A Role for Hepatic Scavenger Receptor Class B, Type I in Decreasing High Density Lipoprotein Levels in Mice That Lack Phosphatidylethanolamine N-Methyltransferase*

Phosphatidylethanolamine N-methyltransferase (PEMT) is a liver-specific enzyme that converts phosphatidylethanolamine to phosphatidylcholine (PC). Mice that lack PEMT have reduced plasma levels of PC and cholesterol in high density lipoproteins (HDL). We have investigated the mechanism responsible for this reduction with experiments designed to distinguish between a decreased formation of HDL particles by hepatocytes or an increased hepatic uptake of HDL lipids. Therefore, we analyzed lipid efflux to apoA-I and HDL lipid uptake using primary cultured hepatocytes isolated from Pemt+/+ and Pemt–/– mice. Hepatic levels of the ATP-binding cassette transporter A1 are not significantly different between Pemt genotypes. Moreover, hepatocytes isolated from Pemt–/– mice released cholesterol and PC into the medium as efficiently as did hepatocytes from Pemt+/+ mice. Immunoblotting of liver homogenates showed a 1.5-fold increase in the amount of the scavenger receptor, class B, type 1 (SR-BI) in Pemt–/– compared with Pemt+/+ livers. In addition, there was a 1.5-fold increase in the SR-BI-interacting protein PDZK1. Lipid uptake experiments using radiolabeled HDL particles revealed a greater uptake of [3H]cholesteryl ethers and [3H]PC by hepatocytes derived from Pemt–/– compared with Pemt+/+ mice. Furthermore, we observed an increased association of [3H]cholesteryl ethers in livers of Pemt–/– compared with Pemt+/+ mice after tail vein injection of [3H]HDL. These results strongly suggest that PEMT is involved in the regulation of plasma HDL levels in mice, mainly via HDL lipid uptake by SR-BI.

found in plasma lipoproteins, bile and lung surfactant (1). All mammalian cells can synthesize PC from choline via the Kennedy (CDP-choline) pathway (2). The rate-limiting enzyme for this synthetic route is CTP:phosphocholine cytidylyltransferase (3). In the liver, an alternative pathway contributes to the biosynthesis of PC. This pathway involves three successive methylations of phosphatidylethanolamine (PE) catalyzed by PE N-methyltransferase (PEMT) and represents one-third of hepatic PC production (4,5). The liver-specific expression of this enzyme has suggested several roles for PEMT-derived PC in hepatic metabolism, including its involvement in lipoprotein metabolism (5)(6)(7)(8) and bile secretion (9,10). In 1997, a mouse lacking PEMT was generated to gain further insight into the function of this enzyme. When fed a chow diet, Pemt Ϫ/Ϫ mice appear normal, and the production of PC from the Kennedy pathway is increased by ϳ50% to compensate for the absence of the PEMT activity (11). Indeed, hepatic levels of PC do not change significantly between the Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice. However, when Pemt Ϫ/Ϫ mice are fed a choline-deficient diet to attenuate the CDP-choline pathway, severe liver damage occurred within 3 days (12). The liver failure can be reversed if choline is fed to the mice prior to the third day of the cholinedeficient diet (13). Interestingly, neither dietary dimethylethanolamine (14) nor propanolamine (15) can substitute for dietary choline when fed to Pemt Ϫ/Ϫ mice, indicating a striking specificity for PC in liver function. It is believed that the PEMT pathway survived throughout evolution to provide PC in situations where choline intake is insufficient, such as pregnancy/ lactation or starvation (12).
More recently, the role of PEMT in very low density lipoprotein metabolism was examined. PEMT deficiency in male mice challenged with a high fat/high cholesterol diet resulted in a 50% decrease in apoB100, PC and triacylglycerol (TG) secretion from the liver (9). These observations are in agreement with earlier work in which levels of secreted apoB100 and lipids (PC and TG) by hepatocytes from Pemt Ϫ/Ϫ mice were 50 -70% lower than from Pemt ϩ/ϩ mice (16). This defect in very low density lipoprotein secretion in the absence of PEMT activity was only evident in males fed an HF/HC diet, showing a genderand diet-specific alteration of lipoprotein metabolism in Pemt Ϫ/Ϫ mice (16).
Several adaptive mechanisms might be involved in regulating PC homeostasis in Pemt Ϫ/Ϫ mice, such as redistribution of cellular PC pools (17), reduction in PC degradation, and/or regu-lation of lipid flux in and out of hepatocytes to maintain PC levels constant and ensure optimal liver functions. In this respect, a significant reduction of cholesterol and PC in plasma HDL is observed in both genders of Pemt Ϫ/Ϫ mice fed an HF/HC diet. Surprisingly, these lower HDL levels were also detected in female Pemt Ϫ/Ϫ mice fed a chow diet (18). This decrease in plasma HDL lipids is intriguing, particularly since a role for PEMT in HDL metabolism has not been previously addressed.
In the last decade, many studies have focused on HDL formation, maturation, and delivery of its associated lipids to the liver. This reverse cholesterol transport (19 -21) is believed to be atheroprotective, since HDL mediates the transport of excess cholesterol from peripheral tissues back to the liver for conversion into bile acids and/or for excretion into bile (22)(23)(24). The generation of HDL particles from poorly lipidated apoA-I involves the efflux of cellular lipids and relies on the activity of ABCA1 (ATP-binding cassette transporter A1). In addition, partially lipidated "nascent" HDL can acquire unesterified cholesterol in an ABCA1-independent manner via other ABC transporters, such as ABCG1 or ABCG4, or via a passive diffusion process (25)(26)(27). Because hepatocytes secrete apoA-I, express ABCA1 on their cell surface, and can generate HDL by both ABCA1-dependent and -independent pathways (28, 29), we considered the possibility that ABCA1-mediated efflux of PC and cholesterol from PEMT-deficient hepatocytes might be decreased as a mechanism for maintaining PC homeostasis in the liver.
It is also now well established that lipids associated with HDL can be delivered to steroidogenic tissues (adrenals, ovary, or testes) and to the liver via the scavenger receptor class B, type I (SR-BI) (30). This HDL receptor is mainly expressed on the cell surface of hepatocytes and steroidogenic cells and is involved in the selective uptake of lipids from plasma lipoproteins, without concomitant degradation of the whole lipoprotein particle (31). Thus, cholesteryl esters and other lipids are segregated from apoA-I, which is recycled back to the plasma compartment, whereas lipids are thought to be targeted to the canalicular membrane for biliary secretion. Although SR-BI has also been described as a promoter of cholesterol efflux (32), the absence of a cholesterol efflux defect from SR-BI-deficient mouse macrophages to HDL suggests SR-BI has antiatherosclerotic effects for reasons other than cholesterol efflux (33). Currently, SR-BI is believed to play a crucial role in the final step of reverse cholesterol transport (i.e. the selective uptake and sorting of plasma HDL lipids) (32,34,35). Thus, another explanation for the reduced levels of plasma HDL in Pemt Ϫ/Ϫ mice might be a compensatory up-regulation of selective lipid uptake from HDL by hepatocytes in an attempt to maintain PC levels.
Thus, using Pemt Ϫ/Ϫ mice as a model, we have studied the hepatic contribution of PC and cholesterol to HDL formation and subsequent lipid uptake by the liver under conditions for which PC biosynthesis was challenged. The results show that in livers and hepatocytes from Pemt Ϫ/Ϫ mice, compared with Pemt ϩ/ϩ mice, the expression of SR-BI is increased by 50%, and the uptake of cholesterol is enhanced.

EXPERIMENTAL PROCEDURES
Materials-HDL (d ϭ 1.07-1.21 g/ml) were obtained by ultracentrifugation of pooled plasma of healthy male and female volunteers (36). The protein fraction of HDL was obtained by delipidation of HDL, and purified apoA-I was obtained by chromatography on DEAE-cellulose (37 . Cholesterol, cholesteryl oleate ester standards, 6-amino-n-hexanoic acid, and dodecyl maltoside were obtained from Sigma. The BCA protein kit, including the bovine serum albumin standard, was obtained from Pierce. Rabbit polyclonal anti-mouse SR-BI (NB 400-101), goat polyclonal anti-SR-BI (NB 400-131), rabbit polyclonal anti-PDZK1 (NB 400-149), and rabbit polyclonal anti-human ABCA1 antibodies were obtained from Novus Biologicals (Littleton, CO). Rabbit polyclonal anti-human apoA-I antibodies were from Biodesigns (Kennebunk, ME). Rabbit polyclonal anti-rat protein-disulfide isomerase antibody was from StressGen (Victoria, Canada). Goat anti-rabbit IgG secondary antibody conjugated to horseradish peroxidase was from Pierce, and the enhanced chemiluminescence detection system was from Amersham Biosciences (Buckinghamshire, UK). Cholesterol and choline-containing phospholipids were analyzed by fast protein liquid chromatography using the Infinity Cholesterol reagent from ThermoDMA (Calgary, Canada) and Phospholipids B kit from Wako Diagnostics (Richmond, VA), respectively. All other chemicals and reagents were from standard commercial sources. Reagents for quantitative PCR analysis of mRNA levels were purchased from Invitrogen. The molecular weight markers for the blue native PAGE were from Amersham Biosciences.
Care and Feeding of Mice-Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice had a mixed genetic background of 129/J and C57BL/6 (11) and were maintained via homozygous breeding in a reversed 12-h light/ dark cycle. All mice were females 10 -14 weeks old. Similar experiments were performed using male mice with comparable outcomes. Since the difference in HDL between the genotypes was greater with females than males (18), the studies concentrated on female mice.
Lipid Mobilization Assays-Hepatocytes from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice were incubated for 16 h at 37°C in the presence of 5 Ci of [ 3 H]glycerol, 1 mM glycerol with or without 20 g/ml apoA-I. The media were collected, cells were scraped into phosphate-buffered saline, and lipids were extracted with CHCl 3 /CH 3 OH (2:1) and then washed with (CH 3 OH/H 2 O/ CHCl 3 /CH 3 COOH (480:470:30:9.6, v/v/v/v) (38). The solvents were evaporated under nitrogen, and lipids were dissolved in CHCl 3 . Phospholipids were separated by thin layer chromatography using a developing solvent of CHCl 3 /CH 3  Analysis of Plasma Lipoprotein Lipid Profiles-Lipoproteins in 12 l of plasma were separated isocratically on a 30 ϫ 1-cm Superose 6 gel filtration fast protein liquid chromatography column (Amersham Biosciences). Eluted fractions were mixed with Infinity cholesterol reagent (ThermoDMA), and cholesterol content was measured at 37°C by spectrophotometric detection at 500 nm. Choline-containing phospholipids were measured postcolumn using the Phospholipids B kit from Wako Diagnostics.
Isolation and Radiolabeling of HDL-Radiolabeled cholesteryl oleoyl ether (50 Ci) and PC (50 Ci) were sonicated in a 37°C water bath in 1 ml of phosphate-buffered saline for 10 min. The donor liposomes were incubated with 25 mg of HDL and cholesteryl ester transfer protein; fresh lipoprotein-deficient human plasma (d Ͼ 1.21 g/ml) was heated at 60°C to inactivate lecithin:cholesterol acyltransferase and was used as a source of cholesteryl ester transfer protein (39). The lipoprotein preparation was mixed at 37°C for 48 h on an orbital shaker at 200 rpm, and HDLs were reisolated by sequential centrifugations (36). The [ 3 H]HDL was allowed to percolate through a heparin-Sepharose column to remove apoE/apoB-containing lipoproteins. After extensive dialysis, the protein concentration was determined (40).

[ 3 H]HDL was kept at 4°C under nitrogen.
Lipid Uptake by Hepatocytes-Hepatocytes from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice were incubated 1 h at 37°C in Dulbecco's modified Eagle's medium containing 75 g of protein/ml of [ 3 H]HDL. The radioactive media were removed, and then cells were washed and incubated with 50 g of protein/ml of unlabeled HDL to eliminate nonspecific binding of HDL to cellular membranes. Cells were scraped, and the lipids were extracted and separated by thin layer chromatography as described above. Radioactivity associated with cell extracts (as [ 3

H]PC and [ 3 H]cholesteryl ether) was measured with a liquid scintillation counter.
In Vivo Lipid Uptake Assays-Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice were fasted overnight, after which they were injected via the tail vein with 200 l of [ 3 H]HDL (ϳ500 g of [ 3 H]HDL protein containing 2000 dpm/g of [ 3 H]cholesteryl ether and 750 dpm/g of [ 3 H]PC). 1 h after the injection, the mice were sacrificed, the livers were removed and homogenized, lipids were extracted, and the associated radioactivity was measured.
Immunoblot Analyses-Equal amounts of proteins (from cellular extracts and from liver homogenates) were separated by electrophoresis on polyacrylamide gels (containing 0.1% SDS) (6% gel for ABCA1 analysis, 12.5% gel for SR-BI, and 10% for PDZK1), and the resolved proteins were transferred onto a nitrocellulose membrane. ABCA1 (1:250 dilution), SR-BI (1:5,000 dilution), and PDZK1 (1:3,000 dilution) were detected using specific primary antibodies. For each immunoblot, protein-disfulfide isomerase (1:2,000 dilution) was used as a loading control. The apoA-I immunoblot (1:10,000 dilution) was performed with fresh plasma, and the membrane was stained with Coomassie Blue to confirm equal loading into each lane of the gel. Densitometric analysis of the blots was performed using Quantity One software.

S Metabolic Labeling of SR-BI in Primary Hepatocytes-
Freshly isolated hepatocytes were incubated at 37°C for 1 h in serum-free medium (Dulbecco's modified Eagle's medium) and then in methionine-free medium. 100 Ci of [ 35 S]methionine/ cysteine cell labeling mix (PerkinElmer Life Sciences) was added to each dish. The pulse period allowed radiolabeling of newly synthesized proteins for 2 h before the 35 S-containing medium was removed. Chase periods ranged from 0 to 12 h. Following the pulse-chase experiment, cells were washed with PBS and lysed in immunoprecipitation lysis buffer (0.63 M Tris-HCl, pH 7.4, 0.75 M NaCl, 25 mM EDTA, 5 mM phenylmethylsulfonyl fluoride, and 5% Triton) containing protease inhibitors. Protein concentration was determined, and equal aliquots of 100 g of cellular protein were prepared for immunoprecipitation of SR-BI. Samples were incubated overnight at 4°C with the anti-SR-BI antibody (NB400-101) to immunoprecipitate SR-BI (1:100). Protein-Sepharose A was added to the samples, and SR-BI was pelleted by centrifugation at 2000 rpm for 3 min and washing thoroughly in between (with buffer containing 0.01 M Tris-HCl, pH 7.4, 2 mM EDTA, 0.1% Triton, and 0.1% SDS). Reducing sample buffer (6 mM Tris-HCl, pH 7.4, 10% glycerol, 2% SDS, and 8 M urea) was added directly to the samples. After boiling for 20 min, samples were loaded onto a 12% polyacrylamide gel. Immunoprecipitated proteins were visualized by immunoblotting with the goat polyclonal anti-SR-BI antibody (NB 400-131). The electrochemiluminescent signal (from the immunoblot analysis) was allowed to dissipate for 2 days before the membranes were exposed at Ϫ80°C during 6 -8 weeks for the 35

S signal (autoradiograph).
Preparation of mRNA Extracts for Quantitative PCR-Livers from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ female mice (ϳ100 g) were homogenized in TRIzol solution, and RNA was precipitated with isopropyl alcohol according to the manufacturer. The iso-propyl alcohol was removed, and 75% ethanol (in water containing diethylpyrocarbonate) was added. RNA concentration was determined by measuring the absorbance, and samples were then kept at Ϫ80°C in diethylpyrocarbonate-water. cDNA preparations were synthesized with Superscript III polymerase (Invitrogen).

RESULTS
Plasma HDL Lipids Are Decreased by PEMT Deficiency-A significant reduction of plasma lipid levels was previously observed in chow-fed Pemt Ϫ/Ϫ mice compared with Pemt ϩ/ϩ mice (18). To investigate the mechanism responsible for this decrease, we first analyzed the lipoprotein profiles of plasma from mice of the two genotypes by fast protein liquid chromatography. The mice had been fasted overnight prior to the analysis. Both cholesterol (Fig. 1A) and choline-containing phospholipids (Fig. 1B) were lower in plasma of mice lacking PEMT compared with mice with PEMT.
ABCA1-mediated Lipid Efflux Is Not Decreased by the Absence of PEMT-We next evaluated the role of PEMT-derived PC in HDL formation. Primary cultured hepatocytes, obtained from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice, were incubated in the presence or absence of poorly lipidated apoA-I (20 g protein/ml), an extracellular acceptor. The efflux of cholesterol and PC from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ hepatocytes was linear with time for up to 24 h and with the amount of apoA-I (up to 200 g/ml) (data not shown). The release of cholesterol from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ hepatocytes into the medium was determined after a 16-h incubation with or without apoA-I. The amount of cholesterol in the culture medium was analyzed by gas-liquid chromatography (Fig. 2A). The difference in the amount of cholesterol recovered in the medium of Pemt ϩ/ϩ and of Pemt Ϫ/Ϫ hepatocytes did not reach statistical significance. The efflux of PC from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ hepatocytes was assessed after radiolabeling of cellular phospholipids with [ 3 H]glycerol (Fig. 2B). The release of [ 3 H]PC from Pemt Ϫ/Ϫ hepatocytes was independent of Pemt genotype and was surprisingly not stimulated by apoA-I.
The Amount of Hepatic ABCA1 Is Not Altered by PEMT Deficiency-ABCA1 is a membrane transporter that promotes HDL formation via the release of cellular lipids to apolipoprotein acceptors in plasma. Livers from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice were homogenized, and hepatic levels of ABCA1 protein were evaluated by immunoblot analysis (Fig. 2C). Densitometric analysis of immunoblots for ABCA1 protein from seven mice of each geno-  type revealed no significant change in ABCA1 levels in the livers from mice lacking PEMT compared with Pemt ϩ/ϩ mice. The results shown in Fig. 2 indicate that the decrease in plasma HDL in Pemt Ϫ/Ϫ compared with Pemt ϩ/ϩ mice is not mediated by a decrease in the activity or expression levels of ABCA1.

SR-BI in Livers of Pemt
Plasma ApoA-I Levels Are Similar in Pemt ϩ/ϩ and Pemt Ϫ/Ϫ Mice-The above data suggest that the lower plasma levels of cholesterol and PC in Pemt Ϫ/Ϫ , compared with Pemt ϩ/ϩ , mice is not due to reduced expression or activity of hepatic ABCA1. Since apoA-I is the major apolipoprotein of HDL, we determined whether or not apoA-I levels were altered by the absence of PEMT. Plasma (2 l) was collected from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice and analyzed for apoA-I by immunoblot analysis. The immunoblot membrane was also stained with Coomassie Blue to show total proteins as a loading control. Although the amounts of HDL PC and cholesterol were significantly lower in mice lacking PEMT (Fig. 1), the levels of apoA-I were not significantly different between Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice (Fig. 3). Moreover, because apoA-I is an exchangeable apolipoprotein that can also be found on other lipoproteins, plasma samples were fractionated by sequential density ultracentrifugation. No difference was observed between the apoA-I level in plasma of Pemt Ϫ/Ϫ mice compared with that of wild type mice, confirming an earlier result (18).
Uptake of Lipids from HDL by Primary Hepatocytes Is Increased in the Absence of PEMT-To investigate whether or not lipid uptake capacity is altered by PEMT deficiency, we incubated hepatocytes from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice with HDL containing [ 3 H]choline-labeled PC and [ 3 H]cholesteryl oleoyl ether, a nonhydrolyzable tracer for cholesteryl esters. The amount of radioactivity associated with cellular extracts was determined (Fig. 4). The uptake of both radiolabeled lipids was linear with respect to time of incubation (up to 240 min) and protein concentration (up to 250 g/ml of radiolabeled HDL) (data not shown). The uptake of [ 3 H]PC was significantly greater (p Ͻ 0.01) in Pemt Ϫ/Ϫ , compared with Pemt ϩ/ϩ , hepatocytes. Similarly, the uptake of [ 3 H]cholesteryl ether was higher (p Ͻ 0.001) in hepatocytes derived from Pemt Ϫ/Ϫ compared with Pemt ϩ/ϩ mice. Thus, PEMT deficiency increases the uptake of lipids from HDL.
In Vivo Hepatic Uptake of HDL Lipids Is Increased by the Absence of PEMT-To determine if the in vivo uptake of HDL lipids was increased by the absence of PEMT, Pemt ϩ/ϩ and  Pemt Ϫ/Ϫ mice were injected with radiolabeled HDL via the tail vein. After 60 min, liver samples were analyzed for the amount of [ 3 H]cholesteryl oleoyl ether and [ 3 H]PC associated with hepatic tissue (Fig. 5). The uptake of [ 3 H]PC was not significantly different (p Ͻ 0.1) in livers from Pemt Ϫ/Ϫ compared with Pemt ϩ/ϩ mice. In contrast, more [ 3 H]cholesteryl ether was recovered in livers of Pemt Ϫ/Ϫ , compared with Pemt ϩ/ϩ , mice (p Ͻ 0.001).
The Amount of Hepatic SR-BI Is Elevated in the Absence of PEMT Activity-SR-BI is a cell surface receptor that binds HDL and mediates the selective uptake of cholesteryl esters and PC from HDL into cells (32). Hepatocytes obtained from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice were collected after the lipid uptake assay, and the level of cellular SR-BI protein was evaluated by immunoblot analysis (Fig. 6A). Densitometry quantification of

. Expression of SR-BI protein is increased in PEMT-deficient hepatocytes. A, representative immunoblot of cellular extracts from
Pemt ϩ/ϩ and Pemt Ϫ/Ϫ hepatocytes (10 g protein/lane). As a loading control, the membrane was also immunoblotted for protein-disulfide isomerase (PDI). B, densitometric analysis was performed on immunoblots from hepatocytes isolated from 10 mice of each genotype. SR-BI and PDI signals were compared and ratios were expressed relative to Pemt ϩ/ϩ samples. C, immunoblots of liver homogenates from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice (35 g of protein/lane). Protein-disulfide isomerase was used as a loading control. D, densitometric analysis of immunoblots from eight mice of each genotype. SR-BI and PDI signals were compared, and ratios were expressed relative to Pemt ϩ/ϩ samples. Values are means Ϯ S.E. *, p Ͻ 0.001. immunoblots for SR-BI from 10 mice of each genotype revealed a significantly higher (1.5-fold, p Ͻ 0.05) level of SR-BI in hepatocytes lacking PEMT than in Pemt ϩ/ϩ hepatocytes (Fig. 6B). Similarly, analysis of liver homogenates from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice revealed a corresponding 1.6-fold (p Ͻ 0.001) increase in SR BI expression in the PEMT-deficient livers (Fig.  6, C and D).
Increased SR-BI Protein in Pemt Ϫ/Ϫ Hepatocytes Is Not the Result of Up-regulated SR-BI mRNA Expression-Our data suggest that the absence of hepatic PEMT activity results in enhanced lipid uptake, most likely via the pathway mediated by SR-BI. Therefore, we analyzed SR-BI protein synthesis and degradation in primary cultures from both genotypes using 35 S metabolic labeling. Following a pulse period of 2 h to allow the radiolabeling of newly synthesized proteins, the degradation of SR-BI protein was analyzed over a time period of 0 -8 h of chase. Fig. 7A shows 35 S metabolic labeling of SR-BI protein in hepatocytes from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice. The top panel represents the autoradiograph generated by the 35 S signal. The bottom panel is an immunoblot analysis of total immunoprecipitated hepatic SR-BI. At the end of the pulse period (time 0), there is more SR-BI protein in Pemt Ϫ/Ϫ hepatocytes (as demonstrated by the immunoblot data) than in Pemt ϩ/ϩ cells. This observation also correlates with a greater production of 35 Slabeled SR-BI protein in Pemt Ϫ/Ϫ hepatocytes. Thus, in agreement with our previous data in whole cell lysates, Fig. 7 shows that total immunoprecipitated and 35 S-labeled protein are more abundant in samples from Pemt Ϫ/Ϫ mice. Interestingly, regardless of their Pemt genotype, hepatocytes did not significantly degrade the newly made 35 S-labeled SR-BI protein over the 8-h incubation. However, total (labeled ϩ unlabeled) protein levels (Fig. 7B) did show a substantial decline during this chase period. Indeed, SR-BI protein expression is ϳ45% lower after 8 h in Pemt ϩ/ϩ cells. It should be emphasized that the decrease is only about 25% in Pemt Ϫ/Ϫ hepatocytes.
In addition to these metabolic studies, hepatic SR-BI mRNA levels were analyzed using real time PCR (Fig. 7C). When compared with their wild type counterparts, the Pemt Ϫ/Ϫ mice did not show any significant difference in SR-BI mRNA transcripts detected in liver homogenates.
Blue Native Polyacrylamide Gel Electrophoresis of Hepatic SR-BI-To characterize the various oligomeric forms of hepatic SR-BI, we used blue native PAGE analysis. This method is based on separation by charge and is believed to achieve higher resolution of protein complexes than other nondenaturing techniques (41). Liver homogenates were electrophoresed on a 4 -20% gradient gel, and SR-BI complexes were visualized by immunoblotting. Fig. 8 is representative of the data obtained from n Ն 6 mice of each genotype, showing two samples from Pemt ϩ/ϩ mice on the left and three samples from Pemt Ϫ/Ϫ mice on the right. A monomeric form of SR-BI is detected at ϳ82 kDa in every lane, although the exact molecular size can only be grossly approximated when using the blue native PAGE. Dimers of SR-BI can be visualized at ϳ160 kDa. Moreover, the blue native PAGE analysis reveals the presence of higher order complexes, probably oligomers of SR-BI associated with itself or with other proteins, which is in agreement with previous reports of the multimerization of SR-BI in different steroido- genic and nonsteroidogenic cells (42,43). It is noteworthy that the size of these complexes does not differ significantly between genotypes, although the total SR-BI present is greater in liver homogenates from Pemt Ϫ/Ϫ mice. The up-regulation of SR-BI protein results in significantly higher levels of SR-BI monomers and, to a lesser extent, of SR-BI dimers (Fig. 8). Indeed, densitometric analysis of our blue native PAGE results confirmed the increase in monomer/dimer ratio (0.49 Ϯ 0.17 for Pemt Ϫ/Ϫ mice compared with 0.12 Ϯ 0.01 for Pemt ϩ/ϩ liver homogenates. This indicates that increases in SR-BI protein expression correspond indeed to a greater lipid uptake potential (Fig. 4) but do not necessarily translate into more oligomeric complexes, although they are also more abundant as well in the absence of PEMT.
The Absence of PEMT Results in Increased Hepatic PDZK1 Protein-PDZK1, a 70-kDa scaffold protein, is important for maintaining hepatic SR-BI steady state levels (44). Therefore, we analyzed PDZK1 expression in homogenates of livers from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice (Fig. 9). As confirmed by the densitometry analysis, the absence of the PEMT pathway induces a 1.5-fold increase in PDZK1 levels (*, p Ͼ 0.005).

DISCUSSION
In this study, we have addressed the mechanism by which PEMT deficiency in mice reduces the levels of cholesterol and PC in plasma (18), a decrease that occurs primarily in HDL (Fig.  1). Two hypotheses were considered based on our current understanding of the mechanisms that can alter the levels of lipids in plasma HDL: (i) decreased hepatic HDL formation as a consequence of a lack of PC biosynthesis or (ii) a possible increase in lipid uptake by the liver to replenish hepatic PC stores.
ABCA-1-mediated Lipid Efflux Is Not Decreased in the Absence of PEMT Activity-HDL removes excess cholesterol from peripheral tissues for delivery to steroidogenic organs and to the liver. However, hepatocytes, in concert with enterocytes, synthesize and secrete most of the apoA-I found in the plasma compartment and express ABCA1, thus making the liver a major contributor to HDL formation (29,45). The assembly of HDL particles from lipid-poor apoA-I requires the efflux of both phospholipids and cholesterol and relies on the action of ABCA1 (46). In humans, the absence of ABCA1 activity results in Tangier disease and near absence of HDL particles with an increased risk of cardiovascular disease (25). To determine if PEMT-derived PC is involved in the lipidation of apoA-I, we investigated HDL formation by Pemt Ϫ/Ϫ hepatocytes. The results show that ABCA1-mediated efflux of cholesterol and PC is unaffected by the absence of PEMT. Moreover, the hepatic expression of ABCA1 is similar in mice of both Pemt genotypes. This is consistent with ABCA1 expression being regulated by cellular cholesterol levels (25), which are unchanged in the livers of Pemt Ϫ/Ϫ compared with Pemt ϩ/ϩ mice (18). Reduction of hepatic ABCA1 activity is associated with decreased plasma apoA-I levels (47), which we did not find (Fig.  3), further supporting the conclusion that absence of PEMT activity does not affect hepatic ABCA1 expression or activity.
Recently, other members of the ABC transporter family have been implicated in mobilization of cholesterol and/or phospholipids from cells to HDL (46,48). ABCG1 mediates the efflux of cholesterol to partially lipidated "nascent" HDL, but studies have suggested that its contribution to hepatic lipoprotein production or clearance is minimal (49,50). ABCA7 has the highest homology known to ABCA1 (ϳ54%) and promotes selective cellular phospholipid efflux to apoA-I (49 -52). It is unlikely that ABCG1 or ABCA7 is responsible for the decreased levels of plasma HDL lipids, since murine hepatocytes have very low levels of ABCG1 mRNA (49,50), and ABCA7 protein is virtually absent from murine liver (45). Thus, the decreased plasma  HDL lipid levels cannot be explained by a reduced efflux of hepatic lipids in Pemt Ϫ/Ϫ mice.
Lipid Uptake from HDL Is Increased in the Absence of PEMT Activity-The final step of reverse cholesterol transport involves the uptake of HDL lipids in the liver via SR-BI, a cell surface receptor that mediates selective lipid uptake. In fact, ϳ60 -80% of plasma cholesterol is cleared from the rodent circulation by the liver (53). After hepatic uptake, cholesterol can be metabolized and excreted into bile by cellular sorting mechanisms that are still under investigation (54 -56).
The results from the lipid uptake experiments strongly suggest that the lower levels of plasma cholesterol and PC can be, in part, attributed to a greater amount of hepatic SR-BI in Pemt Ϫ/Ϫ mice when compared with Pemt ϩ/ϩ mice. This conclusion is also supported by our finding of no significant difference in plasma apoA-I levels in mice lacking PEMT activity. Selective lipid uptake, first described by Glass et al. (23), involves the recycling of apoA-I to the plasma compartment. Accordingly, overexpression of SR-BI is accompanied by a significant decrease in HDL cholesterol, but plasma apoA-I concentration remains unchanged (22). Thus, our findings are consistent with an involvement of SR-BI in regulating plasma lipid levels when hepatic PC biosynthesis is challenged. The present study shows that expression of SR-BI in hepatocytes is up-regulated by the absence of PEMT, and this is accompanied by a significant increase in the ability of Pemt Ϫ/Ϫ hepatocytes to accept HDL cholesterol and PC from the culture medium. Additionally, the in vivo delivery of [ 3 H]HDL lipids to the liver, following tail vein injection, was greater in Pemt Ϫ/Ϫ mice, suggesting that SR-BI might be involved in regulating plasma HDL levels in situations where lipid metabolism is perturbed, as is the case when PEMT is absent.
To gain further insight into the mechanism by which PEMT deficiency results in the induction of hepatic SR-BI, we analyzed mRNA levels by quantitative real time PCR. No difference in the amount of SR-BI mRNA transcripts was detected when liver samples from Pemt Ϫ/Ϫ mice were compared with samples from Pemt ϩ/ϩ mice. There is evidence that hepatic SR-BI expression is not highly controlled at the transcriptional level (57), although hormonal regulation of its promoter has been reported (58,59). Thus, it is likely that hepatic SR-BI levels are mainly determined by protein turnover, cellular targeting, and/or protein-protein interactions (57). Any of these posttranslational regulatory mechanisms could potentially explain the increased protein expression and/or the enhanced lipid uptake observed in Pemt Ϫ/Ϫ hepatocytes. PDZK1, a PDZ domain-containing protein, has been shown to interact with the C-terminal cytoplasmic domain of SR-BI (60,61). It has been proposed that PDZK1 serves as a scaffold for SR-BI to be targeted to the appropriate subcellular localization or to be in close proximity of downstream signaling molecules. PDZK1 might also regulate the rate of retroendocytosis of SR-BI and/or help in selective lipid sorting after internalization of HDL particles. Nonetheless, PDZK1 is believed to be an important regulator of hepatic SR-BI stability, as demonstrated by a 95% reduction of SR-BI protein levels in mice lacking PDZK1 (61,62). Moreover, expression of PDZK1 in SR-BI-expressing cells caused an up-regulation of SR-BI protein without affecting its mRNA levels (63). We thus examined hepatic PDZK1 expression in the absence of PEMT and found greater levels of PDZK1 (1.5-fold) in knock-out cells. Interestingly, this up-regulation also correlates with the extent of the detected increase in hepatic SR-BI protein levels. Although the exact mechanism underlying the increased expression of PDZK1 and SR-BI is still unknown, it appears that a change in cellular PC homeostasis can significantly affect hepatic lipid uptake pathways.
A change in SR-BI protein turnover is also possible under conditions of altered hepatic PC biosynthesis. We aimed to analyze protein synthesis and degradation over time in primary hepatocytes obtained from Pemt ϩ/ϩ and Pemt Ϫ/Ϫ livers using 35 S-labeled methionine/cysteine amino acids to trace newly synthesized proteins. Performing such an experiment was difficult in primary cultures, since after the isolation procedure, hepatocytes do not make significantly more SR-BI. Indeed, SR-BI protein levels decrease consistently over time, starting soon after plating the cells (data not shown). Thus, the experiment was not designed to determine protein degradation, since there was no obvious decline of 35 S signal over a chase period of 8 h. Also, in a study on the intracellular localization of SR-BI, Ahras et al. (64) localized the HDL receptor to late endosomal/ lysosomal compartments. They reported that after an 8-h cycloheximide chase, SR-BI protein was still abundantly found in association with the lysosomal compartment. Their data suggest that this intracellular pool of SR-BI is fairly stable and that SR-BI plays a role in endosomal/lysosomal lipid trafficking and does not undergo significant degradation (64), in agreement with our observations (Fig. 7). The potential existence of two different pools of SR-BI is quite interesting, especially since the internalization of HDL particles appears to be independent of this intracellular SR-BI pool, relying mostly on the cell surface expression of the receptor (i.e. the plasma membrane pool) (64).
Thus, our results clearly show that even with unchanged mRNA expression, there is more SR-BI protein present in liver samples of PEMT-deficient mice. An increase in SR-BI protein levels with unchanged mRNA expression has previously been reported by Nakamura et al. (63). Rats were treated with glucagon, and the results showed increased hepatic SR-BI protein expression without any change in mRNA levels (63). A lack of correlation between protein expression and mRNA levels of SR-BI has also been reported in hepatocytes by Shetty et al. (65). Interestingly, following isolation of hepatocytes, there appeared to be slightly more SR-BI protein synthesis occurring in 2 h of incubation of Pemt Ϫ/Ϫ cells with 35 S-labeling mix compared with their wild type counterparts (Fig. 7).
Since dimerization of SR-BI protein has been demonstrated (43, 66), we analyzed the formation of SR-BI oligomeric complexes in mice lacking the PEMT pathway. The results obtained from the blue native PAGE analysis revealed no difference in the size or the ratios of the complexes formed by SR-BI in the livers of Pemt ϩ/ϩ and Pemt Ϫ/Ϫ mice. In different cell lines, SR-BI has been shown to homodimerize and to form higher order oligomers (42,43,66). The expression and dimerization of SR-BI on the cell surface appears to influence microvillar channel formation in steroidogenic tissues and correlates with a greater selective lipid uptake (67). The effect of SR-BI selfassociation and/or of its oligomeric forms on the efficiency of HDL lipid uptake by the liver, however, has not been investigated. Our study demonstrates a 1.5-fold up-regulation of SR-BI protein; thus, it is not surprising to detect proportionally more SR-BI in the monomeric form and as part of oligomer complexes. Also, we cannot exclude the possibility that the hepatic oligomers detected by blue native PAGE are heterooligomers of SR-BI with other proteins of similar molecular mass, such as the 70-kDa PDZK1 (68).
Since apoA-I remains in the plasma compartment and is available for relipidation, SR-BI appears to protect against atherosclerosis despite the fact that HDL cholesterol levels are lowered by overexpression of SR-BI (69 -74). In the absence of PEMT, the up-regulation of hepatic SR-BI (due to altered PC biosynthetic pathways in the liver) might alter plasma lipids without affecting the amount of plasma apoA-I. Interestingly, elimination of PEMT in mice lacking the low density lipoprotein receptor significantly reduced atherosclerotic lesions after the mice were fed a high fat/high cholesterol diet for 12 weeks. 4 The exact mechanism linking PC biosynthetic pathways to SR-BI has yet to be discovered. A few years ago, our laboratory showed that PEMT-derived PC is preferentially targeted to secretion into bile (9). In the case of PEMT deficiency, it is possible that the liver compensates by up-regulating the uptake of HDL lipids in order to maintain efficient biliary secretion of PC. Thus, it is conceivable that PEMT-derived PC is involved in controlling plasma HDL metabolism. Consistent with the idea that PC homeostasis in the liver can regulate HDL levels, mice in which the level of CTP:phosphocholine cytidylyltransferase ␣ in the liver has been reduced by 85% also exhibit decreased levels of circulating HDL lipids (75). More studies on mice lacking either PEMT or the CDP-choline pathway should provide further insight into how hepatic PC biosynthesis is involved in the regulation of plasma lipoprotein levels.