Cholesterol Substitution Increases the Structural Heterogeneity of Caveolae*

Caveolin-1 binds cholesterol and caveola formation involves caveolin-1 oligomerization and cholesterol association. The role of cholesterol in caveolae has so far been addressed by methods that compromise membrane integrity and abolish caveolar invaginations. To study the importance of sterol specificity for the structure and function of caveolae, we replaced cholesterol in mammalian cells with its immediate precursor desmosterol by inhibiting 24-dehydrocholesterol reductase. Desmosterol could substitute for cholesterol in maintaining cell growth, membrane integrity, and preserving caveolar invaginations. However, in desmosterol cells the affinity of caveolin-1 for sterol and the stability of caveolin oligomers were decreased. Moreover, caveolar invaginations became more heterogeneous in dimensions and in the number of caveolin-1 molecules per caveola. Despite the altered caveolar structure, caveolar ligand uptake was only moderately inhibited. We found that in desmosterol cells, Src kinase phosphorylated Cav1 at Tyr14 more avidly than in cholesterol cells. Taken the role of Cav1 Tyr14 phosphorylation in caveolar endocytosis, this may help to preserve caveolar uptake in desmosterol cells. We conclude that a sterol C24 double bond interferes with caveolin-sterol interaction and perturbs caveolar morphology but facilitates Cav1 Src phosphorylation and allows caveolar endocytosis. More generally, substitution of cholesterol by a structurally closely related sterol provides a method to selectively modify membrane protein-sterol affinity, structure and function of cholesterol-dependent domains without compromising membrane integrity.

Immunofluorescence Microscopy and Western Blotting-IF stainings and filipin permeabilization were performed as in Ref. 17. When used, saponin (0.05%) (Sigma) was present in all steps of IF stainings. Images of Cav1 and filipin stainings were acquired with an inverted Olympus IX-71 microscope equipped with a ϫ60 water objective (N.A. 1.2), and deconvolved using the blind-deconvolution wizard of Huygens Essential before processing figures in ImagePro Plus 5.1. For anti-Cav1 Western blotting, cells or trichloroacetic acid-precipitated gradient fractions were lysed in Laemmli sample buffer and boiled unless noted otherwise. For phospho-Tyr immunoblotting, cells were scraped in boiling 1% SDS, 10 mM Tris-HCl, pH 7.4, 1 mM Na 3 VO 4 , including a protease inhibitor mixture (chymostatin, leupeptin, antipain, and pepstatin A; 25 g/ml each; CLAP, Sigma).
Subcellular Membrane Fractionations-The sucrose density gradient fractionation was preformed according to Ref. 13. For Triton-Optiprep TM fractionation, the cells were collected by centrifugation at 1000 ϫ g for 5 min and incubated for 10 min on ice in 240 l of lysis buffer (25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, 1 mM dithiothreitol, 10% sucrose, and 1% Triton X-100, CLAP). The lysate was then mixed with 480 l of 60% Optiprep TM (Nycomed-Pharma, Oslo, Norway) and overlaid with step gradients of 35,30,25,20, and 0% Optiprep TM in lysis buffer. After centrifugation for 4 h at 40,000 ϫ g at 4°C with a SW60Ti rotor, fractions were collected and used for protein or lipid determinations. The detergent-resistant membranes (DRM) fractions of ts-v-S-MDCK cells were isolated as above except that Na 3 VO 4 was present in all buffers and cells were homogenized by passing for 10 times through 18-gauge needle before mixing with 60% Optiprep TM .
Immunoisolation of Caveolar Membranes-Adipocytes were prepared from epididymal fat of male Wistar rats according to Ref. 28 (20 nmol, 0.05 Ci/mmol), after which the cellassociated dissociations per minute (dpm) were similar for both labels. Caveolar membranes were immunoisolated essentially as described in Refs. 29 and 30. Briefly, cells resuspended in 0.25 M sucrose, 10 mM Tris, pH 7.5, 1 mM EDTA were broken by 10 strokes with a glass Teflon Dounce and the material centrifuged for 15 min at 16,000 ϫ g. The pellet was resuspended in 2 ml of 500 mM Na 2 CO 3 , pH 11, including CLAP, and sonicated three times for 20 s at full power using a MSE Soniprep 150 sonicator. The homogenate was transferred to a SW40 tube and adjusted to 45% sucrose by the addition of 2 ml of 90% sucrose in 20 mM Tris, pH 7.5, 2 mM EDTA and CLAP. The sample was overlaid with 4 ml of 35 and 5% sucrose in the same buffer. After centrifugation (39,000 ϫ g for 24 h at 4°C with an SW40 rotor) the caveolae-enriched membrane fraction at the 5-35% sucrose interphase (fraction 3) was collected. This fraction was divided into two and incubated overnight with or without Cav1 Ab (clone 2234), followed by a 1-h incubation with Dynal beads (Dynal Biotech) linked to anti-mouse IgG and washes according to the manufacturer's instructions. Control pull-down without Cav1 antibodies contained ϳ10% of [ 3  Intensity Analysis of Cav1-GFP Structures-Total internal reflection fluorescence microscopy (TIRFM) images were acquired using an Olympus IX-71 microscope with a PLAPON TIRFM ϫ60/1.45 objective and deconvolved using the blinddeconvolution wizard of Huygens Essential. Cells were kept at 37°C in CO 2 -independent medium (Invitrogen). Cav1-GFP containing structures were automatically detected by defining potential dot regions (31) and then finding pixels with a higher intensity than their 8-pixel neighborhood. Each dot was assumed to approximate a three-dimensional normal distribution, with pixel intensity considered as the third dimension. Total dot intensity was defined by the volume under this multivariate normal distribution. If two or more dots overlapped, they were treated as a mixture distribution of multivariate normal distributions. Each dot had six parameters (location x, location y, covariance matrix with three parameters and volume) that were estimated by a genetic algorithm (32). Dot intensities per cell were plotted in a histogram. Peaks in the intensity histogram were considered to represent different pools of caveolar structures (33). The histograms were cropped, leaving only the first three peaks, and a mixture distribution of three univariate Gaussian distributions was fitted with the algorithm (32). The distance between the first and the third peak in each histogram was normalized by multiplication to convert the data from different cells and experiments to the same scale.
Quantification Cav1-GFP Mobility-Cav1-GFP mobility was analyzed using the same principle as in Ref. 34. HeLa Cav1-GFP cells were imaged by TIRFM using 67 ϫ 67-nm sampling and an exposure time of 500 ms. Per condition at least 20 separate 1-min movies were acquired containing 4 frames per minute. Images were deconvolved as described above and Pearson correlation was calculated for succeeding frames (15-s interval).
Endocytosis Assays-The internalization of BODIPY-lactosylceramide (LacCer, D-erythro isomer) was analyzed essentially as in Ref. 3. Briefly, fibroblasts were incubated with 5 M BODIPY-FL5-LacCer⅐BSA complex in Earle's minimal essential medium for 30 min at 0°C and, after imaging for labeling efficiency, incubated for 5 min at 37°C. Uninternalized label was removed from the PM by washing the cells 6 ϫ 10 min with 5% BSA in Earle's minimal essential medium at 0°C. Images were acquired using an inverted Olympus IX-71 microscope equipped with a ϫ60 water objective (N.A. 1.2), 1 pixel ϭ 232 ϫ 232 nm. Fluorescence intensity per cellular area was determined after background subtraction using a rolling ball radius of 5 pixels (Image J). Samples with a similar labeling efficiency were used for quantification. To analyze Tfn or dextran uptake, HeLa cells were incubated with 50 g/ml of AF568-Tfn (Molecular Probes) for 30 min at 4°C followed by 5 min at 37°C, or with 1 mg/ml of AF488-dextran (10,000 M r ; Molecular Probes) for 20 min at 37°C, prior to acid-stripping as described (35). Control samples treated similarly but without 37°C internalization were included for both dextran and Tfn uptake assays. Imaging and fluorescence intensity analysis were performed as described for BODIPY-LacCer. For integrin cross-linking, HeLa Cav1-GFP cells were incubated with primary and secondary Abs as described (4). The colocalization between ␣2␤1 inte-grin and Cav1-GFP was analyzed from single section confocal images (Leica TCS SP2; HCX PL APO CS ϫ60 1.4 oil objective). Images were processed by using a rolling ball background subtraction (radius 8 pixels) and a 2 ϫ 2 median filter (Image J). After intensity thresholding, Pearson correlation was determined (Imaris Bitplane).
Electron Microscopy-For ultrastructural analysis and quantification, cells grown on glass coverslips were glutaraldehydefixed, osmicated, en block stained with uranyl acetate, and epon embedded as described in Ref. 36. Pre-embedding immunolabeling of fixed and saponin-permeabilized cells using anti-Cav1 primary antibody (C13630, BD Transduction Laboratories) was as described in Ref. 36. Sections were cut parallel or vertical to the coverslip, post-stained with uranyl acetate and lead citrate, and examined using a Tecnai 12 transmission electron microscope (FEI Corp.) operating at 80 kV. Images for quantification were acquired with a ES500W CCD camera (Gatan Corp.).
Statistical Analysis-Data are reported as mean Ϯ S.E. or S.D. Statistical analysis was performed using Microsoft Excel "t Test: two-sample assuming unequal variances" (or "F-test two sample for variances" where indicated). Results were considered significant if p Ͻ 0.05.

RESULTS
Exchange of Cholesterol for Desmosterol-Cholesterol was exchanged for desmosterol by cultivating cells in the presence of LPDS and 10 nM 20,25-diazacholesterol (20,25-DAC), an inhibitor of DHCR24 (37). This treatment did not result in significant accumulation of other cholesterol precursors as detected by Ag ϩ -HPLC (Fig. 1A, and not shown). The cells attained a free sterol composition of ϳ90% desmosterol and ϳ10% cholesterol in 1 week (Fig. 1, A and B; HeLa cells) to 2 weeks (primary human fibroblasts, not shown). Here we refer to such cells as desmosterol cells and cells grown similarly but in the absence of the drug as cholesterol cells. Cells grown with 20,25-DAC and FBS did not accumulate desmosterol, indicating that such cells rely on exogenously supplied cholesterol (Fig. 1B).
The sterol exchange did not result in any noticeable effects on cell growth or viability. Also, we found no significant changes in total sterol or phospholipid composition (Fig. 1C) and no gross alterations in the chain length or unsaturation of major phospholipid species (Fig. 1D). To analyze PM fluidity, we measured TMA-DPH fluorescence anisotropy in HeLa cells as in Ref. 27. Desmosterol and cholesterol cells gave identical anisotropy values (0.32 Ϯ 0.01 versus 0.32 Ϯ 0.01, S.D., n ϭ 3), suggesting that the overall PM fluidity was not altered. In comparison, inhibition of sterol biosynthesis (by 10 M lovastatin, 100 M mevalonate in LPDS for 3 days) resulted in increased membrane fluidity, as indicated by lower anisotropy (0.29 Ϯ 0.01).
Sterol distribution as detected by filipin staining and anti-Cav1 Ab staining patterns were similar between desmosterol and cholesterol cells ( Fig. 2A), as was the distribution of GFPtagged Cav1 (Cav1-GFP) in stably overexpressing HeLa cells (Figs. 4A and 7A). To biochemically assess sterol and Cav1 compartmentalization, membranes were resolved by equilibrium density gradient centrifugation (Fig. 2B). This confirmed a sim-ilar distribution of sterols and Cav1 in desmosterol and cholesterol cells. Moreover, the total amount of Cav1 remained similar (in HeLa cells, Fig.  2C, and in fibroblasts, not shown). Together, these results indicate that exchanging cholesterol for desmosterol did not impair cell viability, general membrane integrity, or composition of other major membrane lipids, nor the gross compartmentalization of sterols or Cav1.
Sterol C24 Double Bond Reduces Affinity for Caveolae-To analyze the association of cholesterol and desmosterol with caveolar membranes, primary rat adipocytes were labeled for 2 days with both [ 3 H]desmosterol and [ 14 C]cholesterol in the presence of 10 nM 20,25-DAC (to prevent the conversion of [ 3 H]desmosterol). Cells were then disrupted by sonication and the caveolae-enriched membrane fraction isolated without detergents as described (Fig. 3A) (29,30). Caveolar membranes were further immunoisolated using Cav1 Abs as described in Ref. 30. We found that both [ 3 H]desmosterol and [ 14 C]cholesterol specifically precipitated with Cav1 (Fig. 3B). However, the amount of co-isolated [ 3 H]desmosterol was 35% lower than that of [ 14 C]cholesterol (Fig. 3B), indicating a lower affinity of desmosterol than cholesterol for caveolar membranes.
Contrary to cholesterol cells, Cav1 was largely excluded from the DRM fraction in desmosterol cells (Fig. 3B HeLa cells; similar results were obtained from primary fibroblasts, not shown). The decrease in Cav1 detergent resistance was fully restored when cells were grown in the presence of the drug and FBS (data not shown), indicating that 20,25-DAC did not act unrelated to the sterol exchange. The fraction of desmosterol partitioning into DRMs was also lower than that of cholesterol (23 Ϯ 3 versus 37 Ϯ 1%; mean Ϯ S.D., n ϭ 3), in accordance with earlier results (13). Cav1 has been described to form SDS-resistant oligomers (38). When the stability of Cav1 oligomers was analyzed by Western blotting, we found that 18 Ϯ 8% of Cav1 immunoreactivity remained as heat-resistant oligomers in desmosterol cells as compared with 43 Ϯ 6% in cholesterol cells ( Fig. 3D; primary fibroblasts, a similar tendency was observed in HeLa cells, not shown). Thus, the detergent resistance of Cav1 and the stability of Cav1 oligomers were decreased in desmosterol cells. These differences could result from a relative increase in the Golgi pool of Cav1 (8). However, as no apparent Golgi accumulation of Cav1 was observed in HeLa Cav1-GFP cells (Fig. 7A), it seems likely that the decreased detergent resistance and oligomer stability of Cav1 result from the lower affinity of desmosterol for caveolae.   MAY 23, 2008 • VOLUME 283 • NUMBER 21

JOURNAL OF BIOLOGICAL CHEMISTRY 14613
Sterol Exchange Alters Caveolar Structure-When HeLa Cav1-GFP cells were analyzed by TIRFM (Fig. 4A), we found the number of caveolar structures to be closely similar in desmosterol and cholesterol cells, as identified by GFP fluorescent dots (77 Ϯ 2 and 72 Ϯ 2 dots per 100 m 2 , respectively; mean Ϯ S.E. n ϭ 14 cells). Using this cell line, PM caveolae were shown to be composed of unit size building blocks containing ϳ144 Cav1 molecules (33): structures composed of 1 unit were characterized as individual caveolae and structures of multiple units as grape-like caveolar clusters. To analyze the effect of sterol exchange on Cav1 composition of caveolae, we developed an automated method for the identification and intensity quantification of Cav1-GFP structures in the TIRF plane (Fig. 4B). Using this method, we found that the distribution of caveolar structures to quantal units was less evident in desmosterol cells (Fig. 4C). Although separate distributions of fluorescence intensity were observed, their internal variation was larger, as indicated by increased peak width. The average width of the first peak, representing individual caveolae, was ϳ30% larger in desmosterol cells (Fig. 4C). This can be explained by a higher variation in the number of Cav1-GFP molecules per individual caveola. By immunoelectron microscopy, Cav1 Abs decorated uncoated PM invaginations. However, both in fibroblasts (Fig.  5A) and HeLa Cav1-GFP cells (not shown), the sterol exchange altered caveolar morphology. In desmosterol cells, uncoated PM invaginations with small openings (20 -45 nm) were less   abundant (Fig. 5A), and there was an increase in the average width and depth of the invaginations ( Table 1). The most striking difference in desmosterol cells was an increased variance in all the measured dimensions (Table 1). This difference was not observed upon culturing the cells in the presence of FBS and 20,25-DAC, indicating that the drug was not directly responsible for the altered morphology. These results imply that the additional double bond in desmosterol results in a higher variation in the morphology and Cav1 composition of individual caveolae and suggest that cholesterol controls caveolar uniformity.
Sterol Exchange Impairs Caveolar Uptake but Facilitates Cav1 Src Phosphorylation-When the mobility of Cav1-GFP structures in HeLa cells was analyzed by TIRFM using a similar method as in Ref. 34, no difference in either basal or vanadatestimulated caveolar mobility was observed between desmosterol and cholesterol cells (Fig. 6A). Also the intracellular mobility of Cav1-GFP vesicles was similar as assessed by fluorescence recovery after photobleaching. 4 To study the effect of sterol exchange on caveolar endocytosis, we analyzed the internalization of BODIPY-LacCer (3). Fibroblasts were labeled with BODIPY-LacCer at 0°C for 30 min and the label internalized for 5 min at 37°C. Uninternalized label was removed from the PM by BSA back-exchange and the remaining intensity quantified. We found that desmosterol cells showed a moderate but significant reduction in BODIPY-LacCer uptake compared with cholesterol cells (Fig. 6B). In comparison, no inhibition in the internalization of the fluid-phase tracer dextran or Tfn, a marker of clathrin-mediated endocytosis, was observed in desmosterol cells (Fig. 6C). Together, these data suggest that desmosterol replacement did not perturb the steady-state or vanadate-stimulated mobility of caveolae but selectively affected caveolar ligand uptake.
When caveolar internalization of Ab-clustered integrins was analyzed as in Ref. 4 we found that in desmosterol cells, the caveolar localization of clustered integrins was slightly reduced, as judged by colocalization of the ␣2␤1 integrin with Cav1-GFP (Fig. 7A). Integrin cross-linking is known to induce Cav1 phosphorylation at Tyr 14 , which is required for caveolar endocytosis (39). However, we found that in desmosterol cells, Cav1 phosphorylation upon integrin cross-linking was not attenuated but was in fact more robust (Fig. 7B). To address the mechanism underlying the increased Cav1 phosphorylation in desmosterol cells, we studied the effect of sterol exchange on the activity of Src kinase using an inducible MDCK cell line overexpressing Src (18). We found that both the basal Cav1 Tyr 14 phosphorylation and in particular, the Src-induced increase in this phosphorylation were more pronounced in desmosterol cells (Fig.  8). In the same cells, the overall degree of Tyr phosphorylation in basal or Src-induced conditions was not significantly increased (Fig. 8). These results indicate that desmosterol replacement selectively facilitated Cav1 phosphorylation by Src and imply that the inhibition of caveolar uptake was not due to impaired Cav1 Tyr 14 phosphorylation.

DISCUSSION
In this study, we exchanged cellular cholesterol for desmosterol by inhibiting DHCR24, to study the effects of sterol replacement on Cav1 and caveolae. Because other post-lanosterol enzymes of cholesterol biosynthesis can operate on substrates containing the C24 double bond, the enzyme inhibition 4 V. Pietiäinen, unpublished observation.  lead to the accumulation of the penultimate precursor of cholesterol, desmosterol (40). After the treatment, the levels of total sterols and major phospholipids remained similar. We cannot exclude the possibility that minor changes in other lipids might contribute to the observed effects. Nevertheless, considering the extent of sterol replacement and the close association of Cav1 with cholesterol, this is unlikely to play a major role. The sterol exchange was found to occur reproducibly and without toxic effects in all the cell types studied, i.e. primary fibroblasts, HeLa, and MDCK cells. Primary human fibroblasts were used for electron microscopy and BODIPY-LacCer uptake assays because of their abundance of caveolae, HeLa Cav1-GFP cells to study dynamic caveolar processes in living cells, and MDCK cells with an inducible Src expression to address Cav1 modification by phosphorylation. First, we found that the sterol C24 double bond reduced the ability of the sterol to associate with Cav1. This was based on the preferential co-precipitation of cholesterol with Cav1 from cells into which differently radiolabeled cholesterol and desmosterol were simultaneously introduced. It should be noted, however, that this result does not allow us to make conclusions about the overall sterol content of caveolae in the two conditions. This would require large scale preparation of immunoisolated caveolae from 20,25-DAC-treated versus untreated animals, for example. The C24 double bond gives rise to additional stress in the sterol tail, favors tilting of the molecule in the bilayer (13), and may thereby sterically hinder interactions with Cav1. The reduced sterol affinity may, in turn, explain the reduced association of Cav1 with DRMs as well as the decreased stability of caveolin oligomers.
Second, we noted that caveolae in desmosterol cells had altered morphology. They were less uniform in dimensions than in cholesterol cells as measured by electron microscopy, and in particular the opening of the invaginations was increased. Moreover, by analyzing the fluorescence intensity of Cav1-GFP structures in HeLa cells we found that the distribution of fluorescence into quantal units was less evident in desmosterol cells. This suggests that the number of Cav1 molecules per individual caveola was more variable (33), agreeing well with our electron microscopic observation of invaginations with less uniform dimensions. Hindrance of sterol-caveolin interaction may partially account for the altered shape of caveolae in desmosterol cells. Given that Cav1 is directly involved in bending the membrane (41), the larger variation in Cav1 molecules per caveola may also contribute.
Third, we found caveolar ligand uptake to be inhibited by ϳ25%, as assessed by endocytosis of BODIPY-LacCer. This lipid analogue is predominantly taken up by a caveolar-related mechanism, as judged by 80 -90% inhibition of BODIPY-Lac-Cer endocytosis in Cav1-depleted cells (42). Simian virus 40 (SV40) enters host cells via caveolae (43) although this is not its exclusive entry route (44). We therefore also studied whether the sterol exchange affects SV40 infectivity, and found cholesterol and desmosterol cells to be similarly infected as judged by T-antigen immunostaining. 4 Moreover, integrins are recycled by a caveolin/lipid raft-dependent mechanism (45, 46) and we found reduced co-localization of ␣2␤1 integrin with Cav1, suggestive of impaired caveolar uptake in desmosterol cells. However, also in this case the difference to cholesterol cells was minor. Considering the marked changes in caveolin-