Conversion of Low Density Lipoprotein-associated Phosphatidylcholine to Triacylglycerol by Primary Hepatocytes*

We have studied the uptake and metabolism of phosphatidylcholine (PC), the major phospholipid of low density lipoproteins (LDL), by cultures of primary hepatocytes. Strikingly, in the absence of the LDL receptor, PC incorporation into hepatocytes was inhibited by only 30%, whereas cholesteryl ether uptake was inhibited by 60-70%. On the other hand, scavenger receptor class B, type I, the other important receptor for LDL in the liver, was found to be responsible for the uptake of the remaining 30-40% of LDL-cholesteryl ether. PC uptake was, however, only partially inhibited (30%) in scavenger receptor class B, type I, knock-out hepatocytes. Once LDL-PC was taken up by hepatocytes, ∼50% of LDL-[3H]oleate-PC was converted to triacylglycerol rather than degraded in lysosomes as occurs for LDL-derived cholesteryl esters. The remainder of the LDL-derived PC was not significantly metabolized to other products. Triacylglycerol synthesis from LDL-PC requires a PC-phospholipase C activity as demonstrated by inhibition with the phospholipase C inhibitor D609 or activation with rattlesnake venom. Small interfering RNA-mediated suppression of acyl-CoA:diacylglycerol acyltransferase 2 (DGAT2), but not DGAT1, decreased the acylation of the LDL-derived diacylglycerol. These findings show that PC in LDL particles is taken up not only by the classical receptors but also by additional mechanism(s) followed by metabolism that is completely different from the cholesteryl esters or apoB100, the other main components of LDL.

LDL-cholesteryl esters (LDL-CE) is associated with enhanced risk of atherosclerosis, leading to cardiovascular diseases, including heart attack and stroke (2,3). The pioneering work of Brown and Goldstein in the 1970s on the cellular basis of cholesterol homeostasis led to the detailed characterization of the uptake and metabolism of LDL following receptor-mediated endocytosis. In this pathway, LDL binds to the LDL receptor (LDLR) (4,5), and the complex is internalized from clathrincoated pits (6). After endocytosis, cholesteryl esters (CE) and apoB100 are hydrolyzed in the lysosomes (5,7,8). The unesterified cholesterol thus produced can be released from the lysosomes and re-esterified (9). On the other hand, the LDLR is mainly recycled to the plasma membrane (10).
It is now clear that LDL also delivers LDL-CE to cells via "selective uptake," a process in which LDL-CE is taken up without intracellular apolipoprotein degradation (11,12). Several proteins appear to be involved in the selective uptake of CE such as lipoprotein lipase and proteoglycans (13), apoE, LDL receptor-related protein/␣ 2 -macroglobulin receptor (14), and scavenger receptor class B, type I (SR-BI) (15,16). A variety of ligands, including native lipoproteins (LDL, high density lipoproteins (HDL), very low density lipoproteins, and chylomicrons) and modified lipoproteins (acetylated LDL, oxidized LDL, and oxidized HDL), can bind to SR-BI (17). Specific amino acid residues of SR-BI are involved in its binding to the various ligands (18). The exact mechanism by which SR-BI mediates selective lipid uptake from lipoproteins is unknown. At least two models have been proposed as follows: the first suggests that selective uptake takes place at the plasma membrane (19) and that SR-BI creates a hydrophobic channel allowing CE to be moved down its concentration gradient into the plasma membrane (20). The second model postulates that lipids are exchanged after internalization of lipoproteins, and later the lipid-depleted particles are then resecreted, resulting in selective lipid uptake (21,22). Regardless of the precise mechanism, LDL-CE is taken up by selective uptake and is degraded in lysosomes (23). On the other hand, HDL-CE is mainly hydrolyzed in an extralysosomal compartment (24), although partial lysosomal involvement was also described (23).
More recently, a new player in cholesterol homeostasis has been described. Proprotein convertase subtilisin/kexin type 9 is a protein that induces degradation of LDLR in the liver and hence adds a new dimension to the regulation of plasma LDL levels (25)(26)(27). Overexpression of proprotein convertase subtilisin/kexin type 9 leads to hypercholesterolemia (28). Conversely, inhibition of proprotein convertase subtilisin/kexin type 9 expression is associated with increased expression of LDLR and therefore low levels of LDL-CE (29).
Despite the considerable attention devoted to studying the metabolism of LDL-CE and apoB100, much less is known about the uptake and metabolism of phospholipids associated with LDL. The fate of PC on HDL and LDL when incorporated into human platelets was investigated. Rapid incorporation of PC was shown to supply platelets and monocytes with polyunsaturated fatty acids (arachidonic acid) for the production of eicosanoids (30 -32). Subsequently, it was demonstrated that phospholipid transfer from LDL to platelets is independent of the high affinity binding of LDL to platelets and was stimulated by wheat germ agglutinin (33). In monocytes, high affinity binding of LDL was shown to involve SR-BI that mediated the selective import of the major lipoprotein-associated phospholipids (34). A role for SR-BI in PC uptake from LDL and HDL by endothelial cells was also demonstrated by Sattler and co-workers (35).

Materials-McArdle
Animal Care-All procedures were approved by Institutional Animal Care Committees at either the University of Alberta or McMaster University and were in accordance with guidelines of the Canadian Council on Animal Care. Male mice (3-6 months old) from C57BL/6 (wild type), LDLR knock-out, SR-BI knock-out, hepatic lipase knock-out, and apoE knockout mice were fed ad libitum a chow diet (LabDiet, PICO laboratory Rodent Diet 20) and were exposed to a 12-h light/dark cycle starting at 8:00 a.m.
Preparation and Labeling of Lipoproteins-LDL was isolated from healthy human donors by ultracentrifugation (density ϭ 1.020 -1.063 g/ml). First, the density of plasma was adjusted to 1.019 g/ml with KBr and the sample centrifuged at 50,000 ϫ g for 24 h at 8°C to remove very low density lipoproteins. The density was adjusted to 1.063 g/ml with KBr, and the sample was centrifuged under the same conditions. The LDL fraction was recovered from the top and dialyzed against 150 mM NaCl, 1 mM EDTA, 100 M diethylenetriaminepentaacetic acid, pH 7.5. LDL were stored under nitrogen at 4°C. Lipoprotein-deficient serum (LPDS) was prepared by adjusting plasma to d ϭ 1.215 g/ml with KBr and centrifugation at 99,000 rpm for 4 h at 8°C in a tabletop Beckman ultracentrifuge, rotor TLA 100.4. LPDS was dialyzed against the buffer above and served as the source of phospholipid transfer protein and CE transfer protein.
Liposomes were prepared by drying 50 Ci of [ 3 H]PC under nitrogen and resuspension of the lipid in 1 ml of phosphatebuffered saline and sonication for 5 min using a tip sonicator (15-s cycles). Some experiments in addition to [ 3 H]PC also used 50 Ci of [ 3 H]cholesteryl ether (CEt), a nonhydrolyzable analog of CE. The resulting small unilamellar vesicles were mixed with LPDS and freshly prepared LDL. EDTA and diethylenetriaminepentaacetic acid were added to a final concentration of 1 mM and 100 M, respectively. The mixture was incubated for 20 h at 37°C in an orbital shaker (250 rpm). After incubation, [ 3 H]LDL was re-purified by ultracentrifugation as described above.
Isolation and Culture of Primary Hepatic Cells-Hepatocytes were isolated from mice after liver perfusion with collagenase as described (37). Cells were plated in 60-mm collagen-coated dishes at a density of 9 ϫ 10 5 cells/dish. Unless stated otherwise, hepatocytes were cultured in DMEM with 10% FBS. Two h after plating, the cultures were rinsed in DMEM containing 10% FBS, and then incubated in the same medium for 12 h. Prior to experiments, cultures were rinsed twice in serum-free DMEM over a 1-h period and then incubated in serum-free DMEM with radiolabeled LDL as described below.
PC LDL Uptake-Hepatocytes were preincubated with or without the specified inhibitors and chemicals for 1 h prior to the addition of 50 g of [ 3 H]LDL protein/ml. Cells were then incubated at 37°C for the indicated times. Media were removed, and cellular monolayers were washed three times with cold phosphate-buffered saline. Cells were harvested; homogenates were sonicated for 10 s; lipids were extracted (38), and phospholipids were separated by TLC (chloroform/methanol/acetic acid/water, 25:15:4:2) until the solvent reached halfway up the plate. The solvent was evaporated from the plate, and the neutral lipids were separated in the solvent heptane/diisopropyl ether/ acetic acid (60:40:4). Following iodine visualization, bands corresponding to specific lipids were scraped, and radioactivity was measured by liquid scintillation counting. Water-soluble compounds were obtained from the aqueous phase after lipid extraction. Because of the low radioactivity of the samples, they were dried under nitrogen; 50 l of water/methanol (1:1) was added, and radioactivity was measured by liquid scintillation counting.
Quantification of Cell Surface LDLR-The surface expression of LDLR was estimated by biotinylation of the cell surface proteins (39). Briefly, hepatocytes were treated with the indicated inhibitors and then washed twice with cold phosphatebuffered saline. Sulfo-NHS-biotin (0.5 mg/ml) in phosphatebuffered saline was added, and cells were incubated for 30 min at 4°C. Cells were washed, and excess biotin was quenched with 100 mM glycine in phosphate-buffered saline. Hepatocytes were lysed with lysis buffer (1% Triton X-100, 4 mM EGTA, 10 mM Tris-HCl, pH 8). After removal of cell debris, biotinylated proteins were collected on neutravidin-agarose. The samples were eluted from beads with boiling SDS electrophoresis sample buffer, resolved by SDS-PAGE, and immunoblotted with anti-LDLR antibody.
Induction of PC-PLC Activity-Snake venom from Crotalus atrox (Western Diamondback Rattlesnake) was used for inducing PC-PLC activity (40). A 2 g/l venom stock solution was treated with 5 mM Me-indoxam (41) for 30 min at room temperature to eliminate all PLA 2 activity. A final concentration of 4 g/ml of venom was incubated with hepatocytes. Cellular toxicity was evaluated by the release of LDH. Briefly, culture media were collected after different treatments and centrifuged at 3,000 ϫ g for 10 min. LDH activity was measured spectrophotometrically at 340 nm by following the oxidation of NADH in the presence of pyruvate. Total LDH per culture was determined from 0.1% Triton X-100 lysates of untreated control cultures. Lipid composition after treatment was analyzed by phosphorous content as described (42).
Treatment with Phospholipases-Phospholipase incubations were performed as described by Hassan et al. (43) with few modifications. Briefly, primary hepatocytes were incubated for 1 h in DMEM containing 50 g of [ 3 H]LDL protein/ml. After washing to remove unbound lipoproteins, hepatocytes were treated with 0.5 units/ml PC-PLC or 1 unit/ml sphingomyelinase (Sigma) for 30 min at 37°C. Media were removed, and cellular monolayers were washed three times with cold phosphate-buffered saline. Cells were harvested; homogenates were sonicated for 10 s, and lipids were extracted. Cellular toxicity under treatment with phospholipases was evaluated by the release of LDH as described above.
Silencing of DGAT1 and DGAT2-Commercial siRNAs specific for DGAT1 and DGAT2 were used. The transfection medium was prepared as follows: 5 l of Lipofectamine and 100 pmol of siRNA-DGAT1 or siRNA-DGAT2 were diluted in 100 l of serum-free and antibiotic-free DMEM. The complex was incubated for 30 min at room temperature and then diluted with 1.4 ml of William's E medium containing 5% FBS.
Two hours after plating, hepatocytes were washed twice with William's E medium containing 5% FBS, and then transfection medium was added. Cells were incubated for 12 h with siRNAcontaining medium, and the medium was changed to William's E medium containing 10% FBS. The cells were incubated 24 h prior to the start of experiments.
RNA Isolation and Real Time PCR-RNA from hepatocytes was isolated using TRIzol reagent according to the manufacturer's instructions. Total RNA was then reverse-transcribed using an (dT) [12][13][14][15][16][17][18] primer and Superscript II reverse transcriptase according to the manufacturer's instructions. DGAT1, DGAT2, and cyclophilin transcripts were detected by real time PCR using a Rotor-Gene 3000 instrument (Montreal Biotech). Reaction mixtures contained 0.25 M of each primer and Platinum SYBR Green qPCR SuperMix-UDG, in a total volume of 25 l. Data analyses were performed using the Rotor-Gene 6.0.19 program (Montreal Biotech).

LDL-PC Avoids Lysosomal
Degradation-LDL particles are well known to be internalized by cells via receptor-mediated endocytosis and sorted to the lysosomes where components are hydrolyzed (1). To study the metabolism of LDL-PC, [ 3 H]choline-labeled PC was used. As shown in Fig. 1A, during an 8-h incubation with primary mouse hepatocytes, radioactivity remained associated mainly with the PC fraction with negligible radioactivity in sphingomyelin or lyso-PC. Surprisingly, only a small fraction of radioactivity was found associated with water-soluble metabolites. Indeed, choline-derived aqueous compounds represented less than 10% of total radioactivity at all time points and were generated at the lysosomes, because chloroquine inhibited the degradation of this minor fraction of LDL-PC (Fig. 1B). This result indicates that most LDL-PC avoided lysosomal degradation, which would have been the expected consequence of being internalized by the LDLR. Chloroquine, as well as monensin, inhibited LDLR recycling to the plasma membrane as demonstrated by a biotinylation assay but neither compound modified the total cellular expression of LDLR (Fig. 1C). SR-BI surface expression was inhibited to some extent by monensin, in agreement with previous reports (44), but not by chloroquine (see below). For that reason, the uptake of LDL-PC was measured in the presence of these inhibitors. The uptake of LDL-CEt was measured as well as a control, because LDL-CE uptake and degradation by lysosomes are well known. As expected, the uptake of LDL-CEt was greatly reduced by chloroquine and monensin after 3 h of incubation ( Fig. 1D), suggesting that CEt is taken up mainly by the LDLR in primary hepatocytes. The uptake of LDL-PC was slightly lower in the presence of chloroquine, although it did not reach statistical significance and was moderately inhibited by monensin (Fig. 1E). Because both inhibitors were likely blocking LDLR recycling, the results suggest that LDL-PC can be delivered to hepatocytes by an LDLR-independent pathway.
The Role of LDL Receptor and SR-BI in LDL-PC Uptake by Mouse Hepatocytes-To determine the role that LDLR plays in LDL-PC uptake, hepatocytes were prepared from both wild type and LDLR knock-out mice. In agreement with the chloroquine experiments (Fig. 1), LDL-CEt uptake was inhibited by Ͼ70% in LDLR knock-out cells ( Fig. 2A). However, LDL-PC uptake was only inhibited by up to 30% in cells lacking LDLR (Fig. 2B). These results suggest that although some LDL-PC is taken up during endocytosis of LDL and is subsequently degraded in lysosomes, most (Ͼ70%) LDL-PC is delivered by alternative pathways. It is noteworthy that neither monensin nor chloroquine induced any further inhibition of LDL-CE or LDL-PC uptake in LDLR knock-out hepatocytes (data not shown), suggesting that there is no LDLR-independent effect of these agents. SR-BI is considered an important receptor not only for the selective uptake of HDL-CE but also for LDL-CE (16). Two approaches were used to assess the involvement of SR-BI in  LDL-PC uptake. First, a blocking antibody was used to inhibit SR-BI activity in wild type hepatocytes. Second, [ 3 H]LDL uptake was measured in SR-BI Ϫ/Ϫ -derived hepatocytes. Using the blocking antibody, the delivery of PC to hepatocytes was inhibited by 20% (Fig. 2C). LDL-PC uptake in SR-BI Ϫ/Ϫ hepatocytes was 30% less than in wild type cells (Fig. 2D). On the other hand, LDL-CEt uptake was inhibited by 40% with both approaches (Fig. 2, C and D). Consequently, as expected LDL-CEt is taken up mainly by the LDLR (60 -70%), whereas the remainder is internalized by SR-BI (30 -40%). In contrast, the uptake of LDL-PC cannot be completely explained by a combination of LDLR (30%) and SR-BI (20 -30%). Therefore, other mechanisms must operate for LDL-PC uptake that are responsible for the remaining 40 -50% of LDL-PC uptake.
One possible uptake pathway might be the hydrolysis of LDL-PC in the medium or bound to the plasma membrane of the hepatocytes, thereby allowing lyso-PC to diffuse into the plasma membrane. Scagnelli et al. (45) showed that PC in reconstituted HDL is hydrolyzed by phospholipase A 1 in rats, presumably hepatic lipase, leading to the production of lyso-PC and fatty acids that were readily taken up by the liver. We therefore inhibited phospholipase activities by using the following compounds: ATK, an inhibitor of cytosolic phospholipase A 2 and Ca 2ϩ -independent phospholipase A 2 (46), tetrahydrolipstatin, which inhibits the activity of mammalian lipases by covalently binding to the active site serine residue (47), and Me-indoxam, a potent inhibitor of secretory phospholipases A 2 (41). None of these inhibitors reduced the delivery of LDL-PC to the hepatocytes (Fig. 2E), suggesting that hydrolysis is not required for the uptake of LDL-PC. Hepatic lipase is involved in lipoprotein metabolism not only in its catalytically active form but also as an enzymatically inactive form that can act as a bridge by binding to the lipoprotein and to the cell surface (48). To test possible noncatalytic involvement of hepatic lipase in LDL-PC uptake, we prepared hepatocytes from hepatic lipase knock-out mice. Only a slight decrease in LDL-PC uptake was observed in knock-out hepatocytes compared with wild type hepatocytes (data not shown). Thus, hepatic lipase has a minor or no effect on LDL-PC uptake by cultured primary mouse hepatocytes.

LDL-PC Is Partially Converted to TG-The studies described above utilized [ 3 H]choline-labeled PC that traced the choline head group of PC. To evaluate the fate of the acyl moieties, PC was labeled with either [ 3 H]oleic acid or [ 3 H]palmitic acid.
Oleic acid is preferentially incorporated in the sn-2 position of PC, whereas palmitic acid tends to be present in the sn-1 position (49). In agreement with the above results (Fig. 1A), the uptake of LDL-PC resulted in negligible formation of [ 3 H]lyso-PC and 3 H-fatty acids. However, unexpectedly the radioactivity recovered in the cells was associated not only with PC but also with TG (Fig. 3, A and B). Both [ 3 H]oleic acid- (Fig. 3A) and [ 3 H]palmitic acid (Fig. 3B)-labeled PC were converted into TG.
In other experiments, we showed that the production of TG from LDL-PC in LDLR Ϫ/Ϫ hepatocytes was inhibited by 30% (data not shown), the same extent of inhibition as LDL-PC uptake (Fig. 2B). On the other hand, a similar degree of inhibition was obtained when SR-BI was absent (data not shown). It is interesting to note that the addition of monensin or chloro-quine induced a significant inhibition of TG production not only in wild type but also in LDLR Ϫ/Ϫ an SR-BI Ϫ/Ϫ hepatocytes (not shown).
[ 3 H]Diacylglycerol (DG) was present in the hepatocytes after incubation with the radiolabeled LDL. The DG-associated radioactivity was ϳ5-10 times lower than [ 3 H]TG suggesting that DG might be an intermediate in this process. Although radioactivity associated with LDL-derived DG was low, [ 3 H]DG levels were always higher than [ 3 H]lyso-PC and [ 3 H]fatty acids, which correspond to the basal background level. Moreover, no significant changes in TG production or PC uptake were observed upon addition of ATK, consistent with the idea that PLA 2 , and hence the release of fatty acids from LDL-PC, is not involved in TG production from LDL-PC (Fig. 3C). In this regard, an excess of unlabeled oleic acid (0.4 mM), which would dilute radiolabeled fatty acids released from PC, did not inhibit [ 3 H]TG production from LDL (Fig. 3D). Thus, hydrolysis of fatty acids from LDL-derived PC is unlikely to be the pathway by which TG is produced from LDL-PC.
A PC Phospholipase C Is Implicated in TG Production-Because DG is an obligatory intermediate in TG synthesis, we studied the involvement of phospholipase C (PC-PLC) in the process. For that purpose D609, a PC-PLC inhibitor (50), was used. TG synthesis from LDL-PC was inhibited 70% by 50 g/ml D609 suggesting that PC-PLC activity is associated with TG production (Fig. 4A). In agreement with this idea, as TG synthesis from LDL-PC was inhibited, there was a concomitant increase in the radioactivity associated with PC (Fig. 4A). Fig.  4A also shows that D609 reduced the uptake of CEt. Thus, hepatocytes dissociate the selective uptake of CE from PC uptake, further demonstrating that LDL-PC uptake occurs independently of LDL-CE uptake.
To confirm that PC-PLC is implicated in TG production from LDL-PC, rattlesnake venom was used as an inducer of PC-PLC (40). As expected, TG production increased markedly (13-fold) upon incubation with rattlesnake venom (Fig. 4B). This induction was reversed by the addition of D609 (Fig. 4B). There was no toxic effect of the venom alone or in combination with D609, ruling out the possibility of an artifact because of compromised cellular viability (data not shown). The total amount of PC (radioactivity in PC ϩ TG) taken up by hepatocytes was greatly stimulated by the presence of the venom. It is possible that the conversion of PC to TG limits the uptake of LDL-PC in hepatocytes. Rattlesnake venom also increased DG production (from 500 Ϯ 10 dpm/mg protein to 6767 Ϯ 153 dpm/mg protein), although this increase was minor compared with the radioactivity in TG. It is important to note that the venom contained no residual PLA 2 activity after treatment with Me-indoxam, because no 3

H-fatty acids nor [ 3 H]lyso-PC was released from [ 3 H]LDL or [ 3 H]oleic acid-prelabeled cells after incubation with Me-indoxam-treated venom (data not shown).
Moreover, rattlesnake venom has no intrinsic PC-PLC activity (data not shown). The implication of PC-PLC activity in TG production from LDL-PC was further confirmed when exoge-nous PC-PLC was added to hepatocytes preincubated with radiolabeled LDL-PC (Fig. 4C). On the other hand, the addition of sphingomyelinase did not induce TG production from LDL-PC (Fig. 4C). A second potential source of cellular PCderived DG is via phospholipase D and the phosphatidic acid phosphohydrolase pathway (51). To determine whether this pathway was operational, hepatocytes were treated with 1-butanol. In the presence of this primary alcohol, phospholipase D catalyzes preferentially a phosphatidyl transfer reaction producing phosphatidylbutanol, which is not hydrolyzed by phosphatidic acid phosphohydrolase. Consequently, the production of DG from PC is inhibited by 1-butanol (52,53). As shown in Fig. 4D, the uptake of LDL-PC was inhibited by 30% in the presence of 1-butanol, and TG production from LDL-PC was inhibited to the same extent. However, the cellular ratio [ 3 H]PC/[ 3 H]TG was the same as cells not treated with 1-butanol, suggesting that the inhibition of TG production by 1-butanol is likely because of the inhibition of LDL-PC uptake rather than an inhibition of LDL-DG production from LDL-PC. Moreover, [ 3 H]phosphatidylbutanol did not accumulate. These results indicate that phospholipase D is not linked to TG production from LDL-PC.
The Last Reaction in TG Production Is Mainly Catalyzed by DGAT2-The above results suggest that PC is degraded to DG and then reacylated to TG by DGAT. To determine which DGAT is involved in TG production from LDL-PC, siRNAs specific for DGAT1 and DGAT2 were used. The siRNAs were effective in reducing mRNA levels of DGAT1 (80%) and DGAT2 (70%) (Fig. 5, A and B). We observed an increase in DGAT1 mRNA when DGAT2 was silenced. No compensatory increase of the DGAT2 isoform was found when DGAT1 was targeted (Fig. 5, A and B). When transfected hepatocytes were  . DGAT2 is used for TG synthesis from LDL-PC. Hepatocytes were transfected 2 h after plating with siRNAs specific for either DGAT1 or DGAT2 or a negative control siRNA. Cells were incubated at 37°C, and medium was changed every 12 h. After 36 h, DGAT1 mRNA (A) and DGAT2 mRNA (B) were measured using cyclophilin as the reference gene. mRNA levels are expressed relative to the control siRNA. C, the remainder of the dishes were incubated with 50 g/ml [ 3 H]LDL for 3 h at 37°C. Lipids were extracted from cells, and radioactivity was quantified. Data are expressed as means Ϯ S.D. of five independent experiments. *, statistically significant differences from control (p Ͻ 0.05) are indicated. incubated with [ 3 H]LDL, TG synthesis was decreased only in cells in which DGAT2 was silenced (Fig. 5C). Silencing of DGAT1 did not change TG production from LDL-PC (Fig. 5C). Thus, DGAT2, but not DGAT1, is involved in the conversion of LDL-PC to cellular TG in mouse hepatocytes.

DISCUSSION
This work is the first detailed study of the uptake and metabolism of LDL-PC in primary hepatocytes. The cell model chosen was primary hepatocytes because liver is the main organ involved in LDL clearance (36,54). Moreover, mouse hepatocytes allow a direct comparison between the uptake and metabolism of LDL in wild type and several knock-out hepatocytes lacking key proteins involved in LDL metabolism. The major findings are as follows: (i) LDL and SR-BI contribute to the delivery of LDL-PC to hepatocytes; (ii) LDL-PC has a unique metabolism different from that of LDL-CE and apoB100; and (iii) ϳ50% of LDL-PC is converted to TG via phospholipase C and DGAT2. Thus, LDL-PC is a novel source of hepatic TG that needs to be taken into account with respect to TG homeostasis in liver.
LDL Receptor and SR-BI Are Important for Delivery of LDL-PC to Hepatocytes-Mice have low levels of LDL, and HDL is the main cholesterol-carrying lipoprotein. Therefore, because of limited amounts of mouse LDL available, we studied the uptake of human LDL-PC. Furthermore, apoE is not present in human LDL; therefore, interaction between LDL and the LDLR-related protein is irrelevant (55).
From our knowledge of the pathway of LDL uptake by the LDLR pathway, the uptake of LDL-PC would be expected to occur mainly by the endocytic pathway and sorted to the lysosomes, where phospholipases A are thought to be the enzymes involved in phospholipid degradation (56). However, prior to our studies, LDL-PC uptake had not been studied in depth. In contrast to the predictions, we have shown that choline-derived water-soluble compounds account for only a small percentage (Fig. 1A) of LDL-PC taken up by hepatocytes. The amount of radioactivity associated with aqueous PC degradation products (e.g. choline, phosphocholine, and glycerophosphocholine) was too low for us to identify them. Nevertheless, chloroquine inhibited the production of the water-soluble products from LDL-PC, suggesting that at least a minor fraction of LDL-PC does undergo degradation in the lysosomes. This result is in agreement with a report of Ishikawa et al. (57) who showed that in smooth muscle cells only a fraction of LDL-PC was degraded in lysosomes. In fact, they found 25.1% of LDL-PC in the lysosome-rich fraction and 24.8% in the cytosol-rich fraction (57). They concluded that LDL-PC was sorted equally between lysosomal and extralysosomal compartments (57).
We found that the absence of LDLR decreased the uptake of LDL-PC by 30% but inhibited LDL-CEt uptake by 70%. The results with LDLR Ϫ/Ϫ mice were confirmed when LDLR surface expression was inhibited with monensin or chloroquine in wild type cells. Therefore, besides LDLR endocytosis, additional mechanisms must be present to explain the uptake of LDL-PC, at least in primary hepatocytes.
A likely candidate for LDL-PC uptake is SR-BI, not only because SR-BI is a receptor for LDL (15,16) but also because SR-BI is involved in LDL-phospholipid uptake in monocytes and cerebrovascular endothelial cells (34,35). Our results with SR-BI Ϫ/Ϫ cells and with a blocking anti-SR-BI antibody showed that 30% of LDL-PC is taken up by this receptor. In contrast, 60 -70% LDL-CEt is taken up by LDLR, and the remaining fraction is taken up by SR-BI.
Phospholipase activity was proposed as an important player in LDL-CE homeostasis, and an LDL-associated PLA 2 activity has been reported (58). Lipoprotein-associated phospholipase A 2 has been linked to key steps in atherosclerosis and atherothrombosis (59). In addition, cytosolic phospholipase A 2 might be present at the plasma membrane, where it could hydrolyze LDL-PC and hence modify the interaction of apoB100 with its receptor(s) (57,60). In this regard, Aviram et al. (61) showed that both lipoprotein lipase and hepatic lipase can lead to an enhanced LDL-LDLR interaction. However, based on the results obtained using several inhibitors for phospholipases and hepatic lipase knock-out hepatocytes, we found that LDL-PC does not undergo significant extracellular hydrolysis prior to uptake.
Although human LDL particles do not contain apoE, it was possible that endogenously synthesized apoE might be released from primary hepatocytes to the external medium and become associated with LDL (62). However, we excluded an involvement of apoE in LDL-PC uptake because LDL-PC uptake was the same in apoE knock-out hepatocytes and wild type hepatocytes. Uptake of LDL-PC by the LDLR-related protein is also unlikely because apoE does not appear to play a role in LDL-PC uptake. On the other hand, proteoglycans are important receptors for TG-rich lipoproteins (63). Even though no changes of LDL were found in mice lacking normal proteoglycans (63), an involvement of proteoglycans in LDL-PC uptake remains a possibility.
Another potential player in LDL-PC uptake is plasma phospholipid transfer protein (64,65). Phospholipid transfer protein can mediate the remodeling of HDL particles, which in turn leads to an enhanced uptake of phospholipids and CE from HDL by the liver (66). Although no information is available regarding LDL-PC, overexpression of phospholipid transfer protein reduces the plasma levels of apoB-containing lipoproteins (67). Therefore, it is plausible that phospholipid transfer protein plays a role in LDL-PC uptake by the liver. We also think that bulk-phase endocytosis is only marginally involved in delivery of PC from LDL to hepatocytes because Liu et al. (54) concluded that only 6% of LDL-cholesterol is taken up via this mechanism. In summary, ϳ50 -60% of LDL-PC uptake by hepatocytes involves LDLR or SR-BI, whereas the remaining 40 -50% occurs via unknown pathways.
Conversion of LDL-PC to TG in Hepatocytes-Both palmitic acid-and oleic acid-labeled LDL-PC are precursors for the synthesis of TG. Hence, fatty acids at both the sn1 and sn2 positions of PC are utilized for TG synthesis. In this regard, it was shown that platelets can selectively internalize LDL phospholipids by an unknown receptor(s) and use them as a source of fatty acids, particularly arachidonic acid, which in turn is involved in the production of eicosanoids (32,34). It is noteworthy that platelets do not express LDLR (34).
Our data imply that TG production from LDL-PC involves a PC-PLC that releases DG. In agreement with this suggestion, Stein et al. (68) observed TG production from an ether analog of PC in liposomes when incubated with rat hepatocytes as well as in in vivo experiments with rats. Furthermore, Igal et al. (69) demonstrated that phospholipid-derived DG produced by PLC can be incorporated directly into TG in Chinese hamster ovary cells without formation of a phosphatidic acid precursor. Moreover, Wiggins and Gibbons (70) found that in very low density lipoproteins TG is produced not only de novo from fatty acid synthesis but also from a pool of cellular phospholipids. At the present time it is not possible to confirm our findings genetically for a role of PC-PLC in conversion of PC to TG because a PC-PLC gene has not been described in mammals. The existence of a PC-PLC activity was strongly implicated not only by the experiments with exogenous bacterial PC-PLC but also by the PC-PLC inhibitor D609. This inhibitor greatly reduced the production of TG from LDL-PC despite an increase in LDL-PC uptake. Besides its well characterized inhibitory activity of PC-PLC, D609 is also known as an inhibitor of sphingomyelin synthase 1 and 2, which display PC-PLC activity and use PC as a substrate to produce sphingomyelin (71); therefore, these enzymes might be good candidates for the PC-PLC activity implicated in LDL-TG production. However, the lack of production of [ 3 H]sphingomyelin from [ 3 H]LDL-PC argues against this possibility. Moreover, D609 did not change the mass of sphingomyelin nor cellular PC during the incubations (data not shown). Some connection between PC and TG may exist because the induction of TG production by rattlesnake venom led to an increase in LDL-PC uptake as though TG production would displace the equilibrium inducing an increase in LDL-PC uptake.
TG production is markedly inhibited by monensin and chloroquine not only in wild type cells but also in SR-BI and LDLR knock-out hepatocytes, which may suggest the requirement of an acidic environment for this process. Because monensin and chloroquine may also inhibit the recycling of some proteins to the plasma membrane, it is plausible that proteins at the plasma membrane are involved at some point in TG production from LDL-PC. Interestingly, it was suggested that PC-PLC might be present at the plasma membrane in some cell types (72). Furthermore, exogenous PC-PLC mimicked the cellular production of TG from LDL-PC (Fig. 4C). It is tempting to conjecture that PC-PLC involved in TG production may be located in the plasma membrane, but this idea needs to be examined in more detail.
The final step in TG production from LDL-PC (i.e. the acylation of PC-derived DG) could, theoretically, be catalyzed by either DGAT1, DGAT2, or both enzymes (73). Although these enzymes catalyze the same reaction, they appear to have different roles in TG homeostasis (74). DGAT1 can catalyze the synthesis of not only TG but also of DG, retinyl esters, and waxes (75). Mice lacking DGAT2 have severe reductions in TG and die soon after birth (74). On the other hand, mice lacking DGAT1 have only moderate reductions in tissue TG, are resistant to diet-induced obesity, and are more sensitive to insulin and leptin (76,77). Notwith-standing these observations, the exact role of each DGAT remains unknown. We used siRNA-mediated gene silencing to demonstrate that DGAT2, but not DGAT1, is involved in TG production from LDL-PC in hepatocytes. Because no LDL-derived DG accumulated in hepatocytes, DG production from LDL-PC is likely the rate-limiting step of the process. The involvement of DGAT2, and not DGAT1, in this process could be because DGAT2 is the predominant isoform in the liver (78). In this regard, Monetti et al. (79) showed in mice overexpressing DGAT in the liver that relatively small increases in DGAT2 mRNA levels resulted in large increases in hepatic TG, whereas large increases in DGAT1 mRNA were not associated with TG accumulation. Interestingly, Choi et al. (80) recently showed that targeting DGAT2, but not DGAT1, with antisense oligonucleotides reduces TG storage in liver and improves hepatic steatosis, insulin signaling, and insulin sensitivity. These findings imply that DGAT2 is more active than DGAT1 in promoting TG synthesis in the liver.
In summary, we demonstrate for the first time that PC present in LDL is taken up not only by the classical lipoprotein receptors LDLR and SR-BI, but also by an additional mechanism(s) followed by a metabolism completely different from that of CE or apoB100 (Fig. 6). Surprisingly, a large proportion of LDL-PC is used for TG synthesis. These findings contribute to the complete description of LDL metabolism. LDL-PC enters the hepatocytes by different pathways such as the LDLR, SR-BI, and unknown receptors(s). A minor portion of PC is degraded in the lysosomes, whereas an important fraction of PC is associated with TG synthesis via the PC-PLC pathway. The rest of PC remains in the cell without further metabolism. The precise location where DG production takes place remains to be determined. Two probable places are as follows: A, plasma membrane; B, endoplasmic reticulum. The last step, i.e. acylation of LDLderived DG by DGAT2, should be restricted to the endoplasmic reticulum where DGAT2 is present. PM, plasma membrane; ER, endoplasmic reticulum.