Cholesterol depletion induces PKA-mediated basolateral-to-apical transcytosis of the scavenger receptor class B type I in MDCK cells

Abstract

Cholesterol-based membrane microdomains, or lipid rafts, are believed to play important, yet poorly defined, roles in protein trafficking and signal transduction. In polarized epithelial cells, the current view is that rafts are involved in apical but not in basolateral protein transport from the trans-Golgi network (TGN). We report here that cholesterol is required in a post-TGN mechanism of basolateral regionalization. Permanently transfected Madin-Darby canine kidney cells segregated the caveolae/raft-associated high-density lipoprotein scavenger receptor class B type I (SR-BI) predominantly to the basolateral domain where it was constitutively internalized and recycled basolaterally. Acute cholesterol depletion did not significantly alter SR-BI internalization, implying a cholesterol depletion-insensitive endocytic process but instead induced its transcytosis through a protein kinase A (PKA)- and microtubule-dependent mechanism. Forskolin also elicited SR-BI transcytosis. The basolateral distribution of endogenous epidermal growth factor receptor remained unaffected. Strikingly, cholesterol depletion induced PKA activity without increasing the cAMP levels. Thus, our results are consistent with a scenario in which cholesterol-based rafts promote internalization and basolateral recycling of internalized SR-BI whereas a PKA pool sensitive to cholesterol depletion mediates SR-BI transcytosis. Regulated transcytosis of SR-BI may provide an additional mechanism to control cholesterol homeostasis. These results disclose relationships between cholesterol-based rafts and PKA activity operating in a post-TGN mechanism of regulated apical-to-basolateral cell surface protein distribution.

Polarized epithelial cells perform a variety of vectorial processes based on their characteristic cell surface division into apical and basolateral domains bearing distinct protein and lipid compositions (1-4). The asymmetric protein distribution is mainly generated and maintained by exocytic pathways emerging from the trans-Golgi network (TGN) and by recycling or transcytotic pathways originating from endocytic compartments (2). Specific sorting signals and a poorly understood decoding machinery direct protein trafficking. Regulatory elements include GTPases and kinases (2, 4, 5), some of them, such as protein kinase A (PKA) (6, 7), shared by signal-transducing systems and thus linking protein trafficking to cellular demands and external stimuli. An intriguing problem concerns the extent to which cholesterol-structured membrane microdomains play a role in the protein-sorting machinery (3, 8, 9).

Cholesterol interacting with sphingolipids forms liquid-ordered phase microdomains in the lateral plane of membranes, called lipid rafts, which include or exclude selected proteins and provide special lipid environments for protein activities (8, 10). Caveolae represent a specialized kind of cholesterol-based raft assembled with caveolin-1 (cav-1) into flask-shaped invaginations of the plasma membrane (8, 11). Caveolar and noncaveolar rafts together with their constituent proteins coisolate by isopycnic floatation as low-buoyant density membrane fractions (10) obtained after cold detergent extraction (8, 12) or detergent-free extraction and sonication (13-15), and both are disrupted by cholesterol depletion (16, 17), currently achieved by cyclodextrins (18). Cholesterol depletion experiments have suggested that raft integrity is needed for protein trafficking along certain steps of the exocytic (12, 19) and endocytic (20-23) pathways as well as for signal transduction (10, 17).

Apical sorting is promoted by information held by exoplasmic or cytosolic domains, or by peptidic or glycosylphosphatidylinositol membrane anchors (1) and can be entirely dependent on proteinaceous sorting features (1, 15, 24, 25) although, for yet unknown reasons, it frequently requires N- and/or O-glycosylation (1, 26), suggesting multiple decoding systems. In contrast, basolateral sorting signals are restricted to relatively short linear sequences located in the cytosolic domain of transmembrane proteins (27), and are decoded by cytosolic adaptor proteins (28, 29).

In Madin-Darby canine kidney (MDCK) cells, newly synthesized plasma membrane proteins are first directly segregated during their transit through the TGN into apical or basolateral exocytic pathways. Then, post-TGN additional sorting events occur to endocytic proteins that, after internalization, follow either recycling pathways to the same plasma membrane domain or transcytotic pathways to the opposite domain (2, 4). Requirements for lipid rafts have been restricted to apical sorting at the TGN (12, 19, 30, 31). Most membrane apical proteins, first epitomized by glycosylphosphatidylinositol proteins (12), associate with lipid rafts (3). Cholesterol depletion diminishes apical transport capacity (19, 30, 31) without affecting basolateral sorting of model proteins such as vesicular stomatitis virus glycoprotein (VSVG) (19) and Na⁺/K⁺-ATPase (32), which do not associate with lipid rafts. However, considering that the basolateral domain contains lipid rafts (33, 34) and that caveolae are mainly formed at the basolateral domain of epithelial cells (16, 35), it seems still possible that cholesterol might play a yet unknown role in basolateral protein regionalization mediated by post-TGN processes.

Here, we used as a model protein the scavenger receptor class B type I (SR-BI), which in nonpolarized cells has been described to reside mostly in caveolae (14). Furthermore, SR-BI mediates high-affinity binding but not degradation of high-density lipoprotein (HDL), as well as selective lipid uptake and cholesterol efflux between HDL and cells (36), without requiring endocytosis (37). We found that, within certain expression levels, SR-BI predominates in the basolateral domain of transfected MDCK cells where it becomes internalized through a cholesterol depletion-insensitive pathway and recycles basolaterally. Unexpectedly, cholesterol depletion induced transcytosis of SR-BI, randomizing its cell surface distribution whereas the basolateral location of the epidermal growth factor receptor (EGFR), another caveolae/raft-associated protein (38, 39), remained unaffected. Such transcytosis required PKA activity and intact microtubules. Strikingly, cholesterol depletion also activated PKA, but without increasing cAMP levels. These results involve cholesterol in basolateral regionalization, participating in a post-TGN process of regulated basolateral-to-apical membrane protein distribution and in a noncanonical mechanism of PKA activation.

Materials and Methods

Expression Plasmid and Stable Transfection. A full-length cDNA encoding 509 residues of human SR-BI receptor was provided by S. Acton (Millennium Pharmaceuticals, Cambridge, MA), and subcloned into the pcDNA3 expression vector (Invitrogen). Transfection of MDCK type-II cells with Lipofectamine Plus (GIBCO/BRL), selection of stable transfected cells with G418, induction of protein expression by sodium butyrate, and immunofluorescence, immunoblot, and/or immunoprecipitation were all made as described (15, 40, 41).

Antibodies and Assays for Protein Expression, Polarity, Endocytosis, and Transcytosis. We used polyclonal antibodies raised against synthetic peptides of intracellular domain residues 495-509 of SR-BI, provided by S. Acton, and against residues 984-996 of EGFR (41). Apical-to-basolateral distribution ratio was assessed by domain-specific biotinylation assays in MDCK cells grown on Transwell chambers (Costar) (15). Reducible sulfo-NHS-SS-biotin (Pierce) and glutathione stripping (41, 42) were used for endocytosis and transcytosis experiments. PKA activity was assessed with the kit SignaTECT (Promega), and cAMP levels were determined by the cAMP enzyme immunoassay system (Amersham Pharmacia) as described (41).

[³H]Cholesterol Labeling of Cells, Extraction with Methyl-β-cyclodextrin (MβCD), and Raft Association. Cholesterol extraction with MβCD was estimated by preincubating cells for 20 h in media supplemented with 3.3 μCi (1 Ci = 37 GBq) [1α,2α(n)-³H]cholesterol (Amersham Pharmacia). After MβCD treatment, we measured [³H]cholesterol released into the media and remaining in the cells by liquid scintillation counting (Beckman) (19, 22). Caveolin-enriched low-buoyant density membranes were isolated as described (13, 15). Antibodies against the caveolar marker cav-1 and the non-raft protein Na⁺/K⁺-ATPase were from Transduction Laboratories (Lexington, KY) and Upstate Biotechnology (Lake Placid, NY), respectively.

Results

Apical-to-Basolateral Distribution of SR-BI in Stably Transfected MDCK Cells Depends on Its Expression Levels. To study the apical/basolateral sorting of SR-BI, we obtained stably transfected MDCK cells in which, similarly to other exogenous proteins (15, 40), SR-BI expression could be induced by 10 mM sodium butyrate. After 24 h of butyrate induction, it became easily detectable as an 82-kDa protein (Fig. 1A). The steady-state cell surface distribution of SR-BI changed from a predominantly basolateral location at the expression levels achieved by 24 h of sodium butyrate induction (Fig. 1B, lanes 1 and 2, and Fig. 1C), to a random distribution when its expression levels increased additionally by 3- to 6-fold after 48 h of butyrate induction (Fig. 1B, lanes 3 and 4). Instead, the endogenous EGFR remained basolateral, thus discarding nonspecific effects of sodium butyrate on the basolateral sorting machinery. We also observed a random distribution when certain colonies of transfected MDCK cells expressing basolateral chimeric SR-BI-green fluorescent protein were treated with sodium butyrate for just 12 h, as well as, when SR-BI was overexpressed by infection with an adenoviral vector (not shown). Thus, in MDCK cells, SR-BI behaves as a basolateral protein with a saturable sorting mechanism. For all additional experiments, we used 24 h of sodium butyrate induction to ensure a predominant basolateral distribution of SR-BI at steady-state.

Fig. 1.

Apical-to-basolateral distribution ratio of SR-BI expressed in transfected MDCK depends on its expression levels. Permanently transfected MDCK cells were treated with 10 mM sodium butyrate to induce exogenous SR-BI expression. After 24 h of butyrate induction, SR-BI was easily detected as the expected 82-kDa protein in immunoblot of total membranes (A), achieving a predominantly basolateral distribution, as showed by apical (Ap) and basolateral (Bl) domain-specific biotinylation (B, lanes 1 and 2) and indirect immunofluorescence (C). After 48 h of induction, its expression level further increased 3- to 6-fold, losing polarity (lanes 3 and 4) whereas the EGFR remained basolateral.

Basolateral SR-BI Associates with Low-Buoyant Density Membrane Fractions in Sucrose Isopycnic Gradients. Current assays for caveolae/lipid raft association assess both insolubility in cold nonionic detergents and low buoyancy (9). Moreover, different classes of lipid rafts have been roughly discriminated by sensitivity to different nonionic detergents (9, 10, 43, 44).

Treatment with 1% Triton X-100, 0.5% Lubrol, or 0.5% Brij-35 solubilized basolaterally expressed SR-BI whereas most cav-1 remained in the insoluble pellet (Fig. 2A). However, an established detergent-free method that coisolates caveolar and noncaveolar rafts base on pH 11 sodium carbonate extraction and sonication (13, 15, 38), showed SR-BI floating together with cav-1 in low-buoyant density membrane fractions of sucrose gradients whereas the non-raft-associated Na⁺/K⁺-ATPase (15, 32) partitioned into higher density fractions (Fig. 2B). Treatment with MβCD for 1 h, which extracts ≈50% of cellular cholesterol (see Fig. 4A), induced a density shift in the distribution of SR-BI and cav-1 from 15-20% to 30% sucrose (Fig. 2C), suggesting that basolateral SR-BI associates with cholesterol-based rafts, which as judged by detergent solubility, might not necessarily correspond to caveolae.

Fig. 2.

Basolateral SR-BI is solubilized in cold nonionic detergents but behaves as a raft-associated protein floating together with cav-1 in low-density membrane fractions. (A) Cells were lysed for 30 min in cold buffer containing 1% Triton X-100, Lubrol, or Brij-35, and the extracts were centrifuged at 100,000 × g for 1 h, to separate soluble (S) and insoluble (I) fractions. In contrast with the insolubility of cav-1, SR-BI was almost completely solubilized. (B and C) Extracts of cells either untreated (B, control) or treated with 10 mM MβCD (C) for 1 h at 37°C were prepared at 4°C in 500 mM sodium carbonate (pH 11), sonicated, and centrifuged to equilibrium in a sucrose gradient. Most SR-BI floated in low-density membrane fractions together with cav-1 (fractions 3 and 4; 15 and 20% sucrose) and shifted toward denser fractions (20-30% sucrose) after cholesterol extraction, indicating association with cholesterol-structured rafts. As control, the non-raft protein Na⁺/K⁺-ATPase distributes in denser fractions.

Fig. 4.

Acute cholesterol depletion by MβCD treatment induces loss of basolateral distribution of SR-BI, but not of EGFR, through a process inhibited by H89 and nocodazol, and mimicked by FSK. (A) Percentage of preloaded [³H]cholesterol (mean and SE) remaining in the cells after extraction with 10 mM MβCD at 37°C. (B and C) Domain-specific biotinylation of MDCK cells grown in Transwell filters, pretreated with 50 μM cycloheximide for 30 min and then incubated in the absence (lanes 1 and 2) or presence of 10 mM MβCD at 37°C. (B) MβCD was added alone for 10, 30, or 60 min (B, lanes 3-8) or preloaded with cholesterol and incubated with cells for 60 min (B, lanes 9 and 10). (C) The cells were treated for 1 h at 37°C with 10 mM MβCD in the absence (lane 3 and 4) or presence of either 50 μM H89 (lanes 5 and 6) or 33 μM nocodazole (lanes 7 and 8) added 30 min before MβCD. Cells were also treated with 50 μM FSK alone (lanes 9 and 10).

Basolateral SR-BI Is Constitutively Internalized by Means of an Endocytic Pathway Insensitive to Cholesterol Depletion and Recycles Basolaterally. The notion that SR-BI is a nonendocytic receptor has been recently challenged (45). Therefore, we performed internalization assays biotinylating the basolateral proteins with the cleavable sulfo-NHS-SS-biotin at 4°C and then stripping the cell surface biotin by reduction with glutathione added basolaterally at different time periods after shifting the temperature to 37°C. Biotinylated SR-BI increased (internalization) during the first 2-10 min (Fig. 3A, lanes 3-5), and then decreased in the next 20 min (lane 6). Because the mass of SR-BI remained unchanged (not shown), such decrement indicates recovery of glutathione sensitivity by basolateral recycling instead of degradation. Cholesterol depletion did not affect significantly SR-BI endocytosis whereas, as expected for a clathrin-mediated endocytosis (41), it inhibited by 90% the EGFR endocytosis (compare lanes 5 and 7). Thus, SRB-I is constitutively internalized through an atypical pathway and recycles basolaterally. Fig. 3 B and C shows a densitometric analysis of this data.

Fig. 3.

SR-BI is constitutively endocytosed via a pathway mostly insensitive to cholesterol depletion, and recycles basolaterally. (A) Transfected MDCK cells expressing SR-BI basolaterally and grown to confluence on Transwell chambers were biotinylated from the basolateral domain with reducible sulfo-NHS-SS-biotin at 4°C and then incubated at 37°C for the indicated periods of time to allow endocytosis (lanes 2 to 7). Neutravidin-agarose precipitation followed by immunoblot of SR-BI and the endogenous EGFR shows an increased resistance to basolateral stripping by glutathione reduction in the first 10 min, indicating internalization of both proteins. Afterward, both proteins recovered glutathione stripping sensitivity while maintaining their total mass (not shown), thus evidencing their basolateral recycling. Lane 1 shows total biotinylated SR-BI and EGFR without biotin stripping. Cells pretreated for 60 min with 10 mM MβCD showed similar SR-BI endocytosis (measured at 10 min; lane 7) whereas EGFR endocytosis was reduced by 90% (compare lanes 5 and 7). Quantitative analysis of the time course (B) and the effect of MβCD (C) was made on digitalized bands.

Cholesterol Depletion Induces Transcytosis of SR-BI Mediated by Microtubules and cAMP-Independent Induction of PKA Activity, and Forskolin (FSK) also Induced SR-BI Transcytosis. Unexpectedly, treatment with 10 mM MβCD for 30 min or 60 min, which provoked loss of ≈40 and 50% in cellular cholesterol, respectively (Fig. 4A), caused SR-BI to lose progressively its basolateral polarity, resulting in 38% and 50% apical localization after 30 and 60 min of MβCD treatment. In contrast, the basolateral distribution of endogenous EGFR remained unchanged (Fig. 4B). MβCD preloaded with cholesterol had no effect (Fig. 4B, lanes 9 and 10), indicating that cholesterol extraction (and not other nonspecific effects of MβCD) caused apical appearance of SR-BI. Apical SR-BI redistribution reached a mean of 48.83 ± 1.83% (mean ± SE; P = 0.00051, Wilcoxon test) of the total after 60 min of MβCD treatment and occurred even under protein synthesis inhibition, suggesting induction of basolateral-to-apical transcytosis.

The extensively studied transcytotic pathway of polyIg receptor is sensitive to PKA activation (7) and requires intact microtubules (46, 47). Similarly, redistribution of SR-BI induced by cholesterol depletion was completely abrogated by both the selective PKA inhibitor H89 and nocodazole whereas it was mimicked by FSK, a direct activator of adenylate cyclase (Fig. 4C). This result prompted us to study the effect of MβCD on the cAMP/PKA regulatory system. Strikingly, MβCD increased the PKA activity severalfold (Fig. 5A), but without any detectable change in cAMP levels (Fig. 5B), even in the absence or presence of high concentrations of the phosphodiesterase inhibitor 3-isobutyl-1-methylxantine (IBMX), which instead increased the response to FSK (Fig. 5C).

Fig. 5.

Acute cholesterol depletion by MβCD activates PKA without increasing intracellular cAMP levels. MDCK cells were incubated at 37°C with 10 mM MβCD or 50 μM FSK. (A) Both treatments increased PKA activity. (B) In contrast with FSK, cells treated with MβCD showed undetectable levels of cAMP levels. (C) In the presence of increasing concentrations of phosphodiesterase inhibitor IBMX, cAMP levels elicited by 1 h of FSK treatment increased progressively whereas, during treatment with MβCD, they remained almost undetectable. Each point represents average and SE.

Direct demonstration of SR-BI transcytosis was obtained by biotinylating the basolateral cell surface at 4°C with cleavable sulfo-NHS-SS-biotin and then stripping the biotin by glutathione reduction from the apical surface after warming the cells to 37°C for 1 h in the absence or presence of 10 mM MβCD or 50 μM FSK. The basolaterally biotinylated SR-BI decreased without detectable degradation whereas the EGFR remained unaffected (Fig. 6 A and B). Such SR-BI-induced transcytosis did not require HDL, occurring similarly in the presence or absence of serum.

Fig. 6.

MβCD- and FSK-induced basolateral-to-apical transcytosis of SR-BI. Transcytosis was assayed by biotinylating the basolateral domain with the cleavable sulfo-NSH-SS-biotin at 4°C, shifting to 37°C to allow trafficking, and then stripping the biotin from proteins appearing at the apical cell surface by reduction with 50 mM glutathione at 4°C. (A) Analysis of the remaining biotinylated proteins by precipitation with neutravidin-agarose and immunoblot shows that 50 μM FSK (lane 2) or 10 mM MβCD (lane 3) for 60 min at 37°C decreased the biotinylation of SR-BI but not of the endogenous EGFR. (B) The cells were incubated for the indicated time periods with 10 mM MβCD, and then SR-BI was immunoprecipitated. Its biotynilation assessed by blotting with neutravidin-HRP already decreased at the 30 min of treatment (upper blots) whereas its total mass analyzed by immunoblot showed no significant degradation (lower blots), thus indicating transcytosis.

Discussion

We have studied the effects of acute cholesterol depletion on the polarized segregation of SR-BI. Our experiments uncovered an example of a cholesterol requirement in basolateral membrane protein regionalization, revealing a post-TGN mechanism of regulated apical-to-basolateral protein distribution that is based on rafts and PKA activity. We also found that cholesterol depletion activates PKA, strikingly, without increasing cAMP to detectable levels, implicating cholesterol-based rafts in a compartmentalized mechanism of PKA regulation.

SR-BI expressed at moderate levels in transfected MDCK cells achieved a predominant basolateral location maintained by a cholesterol-dependent mechanism. Basolateral SR-BI became constitutively internalized via an atypical endocytic route insensitive to cholesterol depletion followed by basolateral recycling. Acute cholesterol depletion by MβCD induced transcytosis of SR-BI to the apical domain, leading to its randomization through a process inhibited by H89 and nocodazole, thus mediated by PKA activity and microtubules. Stimulation of cAMP/PKA by FSK also elicited SR-BI transcytosis. In contrast, neither cholesterol depletion nor FSK altered the basolateral distribution of endogenous EGFR, another caveolae/raft-associated protein (38, 39, 48).

In nonpolarized Chinese hamster ovary cells, detergent-free membrane fractionation in Optiprep density gradients and double immunofluorescence suggested that SR-BI resides predominantly in caveolae (14). As expected for association with cholesterol-structured lipid membrane domains, we observed, in detergent-free fractionation after sonication (13, 15), that basolateral SR-BI floated in low density membrane fractions codistributing with the caveolar marker cav-1 and shifted to higher density fractions on cholesterol depletion. However, by contrast with cav-1, SR-BI was readily solubilized by cold Triton X-100, Lubrol, and Brij-35, suggesting that it might reside in a distinct subclass of noncaveolar raft, a possibility also supported by its endocytic behavior. Basolateral SR-BI was constitutively internalized by a mechanism insensitive to cholesterol depletion, known to inhibit both caveolae and clathrin-coated pit-mediated endocytosis (21, 22). Furthermore, although caveolae mediate transcytosis in endothelial cells and ligand-inducible endocytosis of specific molecules in nonendothelial cells, recent evidence has revealed that, in nonendothelial cells, caveolae are relatively static structures with low potential for constitutive endocytosis (49, 50). Instead, noncaveolar rafts are active mediators of endocytosis even after cholesterol extraction (51). On the other hand, recycling endosomes seem to be enriched in cholesterol-dependent rafts (20, 52-54) participating in endocytic sorting (20). Thus, it is likely that basolateral SR-BI associates with noncaveolar cholesterol-based rafts actively competent in endocytosis and basolateral recycling.

The well characterized transcytotic pathway of polyIg receptor (pIgR) involves basolateral endocytosis via clathrin-coated pits, followed by transport along endosomal compartments to the apical domain, seemingly requiring inactivation of a basolateral sorting signal and intact microtubules (2). Transcytosis of pIgR is mainly constitutive although it is stimulated by both ligand and several kinase activities, including PKA (2). In contrast, transcytosis of SR-BI occurred only on cholesterol removal or FSK stimuli. Basolateral recycling could be directed either by the same basolateral sorting signal primarily operating at the TGN (55, 56) or by a different sorting motif (57) and, at least for certain proteins, requires the recently described epithelial-specific AP-1B adaptor (58). Cholesterol depletion leaves SR-BI-containing rafts still competent in endocytosis, but very likely altered in SR-BI recycling, and/or determines PKA-mediated changes in yet unidentified SR-BI sorting signals, thus promoting SR-BI diversion from a recycling into a transcytotic pathway.

Because SR-BI did not accumulate at the TGN by the currently used 20°C block (not shown), we could not discriminate whether its basolateral delivery from the TGN is also sensitive to cholesterol depletion. However, in contrast with biosynthetic apical pathways, which are inhibited by cholesterol extraction (19, 30), our results suggest that indirect apical pathways arising from endocytic compartments do not require intact cellular cholesterol levels.

The sorting behavior of SR-BI has been controversial, and its relevance for the physiological role of this receptor remains unknown. We observed that high expression levels led to complete loss of SR-BI basolateral polarity without affecting that of endogenous EGFR. Saturation of a limiting element could explain this result as well as the unpolarized distribution of SR-BI overexpressed in mouse liver (59) and the relatively poor basolateral asymmetry previously reported in transfected MDCK cells (45). However, in enterocytes (60, 61) and gallbladder epithelial cells (62), endogenous SR-BI has been found apically distributed, thus implying additional sorting complexities to be elucidated. On the other hand, although recent studies (45) contradict the widely accepted notion that SR-BI is a nonendocytic receptor, there is no evidence that SR-BI can be recycled and transcytosed, as HDL does (63). Our results demonstrate that SR-BI internalization is constitutive in MDCK cells and can lead to either recycling or transcytosis, depending on cellular cholesterol content and PKA activity. Because SR-BI displays intrinsic properties for selective lipid uptake from HDL without requiring internalization (37), endocytic sorting could provide additional control on its cell surface accessibility. Regulated SR-BI transcytosis might have special relevance in the liver where glucagon induces changes in cAMP levels (64) and SR-BI mediates cholesterol uptake and its subsequent excretion into bile (59, 65).

Our observation that cholesterol depletion stimulates PKA activity without increasing cAMP to detectable levels suggests that cholesterol-based microdomains could regulate a particular pool of PKA either independently of cAMP, which would be consistent with the reported inhibition of adenylate cyclase by cholesterol extraction (66), or involving highly compartmentalized cAMP production. Cholesterol efflux has been involved in PKA-mediated capacitation of mammalian sperm (67, 68). There are also reports of cAMP-independent mechanisms of PKA activation involving a release of PKA catalytic subunits from inhibitory complexes with NF-κB/Rel (69) and AKAP110 (70). Cholesterol depletion can alter different transduction pathways by disrupting the interactions of signaling proteins with both cav-1 and cholesterol-based rafts (10, 71). However, it remains unknown whether it releases active catalytic PKA subunits from reported inhibitory interactions with cav-1 (72) or from other complexes. Regardless, because PKA activation is involved in a variety of vesicular transport pathways (ref. 41 and references therein), it should be considered when interpreting cholesterol requirements on protein trafficking assessed by acute cholesterol depletion. The same cautiousness should apply to signaling events.

In summary, our results are compatible with a model in which cholesterol-based lipid rafts promote sorting of internalized SR-BI into a recycling instead of a transcytotic pathway. A pool of PKA controlled by the classical cAMP signaling system as well as by some yet unknown cAMP-independent mechanism sensitive to cholesterol contents mediates SR-BI transcytosis. Regulation of polarized protein distribution by inducible transcytosis adds further possibilities to modulate specific cell surface functions, which in the case of SR-BI may influence cholesterol homeostasis.