Effect of up-regulating individual steps in the reverse cholesterol transport pathway on reverse cholesterol transport in normolipidemic mice.

Cholesterol acquired by extrahepatic tissues (from de novo synthesis or lipoproteins) is returned to the liver for excretion in a process called reverse cholesterol transport (RCT). We undertook studies to determine if RCT could be enhanced by up-regulating individual steps in the RCT pathway. Overexpression of 7alpha-hydroxylase, Scavenger receptor B1, lecithin:cholesterol acyltransferase (LCAT), or apoA-I in the liver did not stimulate cholesterol efflux from any extrahepatic tissue. In contrast, infusion of apoA-I.phospholipid complexes (rHDL) that resemble nascent HDL markedly stimulated cholesterol efflux from tissues into plasma. Cholesterol effluxed to rHDL was initially unesterified but by 24 h this cholesterol was largely esterified and had shifted to normal HDL (in mice lacking cholesteryl ester transfer protein) or to apoB containing lipoproteins (in cholesteryl ester transfer protein transgenic mice). Most of the cholesterol effluxed into plasma in response to rHDL came from the liver. However, an even greater proportion of effluxed cholesterol was cleared by the liver resulting in a transient increase in liver cholesterol concentrations. Fecal sterol excretion was not increased by rHDL. Thus, although rHDL stimulated cholesterol efflux from most tissues and increased net cholesterol movement from extrahepatic tissues to the liver, cholesterol flux through the entire RCT pathway was not increased.

Cholesterol that has been acquired by extrahepatic tissues (from de novo synthesis or lipoproteins) is returned to the liver for excretion in a process called reverse cholesterol transport (RCT) 1 (1)(2)(3). The first step in the RCT pathway is efflux of cholesterol from cell membranes to nascent HDL in interstitial fluid (2,3). Nascent HDL are discoidal pre-␤-migrating complexes of phospholipid and apoA-I (other amphipathic apoproteins such as apoE and apoA-IV may also be present). These particles are secreted by the liver (4,5) and small intestine (6) and are also formed during the metabolism of triglyceride-rich lipoproteins from excess surface material. In addition, lipid-free apoA-I can mediate the efflux of cholesterol and phospholipid from cells generating pre␤-migrating nascent HDL (7,8). ATP-binding cassette transporter 1 (ABCA1) appears to play a key role in this process although the exact mechanism is unclear (9 -12). Cholesterol that is transferred to nascent HDL is esterified by lecithin: cholesterol acyltransferase (LCAT) to cholesteryl esters, which by virtue of their hydrophobicity move into the core of the HDL particle resulting in the formation of ␣-migrating spherical HDL. HDL cholesteryl esters are cleared from plasma mainly by the liver (13)(14)(15)(16). HDL cholesteryl ester uptake by the liver is mediated by the scavenger receptor BI (SR-BI), which selectively transports HDL cholesteryl esters resulting in an HDL particle of reduced size and cholesteryl ester content (17,18). In species with cholesteryl ester transfer protein, a portion of HDL cholesteryl ester is transferred to lower density apoB-containing lipoproteins and ultimately cleared mainly by the liver via the LDL receptor pathway (19). The final step in the RCT pathway is excretion of cholesterol from the liver into bile, either directly or after conversion to bile acids.
Although individual steps in the RCT pathway have been studied in detail, little is known about what regulates cholesterol flux through the entire pathway. The main obstacle to such studies has been the inability to directly quantify cholesterol flux in the RCT pathway in vivo. One approach that has been used in animal models is to quantify the rate of cholesterol acquisition in extrahepatic tissues as a measure of cholesterol flux through the RCT pathway because in a steady state the rate of cholesterol acquisition by extrahepatic tissues (from de novo synthesis and lipoproteins) equals the rate of cholesterol return to the liver for excretion (with the exception of cholesterol that is converted to steroid hormones or lost when cells are sloughed from the gastrointestinal tract or skin). Studies using this approach have shown that plasma HDL cholesterol concentrations can be varied over a wide range as a result of diet or genetic manipulation with no change in the rate of reverse cholesterol transport (20 -22).
The purpose of the current studies was to determine if enhancing individual steps in the RCT pathway increases cholesterol flux through the entire pathway. The RCT pathway involves a series of steps beginning with cholesterol efflux from cells in extrahepatic tissues and ending with the excretion of this cholesterol by the liver and its loss in the feces. If one of these steps is rate-limiting, then up-regulation of this step may result in increased cholesterol flux through the entire pathway. Conversely, if all steps are functioning at maximal capacity or if cholesterol can be diverted out of the pathway at one or more points then up-regulating individual steps may not increase cholesterol flux through the RCT pathway.

EXPERIMENTAL PROCEDURES
Animals-C57BL/6 mice were purchased from Taconic. Mice with targeted disruption of the apoA-I gene (23) were purchased from Jackson * This work was supported National Institutes of Health Grants HL-38049 and HL-47551. 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.
‡ To whom correspondence should be addressed: Laboratories. Homozygous CETP transgenic (24) and hemizygous SR-BI transgenic (25) mice were generously provided by Dr. Alan R. Tall (Columbia University) and have been previously described (24,25). Animals were subjected to light cycling for at least 2 weeks before specific experiments, which were performed during the mid-dark phase of the light cycle. All experiments were performed in nonfasting male mice that were within the age range of 3-6 months. All experiments were approved by the Institutional Animal Care and Research Advisory Committee of the University of Texas Southwestern Medical Center at Dallas.
Recombinant Adenoviruses-The recombinant adenovirus AdCMV-LCAT, carrying a gene encoding mouse LCAT expressed from a human cytomegalovirus promoter, was generated by homologous recombination in 293 cells as described previously (26). The cDNA encoding murine LCAT was generously provided by Dr. Catherine Hedrick (University of Virginia). Generation of the recombinant adenovirus AdCMV-Luc, carrying a gene encoding firefly luciferase (27), AdCMV-7␣, carrying a gene encoding rat hepatic 7␣-hydroxylase (26), and AdCMV-Apo AI, carrying a gene encoding human apoA-I (28) were described previously. Large scale production of recombinant adenovirus was performed by infecting confluent monolayers of 293 cells grown in 15-cm tissue culture plates with primary stock at a multiplicity of infection of 0.1-1.0 (28). Infected monolayers were lysed with Nonidet P-40 when Ͼ90% of the cells showed cytopathic changes, and recombinant virus was purified by precipitation with polyethylene glycol 8000, centrifugation on a discontinuous CsCl density gradient, and desalting by chromatography on Sepharose CL-4B. Purified virus eluting in the void volume was collected, snap frozen in liquid N 2 , and stored at Ϫ80°C until used. Virus titer was determined by plaque assay in monolayer cultures of 293 cells.
Determination of Cholesterol Synthesis Rates-Rates of cholesterol synthesis were measured in vivo using [ 3 H]water. Mice were administered ϳ25 mCi of [ 3 H]water intraperitoneally and then returned to cages under a fume hood as previously described (16). Two hours after the injection of [ 3 H]water, the animals were anesthetized and exsanguinated through the inferior vena cava. Aliquots of plasma were taken for the determination of body water specific activity, and samples of liver and various extrahepatic tissues were taken for the isolation of digitonin-precipitable sterols. The entire remaining carcass was also taken for the isolation of digitonin-precipitable sterols. Rates of sterol synthesis are expressed as the nanomoles of [ 3 H]water incorporated into digitonin-precipitable sterols per h per g wet weight of tissue (nmol/h per g).
Determination of HDL Cholesteryl Ether Transport-Mouse HDL was isolated in the density range of 1.07-1.21 g/ml using sequential preparative ultracentrifugation and standard techniques (29) and labeled with either the intracellularly trapped [1␣,2␣-3 H]cholesteryl oleyl ether (18,30,31) or [cholesteryl-4-14 C]oleate by exchange from donor liposomes as described (13,32). Rates of HDL cholesteryl ether transport were determined using a primed infusion protocol as previously described (16). Animals were administered a priming dose of [ 3 H]cholesteryl ether-labeled HDL intravenously followed by a continuous infusion of the same radiolabeled lipoprotein at a rate determined in preliminary studies to maintain a constant plasma specific activity. The primed infusions of [ 3 H]cholesteryl ether-labeled HDL were continued for 4 h at which time each animal was administered [ 14 C]cholesteryl ester-labeled HDL intravenously (as a marker of the volume of plasma within each tissue) and sacrificed 10 min later. Plasma and tissue samples were assayed for their 3 H and 14 C content as previously described (16). The tissue spaces achieved by the labeled HDL at 10 min ( 14 C dpm/g of tissue divided by the 14 C dpm/l of plasma) and at 4 h and 10 min ( 3 H dpm/g of tissue divided by the steady-state 3 H dpm/l of plasma) were then calculated and have the units of microliters/g. The increase in tissue space over the 4-h experimental time period equals the rate of radiolabeled HDL cholesteryl ether movement into each organ and is expressed as the microliters of plasma cleared of its HDL cholesteryl ether content/h/g of tissue (16).
Determination of Fecal Sterol Excretion-Feces were collected daily for 4 days beginning immediately after the intravenous administration of rHDL. Feces were dried, weighed, and ground. A 1-g aliquot of this material was used to determine total bile acid content by an enzymatic method as previously described (33). A second 1-g aliquot was subjected to alkaline hydrolysis at 120 -130°C for 12 h. After drying the sample, 10 ml of water and 10 ml of ethanol were added. The sample was extracted in 15 ml of petroleum ether to which 1.0 mg of 5-cholestene (Sigma) had been added as an internal standard. The amount of cholesterol, coprostanol, epicoprostanol, and cholestanone in the extracts was quantified by gas chromatography (33). The daily excretion rates of both bile acid and neutral sterol are expressed as micromole/day per 100 g of body weight.
Samples of liver were homogenized in RNA STAT-60 (TEL-TEST, Inc., Friendswood, TX). Total RNA (40 g) was hybridized with 32 Plabeled riboprobes simultaneously at 68°C using the HybSpeed RPA protocol (Ambion Inc). Following RNase digestion, the mRNA-protected 32 P-labeled probes were separated on 8 M urea, 5% polyacrylamide gels together with 32 P-labeled MspI-digested pBR322 size standards. The radioactivity in each band, as well as background radioactivity, was quantified using a phosphorimaging system (Molecular Dynamics Inc., Sunnyvale, CA).
Determination of the Concentration of ApoA-I and Lipids in Plasma and Cholesterol in Liver-Plasma lipoproteins were separated by FPLC using a Superose 6 HR column (Sigma). Two-hundred-l aliquots were collected and used for apoA-I and lipid assays. The following assays were performed on plasma and FPLC column fractions: total cholesterol (Roche Molecular Biochemicals, catalogue number 1127771), free cholesterol (Wako Chemicals, USA, catalogue number 274-47109), phospholipid (Wako Chemicals, catalogue number 990-54009), and apoA-I (Sigma, catalogue number 356-A). Liver cholesterol was quantified by capillary gas-liquid chromatography.
Statistical Analysis-The data are presented as mean Ϯ 1 S.D. To test for differences among groups, one-way analysis of variance was performed. Significant results were further analyzed using the Tukey multiple comparison procedure.

RESULTS
These studies were undertaken to determine if cholesterol flux through the RCT pathway can be increased by up-regulating individual steps in the pathway. The initial step in the reverse cholesterol transport pathway is efflux of cholesterol to circulating acceptor particles. Extrahepatic tissues acquire cholesterol mainly from de novo synthesis (21). The cholesterol biosynthetic pathway is tightly regulated by cholesterol availability (34,35) and the rate of cholesterol synthesis or the cholesterol content of a tissue will be altered in response to changes in sterol influx or efflux. We therefore measured cholesterol synthesis rates and cholesterol concentrations in the extrahepatic tissues under conditions in which individual steps in the reverse cholesterol transport pathway were accelerated. Working backwards from the liver, the final step in the RCT pathway is the conversion of cholesterol into bile acids and their excretion into bile. We stimulated this step in the RCT pathway by overexpressing hepatic 7␣-hydroxylase and determined the effect on cholesterol synthesis rates and concentrations in the extrahepatic tissues. Rates of cholesterol synthesis were measured in vivo 3 days after the intravenous injection of 10 9 pfu of recombinant adenovirus expressing 7␣-hydroxylase from the CMV promoter (AdCMV-7␣) or control virus (AdCMV-Luc) into C57BL/6 mice. Administration of AdCMV-7␣ increased hepatic 7␣-hydroxylase activity by ϳ15-fold as previously described (26,36). Overexpression of hepatic 7␣-hydroxylase lowered plasma VLDL/IDL/LDL cholesterol concentrations but had little effect on plasma HDL cholesterol concentrations as shown in Fig. 1A and no effect on hepatic HDL cholesteryl ether clearance (Fig. 1B). As shown in Fig. 1C, overexpression of 7␣-hydroxylase increased hepatic cholesterol synthesis by 4.5-fold but had no effect on rates of cholesterol synthesis in the extrahepatic tissues. Although not shown, overexpression of hepatic 7␣-hydroxylase had no effect on the cholesterol content of extrahepatic tissues.
The liver takes up HDL cholesteryl esters directly (via SR-BI) or after CETP-mediated transfer to apoB-containing lipoproteins (via the LDL receptor pathway). In the mouse, which lacks CETP, HDL cholesteryl esters are delivered to the liver mainly through the SR-BI pathway. SR-BI transgenic mice have been created that manifest marked overexpression of SR-BI in the liver, rapid clearance of HDL cholesteryl ester by the liver, and a marked reduction in plasma HDL concentrations (25). We confirmed that liver-specific overexpression of SR-BI markedly lowers plasma HDL cholesterol concentrations ( Fig. 2A) as a result of enhanced clearance of HDL cholesteryl ester by the liver (Fig. 2B). The rate of hepatic HDL cholesteryl ether clearance in SR-BI transgenic mice was 21-fold higher than that of SR-BI wild type mice. The marked increase in HDL cholesteryl ether clearance by the liver was due in part to the reduced concentration of HDL cholesteryl ester in plasma; however, clearance rates were still increased 9-fold in SR-BI transgenic mice in which plasma HDL concentrations were normalized by an infusion of unlabeled mouse HDL. As shown in Fig. 2C, rates of cholesterol synthesis were not increased in the extrahepatic tissues of SR-BI transgenic mice; indeed, rates of cholesterol synthesis were significantly suppressed in several tissues. Although not shown, the cholesterol concentration in the extrahepatic tissues shown in Fig. 2C were not altered in SR-BI transgenic animals. These data indicate that liver-specific overexpression of SR-BI does not enhance reverse cholesterol transport and may reduce the rate of cholesterol efflux from some tissues into plasma. It should be noted that we did not specifically look at the adrenal glands in these studies. RCT from the adrenal glands may not be necessary as they convert cholesterol to steroid hormones. Because mouse adrenal glands derive the majority of their cholesterol from HDL, it is likely Adenovirus-mediated gene transfer was used to overexpress 7␣-hydroxylase in the livers of C57BL/6 mice. Studies were performed 3 days after the administration of adenovirus expressing 7␣-hydroxylase (AdCMV-7␣) or luciferase (AdCMV-Luc, used as a control virus). A, distribution of cholesterol among plasma lipoproteins. Plasma from the 2 groups was pooled and lipoproteins size fractionated by FPLC using a Superose 6 HR column. The retention times for mouse VLDL (d Ͻ 1.006 g/ml), IDL/LDL (d 1.02-1.055 g/ml), and HDL (1.07-1.21 g/ml) are indicated. B, hepatic HDL cholesteryl ether clearance rates. C, tissue cholesterol synthesis rates. Each value in panels B and C represents the mean Ϯ 1 S.D. for data obtained in five animals. *, differs significantly from the control value, p Ͻ 0.5. SB, small bowel.

FIG. 2. Effect of overexpressing SR-BI on plasma lipoprotein cholesterol concentrations (panel A), hepatic HDL cholesteryl ether clearance (panel B), and tissue cholesterol synthesis (panel C).
Studies were performed in transgenic mice with liver-specific overexpression of SR-BI (hemizygous) or sib controls. A, distribution of cholesterol among plasma lipoproteins. Plasma from the 2 groups was pooled and lipoproteins size fractionated by FPLC using a Superose 6 HR column. B, hepatic HDL cholesteryl ether clearance rates in SR-BI wild type (wt), SR-BI transgenic (tg), and SR-BI transgenic mice in which plasma HDL concentrations have been raised to normal values by the infusion of unlabeled mouse HDL. C, tissue cholesterol synthesis rates. Each value in panels B and C represents the mean Ϯ 1 S.D. for data obtained in five animals. *, differs significantly from the control value, p Ͻ 0.5. SB, small bowel. that cholesterol synthesis is increased in the adrenal glands of SR-BI transgenic mice to compensate for the decrease in HDL cholesteryl ester uptake that results from the near absence of HDL in the plasma of these mice (25).
Free cholesterol that is transferred from tissues to nascent discoidal HDL is esterified by LCAT leading to the generation of mature spherical HDL. We accelerated this step in the RCT pathway by overexpressing mouse LCAT in the liver using adenovirus-mediated gene transfer and determined the effect on cholesterol synthesis rates and concentrations in the extrahepatic tissues. Rates of cholesterol synthesis were measured 3 days after the intravenous injection of 1 ϫ 10 9 or 2 ϫ 10 9 pfu of recombinant adenovirus expressing mouse LCAT (AdCMV-LCAT) or control virus into C57BL/6 mice. Administration of AdCMV-LCAT resulted in the accumulation of large amounts of LCAT mRNA in the liver (not shown) and in the accumulation of cholesteryl ester-rich HDL particles in plasma that contained mainly apoA-I and apoE (Fig. 3A). However, overexpression of LCAT in the liver had no significant effect on the rate of cholesterol synthesis (Fig. 3, B and C) or the cholesterol concentration (not shown) in extrahepatic tissues.
The initial step in RCT is cholesterol efflux from the plasma membranes of cells in extrahepatic tissues to nascent HDL particles in the interstitial space. We employed two protocols aimed at increasing the availability of nascent HDL. In the first we increased plasma apoA-I concentrations by overexpressing apoA-I in the liver using adenovirus-mediated gene transfer. ApoA-I Ϫ/Ϫ mice were administered 10 9 pfu of recombinant adenovirus expressing human apoA-I from the CMV promoter (AdCMV-apoA-I) or control virus, and the plasma concentration of apoA-I, cholesterol, and phospholipid was determined at 24-h intervals for 3 days. As shown in Fig. 4A, administration of AdCMV-apoA-I increased plasma apoA-I concentrations from 0 to 309 mg/dl; plasma HDL phospholipid and cholesterol concentrations increased to 221 and 119 mg/dl, respectively, with 91% of HDL cholesterol being esterified. In the second protocol, animals were infused with discoidal complexes of phospholipid and apoA-I. These particles (referred to as rHDL) contained human apoA-I and phospholipid in a ratio of ϳ1:5 (w/w) (37) and were generously provided by Dr. Peter G. Lerch, ZLB Central Laboratory, Swiss Red Cross, Bern, Switzerland. ApoA-I Ϫ/Ϫ mice were administered a primed infusion of rHDL at a rate of 1 mg of rHDL apoA-I/h for 3 days through an in-dwelling internal jugular vein catheter that exited the body through a tether and was attached to a fluid swivel. This system allowed the animals free range of motion and access to food and water. As shown in Fig. 4B plasma apoA-I concentrations were similar to those in animals administered AdCMV-apoA-I, whereas plasma phospholipid concentrations were much higher (ϳ1,400 mg/dl). Plasma cholesterol concentrations increased during the 3-day infusion to nearly 1,000 mg/dl with 87% of the cholesterol that accumulated in plasma being unesterified. Fig. 5 shows rates of cholesterol synthesis in the extrahepatic tissues of apoA-I Ϫ/Ϫ mice after 3 days of overexpressing apoA-I or infusing rHDL. As shown in Fig. 5A, overexpressing apoA-I had no significant effect on cholesterol synthesis in any extrahepatic tissue. In contrast, infusion of rHDL markedly increased rates of cholesterol synthesis in most of the extrahepatic tissues that were examined as well as the remaining carcass (Fig. 5B). The greatest relative increases in cholesterol synthesis were seen in lung (6-fold), heart (9-fold), and skeletal muscle (10-fold). Tissue cholesterol concentrations were measured at 3 days and were not altered by the overexpression of apoA-I (not shown) or the infusion of rHDL (Fig. 5C). These studies show that for the same elevation in plasma apoA-I concentrations, an infusion of apoA-I⅐phospholipid complexes is far more effective than overexpressing apoA-I at stimulating cholesterol movement from tissues to plasma.
Further studies were undertaken to characterize the effect of an intravenous bolus of rHDL on plasma lipids and parameters of RCT. We performed these studies in apoA-I Ϫ/Ϫ mice, which have very low background levels of total and HDL cholesterol, and in CETP transgenic mice, which more closely mimic the human situation. Fig. 6 shows changes in plasma lipid and human apoA-I concentrations following the intravenous injection of 6 mg of rHDL apoA-I to apoA-I Ϫ/Ϫ or CETP transgenic mice. In apoA-I Ϫ/Ϫ mice, plasma concentrations of human  (panels B and C). Adenovirus-mediated gene transfer was used to overexpress LCAT in the livers of C57BL/6 mice. Studies were performed 3 days after the administration of adenovirus expressing 1 ϫ 10 9 or 2 ϫ 10 9 pfu mouse LCAT (AdCMV-LCAT) or luciferase (AdCMV-Luc, used as a control virus). A, distribution of cholesterol among plasma lipoproteins. Plasma from each group was pooled and lipoproteins size fractionated by FPLC using a Superose 6 HR column. The FPLC profiles from the 2 luciferase groups were superimposible so mean values are shown. Equal volumes of FPLC fractions corresponding to HDL (fractions 16 -30, 26 -40 and 31-42 for high dose AdCMV-LCAT, low dose AdCMV-LCAT, and control, respectively) were pooled, delipidated, and equal amounts of protein separated on 2-15% gradient polyacrylamide gels. B, tissue cholesterol synthesis rates in animals administered 1 ϫ 10 9 pfu AdCMV-LCAT or AdCMV-Luc. C, tissue cholesterol synthesis rates in animals administered 2 ϫ 10 9 pfu Ad-CMV-LCAT or AdCMV-Luc. Each value in panels B and C represents the mean Ϯ 1 S.D. for data obtained in five animals. *, differs significantly from the control value, p Ͻ 0.5. SB, small bowel. apoA-I and phospholipid equaled 1,662 and 429 mg/dl 1 h after administration of rHDL and had returned to near normal levels by 24 h (Fig. 6A). As shown in Fig. 6C, plasma cholesterol concentrations peaked at ϳ6 h at which time 88% was unesterified. Free cholesterol concentrations returned to near normal values by 24 h although total cholesterol (and thus esterified cholesterol) was still elevated. Basal plasma lipid levels were higher in CETP transgenic than in apoA-I Ϫ/Ϫ mice; however, changes in plasma lipid and human apoA-I concentrations after rHDL (Fig. 6, B and D) were similar to those in apoA-I Ϫ/Ϫ mice. Also shown in Fig. 6 are the changes in liver cholesterol concentrations following the administration of rHDL. In both animal models, liver cholesterol concentrations were significantly reduced 1 h after the administration of rHDL. Liver cholesterol concentrations returned to normal values by 6 h and were significantly elevated at 24 and 48 h before returning again to normal levels at 72 h. Cholesterol concentrations were significantly reduced in many extrahepatic tissues 1-2 h after rHDL administration but were subsequently normal at 6, 24, 48, and 72 h (data not shown). Fig. 7 shows the lipoprotein distribution of plasma cholesterol after administration of 6 mg of rHDL apoA-I. Six hours after the injection of rHDL, cholesterol (88% of which was unesterified) was mainly present in particles having the size of VLDL and IDL/LDL in both apoA-I Ϫ/Ϫ and CETP transgenic mice. By 24 h, cholesterol had shifted into particles having the size of HDL in apoA-I Ϫ/Ϫ mice (Fig. 7A), but remained in VLDL-and IDL/LDL-sized particles in CETP transgenic mice (Fig. 7B). By 48 h, the lipoprotein distribution of plasma cholesterol had returned to that of control animals in both animal models (data not shown). These changes in plasma lipid concentrations are similar to those reported in humans administered rHDL (37).
Rates of tissue cholesterol synthesis were measured in parallel groups of apoA-I Ϫ/Ϫ and CETP transgenic mice 6, 24, and 48 h after the injection of 6 mg of rHDL apoA-I. In both apoA-I Ϫ/Ϫ mice (Fig. 8A) and CETP transgenic mice (Fig. 8B), rates of hepatic cholesterol synthesis were markedly elevated (2-3-fold) at 6 h, were suppressed (Ͻ50% of control values) at 24 h, and had returned to normal levels by 48 h after rHDL administration. Rates of cholesterol synthesis in the extrahepatic tissues generally showed a similar pattern of changes in response to a bolus of rHDL. In most tissues rates of cholesterol synthesis were markedly elevated at 6 h, were suppressed below control values at 24 h, and had returned to normal levels by 48 h. Tissue cholesterol synthesis rates at 72 h did not differ significantly from those at 48 h and time 0 (not shown). LDL receptor, SR-BI, and ABCA1 mRNA levels were measured by RNase protection in liver, lung, spleen, heart, and muscle of individual animals (5 mice/group) at 6, 24, 48, and 72 h after rHDL administration as described under "Experimental Procedures." There were no significant changes in mRNA levels for the LDL receptor, SR-BI, or ABCA1 at any of these time points (data not shown). The data in Fig. 8 indicate that rHDL stimulates cholesterol efflux from most tissues of the body. However, when cholesterol synthesis rates are expressed per whole organ, it becomes apparent that much of the cholesterol entering plasma after rHDL administration is derived from the liver. Fig. 9 shows the changes in cholesterol synthesis rates in liver, extrahepatic tissues, and whole body after rHDL administration. In apoA-I Ϫ/Ϫ mice (Fig. 9A), total body cholesterol synthesis increased 48% (13.7-20.6 mol/h per 100 g) and hepatic cholesterol synthesis increased 77% (5.7-10.1 mol/h per 100 g) 6 h after the administration of rHDL. It can therefore be calculated that the liver accounted for 64% of the increase in total body cholesterol synthesis in response to rHDL. A similar pattern was seen in the CETP transgenic mice (Fig. 9B) where the liver accounted for 50% of the increase in total body cholesterol synthesis in response to rHDL. Cholesterol synthesis was suppressed in many tissues 24 h after rHDL administration as cholesterol that effluxed into plasma was cleared from plasma. In apoA-I Ϫ/Ϫ mice (Fig. 9A), total body cholesterol synthesis was suppressed 33% below control values (13.6 to 9.1 mol/h per 100 g) and hepatic cholesterol synthesis was suppressed 61% (5.7 to 2.2 mol/h per 100 g) 24 h after rHDL administration. Thus, the liver accounted for 78% of the decrease in total body cholesterol synthesis at 24 h. A similar pattern was seen in CETP transgenic mice where the liver accounted for 79% of the decrease in total body cholesterol synthesis at 24 h.
To determine if rHDL administration increased net cholesterol flux through the entire RCT pathway we performed sterol balance studies in apoA-I Ϫ/Ϫ and CETP transgenic mice after rHDL administration. Feces were collected daily (for 4 days) from groups of 5 animals beginning immediately after the intravenous administration of 6 mg of rHDL apoA-I. As shown in Fig. 10, fecal neutral sterol and bile acid output was not affected by rHDL in either mouse model. In other studies mice received daily intravenous injections of rHDL (6 mg of rHDL apoA-I/dose) for 7 days. Feces were collected daily for the determination of bile acids and neutral sterols beginning 2 days before the first injection and ending 2 days after the last injection. These studies showed no effect of rHDL on external sterol balance. 2 Hepatic cholesterol 7␣-hydroxylase is the rate-limiting enzyme in the main bile acid biosynthetic pathway. We measured 7␣-hydroxylase mRNA levels by RNase protection at times 0, 6, 24, 48, and 72 h after rHDL administration and found no significant effect of rHDL at any time point as shown in Table I. DISCUSSION The main objective of these studies was to determine if up-regulating individual steps in the RCT pathway leads to increased cholesterol flux through the entire RCT pathway. The RCT pathway encompasses a series of steps beginning in extrahepatic tissues with the efflux of cholesterol to nascent HDL and culminating in the liver and gastrointestinal tract with the secretion of sterols into bile and their elimination in feces. We show that RCT is not significantly enhanced in mice under conditions in which each of the major steps in the RCT pathway is markedly accelerated. These observations suggest that it will be necessary to up-regulate multiple (possibly all) steps in the RCT pathway if RCT is to be significantly increased. The initial step in RCT can be markedly accelerated by the intravenous administration of apoA-I⅐phospholipid complexes (rHDL) that resemble nascent HDL. Administration of rHDL induced a rapid increase in the plasma concentration of cholesterol (initially unesterified) that was accompanied by a transient decrease in the cholesterol content of many tissues. Enhanced efflux of cholesterol from tissues to plasma triggered a marked increase in tissue cholesterol synthesis that was apparently fully compensatory because tissue cholesterol concentrations were restored to normal values and tissue LDL receptor and SR-BI mRNA levels remained unchanged. The greatest relative increases in cholesterol synthesis (and thus cholesterol efflux) in response to a bolus of rHDL occurred in heart, skeletal muscle, lung, spleen, and liver. Small intestine was relatively resistant to rHDL and preliminary studies showed the brain to be completely unresponsive to rHDL. Overall, total body cholesterol synthesis increased ϳ50% 6 h after the administration of rHDL and the liver accounted for approximately half (CETP transgenic mice) to two-thirds (apoA-I Ϫ/Ϫ mice) of this increase. Thus, the liver was the single most important source of cholesterol entering plasma after the administration of rHDL.
While the tissue source of cholesterol entering plasma in response to rHDL can be determined quantitatively, the precise destination of this cholesterol as it is cleared from plasma is less certain. Plasma cholesterol concentrations peaked ϳ6 h after a bolus of rHDL and at this time most of the cholesterol was unesterified. By 24 h the majority of plasma cholesterol was esterified and present in either HDL (in mice lacking CETP) or LDL (CETP transgenic mice). It is likely that most of this cholesterol was transported to the liver since the liver is the major site for the clearance of cholesterol carried in both HDL (16,18) and LDL (38). However, cholesterol synthesis was suppressed in many extrahepatic tissues at 24 h indicating that these tissues contributed to the clearance of cholesterol that had entered plasma in response to rHDL. Overall, total body cholesterol synthesis was suppressed ϳ33% 24 h after rHDL administration and the liver accounted for nearly 80% of this decrease. Thus, the majority of cholesterol that effluxed into plasma in response to rHDL came from the liver; however, an even greater proportion of this cholesterol may have returned to the liver and if so this would result in a net flux of cholesterol from extrahepatic tissues to the liver. This is supported by the finding that hepatic cholesterol concentrations increased significantly in both apoA-I Ϫ/Ϫ and CETP transgenic mice as effluxed cholesterol was cleared from plasma whereas the cholesterol concentration in extrahepatic tissues remained constant during this time. The current studies were performed in mice maintained on a low cholesterol diet. Under these conditions, basal rates of hepatic cholesterol synthesis are high and the liver can compensate for changes in cholesterol flux by adjusting the rate of de novo cholesterol synthesis (39). It will be important to determine the effects of rHDL under conditions in which hepatic cholesterol synthesis is suppressed by dietary cholesterol or inhibitors of cholesterol synthesis.
Although rHDL clearly mobilized cholesterol from extrahepatic tissues and resulted in the net movement of cholesterol from extrahepatic tissues to the liver, there was no induction of hepatic 7␣-hydroxylase expression or increase in fecal sterol excretion in either apoA-I Ϫ/Ϫ or CETP transgenic mice in response to rHDL administration. These observations, which in- dicate that rHDL administration did not increase cholesterol flux through the entire RCT pathway, contrast with those of Eriksson et al. (40) who reported that fecal sterol excretion was markedly increased in 4 patients with heterozygous FH after the administration of pro-apoA-I⅐phospholipid complexes. At this point we have no adequate explanation for these apparently conflicting observations. The relative dose of rHDL used in our studies was higher than that used in the study by Eriksson et al. (40) resulting in a greater movement of cholesterol from tissues into plasma; however, this should increase the likelihood of detecting an effect of rHDL on fecal sterol excretion if such an effect exists. It is possible that there are inherent species differences in response to rHDL or significant differences in the rHDL preparations (pro-apoA-I versus apoA-I) used in the two studies.
To investigate the role of phospholipid in cholesterol efflux in vivo, we quantified rates of cholesterol acquisition in the extrahepatic tissues of animals overexpressing apoA-I or infused with rHDL. Plasma apoA-I concentrations were similar in the two groups whereas plasma phospholipid concentrations were much higher in those infused with rHDL. Overexpression of apoA-I raised HDL cholesterol concentrations but had no detectable effect on cholesterol efflux from any tissue. In contrast, infusion of rHDL markedly increased cholesterol efflux from most extrahepatic tissues. These results emphasize the important role of phospholipid in promoting cholesterol efflux in vivo. Raising plasma HDL concentrations by overexpressing apoA-I has been shown to protect against atherosclerosis in mouse models (41,42). Our results suggest that the protective effect of apoA-I overexpression is not related to enhanced RCT. It should be noted that free apoA-I-mediated cholesterol efflux has been demonstrated most consistently in cholesterol-loaded macrophages (3,7,8), a cell type that was not evaluated in our studies. It is possible that overexpression of apoA-I results in efflux of cholesterol from foam cells in the arterial wall without causing efflux from any other tissue but this seems unlikely. While the infusion of nascent HDL-like particles markedly increased cholesterol efflux from most extrahepatic tissues, up-regulating other steps in the RCT pathway did not. Thus, liver-specific overexpression of LCAT, SR-BI, or cholesterol 7␣-hydroxylase did not induce cholesterol efflux from any extrahepatic tissue. LCAT plays a key role in the RCT pathway by catalyzing the esterification of free cholesterol that has been transferred from cell membranes to nascent HDL. The role of LCAT in RCT has been studied in transgenic mice overexpressing human LCAT. Francone et al. (43) showed that HDL from LCAT transgenic mice promoted cholesterol efflux from cells more efficiently than control HDL. Moreover, HDL cholesteryl ester flux to the liver was increased in LCAT transgenic compared with control mice (43) suggesting that LCAT overexpression enhanced RCT. In contrast, Berard et al. (44) concluded that RCT was decreased in human LCAT transgenic mice based on turnover studies that suggested the increased HDL cholesteryl ester levels in these animals was the result of decreased clearance by the liver. In the current work overexpression of mouse LCAT increased HDL cholesteryl ester concentrations in a dose-dependent fashion but did not induce cholesterol efflux from any extrahepatic tissue indicating no significant effect on RCT. We have previously shown that the HDL cholesteryl ester uptake pathway in the liver is saturated at normal HDL concentrations in the mouse (16) and hamster (15). Thus increasing plasma HDL cholesteryl ester concentrations to supernormal levels (by infusing normal HDL or LCAT overexpression) cannot increase HDL cholesteryl ester delivery to the liver in these species unless SR-BI or CETP expression is increased (45).
Liver-specific overexpression of SR-BI markedly increases HDL cholesteryl ester clearance by the liver suggesting enhanced RCT (25). If overexpression of SR-BI in the liver does increase the rate of RCT, cholesterol synthesis will increase in extrahepatic tissues to balance the increased flux of cholesterol from these tissues to the liver or the cholesterol content of extrahepatic tissues will fall. However, we found no change in the cholesterol content or increase in rates of cholesterol synthesis in the extrahepatic tissues of SR-BI transgenic mice; indeed, synthesis rates were reduced in several extrahepatic tissues. These observations indicate that liver-specific overexpression of SR-BI does not increase cholesterol flux through the RCT pathway and may inhibit RCT from some tissues. There are several possible reasons why hepatic overexpression of SR-BI failed to enhance RCT. First, although liver-specific overexpression of SR-BI markedly increases HDL cholesteryl FIG. 10. Fecal bile acid and neutral sterol output after rHDL administration. ApoA-I Ϫ/Ϫ and CETP transgenic mice were administered rHDL (6 mg of apoA-I) by intravenous injection. Feces were collected daily for 4 days from groups of mice and fecal bile acids and neutral sterols quantified as described under "Experimental Procedures." Each value for apoA-I knockout mice represents the mean from 2 pools of five animals each. Each value for CETP transgenic mice represents the mean Ϯ 1 S.D. from data obtained in 3 pools of five animals each. Animals received rHDL (6 mg of apoA-I) by intravenous injection. Hepatic mRNA for 7␣-hydroxylase was determined by RNase protection assay using glyceraldehyde-3-phosphate dehydrogenase as an internal control. Total RNA (40 g) was hybridized with 32 P-labeled riboprobes and the mRNA-protected 32 P-labeled probes were separated by polyacrylamide gel electrophoresis. The radioactivity in each band was quantified using a phosphorimaging system. Data are expressed relative to the control value at time 0. Each value represents the mean Ϯ 1 S.D. for data obtained in 5 animals. ester clearance by the liver (25), this results in the virtual elimination of HDL from plasma. Thus, the absolute rate of HDL cholesteryl ester transported to the liver (clearance multiplied by the plasma concentration) may be normal or even decreased in SR-BI transgenic mice. In addition, the very low plasma HDL concentration in SR-BI transgenic mice is associated with decreased LCAT activity, which may impair RCT (25). It is possible that more modest overexpression of SR-BI in the liver may result in enhanced RCT; however, we have been unable to demonstrate this using adenovirus-mediated gene transfer. 3 Liver-specific overexpression of SR-BI (ϳ10-fold) markedly decreased fatty streak development in heterozygous LDL receptor-deficient mice fed an atherogenic diet but this protective effect was likely due to a reduction in the plasma concentration of apoB-containing lipoproteins (46). This conclusion is supported by the observation that liver-specific overexpression of SR-BI (ϳ10-fold) did not decrease the plasma concentration of apoB-containing lipoproteins or inhibit fatty streak development in human apoB transgenic mice fed an atherogenic diet (47). On the other hand, modest overexpression of SR-BI (ϳ2-fold) did inhibit fatty streak development without significantly reducing non-HDL cholesterol levels suggesting that SR-BI can exert an anti-atherogenic effect that is independent of changes in plasma non-HDL cholesterol concentrations and that this protective effect is not directly proportional to the level of SR-BI expression (47). Further studies will be required to sort out the complex inter-relationship among hepatic SR-BI expression, plasma lipoprotein concentrations, RCT, and atherogenesis.
In summary, we show that up-regulation of individual steps in the RCT pathway does not increase cholesterol flux through the entire pathway. These observations suggest that it will be necessary to up-regulate multiple (possibly all) steps in the RCT pathway to significantly increase cholesterol flux from extrahepatic tissues to the liver and into the stool. Although cholesterol flux through the entire RCT pathway is not increased, infusion of nascent HDL-like complexes markedly stimulates cholesterol efflux from most tissues into plasma and may increase net cholesterol flux from extrahepatic tissues to the liver. These findings raise the possibility that rHDL could stimulate cholesterol efflux from foam cells in the arterial wall and induce regression of atherosclerotic plaque. Phospholipidcontaining acceptor particles can mobilize cholesteryl esters from cholesterol-loaded macrophages in vitro despite an increase in de novo cholesterol synthesis (48). Moreover, intravenous administration of phospholipid-containing complexes has been shown to induce regression of pre-existing atherosclerosis in animal models (49 -51). Although many questions remain, enhancing cholesterol efflux from the arterial wall is an attractive approach that would complement current strategies that are directed primarily at reducing cholesterol influx into the arterial wall.