Scavenger Receptor Class B Type I as a Mediator of Cellular Cholesterol Efflux to Lipoproteins and Phospholipid Acceptors*

We recently reported that the rate of efflux of cholesterol from cells to high density lipoprotein (HDL) was related to the expression level of scavenger receptor class B type I (SR-BI). Moreover, the expression of this receptor in atheromatous arteries raises the possibility that SR-BI mediates cholesterol efflux in the arterial wall (Ji, Y., Jian, B., Wang, N., Sun, Y., de la Llera Moya, M., Phillips, M. C., Rothblat, G. H., Swaney, J. B., and Tall, A. R. (1997) J. Biol. Chem. 272, 20982–20985). In this paper we describe studies that suggest that the presence of phospholipid on acceptor particles plays an important role in modulating interaction with the SR-BI. Specifically, enrichment of serum with phospholipid resulted in marked stimulation of cholesterol efflux from cells that had higher levels of SR-BI expression, like Fu5AH or Y1-BS1 cells, and little or no stimulation in cells with low SR-BI levels, such as Y-1 cells. Stimulation of efflux by phospholipid enrichment was also a function of SR-BI levels in Chinese hamster ovary cells transfected with the SR-BI gene. Efflux to protein-free vesicles prepared with 1-palmitoyl-2-oleoylphosphatidyl-choline also correlated with SR-BI levels, suggesting that phospholipid, as well as protein, influences the interaction that results in cholesterol efflux. By contrast, cholesterol efflux from a non-cell donor showed no stimulation consequent to phospholipid enrichment of the serum acceptor. These results may help to explain observations in the literature that document an increased risk of atherosclerosis in patients with depressed levels of HDL phospholipid even in the face of normal HDL cholesterol levels.

The deposition of lipids in vessel walls and the development of atheromas are thought to be opposed by removal of the cholesterol and transport back to the liver for secretion, a process known as reverse cholesterol transport (1). Efflux of cellular free cholesterol from peripheral cells to the acceptor particles is the first step of this process. Although the mechanisms of cellular cholesterol efflux remain to be clarified, certain classes of high density lipoproteins (HDL) 1 are believed to play a key role in this process (2,3).
Two general models have been proposed regarding cholesterol removal from cells: a nonspecific, aqueous diffusion pathway and a specific pathway that may involve a cell surface receptor (3)(4)(5)(6). In the aqueous diffusion model, cholesterol molecules first desorb from the cell membrane and are then incorporated into acceptor particles after traversing the intervening aqueous layer by diffusion. The theoretical aspects of this mechanism have been extensively reviewed (4,7). Although this model was derived from the studies of free cholesterol transport between model membranes (unilamellar vesicles) (8), the kinetics of cholesterol transfer from living cells to phospholipid-containing acceptors can be adequately described by this model (9). In this process, free cholesterol transfers between donor and acceptors bidirectionally and, for net cholesterol removal to occur, a concentration gradient must be established between the cell surface and the acceptors. This gradient depends upon many properties of both the cell membrane and cholesterol acceptors, such as the cholesterol and phospholipid content, the arrangement of the cholesterol domains within the cell membrane, and the sizes or numbers of acceptors present in the aqueous solution (4, 10 -15). When an acceptor particle has a large capacity to absorb cholesterol (a cholesterol "sink") and is present in excess, the rate-limiting step for cholesterol removal is desorption of cholesterol from the plasma membrane (4).
The second model involves a pathway by which an acceptor promotes cholesterol efflux by interaction with cell surface sites (3,6,16,17). Lipid-free apoA-I or lipid-poor pre-␤-HDL particles are believed to operate by this mechanism. Through reversible interactions with the plasma membrane, apoA-I not only solubilizes cholesterol and phospholipid directly from the plasma membrane but also stimulates mobilization of pools of cholesterol that are readily accessible to esterification by acyl-CoA:cholesterol acyltransferase (16, 18 -20), an enzyme localized to the rough endoplasmic reticulum. Both lipid-free apoA-I and lipid-poor pre-␤-HDL particles presumably function effectively as natural shuttles to transport cholesterol between cells and larger ␣-HDL, since they have a relatively lower cholesterol capacity compared with ␣-HDL (1,21).
Receptor-mediated endocytosis plays an important role in lipoprotein metabolism. Although the LDL receptor is well characterized, the metabolism of HDL by any receptor-mediated pathway is still unclear. Several cell membrane proteins have been described that bind to HDL (for reviews, see Refs. 4 and 6), but the physiological implications are obscure. Re-cently, a member of the scavenger receptor family, the class B scavenger receptor SR-BI, was shown to bind HDL with high affinity (22). This receptor can also bind to many other ligands, such as native LDL, modified proteins (acetylated LDL, oxidized LDL, maleylated bovine serum albumin), and anionic phospholipids (23,24). However, unlike other scavenger receptors such as the LDL scavenger receptor, which binds to its ligand and internalizes the whole particle, SR-BI in certain cells can bind HDL reversibly and mediate selective cholesteryl ester uptake, leaving the HDL particles largely intact (22). SR-BI has highest expression in the adrenal gland, ovary, testis, and liver, where the selective uptake is important to the function of these tissues (25,26). It has also been demonstrated that in vivo the level of SR-BI is under feedback regulation in response to changes of cholesterol stores (27).
Recently, work in our laboratories showed that SR-BI overexpression can markedly enhance the bi-directional cholesterol flux in transfected CHO cells. A strong correlation between the rate of efflux to HDL or serum and the expression levels of SR-BI was also observed in several other cell lines (28). Furthermore, we found that SR-BI mRNA is expressed in the thickened intima of atheromatous aorta, suggesting a potential role of SR-BI in cholesterol flux in the arterial wall. Since SR-BI binds to a variety of lipoproteins that contain different apolipoprotein components and binds to anionic phospholipids (23,24), it is possible that phospholipids on the surface of lipoprotein particles play an important role in mediating the binding. Therefore, in this paper, we correlated the ability of PL to potentiate efflux, which was either incorporated into the lipoprotein fractions of serum or into small unilamellar vesicles, with the expression levels of SR-BI in various cell types.

Methods
Preparation of DMPC Multilamellar Vesicles (MLV)-DMPC was dissolved in methanol, and the solvent was evaporated under a stream of N 2 at room temperature; any remaining solvent was removed by vacuum overnight. Buffer (150 mM NaCl, 10 mM Tris, pH 7.4) was added, and the sample was warmed to 40 -50°C. The lipid was then vortexed vigorously for 1 min, placed in a low power sonication bath for 10 min, and vortexed again for 30 s to obtain turbid dispersions of multilamellar vesicles.
Preparation of POPC Small Unilamellar Vesicles (SUV)-POPC-SUV were made using the sonication technique as described previously (9). POPC in chloroform was dried under a stream of N 2 to form a thin film on a test tube wall and then placed in a vacuum for 2 h to remove any remaining solvent. Buffer (150 mM NaCl, 10 mM Tris, 1 mM EDTA, pH 7.4) was added to bring the concentration of phospholipid to 10 mg/ml and vortexed to generate MLV. SUV were prepared from MLV by sonicating at 0°C for 15 min followed by a 1-min cooling period for several cycles until the dispersion was almost clear. After sonication, the vesicle dispersion was centrifuged for 2 h at 40,000 rpm (Beckman 50 Ti fixed angle rotor) to remove any titanium particles and large multilamellar vesicles.
Modification of Serum with DMPC-MLVs-Serum was heated at 56°C for 30 min in the presence of 2 mM 5,5Ј-dithiobis(2-nitrobenzoic acid) to inactivate complement, which is toxic to some cell types, such as L cells (30) and to inhibit lecithin:cholesterol acyltransferase. The serum was then incubated with DMPC-MLV at 24°C for 2 h at a ratio of 4 mg of DMPC/ml of serum. After incubation, aliquots were removed and frozen at Ϫ70°C. The phospholipid concentration for serum and DMPC-modified serum were 1.5 and 4.8 mg/ml, respectively. Since our preliminary studies showed that freezing and heat treatment did not influence the ability of serum or DMPC-modified serum to remove cholesterol from Fu5AH cells, a large amount of serum or DMPCmodified serum were prepared and used throughout this study to minimize the experimental variability.
Cell Culture-Rat hepatoma cells (Fu5AH) were grown in Eagle's minimal essential medium supplemented with 5% calf serum. Mouse L cell fibroblasts were grown in Eagle's minimal essential medium supplemented with 10% FBS. Two lines of mouse adrenal tumor cells, Y1 and Y1-BS1, which differ in their expression levels of SR-BI (28) were grown in Ham's F-10 with 12.5% horse serum and 5% FBS. Chinese hamster ovary cells (CHO-K1) were grown in Ham's F-12 with 7.5% FBS, and three transfected CHO cells were grown in Ham's F-12 with 5% FBS. Mouse J774 macrophages were grown in RPMI 1640 with 10% FBS. Mouse RAW 2.1 macrophages were grown in Dulbecco's modified Eagle's medium with 10% FBS. All media were supplemented with 50 mg/ml gentamicin except for the transfected CHO cells, which were supplemented with geneticin at a concentration of 0.3 mg/ml. Cells were plated in 24-well plates 3 days before the experiment and labeled for 2 days with 1-2 Ci of [ 3 H]cholesterol in 1 ml of growth medium to obtain confluent radiolabeled cell monolayers. Before the experiments, the labeling medium was replaced with equilibrating medium for 18 h to allow the equilibration of labeled cholesterol among cellular pools. Sandoz 58035, an inhibitor of acyl-CoA:cholesterol acyltransferase, was present at a concentration of 1 g/ml during the labeling, equilibration, and efflux periods in all experiments to prevent sequestration of the label in an intracellular cholesteryl ester pool. More than 95% total cholesterol was in free cholesterol form in all cell types after the experiments. Y1-BS1 cells were provided by Dr. David L. Williams.
Cholesterol Efflux Assay-The cholesterol efflux ability of the serum, DMPC-modified serum, M␤CD-supplemented serum, HDL 3 , and POPC-SUV were evaluated by methods described previously (31). Cholesterol acceptors included serum or DMPC-modified serum, which were customarily diluted in efflux medium to 5% that of the original concentration; SUV was diluted to 1 mg/ml and HDL 3 was diluted to 75 g/ml with medium before the experiment. In some samples, M␤CD was added to the 5% serum before the experiment at a concentration of 0.05 mM. The different cholesterol acceptors were incubated with [ 3 H]cholesterol-labeled cells for various time periods at 37°C, and 100-l aliquots were taken to monitor efflux at indicated time points. These aliquots were filtered through 0.45-m Multiscreen (96 screens) filtration plates (Millipore), and 75-l aliquots were then counted to determine the release of labeled cholesterol from the cells. The total [ 3 H]cholesterol in the cells was extracted with isopropanol from dried cell monolayers that had been washed with phosphate-buffered saline.
In some experiments, cholesterol-impregnated glass fiber filters (Corning) were used as an inert cholesterol donor. Filters were cut to fit multiwell plates, 50 l of an ethanol cholesterol solution was applied to each filter, and the filters were dried overnight. Each filter contained 0.5 Ci of [ 3 H]cholesterol and 5 g of cholesterol mass, which is similar to the average cholesterol content of the cell monolayers used in the efflux assays. The filters were then exposed to 2 ml of culture medium containing the various acceptors tested with the cells. The amount of radiolabeled cholesterol released to the medium was expressed as the fraction of the total radioactive cholesterol present initially in the cells or filters; culture medium without added acceptors was used as a control for all experiments. The net movement of cholesterol mass between cells and acceptors was not monitored in this study. Each data point was the result of triplicate determinations from one or two different experiments.
Immunoblot Analysis-Cells grown in monolayers were homogenized by a N 2 cavitation technique using N 2 (300 p.s.i) for 30 min before the release of pressure. Protease inhibitors (0.5 g/ml leupeptin, 1 g/ml aprotinin, 1 g/ml pepstatin A, 0.2 mM phenylmethylsulfonyl fluoride and 1 mM EDTA) were present in the homogenizing buffer (100 mM Tris). Cell membranes were collected by ultracentrifugation (50 Ti rotor, 32,000 rpm for 1 h), and proteins were resolved by SDS-PAGE. Immunoblotting was performed as described previously (28) using antiserum to rodent SR-BI (27) followed by chemiluminescence detection (Amersham Pharmacia Biotech) and densitometric quantitation.
Kinetic Analysis-The efflux data were analyzed as originally proposed by Johnson et al. (32) and as described in detail for this system by Davidson et al. (13). Briefly, the kinetic analysis assumed a closed system where unesterified, free cholesterol exists in one of two kinetic pools, the cellular or the acceptor pool. The equilibration of cholesterol between these two pools was fitted to the single exponential equation: Y ϭ H 1 e Ϫgt ϩ H 2 . Y represents the fraction of radiolabeled cholesterol remaining in the cells, t is the incubation time, H 1 is a pre-exponential term that reflects the fraction of cellular free cholesterol that is lost to the medium at steady state, g is the sum of the rate constants for efflux (k e ) and influx (k i ), and H 2 is a constant that represents the fraction of labeled cell cholesterol that remains associated with the cells at equilibrium due to a constant retrograde flux of free cholesterol from the extracellular acceptor back into the cells. H 1 , g, and H 2 are variables that were fitted to the experimental data by nonlinear regression. The half-times of efflux were calculated as t 1/2 ϭ ln 2/k e , where k e ϭ g ϫ H 1 . V max and EC 50 values for the efflux of cholesterol from CHO cells (Fig.  7) were calculated using the equation

RESULTS
Free Cholesterol Efflux to Serum, HDL 3 , and SUV from Different Cell Types-To compare the ability of different cells to donate cholesterol to various acceptors, human serum, HDL 3 and POPC-SUV were incubated with [ 3 H]cholesterol-labeled cell monolayers for 8 h. Fig. 1 shows the time course of cholesterol efflux from two cell types that represent two extreme patterns of efflux. It can be seen that after 8 h of incubation 40% of free cholesterol was released from Fu5AH cells to serum, whereas only about 20% was released from Y1 cells. Efflux to HDL 3 and SUV is also faster in Fu5AH cells compared with Y1 cells. POPC-SUV, at a concentration of 1 mg of PL/ml of efflux medium, was able to promote cholesterol efflux, comparable to HDL 3 at 75 g of protein/ml. These data agree with previous work from this laboratory that demonstrated that the Fu5AH cells release free cholesterol much faster than other cells (33).
Correlation between Efflux to 5% Serum and to HDL 3 or SUV-The results obtained from Fu5AH and Y1 cells suggested that similar efflux patterns were obtained whether the acceptors were serum, HDL 3 , or POPC-SUV. To examine this further, a variety of cell types were studied that spanned a range of efflux from fast to slow. Serum (5%), HDL 3 (75 g of protein/ml) or POPC-SUV (1 mg/ml) were incubated with cholesterol-labeled cells for 8 h at 37°C, and aliquots were removed at various time points. Rate constants (k) for efflux to different cholesterol acceptors were obtained from the single exponential curve fit to the time course data. As shown in Fig.  2, there is a very strong linear correlation (r 2 ϭ 0.98) between efflux to serum and to HDL 3 in all the cell types examined. Surprisingly, a strong correlation was also observed between serum and POPC-SUV with a correlation coefficient of 0.94, which suggests that a common cellular component may be involved in mediating the cholesterol efflux to serum and POPC-SUV. This finding is consistent with our previous data that HDL-PL is the component of HDL that best reflects the efflux efficiency of the serum (34,35).
Effect of SR-BI Expression on Cholesterol Efflux-Recently, the scavenger receptor SR-BI was found to bind HDL with high affinity (22). Previously, data from our laboratories have suggested that this receptor might contribute to cholesterol efflux by promoting bi-directional flux (28). Having characterized the cholesterol efflux properties of a variety of cells, the SR-BI expression level was measured by preparing cell membrane proteins and total RNA for Western blot and RNase protection assays, followed by densitometry of chemiluminescence images; mouse liver membranes were used as a control in all assays. As shown in Fig. 3, the expression level of SR-BI was found to vary significantly among different cell types and appears to be correlated with the efflux of cell cholesterol to both serum and POPC-SUV. Panel A shows a combination of data previously reported for serum as acceptor (28), to which additional analyses and cell types have been added; the plot shows the saturation effect at high levels of SR-BI (Ͼ1,000 units), whereas the inset emphasizes the linear relationship at low levels of SR-BI; similar results were found for POPC-SUV (panel B). Note that Y1 cells, although adrenal-derived, were found to have slow efflux in response to serum, HDL, or PL vesicles (Fig. 1) and very low SR-BI expression, whereas Fu5AH cells have the highest rate of efflux and SR-BI levels among various cell types studied (Fig. 3). Another line of the adrenal-derived Y1 cells (Y1-BS1) had a moderate SR-BI expression level and intermediate efflux values (Fig. 3).
Effect of Phospholipid Enrichment of Serum on Cholesterol Efflux-Previously we reported that PL-modification could greatly enhance the ability of serum to remove cholesterol from Fu5AH cells (36). To investigate the generality of this phenomenon, we screened several cell types using a single preparation of serum and DMPC-modified serum as cholesterol acceptors. Fig. 4 shows that two different cell types respond very differently to PL modification, with 155% stimulation of efflux by PL in Fu5AH cells (Fig. 4, panel A) and 0% in Y1 cells (Fig. 4, panel  B). Fig. 5 summarizes the incremental response to PL modification of serum relative to SR-BI expression levels in seven different cell types. The percentage cholesterol efflux stimulation by phospholipid supplementation was expressed as the change in the first order rate constant (k e ), and the data suggest that cells with higher expression levels of SR-BI not only release free cholesterol faster to serum but are more affected by PL enrichment of serum. For unknown reasons, the mouse L cell fibroblast appears to respond to PL enrichment out of proportion to its SR-BI expression level when compared with the other cell types.
Effect of Cyclodextrin on Cholesterol Efflux to Serum-The data suggest that small particles such as HDL 3 or POPC-SUV interact with the SR-BI on cell membranes to facilitate cholesterol efflux and that the reaction of serum with DMPC enhances this interaction to further stimulate efflux. It has been proposed that certain components of serum (natural shuttles) promote efflux by transferring cell cholesterol to large acceptors (sinks) (21,37). To study whether all shuttles interact with the SR-BI, studies were performed with a nonphysiological shuttle. ␤-Cyclodextrins are water-soluble compounds with a hydrophilic external face and a hydrophobic cavity that is capable of dissolving hydrophobic compounds and thus enhancing their solubility in aqueous solutions (38,39). It has been demonstrated recently that cyclodextrins are very efficient in stimulating the removal of cholesterol from a variety of cells in culture (40,41). At lower concentrations (Ͻ1 mM), ␤-cyclodextrins have been shown to function as a shuttle that catalyzes the exchange of cholesterol between cells and acceptors in the medium (21). In our studies, serum was enriched with either cyclodextrin (0.05 mM) or with DMPC (4 mg/ml serum) to study their effect on efflux in various cell types. Data shown in Table  I suggest that the efflux ability of serum is stimulated by M␤CD in all cell types, as contrasted with PL stimulation, which was shown to enhance efflux only in certain cell types. This result is consistent with the hypothesis that cyclodextrins, because of their small size, can access the cell membrane readily so that cholesterol molecules can diffuse directly into the cavity of the cyclodextrin molecules, a process that does not depend upon the existence of receptors such as SR-BI.
Cholesterol Efflux from SR-BI-transfected Cells-To confirm the role of SR-BI in cellular cholesterol efflux, transfected CHO cells with low or high expression levels of SR-BI were also used for the study of efflux. Fig. 6A shows that efflux to all different acceptors except M␤CD was increased in SR-BI-transfected cells as compared with the control CHO cells, and the differences between the efflux from CHO-high cells and CHO-control cells is significant at p Ͻ 0.05 or p Ͻ 0.0001. However, the magnitude of stimulation resulting from increased SR-BI expression was greatest for DMPC-modified serum. Phospholipid enrichment of serum enhanced efflux in all cells, but the percent stimulation over serum alone was greater for the high expression CHO cells (70%) than for the control CHO cells (45%), which again suggests a role for PL in the interaction of HDL with SR-BI. An inverse correlation between efflux stimulated by PL and by M␤CD is also observed (Fig. 6B). This may be explained by the fact that efflux from CHO-high cells to serum alone is already very fast, so that the stimulation of efflux by M␤CD is relatively low.
To further establish the importance of phospholipid, experiments were performed to study the dependence of cholesterol efflux upon the concentration of POPC-SUV in the medium. Data in Fig. 7A show that efflux to POPC-SUV is increased as the PL concentration increases in both CHO-control cells and CHO-high cells. However, the rate of cholesterol release to POPC-SUV is faster from CHO-high cells than that from CHOcontrol cells. The differences between efflux from CHO-control cells and CHO-high cells are statistically significant (p Ͻ 0.05) at all PL concentrations examined, and V max values of efflux from CHO-control cells and CHO-high cells are 0.071 Ϯ .0005 and 0.114 Ϯ .007, respectively (p Ͻ 0.005). The same efflux values have been replotted in Fig. 7B, in which the phospholipid concentration is presented on a log scale, and the k e values observed at the highest POPC-SUV concentration (10 mg/ml) for both control and high SR-BI-expressing CHO cells are made to overlap. In this form it can be readily seen that the two curves are essentially identical and that the concentrations of POPC-SUV that give half-maximal efflux (EC 50 ) are not statistically different (CHO-control ϭ 665 Ϯ 178 g of PL/ml; CHO-high ϭ 972 Ϯ 199 g of PL/ml) (p Ͼ 0.05).
Cholesterol Efflux from an Inert Donor-To further characterize the importance of SR-BI in cholesterol efflux stimulated by PL, we developed a model system that was completely devoid of SR-BI. In these experiments, we used cholesterol-im-   pregnated glass fiber filters labeled with [ 3 H]cholesterol as an inert cholesterol donor. The radiolabeled glass fiber filter was incubated with 2 ml of efflux medium containing 5% serum or 5% serum supplemented with either DMPC (4 mg/ml serum) or M␤CD (0.05 mM) for various times in a humidified incubator at 37°C. Fractions were collected, and the efflux was determined as described for the cholesterol efflux from cells. As compared with serum alone, no efflux stimulation was observed by DMPC modification of serum (Fig. 8). However, efflux increased 3.5fold after M␤CD supplementation of serum. A dose response of efflux to HDL 3 was also observed (data not shown).

DISCUSSION
It is generally accepted that efflux of cholesterol from peripheral cells to HDL is the first step in reverse cholesterol transport. However, the rate of cholesterol release varies significantly among different cell types (9,33,42). This may be for several reasons, such as variability in the fluidity or cholesterol content of the plasma membrane (43), in the thickness or composition of the extracellular matrix, which might prevent the access of cholesterol acceptors to the plasma membrane, or in different expression levels of a lipoprotein receptor, such as SR-BI (44), in the plasma membranes of different cell types.
In our studies, we specifically measured the release of labeled cholesterol from cells and found that SR-BI can enhance cholesterol efflux to various acceptors (Figs. 3, 4, and 6), probably through increasing the local concentration of acceptors close to the cell surface, thus facilitating the bi-directional movement of cholesterol and phospholipids (28). Our data (Figs. 2, 3, and 7) also demonstrate a key role for phospholipid in the efflux process, possibly by mediating the interaction between HDL and SR-BI. It has been demonstrated that the efflux ability of serum correlates best with HDL-PL (34,35) and that efflux can be markedly enhanced by PL supplementation when Fu5AH cells are used as the cholesterol donors (36). As can be seen from Fig. 1, the relative efflux efficiency of different acceptors is similar for different cell types, although the cells exhibit different rates of cholesterol release. However, the change in efflux in response to PL supplementation of serum varies significantly among cell types (Figs. 5 and 6) and generally corresponds to the expression level of SR-BI. Further evidence for the role of phospholipid comes from the fact that efflux promoted by apolipoprotein-free POPC-SUV correlated well with the expression level of SR-BI (Fig. 3) and that efflux is significantly higher in SR-BI high expression CHO cells compared with control CHO cells (Figs. 6 and 7) at all PL concentrations (50 g/ml to 10 mg/ml). These results imply that POPC-SUV can interact with SR-BI. On the other hand, it has been reported that phosphatidylcholine or sphingomyelin liposomes cannot compete with anionic phospholipids such as phosphatidylserine-and phosphatidylinositol-containing liposomes for binding to the receptor (24). The interaction of POPC-SUV with SR-BI is weak, as indicated by EC 50 values for efflux to POPC-SUV ranging between 600 and 900 g of PL/ml but presumably sufficient to sequester acceptors near the cell surface, thereby increasing their local concentration and promoting lipid movements. The marked differences between control and SR-BI-expressing cells in the V max values for efflux and the similarities in EC 50 values (Fig. 7) is consistent with a model in which receptors contributing to cholesterol efflux are increased in the transfected CHO cells, but the affinity of the phospholipid vesicles for the receptor is the same. However, the cholesterol efflux to serum or POPC-SUV is not found to be proportional to the amount of SR-BI present in the cell membrane (Figs. 3 and 7). The change of efflux ability is more sensitive in the relatively low SR-BI range. This suggests that, at very high expression levels of SR-BI, the potential efflux is so fast that another step in cholesterol efflux, such as movement of cholesterol from an intracellular pool to the plasma membrane or from a slow pool of cholesterol to a fast pool within the cell membrane, may become rate-limiting; alternatively, it is possible that all of the expressed protein is not functional, since the immunoblotting assay used in the study can only detect the existence of SR-BI protein in total cell membrane but not its activity.
In addition, we examined the effect of methyl-␤-cyclodextrin in stimulating the cholesterol efflux of serum and compared it with that of PL supplementation. M␤CD at a low concentration was used in our studies as an artificial cholesterol shuttle to catalyze the exchange of cholesterol between cell membranes and serum (21). From Table I, we can see that the efflux ability of serum was increased by M␤CD to some extent in all cell types but particularly in those cells that express low levels of SR-BI. These water-soluble cyclodextrins are likely to closely approach the cell membrane and can remove cholesterol by reducing the activation energy for cholesterol efflux from approximately 20 kcal/mol to around 6 kcal/mol (41); being synthetic molecules, they are unlikely to interact with a cell surface receptor.
Our data show that all cells release cholesterol to serum at a basal level. The rate of basal efflux to serum can be enhanced by either PL or M␤CD supplementation of serum. Since stimulation of efflux by M␤CD is not dependent upon the expression levels of SR-BI, the efflux stimulation by M␤CD is expected to be more pronounced in cell types expressing lower levels of SR-BI and exhibiting slower rates of cholesterol release; this prediction is in agreement with our data (Table I and Fig. 6).
PL supplementation could stimulate efflux by several possible mechanisms. First, after incubation with PL, HDL particles are enriched with PL, which generates a larger surface area for cholesterol solubilization; second, new HDL-like particles are generated, which increases the numbers of cholesterol acceptors; and third, phospholipid modification of serum may change the structure of HDL particles, changing their interaction with a receptor. Since not all cells are responsive to PL supplementation of serum and since the rate of release of cholesterol from an inert donor is not enhanced by phospholipids (Table I and Fig. 8), our data imply that increased cholesterol efflux induced by PL supplementation is not simply due to an increased capacity of the modified serum to hold cholesterol but reflect an enhancement of some specific interaction required for cholesterol removal. Since phospholipid modification is believed to remove apolipoprotein from existing HDL to form new complexes (36,45,46), it is possible that formation of a pre-␤ 1 -HDL type of complex, which has been reported as playing a role in cholesterol efflux (1,47,48), explains the enhancement of efflux. However, our previous studies have shown that both DMPC and bovine brain sphingomyelin enhance the efflux ability of serum to a similar extent after both long (36) and short term (data not shown) incubation with cells, but pre-␤-HDL-like particles were generated only by DMPC modification (36). These data suggest that pre-␤-HDL may not play a crucial role in efflux stimulated by phospholipids. Based on previous studies using low concentration of cyclodextrins as cholesterol shuttles (transporters) and phospholipid vesicles as cholesterol sinks, we proposed that native lipoproteins can serve as physiological sinks and shuttles (21,37). Thus, small HDL particles such as pre-␤-HDL or lipid-poor apoproteins might function as shuttles, whereas larger particles such as ␣-HDL or LDL would participate in the early steps of reverse cholesterol transport as cholesterol sinks (21,37). The observations that phospholipid enrichment of serum stimulates SR-BI-mediated cholesterol efflux and that the stimulatory effect of cyclodextrin shuttles is reduced as SR-BI levels increase suggest that interaction with SR-BI expressed on cells promotes the direct exchange of cholesterol between cells and the larger, less efficient lipoproteins, which function as cholesterol sinks. As a result of this enhanced direct exchange of cholesterol between the plasma membrane and lipoprotein particles, the need for cholesterol shuttles such as cyclodextrins or pre-␤-HDL would be reduced. With regard to the third proposed mechanism, we have previously demonstrated that there is no single HDL particle that serves as the sole cholesterol acceptor from Fu5AH cells (31). Since SR-BI is a multiligand lipoprotein receptor (22,23,49), the function of PL may be to generally modify a number of lipoprotein species in a way that changes their interaction with SR-BI. It has been reported recently that apolipoproteins such as apoA-I, apoA-II, and apoC-III can directly mediate binding to SR-BI (50). Here we provide evidence that HDL-PL can modulate the functional interaction of HDL to SR-BI. The nature of the critical change in HDL structure resulting from PL supplementation remains to be elucidated.
In summary, our data indicate that the SR-BI receptor can participate in lipid efflux, and that phospholipids may have a major role in mediating this process; furthermore, it appears that cells can differ widely in their responsiveness to phospholipids, apparently as a reflection of their levels of expression of the SR-BI protein. The physiological role of the phospholipid component of HDL has attracted much attention recently. There is increasing evidence in the literature that lower levels of HDL phospholipid can put individuals at risk for heart disease even in the presence of a normal HDL cholesterol level. Thus, it has been demonstrated that a low plasma phosphatidylcholine/free cholesterol ratio is significantly correlated with human ischemic vascular disease (51), that serum phospholipid fractions differ significantly in individuals with heart disease as compared with controls (52), and that patients with coronary artery disease have a lower HDL phospholipid content then normal (53). More recently, Bagdade et al. (54) have shown that the free cholesterol to phospholipid ratio is altered in HDL fractions obtained from both men and women with insulin-dependent diabetes mellitus. Also, Cavallero et al. (55) have shown that patients with Type II diabetes have LpA-I particles with aberrant size and composition, including a decrease in phospholipid content, and that these particles are less efficient in promoting cell cholesterol efflux from Ob 1771 cells. These findings lead us to propose that abnormalities in HDL function in disease states may be related to changes in HDL phospholipid composition, possibly involving changes in its interaction with SR-BI in vivo. Since the cells with the highest levels of SR-BI are ones for which influx of cholesterol would be important, the effect on efflux appears paradoxical and may simply reflect the facilitation of flux in both directions. However, PL supplementation of serum may change the affinity of HDL to SR-BI and thus promote the net efflux between the cells and lipoproteins, especially since the efflux process appears exquisitely sensitive at the low expression range of SR-BI.