The Assembly of Very Low Density Lipoproteins in Rat Hepatoma McA-RH7777 Cells Is Inhibited by Phospholipase A2Antagonists*

In McA-RH7777 cells, the oleate-stimulated assembly and secretion of very low density lipoproteins (VLDL) was associated with enhanced deacylation of phospholipids, which was markedly decreased by inactivation of the cellular phospholipase A2. Treatment of the cells with antagonists or antisense oligonucleotide of the Ca2+-independent phospholipase A2 (iPLA2) significantly inhibited secretion of apoB100-VLDL and triglyceride. Similar inhibitory effect of the iPLA2 antagonists was observed on apoB48-VLDL secretion, but secretion of high density lipoprotein particles (such as apoAI- and apoB48-high density lipoprotein) or proteins in general was unaffected. The iPLA2 antagonist did not affect the synthesis of apoB100 or triglyceride, nor did it affect the activities of phospholipase D, phosphatidate phosphohydrolase, or microsomal triglyceride transfer protein. Inactivation of iPLA2 resulted in impaired apoB100-VLDL assembly as shown by decreased apoB100-VLDL and triglyceride within the microsomal lumen, with concomitant increase in apoB100 association with the microsomal membranes. The inhibitory effect of iPLA2antagonists on apoB100-VLDL assembly/secretion could be abated by pretreatment of cells with oleate. Analysis of molecular species of microsomal phosphatidylcholine and phosphatidylethanolamine by electron spray tandem mass spectrometry revealed that the enrichment of oleoyl moieties was altered by the treatment of iPLA2 antagonist. These results suggest that the oleate-induced VLDL assembly/secretion may depend upon the establishment of membrane glycerolipids enriched in oleoyl chain, a process mediated by the iPLA2 activity.

Very low density lipoproteins (VLDL) 1 are assembled in the liver and secreted as triglyceride (TG)-rich particles. Each VLDL particle contains various amounts of TG and a single copy of apoB100. The rat liver synthesizes a truncated form, apoB48 (collinear with the N-terminal 48% of apoB100), that also has the ability to assemble VLDL (1). The subcellular compartment where VLDL is assembled has not been unambiguously determined. Immunohistochemical and electronmicroscopic studies proposed that rat liver VLDL was assembled at the junction of rough and smooth endoplasmic reticulum (ER) (2). Although apoB immunostaining was observed within rough ER, VLDL-sized lipid staining entities lacking apoB were found within the lumen of smooth ER (2). Biochemical studies, however, suggested that association of lipid with apoB could occur not only post-translationally (3) but also during apoB100 translation (4 -7). Recent experiments with improved techniques have shown that in McA-RH7777 cells, assembly of apoB100- (8,9) and apoB48-VLDL (3,10) are both achieved post-translationally and require the activity of microsomal triglyceride transfer protein (MTP). These studies support the model known as "two-step" assembly, which theorizes that the initial product is a primordial dense particle that is subsequently assembled with bulk TG to form a mature VLDL. The stepwise VLDL assembly also occurs in primary hepatocytes (7,11).
The post-translational model for VLDL assembly is established on the basis of kinetic analysis that observes the time taken by apoB and TG to be assembled. It is noticed that there exists a temporal delay between the moment when apoB translation reaches completion and the time when VLDL is matured. In the case of apoB100, assembly of apoB100-VLDL 1 is undetectable until 15-20 min after apoB synthesis (9). Such a lag period was also observed for TG incorporation into hepatic apoB48-VLDL (7). Since ER is a highly dynamic organelle that continuously undergoes vesiculation and fusion, the delayed apoB100-VLDL 1 maturation may reflect the time needed for the movement of apoB100 (and TG as well) from the site of synthesis to the site of assembly. Data suggesting the involvement of ER vesiculation in VLDL assembly are obtained from studies with brefeldin A (10,12). At a low dose of brefeldin A, assembly of VLDL is inhibited but assembly of dense particles remains normal. Thus, membrane trafficking mediated by vesiculation and fusion may be essential for bulk TG incorporation into VLDL. It has been recognized that changes in membrane lipid composition regulated by lipid catalyzing enzymes dramatically affect membrane vesiculation and fusion (13). Therefore, it is important to understand the role of enzyme activities that remodel biological membranes in VLDL assembly.
Two families of cytosolic phospholipase A 2 (PLA 2 , EC 3.1.1.4.) have been characterized, namely cPLA 2 (␣, ␤, and ␥ forms) and iPLA 2 . Although the ␤ form of cPLA 2 is expressed in the liver (14), hepatic expression of the ␣ (15) and ␥ (16) form is rather low. The iPLA 2 also consists of multiple isoforms, which are derived from alternative splicing of a single gene (17). Unlike cPLA 2 , iPLA 2 does not have a Ca 2ϩ -dependent lipidbinding domain (18,19), but carries repeating motifs homologous to the integral membrane protein-binding domain of ankyrin (20 -22). The iPLA 2 displays broad substrate specificity and can be inhibited by a suicide substrate, bromoenol lactone (BEL) (23). Functional roles ascribed to iPLA 2 include phospholipid remodeling (24,25) and Golgi membrane tubulation (26,27), among others (28,29). It is suggested that phospholipid turnover may be associated with VLDL secretion in rat liver cells treated with oleate (30). A potential role of phospholipid turnover in VLDL secretion was first suggested by a work (31) that showed ϳ50% of VLDL-TG secreted from rat liver might use fatty acids derived from phospholipid deacylation. However, it remains to be determined if phospholipid turnover plays a role in the process of bulk TG incorporation into VLDL. Here, we examined the role of iPLA 2 -mediated phospholipid turnover in oleate-induced hepatic VLDL assembly/secretion. Cell Culture and Metabolic Labeling-Cells expressing human apoB100 (hB100) (32) or human apoB48 (hB48) (10) were cultured in Dulbecco's modified Eagle's medium (DMEM) containing 20% fetal bovine serum and 200 g/ml G418.
In metabolic labeling experiments, cells were treated with 0 -100 M PLA 2 antagonists (i.e. BEL or MAFP) for 30 min and labeled with 100 Ci/ml [ 35 S]methionine/cysteine for up to 2 h in a methionine-and cysteine-free DMEM containing 20% fetal bovine serum and 0.4 mM oleate. Sometimes, the cells were treated with oleate prior to PLA 2 inhibition as indicated in figure legends. The medium was collected after labeling and subjected to cumulative rate flotation (9) to resolve apoB100-VLDL 1 (S f Ͼ 100) and apoB100-VLDL 2 (S f 20 -100) from intermediate density lipoproteins (IDL), low density lipoproteins (LDL), and high density lipoproteins (HDL). The [ 35 S]-B100 in each fraction was recovered by immunoprecipitation using a polyclonal antiserum that was raised against human LDL but cross-reacted with rat apoB, and analyzed by PAGE/fluorography as described previously (9). Secretion of endogenous rat 35 S-apoA-I from the hB100-transfected cells were also determined by immunoprecipitation with an anti-rat apoA-I antiserum. The effect of PLA 2 antagonists on endogenous rat apoB100 secretion was determined using non-transfected McA-RH7777 cells. Secretion of hB100 mass was assessed by immunoprecipitation of hB100 from fractionated medium samples with a polyclonal anti-apoB antiserum, followed by immunoblotting with monoclonal antibody 1D1. Inhibitors of PLA 2 were solubilized in dimethyl sulfoxide (Me 2 SO) at a concentration of 40 mM, stored at Ϫ20°C prior to use. All controls contained equal volume of Me 2 SO. Subcellular Fractionation and Trypsin Digestion-Cell homogenization, preparation of microsomes and cytosol, and trypsin digestion of microsome-associated apoB were performed as described previously (9).
Lipid Labeling-Cells in six-well dishes were pulse-labeled with [ 3 H]oleic acid (2 Ci/ml) or [ 3 H]glycerol (3 Ci/ml) for 3 h, and the labeled lipids were chased for indicated times in the presence or absence of 0.4 mM oleate. The effect of iPLA 2 antagonists on the generation of glycerophosphocholine (GPC) was determined using cells labeled with Enzyme Assays-For PLA 2 , phosphatidate phosphohydrolase (PAPH), and MTP activity assays, cells were treated with BEL (100 M) for 30 min, followed by incubation with oleate (0.4 mM) plus BEL (100 M) for 1 h. Cells that were incubated only with or without oleate were used as controls. The PLA 2 activity in cell homogenates was measured using [ 3 H]palmitoyl-PC as substrate in the presence or absence of 10 mM CaCl 2 as described previously (33). Dithiothreitol (0.5 mM) was present when PLA 2 activity was determined in the absence of CaCl 2 (33).
The PAPH activity in cytosol or microsomes was determined by using N-ethylmaleimide to distinguish between PAPH and the membranebound lipid phosphate phosphohydrolase (34,35).
The TG transfer activity of MTP in cell homogenates was measured as described previously (36) using [ 14 C]triolein as substrate.
The activity of phospholipase D was determined by transphosphatidylation assay (37). Cells were labeled with [ 3 H]myristic acid (1 Ci/ml) for 18 h. After labeling, the cells were treated with iPLA 2 antagonists (100 M) for 30 min, and incubated with butan-1-ol (0.5%) for an additional 30 min. Formation of [ 3 H]phosphatidylbutanol at the end of incubation was quantified.
Oligonucleotide Treatment-An antisense oligonucleotide of the rat iPLA 2 cDNA was designed starting from the initiation site: 5Ј-AG-GCGTCCAAAGAACTGCAT-3Ј. Control sense oligonucleotide was 5Ј-ATGCAGTTCTTTGGACGCCT-3Ј. Both were synthesized by Life Technologies, Inc. Cells (20% confluence) were incubated with 10 M oligonucleotides (freshly prepared prior to use) in serum-free DMEM for 4 h. A final concentration of 20% fetal bovine serum was then added (38), and the cells were cultured for 48 h prior to experiments.
Tandem Mass Spectrometry-Equal aliquots of lipid extracts were dissolved in methylene chloride/methanol/H 2 O (45/45/10, by volume) and applied to a Micromass Quattro II triple quadruple mass spectrometer (Micromass) at a flow rate of 6 l/min (39). Data were acquired using MassLynx NT software (Micromass). A mixture of di-14:0, di-16:0, 16:0 -18:1, and di-20:4 PC standards and a separate mixture of di-16:0, di-18:0, and 18:0 -22:6 PE standards of equal molar concentration was used to determine instrument settings and concentration for optimal signal intensities. Standards and samples contained 1% of formic acid for positive ion analysis. The PC species were detected by precursor scanning for molecules generating an ion fragment of 184 mass units (phosphocholine moiety) in the positive ion mode at collision energy of 25 V. The PE species were detected by scanning for molecules with a loss of neutral fragment 141 (phosphoethanolamine moiety) in the positive ion mode at collision energy of 20 V. The fatty acid composition of each molecular species was determined by daughter ion analysis in the negative ion mode.

Oleate-induced VLDL Assembly/Secretion and Phospholipid
Deacylation-We used cumulative flotation ultracentrifugation technique (40) to resolve B100-VLDL 1 (S f Ͼ100) and B100-VLDL 2 (S f 20 -100) from IDL/LDL and HDL that were synthesized by hB100-transfected McA-RH7777 cells. The high level expression of the recombinant protein makes hB100 more readily detectable than the limited amount of endogenous rat apoB. In these cells, VLDL 1 assembly in the microsomal lumen ( Fig. 1A, left panels) and its secretion into medium (Fig. 1A, right panels) were stimulated by oleate. The enhanced VLDL 1 assembly/secretion were associated with increased turnover of cellular PC and PE as shown by pulse-chase experiments (Fig.  1B,  The [ 3 H]oleoyl moiety released from phospholipid is either in free fatty acid form (from the action of phospholipase A (PLA)) or in glycerolipid form such as diacylglycerol (from the action of PLD followed by PAPH). Two pieces of evidence suggested that the PLD pathway might play a limited role. First, oleate ex-erted no effect on the change of radioactivity of [ 3 H]glycerollabeled PC or PE during chase (Fig. 1C, top panels), although it did stimulate incorporation of [ 3 H]glycerol into TG and increase secretion of [ 3 H]glycerol-labeled PC, PE, and TG (Fig.  1C, bottom panels). Second, transphosphatidylation assay showed that PLD activity was unchanged by oleate in cells treated with or without phorbol myristate acetate (a compound known to stimulate PLD activity (Refs. 25 and 37)) ( Table I, left  two columns). Thus, the observed phospholipid turnover is likely catalyzed by a PLA-type activity. Treatment with PLA 2 antagonists also inhibited VLDL secretion. Preliminary experiments showed that the IC 50 of BEL to inhibit VLDL-hB100 and VLDL-TG secretion was ϳ60 M (data not shown). At 100 M, treatment with BEL decreased secretion of 35 S-B100 as VLDL 1 and VLDL 2 by 70% and 50%, respectively, but secretion of IDL/LDL was less affected (Fig.  3A, BEL ϩ OA). Similar decreases in 35 S-B100 secretion as VLDL 1 (by 60%) or VLDL 2 (by 45%) were observed in MAFPtreated cells (Fig. 3A, MAFP ϩ OA). The inhibited VLDL secretion was confirmed by lipid labeling experiments, where treatment with BEL or MAFP diminished [ 3 H]glycerol-labeled TG secretion by more than 50% (Fig. 3B). Most of the secreted [ 3 H]TG was associated with VLDL 1 and VLDL 2 (data not shown). The inhibitory effect of BEL on VLDL-TG secretion was also manifest in pulse-chase experiments where TG was labeled with [ 3 H]oleate (Fig. 3B, inset). At 100 M, BEL did not affect secretion of endogenous apoA-I as HDL (Fig. 3C), nor did it affect secretion of total trichloroacetic acid-insoluble 35 Slabeled proteins (data not shown). The inhibitory effect of BEL treatment on hB100-VLDL secretion was also observed semiquantitatively by immunoblot analysis (Fig. 3D). Moreover, similar inhibitory effect of BEL occurred to the secretion of endogenous rat 35 S-B100-VLDL (Fig. 3E). Pulse-chase analysis showed that the secretion of total 35 S-B100 was inhibited by ϳ15% at the end of 2-h chase (Fig. 3F).
The effect of BEL treatment on VLDL secretion was also observed in cells expressing hB48 that secreted both hB48-VLDL and hB48-HDL, where increasing dose of BEL (0 -100 M) decreased hB48-VLDL but not hB48-HDL secretion (Fig. 4,  A and B). Thus, BEL preferentially inhibits the secretion of TG-rich particles such as B100-VLDL 1 and B48-VLDL with little effect on the general secretory pathway. A point of note is that, although iPLA 2 activity is essential for VLDL secretion, it does not appear to be limiting nor is it sufficient. Transfection of iPLA 2 (20) either transiently or stably into McA-RH7777 cells resulted in a 2-fold increase in iPLA 2 activity (50 versus 25 units/mg of cell protein), but it did not enhance B100-VLDL 1 secretion (data not shown). Similarly, expression of the iPLA 2 in HepG2 cells also failed to induce B100-VLDL 1 secretion, even though the iPLA 2 activity was increased in transfected cells (88 versus 1.3 units/mg of cell protein) to a level comparable to that in McA-RH7777.
Since BEL inhibits preferably iPLA 2 over cPLA 2 and other secreted forms of PLA 2 (41), the above data suggest an effect more attributable to iPLA 2 . However, BEL was reported to inhibit the activity of PAPH as well (42). Hence, we measured PAPH activity in McA-RH7777 cells and found that neither the cytosolic nor the microsomal-associated PAPH activity was inhibited by treatment of 100 M BEL (Table I, third and fourth columns from left) or MAFP (data not shown). Additionally, addition of BEL at various concentrations up to 100 M directly into the enzyme assay mixture did not affect PAPH activity (data not shown). In contrast, the PLA 2 activity was potently inhibited by BEL in a manner independent of Ca 2ϩ (Table I, fifth and sixth columns from left). Furthermore, treating cells with antisense oligonucleotides of iPLA 2 decreased iPLA 2 activity by 50% as compared with controls (Fig. 5A), and decreased secretion of B100-VLDL 1 and B100-VLDL 2 by 50% and 60%, respectively (Fig. 5B). These results combined argue convincingly for a requirement of iPLA 2 in hepatic phospholipid turnover and in VLDL secretion.
Inactivation of iPLA 2 Impairs VLDL Assembly-Assembly of VLDL 1 is achieved post-translationally in McA-RH7777 cells (9), which can be demonstrated by pulse-chase experiments with [ 35 S]methionine/cysteine (20-min pulse and 50-min chase in the presence of oleate) (Fig. 6A, left). Treatment of BEL did not affect the synthesis of total B100 or the association of B100 with IDL/LDL particles, but decreased B100-VLDL 1 assembly by 60% at the end of chase (Fig. 6A, arrows in ϩBEL (Fig. 6B, middle and right). However, the PLA 2 antagonists decreased accumulation of [ 3 H]TG in the microsomal lumen by 50% (Fig. 6B, left). The decreased [ 3 H]TG in the microsomal lumen was not a result of compromised MTP   Fig. 1B. C, density distribution of medium rat 35 S-apoA-I. D, hB100expressing cells were treated with BEL for 30 min followed by incubation with oleate for 2 h. Medium hB100 was detected by immunoprecipitation with anti-apoB antiserum, followed by immunoblot using monoclonal anti-human apoB antibody 1D1. E, analysis of medium rat 35 S-B100 from non-transfected McA-RH7777 cells was similar to that in panel A. F, hB100-expressing cells were pulse-labeled with [ 35 S]methionine/cysteine in the absence or presence of BEL for 1 h, and 35 S-B100 was chased with oleate Ϯ BEL for up to 4 h. Medium 35 S-B100 was analyzed by immunoprecipitation followed by SDS-PAGE/fluorography. activity, because BEL did not affect the TG transfer activity of MTP (Table I, right column). (Addition of BEL directly into the TG transfer assay mixture also exerted no effect on MTP activity (data not shown).) The inhibited VLDL assembly by BEL within microsomal lumen was not associated with impairment in translocation of apoB across the ER membranes. Trypsin digestion assay of microsomes showed that there was no difference between control (Fig. 6C, control, lanes 2, 5, and 8) and BEL-treated cells (Fig. 6C, ϩBEL, lanes 2, 5, and 8) in gaining trypsin resistance of 35 S-B100 during the entire chase period (Fig. 6C).
Inactivation of iPLA 2 Results in Accumulation of ApoB on Microsomal Membranes-Treatment with BEL resulted in a small, but reproducible, increase (about 5-10%) in cell-associated 35 S-B100, as determined by pulse-chase analysis (Fig. 7A). Since previous studies suggested that VLDL was assembled from apoB precursors attached to the microsomal membranes (8,43), we sought to determine if BEL treatment resulted in increased membrane association of apoB. The cells were labeled with [ 35 S]methionine/cysteine for 2 h, and the association of 35 S-B100 with membranes (Fig. 7B) and within the lumen (Fig. 7C) was examined. In BEL-treated cells, there was indeed an increased (by ϳ10%) association of 35 S-B100 with membranes over the controls (Fig. 7B, lanes marked by arrows). At the end of 2 h of labeling, 35 S-B100 associated with lumenal VLDL 1 and VLDL 2 was not as abundant as that seen in pulsechase experiments (Fig. 6A). However, the inhibitory effect of BEL on lumenal 35 S-B100-VLDL 1 was still observable (Fig. 7C,  lanes marked by arrows). It was reported that the membraneassociated 35 S-B100 could be dislodged from membranes by deoxycholate (8). We confirmed this observation and found that deoxycholate removed approximately 75% of the membrane associated 35 S-B100 (Fig. 7B, lanes labeled Dox/KCl) and increased lumenal 35 S-B100 (Fig. 7C). Comparison of lumenal 35 S-B100 derived from sodium carbonate treatment in the absence (Fig. 7C, top) or presence of deoxycholate (Fig. 7C, bottom) showed that the recovery of lumenal 35 S-B100 from deoxycholate-treated samples was markedly increased in IDL/ LDL and HDL fractions. Quantitative analysis (Fig. 7D) showed that there was at least 2-fold increase in 35 S-B100 (mostly in IDL/LDL and HDL form) that were associated with microsomal membranes in BEL-treated cells. Thus, the activity of iPLA 2 may play a role in the release of membrane-associated B100 precursors (i.e. in IDL/LDL forms) into microsomal lumen to form mature VLDL.
Effect of Pretreatment with Oleate and iPLA 2 Antagonists on VLDL Assembly and Microsomal Phospholipid Species-To gain insight into the mechanism by which iPLA 2 antagonist inhibits oleate-induced VLDL assembly, we tested the effect of After BEL pretreatment, oleate was present throughout the experiment. BEL was present in both pulse and chase (right). Trypsin digestion assay of the isolated microsomes was performed (9), and 35 S-B100 were detected by SDS-PAGE/fluorography. iPLA 2 inhibition on cells that had been pretreated with oleate. When cells were incubated with oleate for 30 min prior to iPLA 2 inactivation, the inhibitory effect of BEL treatment on apoB100-VLDL 1 assembly (Fig. 8A, left) or secretion (Fig. 8A, right) was no longer observable. Likewise, secretion of [ 3 H]TG, induced by oleate (Fig. 8B, left, ϩOA versus ϪOA) was not affected by BEL treatment from cells that had been incubated with oleate prior to BEL treatment (ϩOA3ϩBEL). However, in cells that had been pretreated with the antagonist, the oleate-induced [ 3 H]TG secretion was attenuated significantly (ϩBEL3ϩOA). The altered BEL treatment protocol had little effect on the oleate-stimulated incorporation of [ 3 H]glycerol into cell TG (Fig. 8B, right). Thus, the iPLA 2 activity becomes nonessential for VLDL assembly/secretion in cells that have been treated with oleate. To examine if this effect was oleatespecific, we compared oleate with eicosapentaenoic acid, an n-3 fatty acid. Although treatment with eicosapentaenoic acid (0.4 mM) resulted in accumulation of cell-associated [ 3 H]glycerollabeled TG to a level similar to that treated with oleate as shown in Fig. 6B, eicosapentaenoic acid did not prevent inhibition of VLDL secreted exerted by BEL treatment (data not shown).
The above results showing differential effect between oleate and eicosapentaenoic acid suggest that VLDL assembly is dependent of proper lipid composition, which may be affected by inactivation of iPLA 2 . We therefore determined molecular species of microsome-associated PC and PE under different treatment conditions with oleate and/or iPLA 2 antagonists. Electron spray tandem mass spectrometry analysis showed that incubation with oleate increased total cellular PC and PE masses by 60% and 110%, respectively. However, oleate did not increase the mass of PC (oleate, 1.40 ϫ 10 9 ; control, 1.42 ϫ 10 9 ; arbitrary units/mg of protein) or PE (oleate, 0.90 ϫ 10 7 ; control, 1.12 ϫ 10 7 ; arbitrary units/mg of protein) associated with the microsomal membranes. This result derived from mass spectrometry analysis agrees with our previous chemical measurement (9) that oleate increased incorporation of [ 3 H]glycerol into PC (an indication of increased synthesis) but did not affect PC mass in the microsomal membranes. However, oleate treatment resulted in enrichment of species with 18:1/18:1 in PC (by 130%) and 18:1/18:1 in PE (by 24%) as compared with controls ( Fig. 9, A and B). Species of PC and PE with other diacyl chains (e.g. 16:0/18:1 or 18:1/18:2) were less affected by oleate treatment. Treatment of cells with BEL caused a moderate decrease in microsomal PC species with oleoyl chain and markedly decreased the oleoyl-containing PE species (BEL3OA versus OA). Preincubation of cells with oleate prevented the decrease in oleoyl-containing PC and PE species caused by BEL treatment (OA3BEL versus BEL3 OA). These results suggest that oleate pretreatment may have primed the microsomal phospholipids that are essential for VLDL assembly and have achieved an effect that is otherwise dependent upon iPLA 2 action. The effect of oleate and iPLA 2 antagonist on other microsomal phospholipids (such as phosphatidylserine or phosphatidylinositol) was not determined. DISCUSSION Multiple lines of evidence from the current work suggest that the assembly of VLDL requires iPLA 2 activity. The inhibitory effect of iPLA 2 antagonists (chemical inhibitors or antisense oligonucleotides) is rather specific to the assembly and secretion of TG-rich VLDL (e.g. B100-VLDL 1 and B48-VLDL) but not the secretion of small lipoproteins such as HDL. The specific inhibitory effect of iPLA 2 antagonists toward VLDL assembly is reminiscent of observations derived from studies under many other pathophysiological or pharmacological conditions where VLDL secretion is impaired but HDL secretion is normal. Several notable examples related to phospholipid me- Oleate was present throughout the experiment. After oleate pretreatment, BEL was present in both pulse and chase (ϩBEL). 35 S-B100 in microsomal lumen (left) and medium (right) was analyzed as in Fig. 1A tabolism include inhibition of PC synthesis by choline deficiency (44), inhibition of PE methylation by bezafibrate (45), and treatment with monomethylethanolamine (46). Under these conditions, an unidentified step essential for bulk TG incorporation into VLDL is impaired, but the normal ER-to-Golgi trafficking remains functional. Thus, the current study of iPLA 2 -mediated phospholipid turnover, together with previous observations, highlight the existence of a possible vesicular transport system that is specialized for TG-rich VLDL assembly/secretion.
It is now clear that bulk TG incorporation into VLDL is achieved post-translationally (3,9,10,47). However, less clear is the mechanism by which bulk TG is incorporated. Abundant TG availability is essential, but it alone is not sufficient to drive VLDL assembly, as exemplified by studies with hepatic cells treated with n-3 fatty acids (48 -50) or insulin (51,52), where active TG synthesis does not result in VLDL production. In certain hepatoblastoma cell lines (e.g. HepG2 cells), TG synthesis can be effectively stimulated by oleate, but formation of VLDL is not achieved (53). Likewise, PLA 2 antagonists can impair VLDL assembly in McA-RH7777 cells, while TG synthesis remains unaffected (Fig. 6, A and B). Hence, factors other than TG availability must play an important role in facilitating VLDL assembly. Over the past several years, our understanding of the requirements for hepatic VLDL assembly/ secretion has been advanced significantly by extensive biochemical and genetic studies, which have established an important role for various proteins (e.g. MTP (Ref. 36), apoB (Ref. 54), and brefeldin A-sensitive factors (Refs. 10 and 12)) as auxiliary factors of VLDL assembly. The current observation that PLA 2 inhibition results in impaired bulk TG incorporation into VLDL in the face of normal apoB synthesis and apoB translocation across the microsomal membrane indicates that the process is important at the post-translational stage. These observations, together with the results that inhibition of iPLA 2 is associated with decreased microsomal lumenal B100-VLDL 1 and increased B100 association with microsomal membranes (Figs. 6 and 7), suggest strongly that PLA 2 is a novel auxiliary factor of VLDL assembly. The concept of PLA 2 activity acting as an auxiliary factor for hepatic VLDL assembly is also suggested by our data that intracellular phospholipid deacylation is augmented by oleate used to stimulate VLDL assembly/ secretion. Data showing that a variety of iPLA 2 antagonists potently block PC deacylation and B100-VLDL 1 or B48-VLDL secretion provide additional evidence supporting this concept.
The mechanism whereby iPLA 2 acts as such an auxiliary factor of VLDL assembly is unclear. To date, two roles have been ascribed to phospholipid metabolism in the biosynthesis of VLDL. The first role concerns phospholipid as a constituent of lipoproteins; thus, active synthesis of phospholipids, particularly PC, provides substrates for VLDL assembly (44,55). The second role highlights that phospholipid acts as a fatty acid reservoir; therefore, through its turnover, phospholipid supplies fatty acyl moieties for TG synthesis (30). Several pieces of evidence derived from the current work suggest that oleate and iPLA 2 antagonists do not exert their effect on VLDL assembly merely by affecting PC synthesis or substrate channeling between PC and TG. First, although exogenous oleate increased PC synthesis (measured by increased incorporation of [ 3 H]glycerol) was observed previously (9) and in current studies (data not shown), there was no change in microsomal membraneassociated PC mass as quantified by both mass spectrometry analysis (the present study) and by chemical measurement (previous observation in Ref . 9). Therefore, the enhanced incorporation of [ 3 H]glycerol into PC upon oleate treatment reflects not so much synthesis as turnover. Second, although oleate indeed enhanced channeling of oleoyl moieties from PC to TG (Fig. 1B), which agreed with previous observations made using oleate-treated primary rat hepatocytes (3), inactivation of iPLA 2 by antagonist did not alter this oleoyl chain channeling into TG (data not shown). In fact, under no circumstances had iPLA 2 inactivation affected TG synthesis. Therefore, the impaired VLDL assembly/secretion by iPLA 2 antagonist is not simply a consequence of TG availability. Third, the inhibitory effect of BEL on VLDL assembly was observed only when the antagonist was added before but not after oleate treatment (Fig. 8). This result is critical and suggests that changes in phospholipid composition or membrane structure may be more important than phospholipid synthesis in facilitating bulk TG incorporation into VLDL. The finding that the species of microsome-associated phospholipids, particularly that of oleoylenriched PE (Fig. 9), can be significantly influenced by oleate or PLA 2 antagonist raises the possibility that priming of membrane milieu plays an important role in the process of VLDL assembly.
Changes in phospholipid composition render significant biological consequences. Alteration in membrane shape (56) and membrane curvature (57) is associated with membrane vesiculation and tubulation (13,58). Glycerolipids containing monounsaturated acyl chains (e.g. oleoyl chain) are shown to activate proteins essential for the coating of budding vesicles (59) and membrane fusion (60). In addition to their direct effect on membrane curvature, distinct phospholipid species can regulate membrane fusion by changing the secondary structure of peptide in fusion protein (61). It is therefore conceivable that deacylation/reacylation of phospholipid, induced by oleate, provides a means for oleoyl chain to be incorporated into phospholipid species and for subsequent generation of lysophospho- lipid, diacylglycerol, and phosphatidic acid. Generation and proper localization of these oleoyl-rich glycerolipid species may be important for the event of bulk TG incorporation into VLDL. Once such a primed membrane is established (e.g. in the case of pretreatment of cells with oleate), inhibition of PLA 2 activity with antagonists may not exert an acute effect on bulk TG incorporation into VLDL. The activity of PLA 2 , including iPLA 2 , present in cytosol of cells, is shown to participate in endosome fusion of macrophages (62), in the Golgi tubulemediated retrograde trafficking to the ER, and in the maintenance of Golgi complex architecture in rat hepatocytes (26,27). Our current results, although inconclusive, reveal a possible function of iPLA 2 -mediated phospholipid remodeling in VLDL assembly/secretion. However, whether the impaired VLDL assembly/secretion by iPLA 2 inhibition is the results of perturbed membrane movement essential for bulk TG incorporation remains to be determined.