The Activity of Microsomal Triglyceride Transfer Protein Is Essential for Accumulation of Triglyceride within Microsomes in McA-RH7777 Cells

Previously, based on distinct requirement of microsomal triglyceride transfer protein (MTP) and kinetics of triglyceride (TG) utilization, we concluded that assembly of very low density lipoproteins (VLDL) containing B48 or B100 was achieved through different paths (Wang, Y., McLeod, R. S., and Yao, Z. (1997) J. Biol. Chem. 272, 12272–12278). To test if the apparent dual mechanisms were accounted for by apolipoprotein B (apoB) length, we studied VLDL assembly using transfected cells expressing various apoB forms (e.g. B64, B72, B80, and B100). For each apoB, enlargement of lipoprotein to form VLDL via bulk TG incorporation was induced by exogenous oleate, which could be blocked by MTP inhibitor BMS-197636 treatment. While particle enlargement was readily demonstrable by density ultracentrifugation for B64- and B72-VLDL, it was not obvious for B80- and B100-VLDL unless the VLDL was further resolved by cumulative rate flotation into VLDL1 (Sf > 100) and VLDL2 (Sf 20–100). BMS-197636 diminished B100 secretion in a dose-dependent manner (0.05–0.5 μm) and also blocked the particle enlargement from small to large B100-lipoproteins. These results yield a unified model that can accommodate VLDL assembly with all apoB forms, which invalidates our previous conclusion. To gain a better understanding of the MTP action, we examined the effect of BMS-197636 on lipid and apoB synthesis during VLDL assembly. While BMS-197636 (0.2 μm) entirely abolished B100-VLDL1 assembly/secretion, it did not affect B100 translation or translocation across the microsomal membrane, nor did it affect TG synthesis and cell TG mass. However, BMS-197636 drastically decreased accumulation of [3H]glycerol-labeled TG and TG mass within microsomal lumen. The decreased TG accumulation was not a result of impaired B100-VLDL assembly, because in cells treated with brefeldin A (0.2 μg/ml), the assembly of B100-VLDL was blocked yet lumenal TG accumulation was normal. Thus, MTP plays a role in facilitating accumulation of TG within microsomes, a prerequisite for the post-translational assembly of TG-enriched VLDL.

The major function of hepatic very low density lipoproteins (VLDL) 1 is to deliver triacylglycerol (TG) from the liver to peripheral tissues. Each VLDL particle contains a single copy of apolipoprotein B (apoB) and variable amounts of TG (1). There are two forms of apoB proteins present in the plasma, the full-length B100 and a truncated B48 collinear with the N-terminal 48% of B100 (2). In humans, B100 and B48 are produced in the liver and intestine, respectively. In rats, however, the liver produces both B100 and B48, and both forms have the ability to assemble VLDL (1). Accumulating evidence suggests that formation of B100-VLDL (3) and B48-VLDL (4 -7) is accomplished through two steps. Known as the "twostep" model, it postulates that the initial product is a primordial small, dense particle, formed during or immediately after apoB translation in the endoplasmic reticulum (ER). Subsequently, bulk lipid is incorporated into the primordial particle to form a mature VLDL (7). Under certain conditions, such as insufficient lipid supply or treatment of a low dose of brefeldin A (BfA), the assembly/secretion of mature VLDL are impaired whereas those of small, dense particles remain normal (7,8). Apparently, multiple factors are involved for the maturation of VLDL.
One factor that plays an obligatory role in VLDL assembly is microsomal triglyceride transfer protein (MTP) (9). Defective MTP is the cause of abetalipoproteinemia, a recessive genetic disorder manifested by extremely low concentrations of plasma apoB (10,11). MTP is a heterodimer that consists of protein disulfide isomerase and the 97-kDa catalytic subunit (9). The 97-kDa subunit is expressed mainly in intestine and liver, and has the ability to catalyze the transfer of various lipids between lipid surfaces in vitro (12). Physical interaction between MTP and apoB has been observed (13)(14)(15)(16)(17)(18)(19), but it is not clear if MTP also catalyzes lipid transfer onto apoB in vivo. Inactivation of MTP in hepatic or intestinal cells invariably impairs apoB assembly and secretion (7,20,21). On the other hand, coexpression of the MTP 97-kDa subunit with recombinant apoB facilitates assembly of small, dense lipoproteins containing apoB in heterologous cells (22)(23)(24)(25). Although these studies have provided strong evidence that MTP is essential for apoB secretion as lipoproteins, our knowledge on the mechanism by which MTP facilitates VLDL assembly is incomplete and controversial.
Initial transfection studies with non-hepatic cell lines (22)(23)(24)(25) suggested that the major function of MTP was to assist translocation of apoB across the membrane of ER. Since it was believed that the formation of primordial dense particles took place during apoB translocation, some researchers postulated that MTP activity was essential only in the early stage of apoB lipidation (26). In the recent work showing post-translational B100-VLDL assembly, the notion that MTP was required for the "early phase" of assembly but not for bulk TG incorporation at the "late phase" was reinforced (3). Although plausible that MTP acts as a facilitator for apoB translocation, other studies cast some doubts on this conclusion. Experiments with cells (27) or microsomal membranes (28) that lack MTP expression have shown that apoB translocation and assembly into a primordial particle do not need MTP activity. Furthermore, emerging evidence suggests that the demand for MTP is much greater for bulk TG incorporation during VLDL assembly than that for primordial particle formation (7,21). These data raise the possibility that MTP may play a role other than assisting apoB translocation.
Significant new insight into the requirement of MTP activity in VLDL assembly has been gained using stable McA-RH7777 cells. McA-RH7777 cells are the only hepatoma available that produces authentic VLDL but the level of endogenous apoB expression is low. Fortunately, these cells are suitable for stable transfection with recombinant apoBs. Studies conducted so far have consistently demonstrated that the ability of recombinant human B100 and B48 to assemble lipoprotein is indistinguishable from that of endogenous apoBs (29,30). Working with B48-transfected cells, we previously observed marked differences in the kinetics of TG utilization and in the sensitivity toward MTP inhibition between B48-and B100-VLDL assembly/secretion (7). On the basis of these observations, we concluded that dual mechanisms operated in McA-RH7777 cells; while B48-VLDL assembly was achieved post-translationally, assembly of B100-VLDL was co-translational (7). However, in these studies, no attention was given to the heterogeneity of B100-VLDL; hence, they were dealt with as a uniform fraction (d Ͻ 1.02 g/ml) without consideration of size differences (7). Because of this technical caveat, potential size enlargement accompanied with bulk TG incorporation into B100-VLDL was overlooked.
In the current work, we studied VLDL assembly using transfected cells expressing various apoB forms (e.g. B64, B72, B80, and B100) to inquire if there were indeed dual mechanisms for the assembly of VLDL containing small or large apoB. Moreover, we resolved VLDL subclasses by size using cumulative rate flotation to determine if MTP activity was required for the formation of large, TG-rich VLDL. The present study has yielded a unified model that accommodates VLDL assembly with all apoB forms and nullifies the conclusion of dual assembly mechanisms. Also, we have found that MTP activity is crucial in the maintenance of a metabolically dynamic microsomal TG pool, which is essential for bulk TG incorporation during the final stage of B100-VLDL assembly. MTP Activity Assay-The TG transfer activity of MTP was measured using [ 14 C]triolein and the deoxycholate extract of total cell homogenate according to the published protocol (10). Since the extensive dialysis required for the deoxycholate extraction precluded the assessment of BMS-197636 (a reversible inhibitor) effect on MTP, an alternative protocol (referred to as sonication extraction) was developed in which the cell homogenate was sonicated for 2 min at 4°C. After centrifugation (400,000 ϫ g, 16 min) of the sample, the supernatant that contained the MTP activity was used for TG transfer assay. The MTP activity in McA-RH7777 cells measured by the two protocols was comparable (data not shown). Subcellular Fractionation-Metabolically labeled cells (100-mm dish) were suspended in 2 ml of Tris-sucrose buffer (10 mM Tris-HCl, pH 7.4, and 250 mM sucrose) that was supplemented with protease inhibitors (0.1 mM leupeptin, 0.1 mM phenylmethylsulfonyl fluoride, 100 unit/ml aprotinin, and 40 g/ml ALLN). The cells were homogenized using a ball-bearing homogenizer (20 passages) (33), and the postnuclear supernatant was obtained by centrifugation (10,000 ϫ g, 4°C, 10 min). Heavy microsomes were isolated from the postnuclear supernatant by centrifugation using a microcentrifuge (16,000 ϫ g, 4°C, 15 min), and subsequently the light microsomes and cytosol were separated by centrifugation using a Beckman TLA-100.4 rotor (400,000 ϫ g, 4°C, 16 min). The heavy microsomes were enriched with ribosomal RNA as determined by ethidium bromide staining of total RNA separated by agarose gel electrophoresis. In some experiments, the heavy and light microsomes were combined (referred to as total microsomes).

Materials-ProMix
Trypsin Digestion-Heavy microsomes were resuspended in 200 l of Tris-sucrose using a 25-gauge needle (five passages). Aliquots (60 l) of the samples were incubated (30 min on ice) with or without 50 g/ml trypsin and 1% (v/v) of Triton X-100. Trypsin digestion was terminated by the addition of soybean trypsin inhibitor (final concentration of 0.5 mg/ml). The samples were solubilized with an equal volume of lysis buffer (1% (v/v) Triton X-100, 1% (w/v) deoxycholate, and 2% (w/v) SDS) for 16 h and diluted to 0.2% (w/v) SDS prior to immunoprecipitation with an anti-apoB antiserum (Roche Molecular Biochemicals). [ 35 S]ApoB was analyzed by PAGE/fluorography as described previously (7).
Sodium Carbonate Treatment-The total microsomes were rinsed twice with Tris-sucrose buffer to minimize cytosol contamination. The lumenal contents were released from total microsomes with 0.1 mM sodium carbonate, pH 11.3, by gentle mixing using a nutator for 30 min at room temperature. The lumenal contents were separated from microsomal membranes by ultracentrifugation (400,000 ϫ g, 4°C, 16 min).
Ultracentrifugation of Lipoproteins-Conditioned media or the lumenal contents of microsomes were fractionated by cumulative rate flotation ultracentrifugation. Four ml of the sample was adjusted to d ϭ 1.10 g/ml with KBr, and loaded onto the bottom of a Beckman SW40 centrifuge tube. The sample was overlaid with 3 ml of d ϭ 1.065 g/ml NaBr, 3 ml of d ϭ 1.02 g/ml NaBr, and 3 ml of d ϭ 1.006 g/ml NaCl. After ultracentrifugation (40,000 rpm, 20°C, 148 min), VLDL 1 (S f Ͼ 100) was collected from the top 1 ml of the gradient. After an additional ultracentrifugation (37,000 rpm, 15°C, 18 h), VLDL 2 (S f 20 -100) and other lipoproteins were collected from top to bottom into 12 fractions. [ 35 S]ApoB in each fraction was immunoprecipitated as described previously (7). In some experiments, the samples were fractionated by sucrose density gradient ultracentrifugation or sequential flotation ultracentrifugation (d ϭ 1.02 g/ml) as described previously (7).
Analysis of Lipids-[ 3 H]Lipids were extracted with CHCl 3 /CH 3 OH/ H 2 O (4:2:3; by volume) from cytosol, microsomal membrane, and micro-somal lumen, and separated by TLC as described previously (7). The radioactivity associated with individual lipid species was quantified by liquid scintillation counting. The recovery of [ 3 H]TG was nearly 100% for cytosol, microsomal membrane, and microsomal contents as determined by monitoring radioactivity associated with each fraction in comparison to that associated with total postnuclear supernatant. In some experiments, human and rat B100-VLDL were recovered by immunoadsorption under non-denaturing conditions as described previously (7). [ 3 H]Lipids associated with B100-VLDL were extracted and analyzed by TLC as described above.
Other Assays-Protein (34) and TG mass (35) were determined by published methods. Phosphatidylcholine (PC) mass was quantified using an assay kit (Roche Molecular Biochemicals) according to the manufacturer's instruction.

A Unified VLDL Assembly Model for All ApoB Forms-Our
previous study suggested that assembly of B100-VLDL and B48-VLDL took different paths (7). Here we systematically analyzed assembly/secretion of VLDL with B64, B72, B80, or B100 in an attempt to determine if the apparent dual mechanisms were accounted for by apoB length. Under basal culture conditions (i.e. DMEM ϩ 20% serum), the buoyant density of secreted apoB-lipoproteins, as expected, was inversely related to apoB length. Thus, B64, B72, and B80/B100 were found predominantly in fractions with densities resembling those of  high density lipoproteins (HDL), low density lipoproteins (LDL), and VLDL, respectively (Fig. 1A, panels labeled -OA). Upon supplementation with oleate, however, all apoB forms were secreted as VLDL (Fig. 1A, panels labeled ϩOA). The enlargement of lipoprotein size was readily demonstrable for B64 and B72 by ultracentrifugation in a sucrose density gradient, but it was less evident for B80 and B100 because these proteins were already observed in VLDL (d Ͻ 1.02 g/ml) fraction even in the absence of oleate. Likewise, while the effect of MTP inhibitor BMS-197636 (0.2 M) on oleate-stimulated [ 35 S]B72-VLDL secretion was certainly observable (Fig. 1B, a), evidence of the BMS-197636 effect on [ 35 S]B100-VLDL secretion was not as clear although still discernible (Fig. 1B, b). (The inhibitor BMS-197636, unlike the photoactivated inhibitor BMS-192951 used previously (7), did not require ultraviolet treatment yet inhibited B100 secretion within 10 min of administration (data not shown).) These results led us to suspect that the supposed co-translational B100-VLDL assembly was a misinterpretation of data resulting from inadequate resolution of B100-VLDL (7), and that there was nothing but an imaginary dual mechanism for VLDL assembly in McA-RH7777 cells.
In the following experiments, we used cumulative rate flotation technique to resolve B100-VLDL 1 (S f Ͼ 100) and -VLDL 2 (S f 20 -100) from intermediate density lipoproteins (IDL)/LDL. Fig. 2A shows that, without oleate, B100 was secreted as VLDL 2 and IDL/LDL, and that secretion of VLDL, particularly VLDL 1 , was stimulated by exogenous oleate. (Secretion of B100-VLDL 1 was induced by oleate in a dose dependent manner (from 0.1 to 0.4 mM) with maximum stimulation at 0.4 mM (data not shown).) Separation of B100-VLDL subclasses by gradient gel electrophoresis under non-denaturing conditions confirmed the size difference between VLDL 1 and VLDL 2 (Fig.  2B, lanes 1-4), which was not as clear-cut as the quantum leap in particle size between B48-HDL and B48-VLDL (Fig. 2B,  lanes 5-8) Thus, the oleate-induced size enlargement of B100-VLDL (equivalent to a 2-fold increase in diameter) was accompanied with bulk TG incorporation, which was observable only after VLDL 1 and VLDL 2 were resolved.
Previous pulse-chase experiments suggested that assembly of B48-VLDL utilized pre-existing TG but B100-VLDL did not (7). Here we re-examined the utilization of pre-existing TG for B100-VLDL assembly with the same pulse-chase protocol (7) but using the cumulative flotation to separate VLDL 1 and VLDL 2 . Fig. 2C shows that exogenous oleate increased secretion of [ 3 H]oleate-labeled TG as B100-VLDL 1 by 5-fold and concomitantly decreased its secretion as B100-VLDL 2 by 30%. Oleate also increased secretion of [ 3 H]PC as B100-VLDL 1 by 8-fold, yet had no effect on [ 3 H]PC associated with B100-VLDL 2 (Fig. 2C). Thus, through the improved resolution of B100-VLDL particles, oleate-induced bulk TG incorporation, the hallmark of VLDL assembly irrespective of apoB length, became manifest.
To inquire whether B100-VLDL, like B48-VLDL, is also assembled post-translationally, we performed metabolic labeling (Fig. 3A) and pulse-chase experiments (Fig. 3B) under conditions that were optimal for VLDL assembly (i.e. 0.4 mM oleate was supplemented throughout the experiment). Fig. 3A shows that at the end of 10 or 20 min of labeling, [ 35 S]B100 was found mainly in IDL/LDL fraction with small amount in VLDL 2 fraction within the microsomal lumen. It also shows that VLDL 1 -B100 became detectable at 40 min of labeling. Since translation of B100 completes within 20 min (36), the delayed B100-VLDL 1 assembly suggests a post-translational mechanism. Next, we inquired whether or not B100-VLDL 1 assembly required MTP activity by pulse-chase analysis. After 20 min of pulse, the association of [ 35 S]B100 with VLDL 1 gradually became apparent within microsomal lumen at 15, 45, and 60 min of chase (Fig. 3B). Inclusion of BMS-197636 in the chase medium completely abolished the formation of B100-VLDL 1 (Fig. 3B, bottom). These combined results, reminiscent of the post-translational TG incorporation and MTP requirement observed for B48-VLDL assembly (7), provide evidence that B100-VLDL 1 assembly is achieved via a path similar to that for B48-VLDL.
The requirement of MTP for B100-VLDL assembly was also manifest by examining the effect of BMS-197636 on lipoprotein secretion. With doses increasing from 0.05 to 0.5 M, BMS-197636 not only diminished secretion of [ 35 S]B100 to Ͻ15% of control, but also decreased the size of B100-lipoproteins from VLDL to HDL (Fig. 4A and B). The BMS-197636 doses that abolished secretion of B100-VLDL 1 and -VLDL 2 were 0.05 M and 0.2 M, respectively, corresponding to 62% and 50% of normal MTP activity (Fig. 4C). (Although treatment with 0.2 M BMS-197636 resulted in only partial inhibition of cell-associated MTP activity, adding the same concentration of BMS-197636 to the assay mixture entirely abolished MTP activity (see Fig. 4C, closed square).) In contrast to that of B100, secretion of B48 (mainly as HDL) was relatively unaffected by BMS-197636 (Fig. 4B). These data showed clearly that the demand for MTP activity is positively correlated with the extent of apoB lipidation. The inhibited B100-VLDL secretion by BMS-197636, as expected, was accompanied with decreased VLDL-TG secretion (Fig. 4D). However, under no circumstances did BMS-197636 affect synthesis or accumulation of cell TG and PC (Table II).
MTP Activity Is Required for Mobilizing TG into Microsomes-Knowing that BMS-197636 did not affect TG synthesis, we hypothesized that it might impair TG mobilization into microsomes. To test this hypothesis, we determined the effect of BMS-197636 on the distribution of TG among different subcellular compartments. Of total microsomal TG, approximately 74% of 3 H-labeled TG (Fig. 5A, top) and 62% of TG mass (Fig.  5A, bottom) were found in the membranes with the remainder in the lumen in oleate-treated cells. (The high proportion of membrane-associated TG in hepatic microsomes was observed previously with rat (37) and rabbit hepatocytes (38).) Treatment with BMS-197636 entirely abolished the oleate-induced accumulation of TG within microsomal lumen, and also decreased the amount of TG associated with microsomal membrane by 30%. A small (ϳ10%) but reproducible increase in cytosolic TG by BMS-197636 treatment was observed, the quantity of which was equivalent to the decrease in total mi-crosomal TG. BMS-197636 had little effect on the distribution of 3 H-labeled PC (Fig. 5B, top) or PC mass (Fig. 5B, bottom) among different subcellular compartments.
The decrease in microsomal TG could either be a direct result of inhibited TG influx or else a consequence secondary to inhibited VLDL assembly. Examination of the BMS-197636 effect on TG influx showed that accumulation of [ 3 H]TG in microsomal lumen and membrane was decreased by Ͼ60% and 40%, respectively, during the first 30-min labeling (Fig. 6A, a  and b). Little difference in cytosolic [ 3 H]TG (Fig. 6A, c) and no secretion of [ 3 H]TG were observed during this period (Fig. 6A,  d). The inhibitory effect of BMS-197636 on lumenal [ 3 H]TG accumulation and on [ 3 H]TG secretion was also apparent at the end of 1 h of labeling. Thus, BMS-197636 specifically impairs the accumulation and attainment of newly synthesized TG within the microsomal lumen. When [ 3 H]TG accumulation in the microsomal lumen was plotted against the dose of BMS-197636 (Fig. 6B), an inverse relationship, similar to the residual MTP activity within the cells as a function of MTP inhibitor (Fig. 4C), was observed. The abolished B100-VLDL 1 assembly/ secretion (Fig. 4A) coincided with 55% decrease in lumenal [ 3 H]TG accumulation at 0.05 M BMS-197636 (Fig. 6B), sug-  gests strongly that the demand of MTP activity for B100-VLDL assembly correlates closely with the influx of TG into microsomal lumen.
The alternate possibility of diminished lumenal [ 3 H]TG being a consequence of impaired VLDL assembly was tested using cells treated with a low dose of BfA (0.2 g/ml). As shown previously (7,8), BfA effectively blocked bulk TG incorporation into VLDL, which was confirmed here by its effect on B100-VLDL assembly within the microsomal lumen (Fig. 6C). Under this condition, however, influx of [ 3 H]TG into microsomes and [ 3 H]TG secretion decreased marginally as compared with control (i.e. no BfA) (Fig. 6A, a, b, and d). The un-impaired accumulation of lumenal TG by BfA treatment was confirmed by prolonged lipid labeling (Fig. 6D, a) and mass measurement (data not shown). In these cells, however, the [ 3 H]TG associated with microsomal lumen (Fig. 6D, a) and secreted into the medium (Fig. 6D, b) were also sensitive to BMS-197636 treatment, providing another evidence that MTP activity is essential for TG accumulation within microsomes. Thus, the influx of TG into microsomal lumen may not be tightly coupled with VLDL assembly. Together, these data suggest that the diminished lumenal [ 3 H]TG upon BMS-197636 treatment is unlikely attributable to an inhibited VLDL assembly.
Finally, we tested (by pulse-chase experiments) the possibility, although unlikely, that the decreased lumenal TG accumulation by BMS-197636 was the result of impaired B100 translocation across the ER membrane (Fig. 7). The experiment was done under conditions where B100-VLDL assembly was maximized (with exogenous oleate) yet degradation of B100 was minimized (with ALLN). Between control and BMS-197636 (0.2 M)-treated cells, equal amounts of full-length [ 35 S]B100 were found during 0, 15, and 30 min of chase (Fig. 7, compare  lanes 1, 4, and 7 between A and B). Thus, MTP inhibition per se does not affect apoB translation. We then determined translocation of pulse-labeled (15 min) [ 35 S]B100 during chase by trypsin digestion of the isolated microsomes. In control cells 40 -50% of [ 35 S]B100 was sensitive to trypsin at 0, 15, and 30 min of chase (Fig. 7A, lanes 2, 5, and 8). Under these conditions, inhibition of MTP did not have an effect on the attainment of trypsin resistance in [ 35 S]B100 (Fig. 7B, lanes 2, 5, and 8). (The integrity of microsomal vesicles was verified by nearly 100% trypsin resistance of the ER-resident protein disulfide isomerase (data not shown).) Fragments of apoB that were resistant to exogenous trypsin (indicated by a bracket to the right of lanes 2 of Fig. 7, A and B) were observed whose intensity decreased with time (compare 0, 15, and 30 min of chase). These fragments might be derived from partially translocated B100 as reported previously (39). Thus, decreasing MTP activity by half (at 0.2 M BMS-197636) did not impede B100 translation/translocation. DISCUSSION Previously, based on the apparent differences in the kinetics of TG utilization into B100-VLDL and B48-VLDL, we concluded that the two VLDL species were assembled through different paths (7). Although the TG kinetics lends some plausibility to this conclusion, it does not stand up well to scrutiny. Attempts to delineate these distinct mechanisms using various C-terminal truncated apoB variants in this study failed to detect any multiplicity in the assembly pathway. Rather, a unified model that can accommodate VLDL assembly with all apoB forms (i.e. B48, B64, B72, B80, and B100) has emerged which invalidates our previous conclusion. By revealing the enlargement of VLDL particle size using cumulative rate flotation together with monitoring TG content ( Fig. 2 and Table  I), the oleate-inducible, MTP-dependent, post-translational TG incorporation becomes clearly demonstrable for B100-VLDL formation. Two critical points of this model are noteworthy.
The first point concerns that assembly of VLDL, regardless the length of apoB with which it associates, is achieved post-translationally. The second point highlights the fact that in McA-RH7777 cells, formation of TG-rich VLDL (e.g. B100-VLDL 1 ) will not occur unless the medium is supplemented with exogenous oleate. Under conditions where oleate was absent ( Fig.  2A) or where oleate was supplemented but MTP activity was partially inhibited (Fig. 4A, 0.05 M BMS-197636), the largest B100-VLDL particle that could be observed was VLDL 2 but not VLDL 1 . Thus, the hallmark of the post-translational VLDL assembly is the incorporation of bulk TG, which occurs only when lipid supply is sufficient (i.e. upon exogenous oleate supplementation) and is absolutely dependent on normal MTP activity.
Like in most cases where novel information comes about as a result of technical development, demonstration of the discontinuous, post-translational B100-VLDL 1 assembly becomes possible with the use of cumulative rate flotation technique to separate B100-VLDL subclasses (Fig. 3). This is because the previously commonly used density ultracentrifugation technique, although satisfactory for revealing the two-step process for B48-VLDL assembly (4,7,33), does not allow detection of size changes in VLDL particles containing large apoB (i.e. B80 or B100). Only by separating B100-VLDL 1 from B100-VLDL 2 were we able to show unequivocally that B100-VLDL 1 assembly shares three features common to B48-VLDL assembly: (a) absolute dependence of exogenous oleate, (b) post-translationally, and (c) extreme sensitivity to MTP inhibition. Employment of the cumulative rate flotation technique has also revealed that assembly of small B100-lipoprotein particles (i.e. B100-IDL/LDL and B100-VLDL 2 ) is achieved post-translationally (Fig. 3A). Under conditions where MTP activity was decreased to lower than 40% of control, the only observed B100lipoproteins in media were those of HDL size (Fig. 4A). The enlargement of secreted particles, shown as a gradual conversion of B100-HDL into B100-VLDL 2 (Fig. 4A), is correlated closely with increased residual MTP activity (Fig. 4C). Thus, the governing factor in apoB-lipoprotein core enlargement is the incorporation of bulk TG.
An important observation made in this study is that the accumulation and attainment of TG within microsomal lumen is a function of MTP activity. Experimental evidence supporting this conclusion includes (a) measurement of metabolically labeled TG (Figs. 5A and 6A) and of TG mass (Fig. 5A), and (b) demonstration of a direct correlation between the attainment of lumenal TG and the MTP activity (Fig. 6B). In principle, the steady state level of lumenal TG is determined by the rate of influx of TG through mobilization and the efflux via secretion. Since MTP inactivation inhibits TG secretion, yet it does not inhibit TG synthesis, the decreased influx TG is the most likely explanation for the diminished lumenal TG content. Although the origin of TG being mobilized into microsomal lumen is not determined in the current study, our result does suggest that the maintenance of a microsomal TG pool, regardless which being derived from de novo synthesis or from hydrolysis-reesterification of a storage pool, requires normal MTP activity. Another important, although less conclusive, result derived from this study is the demonstration that accumulation of lumenal TG can be independent of VLDL assembly in BfAtreated cells (Fig. 6, A and C). Observation of "TG particles" within the secretory compartments in cells where VLDL assembly is abolished (by genetically inactivating apoB expression) has been reported recently (40). Our data are in accord with this observation and provide indirect evidence that the MTP-mediated TG mobilization may lead to the formation of apoB-free "TG droplets" (41). A point of note is that in mice in which hepatic MTP was inactivated by gene targeting, lipid droplets were completely absent within the secretory compartment (42). These in vitro and in vivo findings together argue strongly that accumulation of bulk TG within microsomes is associated with normal MTP activity. A major challenge ahead to this theory, however, is to characterize these "TG particles" biochemically and to demonstrate that they are indeed precursors for VLDL assembly.
The core difficulty with all VLDL assembly models harks back to the old question concerning the path that is taken by bulk TG eventually being incorporated into apoB. The current studies indicate that MTP may play a role in facilitating the accumulation and attainment of TG in the microsomes. However, they did not address the question of whether MTP activity is required for incorporation of the microsomal TG into VLDL. It is clear that the demand for MTP is much greater for the assembly of TG-rich VLDL than for the assembly of dense particle. For instance, VLDL 1 assembly was abolished at 0.05 M BMS-197636, 4 times lower than the dose needed to abolish VLDL 2 assembly (Fig. 4A). These results are nicely in accord with previous data that formation of B48-VLDL also exhibits higher demand for MTP activity than that of B48-HDL (7). In addition, measurement of the residual MTP activity in BMS-197636-treated cells has shown a positive correlation between MTP activity and lumenal TG content in microsomes (Figs. 4C and 6B). These combined observations suggest the existence of a particular "threshold" MTP activity for the maintenance of sufficient lumenal TG, which is required for efficient assembly of TG-rich VLDL. Thus, the requirement of MTP activity is determined by the amount of TG to be utilized for lipoprotein assembly. The greater amount of TG to be recruited, the more MTP activity is required, and the more sensitive toward MTP inhibition.
A recent report suggested that MTP inhibition did not affect post-translational B100-VLDL assembly (3). In that study, McA-RH7777 cells cultured under basal condition (DMEM ϩ 20% serum) were treated with high dose BfA to arrest B100lipoprotein assembly at HDL stage. By removing BfA from and supplementing oleate into the medium, secretion of B100 lipoproteins resumed. Inhibition of MTP at this stage decreased secretion of total B100 by 25%, but it did not block conversion of the B100-HDL into B100-VLDL (3). Thus, it was concluded that the post-translational B100-VLDL assembly is achieved through a process insensitive to MTP inhibition. However, three caveats associated with these experiments are noteworthy and could potentially jeopardize the conclusion. First, the reported "B100-VLDL" whose assembly was insensitive to MTP inhibition is not B100-VLDL 1 . It is apparent from the current work that, under basal culture condition without exogenous oleate, only low level of TG accumulates within the microsomal lumen and only small, dense B100-lipoproteins (such as B100-IDL/LDL and B100-VLDL 2 ) are produced (Fig. 2). Even in the presence of exogenous oleate, accumulation of TG within microsomal lumen (Fig. 5A) or assembly of B100-VLDL 1 (Fig. 3B) does not occur if the activity of MTP is inhibited. Second, the report was unaware of that the demand for MTP is determined by the amount of TG loaded into B100-lipoprotein. It is clear from the present study that assembly of small B100-lipoproteins is much more resistant to MTP inhibition than that of B100-VLDL 1 . Since accumulation of TG within microsomal lumen is shown unaffected by BfA treatment (Fig. 6A), conceivably the reported "B100-VLDL" formation after BfA removal could result from recruitment of residual microsomal TG into B100-IDL/LDL and B100-VLDL 2 particles. Third, conclusion that MTP inhibition has no effect on VLDL assembly was drawn solely on the basis of measuring B100 associated with VLDL without considering the amount of TG incorporated. It is evident from the current study that B100-VLDL secreted from McA-RH7777 cells upon oleate treatment is heterogeneous, and that the majority of secreted TG is associated with a small portion of B100 (ϳ25% of total) in the VLDL 1 fraction (Table I). Thus, the requirement of MTP activity for B100-VLDL 1 assembly would not be revealed unless the bulk TG incorporation is measured. For these reasons, caution must be exercised in concluding that MTP activity is not required for bulk lipid incorporation during B100-VLDL assembly.
In summary, we have demonstrated that MTP plays an important role in the accumulation and attainment of bulk TG within the microsomal lumen, an event that can be separated from TG incorporation into mature VLDL but represents an indispensable requisite for the oleate-induced, post-translational assembly of VLDL.