [ 3 H]folic uptake experiment, each point is the average of three separate measurements where the standard deviation was less than 1%.

FIG. 5. Progesterone causes redistribution of caveolin to an internal membrane compartment.
A, human fibroblasts were incubated in the presence (Progesterone and Wash-out) or absence (Control) of progesterone plus 0.5 mg/ml fatty acid-free BSA for 1 h at 37°C. One set of cells treated with the progesterone was incubated an additional hour at 37°C in the absence of progesterone (Wash-out). Cells were processed to isolate caveolae up to the first OptiPrep gradient step. Total amount of protein in each fraction from this gradient was separated on polyacrylamide and immunoblotted with an anti-caveolin IgG as described. B, human fibroblasts preincubated in the presence of 50 g/ml cycloheximide for 3 h were incubated in the presence (Progesterone, Wash-out, ProgesteroneϩT, Wash-outϩT, ProgesteroneϩTϩSDS) or absence (Control, ControlϩT) of 10 g/ml progesterone plus 0.5 mg/ml fatty acid-free BSA for 1 h. One set of progesterone-treated cells was subsequently washed and incubated an additional 1 h at 37°C in the absence of progesterone (Wash-out, Wash-outϩT). Cells were fractionated into caveolae membrane (CM), non-caveolae membrane (NCM), plasma membrane (PM), internal membranes (IM), cytosol (Cytosol) and postnuclear supernatant fraction (PNS). Equal amounts of protein from each fraction were either immunoprecipitated directly (Control, Progesterone, Wash-out) with anti-caveolin IgG or first incubated in the presence of 300 g/ml trypsin (ControlϩT, ProgesteroneϩT, Wash-outϩT, and ProgesteroneϩTϩSDS) for 30 min on ice before immunoprecipitation. Each immunoprecipitate was separated on polyacrylamide gels and immunoblotted with anti-caveolin IgG. We first used immunoblotting to determine the location of caveolin in progesterone-treated cells (Fig. 5A). Fibroblasts were either not treated (Control) or incubated in the presence of progesterone for 1 h at 37°C (Progesterone and Wash-out).
One set was analyzed immediately (Progesterone), while the other was incubated further in the absence of progesterone (Wash-out). The plasma membrane from each set was isolated and fractionated on OptiPrep 1 gradients in order to evaluate the distribution of caveolin in the whole membrane (28). In control cells, most of the caveolin was in the top seven fractions, which is the normal distribution of the protein (28). There was markedly less caveolin in the same fractions of progesteronetreated cells, although all of the protein that remained was in the caveolae fractions. The level of caveolin returned to normal once the progesterone was removed from the medium. Progesterone, therefore, causes a reversible loss of caveolin from the plasma membrane. We used an immunoprecipitation assay to see if progesterone caused caveolin to accumulate in the ER (Fig. 5B). All treatments were carried out on cells cultured in the presence of protein synthesis inhibitors. Most of the caveolin was in the caveolae fraction (CM) of control human fibroblasts (Control) and very little in the internal membrane fractions containing ER and Golgi apparatus (IM). There was a dramatic reduction in the amount of caveolin in caveolae (CM) after progesterone treatment (Progesterone) and a corresponding increase in the amount precipitated from internal membranes (IM). The distribution of caveolin returned to normal after progesterone was removed (Wash-out). The caveolin in the internal membrane fraction of progesterone-treated cells (ProgesteroneϩT) was re-sistant to trypsin treatment, while the caveolin at the cell surface (ControlϩT and Wash-outϩT) was completely digested by the protease. If the internal membrane fraction from progesterone-treated cells was permeabilized with ionic detergents, however, the caveolin was degraded by the trypsin (progesteroneϩTϩSDS). Therefore, progesterone causes caveolin to reversibly accumulate in ER and Golgi membranes. This suggests that caveolin movement is linked to the flow of cholesterol between ER and caveolae.
Caveolin might function as a shuttle protein that moves cholesterol from the ER to the cell surface. In this case, the expression of caveolin should effect the rate and direction of cholesterol transport to the cell surface. We measured the cholesterol level of caveolae fractions prepared from lymphocytes expressing or not expressing caveolin. A lymphocyte variant of L1210 cells that expresses the folate receptor (designated L1210-JF cells; Ref. 26) was transfected either with vector alone or a cDNA for caveolin (Fig. 6). Permanent transformants were selected and maintained as stable cell lines. Anti-caveolin immunoblots of the whole cell lysate from parent cells (L1210-JF) and mock-transfected cells (mock) did not detect any caveolin. By contrast, cells transfected with the cDNA to caveolin (caveolin) had a strong reactive band, comparable in intensity to the band obtained from MA104 cell lysates (MA104). Caveolae isolated from either parent cells or mock-transfected cells were not enriched in [ 3 H]cholesterol compared to non-caveolae membrane (Table III)  The rate of new cholesterol transport to the cell surface was significantly faster in cells expressing caveolin (Fig. 7). We used a 1-h pulse of [ 3 H]acetate at 14°C, followed by a chase at 37°C in the absence of label to measure the rate of [ 3 H]cholesterol appearance at the cell surface. The initial rate of transport to plasma membrane (Ç) in both parental cells (A) and mock-transfected cells (B) was ϳ 9 ϫ 10 5 dpm/min. Cells expressing caveolin, on the other hand, had initial transport rates  of ϳ33 ϫ 10 5 dpm/min. Moreover, only the caveolae fraction (Ⅺ) from transfected cells acquired significant cholesterol during the chase. In addition to restoring the cholesterol level of caveolae to near normal, we found that expression of caveolin had other effects. Although the number of invaginated caveolae per cell was low compared to fibroblasts and MA104 cells (Table IV), 36% of caveolin-expressing L1210-JF cells had typical omegashaped membrane structures, while non-expressing cells had none. Caveolin-expressing cells also had a higher density of folate receptors in the caveolae fractions (Fig. 8). Less than 5% of the receptors were in the caveolae fraction of parent (control) and mock-transfected (mock) cells compared to 35% in cells expressing caveolin (caveolin). Expression of caveolin, therefore, appears to have a direct effect on caveolae structure and organization.

DISCUSSION
Cholesterol Flows through Caveolae-Previous studies have found that cholesterol reaches the cell surface within 10 -20 min after synthesis in the ER (14,16) and appears to move directly to the plasma membrane without passing through the Golgi apparatus (16). We have now determined that caveolae are the initial site on the cell surface where this new cholesterol appears. A population of cholesterol-rich membranes with the properties of a membrane intermediate carrying cholesterol from the ER have been detected in several studies (16,33). This membrane, like caveolae membrane, has a light buoyant density on sucrose gradients. Therefore, new cholesterol may reach the plasma membrane in vesicles targeted to caveolae or through contact sites between cholesterol-rich regions of ER membrane and either caveolae or caveolae-derived vesicles (e.g. plasmalemmal vesicles; Ref. 34).
Several studies have implicated caveolin in intracellular cholesterol traffic (12,19,22). Two observations in the current study suggest caveolin functions in rapid cholesterol transport to the cell surface. First, progesterone lowered the cholesterol content of caveolae and at the same time caused caveolin to accumulate in internal membranes. The internal caveolin was protease-resistant, indicating it had moved to the lumenal side of the ER-Golgi membrane, as it does when plasma membrane cholesterol is oxidized by cholesterol oxidase (12). New cholesterol synthesis was required for the cholesterol levels in caveolae to return to normal after the removal of progesterone, so cholesterol did not accumulate in the presence of the steroid. This is curious because progesterone blocks cholesterol synthesis by preventing the conversion of cholesterol precursors (35,36), as well as several oxidized sterols (35), into cholesterol while stimulating 3-hydroxy-3-methylglutaryl-coenzyme A reductase activity (35). Most likely cholesterol was synthesized from these intermediates (35) after progesterone was removed, yet none of it appeared in caveolae. This suggests progesterone directly causes the mislocation of both cholesterol intermediates and cholesterol in the cell. On the other hand, we do not know if inhibiting cholesterol synthesis with compactin might also cause caveolin accumulation in internal membranes and lower caveolae cholesterol. A second and more direct indication of caveolin involvement in cholesterol transport is that lymphocytes expressing caveolin have a 4.4-fold higher concentration of cholesterol in the caveolae fraction than control cells. Furthermore, they transport sterol to the plasma membrane ϳ 4 times more rapidly.
One of the remarkable findings in this study was that newly synthesized cholesterol appears to flow through caveolae membrane. When we allowed the labeled cholesterol pool to accumulate in the ER at 14°C, for example, a bolus of [ 3 H]cholesterol first appeared in the caveolae fraction after the cells were shifted to 37°C and then immediately migrated into the surrounding membrane. Likewise, the level of [ 3 H]cholesterol in caveolae rapidly declined when the supply of [ 3 H]acetate was interrupted or when the flow of cholesterol into the caveolae was blocked by progesterone. There even appears to be a mechanism for exporting cholesterol from caveolae to the surrounding membrane in the absence of new cholesterol synthesis. We also did not detect any movement of cholesterol back into caveolae from the surrounding membrane. Vectorial movement of cholesterol through caveolae may contribute to the overall organization and function of this membrane domain.
The cholesterol oxidase-induced movement of caveolin to the Golgi apparatus provided the first clue that molecular traffic occurs directly between caveolae and ER (12). We have now detected molecular movement in the opposite direction, and it appears to involve caveolin. Two-way transfer of molecules between the plasma membrane and the ER may be an important route of molecular exchange between the cell and its environment. The endoplasmic reticulum in most cells is orga- nized into a network that extends beneath the entire plasma membrane (37), so the travel distance between caveolae and ER compartments is quite short. The ER houses a number of enzymes involved in various aspects of lipid metabolism and storage and is the major site of membrane bilayer synthesis. It makes sense, therefore, that there would be a mechanism for direct delivery of lipid intermediates to the ER. From this perspective, lipids such as fatty acids and cholesterol would be among the molecules expected to be internalized by caveolae. Caveolin may be one member of a class of caveolae molecules that facilitate uptake and excretion of lipids, as well as other hydrophobic molecules, by the cell.
Cholesterol and Caveolae Function-The caveolae membrane fraction in lymphocytes not expressing caveolin behaved exactly the same during cell fractionation as caveolae that contain caveolin. Although we have not carried out a detailed molecular analysis, we know this fraction was low in both cholesterol and the GPI-anchored folate receptor compared to caveolae fractions from cells expressing caveolin. Nevertheless, the membrane must contain sorting information for caveolin because this was the exclusive location for the molecule in cells transfected with caveolin cDNA. An important next step will be to determine what effect caveolin and cholesterol have on the presence in the caveolae fraction of signal-transducing molecules such as tyrosine kinase receptors (38,39), heterotrimeric GTP-binding proteins (1,40) and non-receptor tyrosine kinases (1,38,41) and whether or not these molecules function properly in signal transduction. Non-receptor tyrosine kinases have been found in Triton X-100-insoluble membrane fractions from lymphocytes lacking detectable caveolin, and kinase activity is stimulated in these cells by antibodies against GPI-anchored proteins (42). Still, caveolin as well as cholesterol may have important modulatory influences on signal integration in caveolae.
We confirmed previous observations (43) that cells transfected with the caveolin cDNA tend to have invaginated caveolae similar in overall morphology to those found in fibroblasts and endothelial cells. We also found that the caveolae fraction from these cells had considerably more folate receptor than control cells. This is the third set of experimental evidence (6,7,29) that cholesterol is important for organizing GPI-anchored proteins in caveolae. Nevertheless, we were unable to detect folate internalization in these cells (data not shown). Either cholesterol is necessary but not sufficient for caveolae internalization or the cells did not express high enough levels of caveolin to support receptor internalization. These cells express many more receptors than an MA104 cell, the standard cell for these assays, so at best, internalization would be expected to be inefficient.
Progesterone is a new reagent for inhibiting potocytosis. The mechanism of action is quite different from other inhibitors that have been identified (20,27). Progesterone appears to inhibit by reducing the cholesterol level of caveolae. In part this causes the dispersal of GPI-anchored proteins in the plane of the membrane, but it also reduces the number of invaginated caveolae. If the number of invaginated caveolae is a measure of how many caveolae are cycling (27), then both folate receptor unclustering and loss of internalization contribute to the inhibition of folate uptake. Progesterone could be a useful tool for assessing caveolae function in other cell types.