Hepatic Scavenger Receptor BI Promotes Rapid Clearance of High Density Lipoprotein Free Cholesterol and Its Transport into Bile*

The clearance of free cholesterol from plasma lipoproteins by tissues is of major quantitative importance, but it is not known whether this is passive or receptor-mediated. Based on our finding that scavenger receptor BI (SR-BI) promotes free cholesterol (FC) exchange between high density lipoprotein (HDL) and cells, we tested whether SR-BI would effect FC movement in vivo using [14C]FC- and [3H]cholesteryl ester (CE)-labeled HDL in mice with increased (SR-BI transgenic (Tg)) or decreased (SR-BI attenuated (att)) hepatic SR-BI expression. The initial clearance of HDL FC was increased in SR-BI Tg mice by 72% and decreased in SR-BI att mice by 53%, but was unchanged in apoA-I knockout mice compared with wild-type mice. Transfer of FC to non-HDL and esterification of FC were minor and could not explain differences. The hepatic uptake of FC was increased in SR-BI Tg mice by 34% and decreased in SR-BI att mice by 22%. CE clearance and uptake gave similar results, but with much slower rates. The uptake of HDL FC and CE by SR-BI Tg primary hepatocytes was increased by 2.2- and 2.6-fold (1-h incubation), respectively, compared with control hepatocytes. In SR-BI Tg mice, the initial biliary secretion of [14C]FC was markedly increased, whereas increased [3H]FC appeared after a slight delay. Thus, in the mouse, a major portion of the clearance of HDL FC from plasma is mediated by SR-BI.

Plasma high density lipoprotein (HDL) 1 plays a key role in maintaining cholesterol homeostasis. Epidemiological studies demonstrated a strong inverse correlation between HDL levels and the risk of coronary artery disease (1). Although detailed mechanisms remain uncertain, it has been proposed that HDL promotes reverse cholesterol transport by facilitating transfer of cholesterol from peripheral tissues to the liver for secretion into bile (Ref. 2; see Ref. 3 for a recent review). Early studies using HDL and low density lipoprotein (LDL) with radiolabeled free cholesterol (FC) showed that FC in HDL is the preferred source for biliary cholesterol (4 -8). Recent work with plant sterols provided further evidence that HDL is the preferred carrier for the transport of cholesterol into bile (9). Based on the observations that HDL FC is preferentially utilized for biliary secretion and that, in tissue culture studies, HDL, but not LDL, selectively binds FC (10), Schwartz et al. (5) predicted that a cell-surface HDL receptor might be involved in the hepatic uptake of HDL FC. As a counterpoint to this idea, free cholesterol exchanges readily between lipoproteins and cells, suggesting that passive FC uptake by the liver might be possible.
Scavenger receptor BI (SR-BI) has recently been identified as an authentic HDL receptor that mediates the selective uptake of HDL cholesteryl ester (CE) (11) and bi-directional transfer of FC between HDL and cells (12,13). Its tissue distribution (11,14) and regulatable expression in the adrenal gland, testis, and ovary (14,15) indicate that the receptor plays an important physiological role in cholesterol metabolism (16,17). Hepatic overexpression of SR-BI leads to decreased HDL levels in mice (18 -20) due to accelerated hepatic uptake of HDL CE and subsequently increased HDL CE and protein catabolism (19). On the other hand, decreased expression of SR-BI in gene-targeted mice results in increased HDL levels (21,22). Furthermore, SR-BI mRNA is expressed in thickened intima of atheromatous aorta (12), and the receptor suppresses the development of atherosclerosis in SR-BI transgenic (SR-BI Tg)/LDL receptor-deficient compound mice fed the Paigen diet (23).
In our previous work, we found that, in transfected Chinese hamster ovary cells, SR-BI mediates the cellular uptake of HDL FC as well as CE (12). Moreover, SR-BI promotes cholesterol efflux from cells to HDL (12,13) or protein-free phospholipid (PL) vesicles (13), and the efflux rates correlate with the expression level of SR-BI in different cell lines (Refs. 12 and 13; see Ref. 24 for review). In mice with hepatic overexpression of SR-BI, the biliary concentration of cholesterol is increased (18,25). These results led us to hypothesize that SR-BI plays a physiological role in vivo in promoting the hepatic uptake of HDL FC and facilitating the secretion of cholesterol into bile. We tested this hypothesis in mice with increased or attenuated expression of SR-BI.

MATERIALS AND METHODS
Animals-SR-BI Tg and SR-BI att mice were generated as described previously (19,22). ApoA-I knockout (apoA-I0) mice were purchased from Jackson Laboratory. The SR-BI Tg mice used have been backcrossed with C57BL/6J mice five times, and the apoA-I knockout mice backcrossed with C57BL/6J mice Ͼ10 times. The SR-BI att mice were in a mixed background of BALB/c/ByJ and C57BL/6J (22). The animals were 6 months old, weighed 25-35 g, and were housed at 22°C under a constant light/dark cycle (7:00 a.m. to 7:00 p.m.) with ad libitum access to water and rodent chow. Before the experiments, animals were fasted * This work was supported by National Institutes of Health Grants HL58033 and HL54591 and in part by a grant from Eli Lilly Inc. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. for 10 -12 h (11:00 p.m. to 9:00 -11:00 a.m.).
Clearance of HDL Cholesterol-HDL (d ϭ 1.063-1.21) was provided from human plasma by sequential ultracentrifugation and was duallabeled with [4][5][6][7][8][9][10][11][12][13][14] (26). The specific radioactivity of the dual-labeled HDL was ϳ55,000 cpm [ 14 C]FC/g of HDL FC and 36,000 cpm [ 3 H]CE/g of HDL CE. Animals were lightly anesthetized and injected with dual-labeled HDL in 0.1 ml of phosphate-buffered saline via the femoral vein. Blood was collected periodically via the retro-orbital plexus; the plasma was immediately separated; and the HDL fraction was rapidly obtained by precipitating LDL and very low density lipoprotein with heparin/Mn 2ϩ (HDL cholesterol reagent, Sigma). The radioactivity in HDL and the plasma was determined with a ␤-counter. The radioactivity in the non-HDL fraction was obtained by subtracting the radioactivity in HDL from that in the plasma. The radioactivity of 0.5 min post-injection is defined as 100% of injected radioactivity. Based on the clearance curves of HDL tracers (see Fig. 1A), fractional clearance rates (FCRs) were calculated using a two-compartment model as described (27).
To measure esterification of HDL FC in the plasma, plasma samples were collected periodically and lipid-extracted (28); FC and CE were separated by TLC; and radioactivity was measured. The esterification ratio is expressed as the percentage of cholesterol converted to ester.
Uptake of HDL Cholesterol by Liver and Hepatocytes-At the end of clearance experiments (usually 1 h post-injection), the liver was perfused thoroughly via the portal vein with phosphate-buffered saline. The liver was blotted dry, weighed, and stored at Ϫ80°C. A 200-mg piece was later homogenized and lipid-extracted (29), and radioactivity was determined. The total accumulation of HDL [ 14 C]FC in the liver of each individual mouse was calculated from the total weight of the organ.
Primary hepatocytes were isolated by a two-step collagenase perfusion (collagenase type I (Worthington), ϳ0.3 mg/ml in Hanks' balanced solution supplemented with 10 mM Hepes, pH 7.4) (30). To prevent nonspecific digestion, protease inhibitors (Complete TM , EDTA-free tablets, Roche Molecular Biochemicals) were added to the perfusion buffer according to the manufacturer's instruction. Viable cells were obtained after low speed centrifugation in Percoll (Sigma). The cells were then washed, resuspended, and maintained on a six-well plate in Williams' E medium supplemented with 7% fetal bovine serum, insulin/transferrin/ sodium selenite (Life Technologies, Inc.), 30 mM pyruvate, and penicillin/streptomycin at 37°C under 5% CO 2 . The HDL cholesterol uptake experiment was performed the next day with 0.5% bovine serum albumin in Williams' E medium. After incubation with dual-labeled HDL, cells were washed, and the cell-associated radioactivity was measured (12).
Biliary Secretion of HDL Cholesterol-The animals were anesthetized with ketamine (140 g/g) and xylazine (14 g/g) intraperitoneally. The abdomen was opened; the distal part of the common bile duct was ligated; and the gallbladder was incised and cannulated. The dual-or [ 3 H]CEth-labeled HDL was injected, and bile samples were collected periodically while the animals were placed under a heating lamp to maintain body temperature. The radioactivity in bile was determined by either direct counting (no more than 40 l of bile in 10 ml of scintillation mixture) or counting after lipid extraction and TLC.
Statistical Analysis-Statistical analysis was performed by twotailed Student's t test for unpaired data.

Effect of SR-BI on Clearance of HDL FC in Plasma-
To explore the role of SR-BI in the metabolism of HDL FC, the clearance of HDL cholesterol was studied in mice with different levels of hepatic SR-BI expression. [ 14 C]FC-and [ 3 H]CE-labeled HDL was injected intravenously, and blood samples were obtained at different time points. To avoid ex vivo transfer of free cholesterol radioactivity between lipoproteins, HDL was rapidly separated from other plasma components by precipitation (7,31). As shown in Fig. 1A, the clearance of HDL [ 14 C]FC was dramatically faster in SR-BI Tg mice and slower in SR-BI att mice. Fifty percent of HDL FC was removed from the plasma within 1.8 min (t1 ⁄2 ) in SR-BI Tg mice, compared with 3.2 and 7.0 min in WT and SR-BI att mice, respectively. To determine whether the increased clearance of HDL FC in SR-BI Tg mice was due simply to a decreased level of HDL, the clearance of HDL FC was also measured in apoA-I0 mice, which have similarly decreased levels of HDL (32), but normal hepatic SR-BI expression (14). Despite dramatically decreased plasma HDL levels, apoA-I0 mice exhibited a similar clearance of HDL FC compared with WT mice, especially in the initial stage (t1 ⁄2 ϭ 3.0 min) (Fig. 1A).
To determine whether the accelerated clearance of HDL FC in SR-BI Tg mice was caused by rapid transfer of HDL FC to other plasma lipoproteins, the clearance of [ 14 C]FC in the non-HDL fraction of the plasma was measured. As shown in Fig. 1B, the activity of [ 14 C]FC in the non-HDL fraction was relatively low and similar in mice of different genotypes, indicating that transfer of HDL [ 14 C]FC to non-HDL lipoproteins could not account for differences in the rapid disappearance of HDL [ 14 C]FC.
The FCRs of HDL FC in different mice were calculated using a two-compartment model (27) and are shown in Table I. The FCR was increased by 292% in SR-BI Tg mice and decreased by 54% in SR-BI att mice compared with WT mice (p Ͻ 0.01 for both). The FCR was increased by 66% in apoA-I0 mice (p Ͻ 0.01) compared with WT mice, reflecting the strong influence of the final time point in the model. However, when the initial rate of clearance was calculated, the value was almost the same for apoA-I0 and WT mice (15.8 versus 16.3 pools/h), but the increase in SR-BI Tg mice and the decrease in SR-BI att mice remained (by 72 and 53%; p ϭ 0.007 and 0.00001, respectively) ( Table I). The data indicate that the observed differences in HDL FC clearance are related primarily to SR-BI expression and not to HDL pool size. To assess the possibility of HDL cholesterol being rapidly taken up by the liver as ester, the esterification rate of HDL FC in plasma was measured. The esterification of FC was negligible 0.5 min after HDL injection: 0.6 and 0.4% of HDL [ 14 C]FC was esterified in WT and SR-BI Tg mice, respectively. At 4 min, only 2.1 and 1.6% of [ 14 C]FC was converted to ester, whereas by that time, Ͼ60 and 80% of [ 14 C]FC had already been removed from HDL in WT and SR-BI Tg mice, respectively (Fig. 1A). Cholesterol esterification is decreased in SR-BI Tg mice due to a functional deficiency of lecithin:cholesterol acyltransferase (19). Therefore, the rapid clearance of HDL [ 14 C]FC in SR-BI Tg mice was not a result of the hepatic uptake of esterified [ 14 C]cholesterol from HDL.
The clearance of HDL CE was measured simultaneously with the clearance of HDL FC (data not shown). Consistent with our earlier observations (19,22), the clearance of HDL [ 3 H]CE was accelerated in SR-BI Tg mice (t1 ⁄2 ϭ 0.4 h) and decreased in SR-BI att mice (t1 ⁄2 ϭ 2.4 h) compared with WT mice (t1 ⁄2 ϭ 1.1 h); the clearance of HDL [ 3 H]CE in apoA-I0 mice (t1 ⁄2 ϭ 0.9 h) was similar to that in WT mice. In all cases, HDL CE was cleared much slower than HDL FC.
Effect of SR-BI on Hepatic Uptake of HDL FC-SR-BI Tg mice overexpress SR-BI specifically in the liver, whereas SR-BI att mice have 50% decreased hepatic expression of SR-BI. To determine the effect of SR-BI expression on the hepatic uptake of HDL [ 14 C]FC from the plasma, the accumulation of [ 14 C]FC in the liver was measured. As shown in Fig. 2, ϳ76% of [ 14 C]FC was taken up by the liver in SR-BI Tg mice 1 h after HDL was injected, a 34% increase compared with WT mice (p Ͻ 0.05). In contrast, SR-BI att mice exhibited a 22% decrease in the hepatic uptake of HDL FC (p Ͻ 0.05). Compared with WT mice, the uptake of HDL [ 14 C]FC by the liver in apoA-I0 mice was not significantly different (47% versus 42%). Similarly, the hepatic uptake of HDL [ 3 H]CE was increased by 346% in SR-BI Tg mice and decreased by 63% in SR-BI att mice (p Ͻ 0.01 for both), consistent with earlier reports (19,22). The uptake of HDL CE in apoA-I0 mice was not significantly different from that in WT mice (10.3% versus 9.4%).
In vitro experiments were also performed to compare the uptake of HDL cholesterol by primary hepatocytes with different expression levels of SR-BI. The SR-BI protein levels in primary hepatocytes from SR-BI Tg mice, assessed at the end of the experiments, were ϳ6-fold higher than in hepatocytes from control mice (data not shown). During a 1-h incubation, hepatocytes in primary culture from SR-BI Tg mice exhibited a 2.2-fold increase in the uptake of HDL [ 14 C]FC compared with WT hepatocytes; the uptake of HDL [ 3 H]CE was increased by 2.6-fold in SR-BI Tg hepatocytes (Fig. 3). Thus, the cellular uptake of HDL FC and CE by hepatocytes was stimulated by increased levels of SR-BI expression. Taken together, both in vivo and in vitro data demonstrate clearly that SR-BI plays a critical role in the hepatic uptake of HDL FC as well as HDL CE from the circulation.

Effect of SR-BI on Biliary Cholesterol
Secretion-To explore the role of SR-BI in biliary cholesterol secretion, dual-labeled HDL was injected, and the transport of HDL cholesterol into bile was compared in different mice. The rise in biliary [ 14 C]FC radioactivity paralleled the fall in plasma HDL [ 14 C]FC radioactivity and reached a plateau within 17.5 min (Fig. 4A). As shown in Fig. 4, in SR-BI Tg mice, HDL [ 14 C]FC radioactivity appeared more rapidly in bile than HDL [ 3 H]CE radioactivity. At 2.5 and 7.5 min, the secretion of HDL [ 14 C]FC into bile was increased by 5.1-and 1.6-fold, respectively, compared with WT mice (p Ͻ 0.05 for both). In contrast, the initial secretion rate of cholesterol derived from HDL [ 3 H]CE was not significantly different in SR-BI Tg mice versus WT mice, but was significantly increased at later time points (Fig. 4B). SR-BI att mice tended to have decreased biliary secretion of HDL FC and CE radioactivity, although the difference between SR-BI att and WT mice was not significant (Fig. 4) (27)  The results indicate that, in mice, a significant amount of HDL CE is secreted into bile as an intact molecule, although most of HDL CE is hydrolyzed before secretion.

DISCUSSION
This study provides the first evidence for an in vivo role of SR-BI in the metabolism of HDL FC. SR-BI promotes the uptake of HDL FC by primary hepatocytes and facilitates the rapid clearance of HDL FC by the liver. The magnitude of these effects was large ( Fig. 1 and Table I), suggesting that a major part of the clearance of FC from plasma is receptor-mediated rather than passive. Furthermore, SR-BI overexpression enhances the secretion of HDL cholesterol radioactivity into bile, including both the initial secretion of HDL FC and the slower secretion of FC and CE derived from HDL CE. Our study provides strong evidence that the rapid clearance of HDL FC by the liver and its subsequent appearance in bile (5) can be mediated by SR-BI.
HDL plays a central role in cholesterol homeostasis. It has been recognized for Ͼ2 decades that HDL FC is the preferred substrate for biliary steroid (see Ref. 33 for review). Schwartz et al. (4,34) first demonstrated that the cholesterol used for biliary steroid secretion is derived substantially from plasma lipoprotein FC rather than CE. By injecting [ 3 H]FC-or [ 14 C]FC-labeled HDL and LDL simultaneously into bile fistula patients, these investigators further showed that a much larger fraction of HDL FC than LDL FC is used for biliary secretion and that the radioactivity of HDL FC appears and peaks in bile much faster than that of LDL (5,6). Similar results were obtained in squirrel monkeys (7) and rats (8,35). However, the molecular basis for this preference of HDL FC was not clear. We now provide evidence that a cell-surface HDL receptor is involved. SR-BI has high affinity for HDL (11) and greatly accelerates the transport of HDL FC across the liver into bile (Figs. 1A and 4A). SR-BI also binds to LDL and possibly other lipoproteins, but with a lower affinity (36), which may account for the less favorable utilization of FC from other lipoproteins for biliary secretion (33). Although only 4 -5% of HDL cholesterol (FC ϩ CE) radioactivity appeared in bile, based on the relatively large HDL cholesterol pool size and assuming a similar specific activity of HDL and the liver cholesterol pool that acts as a precursor of biliary cholesterol, this could be enough to explain the output of biliary cholesterol in SR-BI Tg mice, which is increased by 50%. Several studies suggested that, in rats (52) and primates (34,53,54), HDL CE makes a minor contribution to the precursor pool of biliary cholesterol. Results from our study indicate that, at least in mice, cholesterol derived from HDL CE can be secreted into bile after hydrolysis and make a contribution equivalent to that of FC. Furthermore, HDL CE can also be delivered to bile in the intact form, although, in the human, almost all biliary cholesterol exists as FC (55).
The mechanism by which SR-BI mediates the uptake of FC or CE by cells is poorly understood. The mere tethering of HDL or an acceptor to the plasma membrane by SR-BI is essential but probably not sufficient to produce maximal cholesterol transfer between the two surfaces. Comparative studies indicated that other class B scavenger receptors such as CD36 and a splicing variant of SR-BI (SR-BII) bind HDL with similar affinity, but facilitate FC efflux as well as selective lipid uptake to a much reduced extent (43)(44)(45)(46). In contrast, various PL vesicles with a wide range of affinities for SR-BI promote cholesterol efflux to a remarkably similar extent (43). On the other hand, the efflux rates of FC to HDL or PL vesicles correlate strongly with the levels of SR-BI expression in several different cell lines (12,13), indicating that the highly efficient promotion of cholesterol transfer between HDL and the cell plasma membrane is a unique property of SR-BI. Structural and functional studies from several laboratories suggest that the extracellular domain of SR-BI is critical for efficient lipid transfer (45,46), although the C-terminal cytoplasmic domain may also play an important role (46). Although the exact roles the domains play remain unclear, Rothblat and co-workers (24,43) observed recently that SR-BI increases the susceptibility of plasma membrane cholesterol to cholesterol oxidase and changes the kinetics of cholesterol efflux to cyclodextrins. It was further proposed that SR-BI may induce redistribution of cholesterol in the plasma membrane and create microdomains with different degrees of cholesterol enrichment that favor transfer of cholesterol from/to the membrane (24,43). Caveolae may be one of these candidate cholesterol-rich membrane domains since they are involved in intracellular cholesterol trafficking (47) and also mediate cholesterol efflux (48,49). SR-BI has been shown to be palmitoylated (50) and to colocalize with caveolae (44,50). Recently, Graf et al. (51) demonstrated that, during selective uptake, HDL CEth is initially transferred to caveolae and that the transfer requires SR-BI. The initial step of HDL FC transfer into bile could involve transfer of HDL FC into caveolae. SR-BI enhances the exchange of cholesterol between the surface of HDL and the cell. The direction of net flux will largely depend on the chemical gradient between the two surfaces. SR-BI may be expressed at both sinusoidal and canalicular membranes of hepatocytes (18). On the sinusoidal side of hepatocytes, SR-BI facilitates the net transfer of FC from HDL to the plasma membrane, whereas on the canalicular side, SR-BI might promote the release of FC from the membrane to such acceptors as PL vesicles. The excretion of HDL cholesterol into bile is thought to be facilitated by the biliary secretion of PL at the canalicular membrane (39), a process now known to be mediated by a PL translocase, the ATP-dependent Mdr2 P-glycoprotein (40). Also the secretion of HDL FC into bile can be stimulated by the luminal bile salt-mediated solubilization of cholesterol independent of PL secretion (41). The secretion of bile salts at the canalicular membrane was recently shown to be mediated mainly by another member of the ATP-binding cassette transporter superfamily, the sister of P-glycoprotein (42). Hepatic overexpression of SR-BI causes increased biliary secretion of cholesterol without a concomitant increase in PL or bile salt secretion (18,25). Based on the evidence available, a working model would be that, through an energy-dependent process, a membrane transporter at the canalicular membrane (a PL flippase or a bile salt export pump) generates a FC chemical gradient from blood to bile; SR-BI may promote the transport of HDL cholesterol along this gradient at the sinusoidal and/or canalicular membrane by binding HDL or a cholesterol acceptor to the membrane.
This study has provided further evidence for a major role of SR-BI in reverse cholesterol transport. HDL is the preferred acceptor of cholesterol from peripheral tissues (31,37). Our earlier work demonstrated that SR-BI mediates HDL-dependent cellular cholesterol efflux and may facilitate cholesterol efflux in the arterial wall (12,13). The current results indicate that SR-BI also stimulates the hepatic uptake of HDL FC and its transport into bile. It is possible that SR-BI catalyzes free cholesterol transfer down a gradient resulting from bile formation. SR-BI may promote reverse cholesterol transport by facilitating both the initial and final steps in the process.