Apolipoprotein B100 Exit from the Endoplasmic Reticulum (ER) Is COPII-dependent, and Its Lipidation to Very Low Density Lipoprotein Occurs Post-ER*

Hepatic apolipoprotein B100 (apoB100) associates with lipids to form dense lipoprotein particles in the endoplasmic reticulum (ER) and is further lipidated to very low density lipoproteins (VLDL). Because the VLDL diameter can exceed 200 nm, classical ER-derived vesicles may be unable to accommodate VLDLs. Using hepatic membranes and cytosol to reconstitute ER budding, apoB100-containing vesicles sedimented distinct from those harboring more typical cargo but contained Sec23. Moreover, ER exit of apoB was inhibited by dominant-negative Sar1. Budding required Sar1 regardless of whether oleic acid (OA) was added to stimulate apoB lipidation; therefore, either large apoB100-lipoproteins reside in secretory vesicles, or full lipidation occurs post-ER. Using membranes from cells incubated in the presence or absence of OA, we determined that apoB100-lipoproteins in ER vesicles had not become lipidated to VLDLs. VLDL particles resided in the Golgi, but not the ER, after fractionation of OA-treated cells. We conclude that apoB100-lipoproteins exit the ER in COPII vesicles, but under conditions favorable for VLDL formation final lipid loading occurs post-ER.

cargo capture is how cargo larger than this size can be transported beyond the ER.
One such cargo is the class of lipoproteins containing apolipoprotein B (apoB). ApoB has two forms, apoB48 (expressed in mammalian intestine and in rodent liver) and apoB100 (expressed in mammalian liver) (8). Lipoproteins transport waterinsoluble lipids, such as triglycerides, cholesterol, and cholesteryl esters in the plasma. The apoB-containing particles are chylomicrons (of intestinal origin), very low density lipoprotein (VLDL; of hepatic origin), and low density lipoprotein (LDL; a product of VLDL metabolism), and they all promote atherosclerosis. Depending on the amount of neutral lipid (cholesteryl ester and triglyceride) associated with chylomicrons and VLDL, the secretory pathways of the intestine and liver, respectively, have to accommodate lipoproteins that often possess diameters well in excess of Ͼ100 nm. In fact, it has been suggested that novel transport processes may be required for apoB-lipoprotein intracellular trafficking (9).
Another formidable barrier to transport these lipoproteins regards the complexity of the particle assembly process. In addition to apoB, cholesteryl ester, and triglyceride, phospholipids, cholesterol, and other apolipoproteins are also components of these particles. In the liver, the initial assembly involves the co-translocational association of apoB polypeptides with lipids, which is mediated by the microsomal triglyceride transfer protein and results in the formation of a relatively small (ϳ20 nm) dense lipoprotein particle. The other components, including additional lipids, are added subsequently. The outcome is the production of mature VLDL particles of 80 -200 nm in diameter (8). Although experimental data suggest that final maturation of VLDL takes place in the ER (10 -13), other results suggest that this event may occur in a post-ER or Golgi compartment (14 -17).
In the present report, we investigated both the nature of the ER exit process of hepatic apoB100-lipoprotein particles and the site of VLDL formation. These studies were performed in rat hepatoma cells (McArdle RH-7777), which have many features of primary hepatocytes (18), and in cell-free systems using enriched, intracellular components. In addition, we regulated the size of the mature, secreted, lipoprotein particles by maintaining cells in either a basal or a lipid-rich medium to stimulate triglyceride synthesis and thereby increase lipid loading. We found that ER exit of apoB100-lipoproteins employs a COPII-dependent mechanism, although the transport vesicles carrying apoB100 had a number of features distinct from those containing more typical cargo. In addition, we found that independent of the final size of the lipoprotein, relatively small, dense apoB100-lipoproteins exit the ER, and under conditions that support VLDL formation, additional lipid loading occurs post-ER.

EXPERIMENTAL PROCEDURES
Materials-pTarget-Sar1 T39N-HA plasmid was provided by J. Lippincott-Schwartz (National Institutes of Health), and a plasmid encoding Sar1 wild type (WT) in a pET3a vector and rabbit anti-Sec23 antibody were provided by W. Balch (The Scripps Research Institute). Rat liver Golgi membranes were provided by D. Sabatini (New York University School of Medicine). Rabbit anti-albumin and rabbit antitransferrin antibodies were obtained from Bethyl Laboratories Inc. (Montgomery, TX). Mouse anti-HA monoclonal antibody (HA11) and mouse anti-mannosidase II antibody were purchased from Covance (Princeton, NJ). Mouse anti-PDI and rabbit anti-calnexin antibodies were obtained from StressGen (Victoria, Canada), and rabbit anti-COPII and mouse anti-TGN38 antibodies were purchased from Affinity Bioreagents (Golden, CO). Rabbit anti-␣ 2 -macroglobulin antiserum was obtained from Biodesign International (Saco, ME). Mouse anti-syntaxin 6 and mouse anti-Vti1a antibodies were from BD Biosciences. Protein A-Sepharose was purchased from Amersham Biosciences. Cell culture media and related supplies were purchased from Invitrogen and Cellgro (Herndon, VA). Restriction enzymes were obtained from New England Biolabs (Beverly, MA). Other reagents were purchased from Sigma unless stated otherwise.
A DNA fragment containing an HA-tag sequence was inserted into the pET3a-Sar1 WT plasmid at the C terminus of Sar1 WT sequence, using ClaI and EcoRI cloning sites. DNA fragment encoding Sar1 WT-HA was subcloned from the pET3a vector using KpnI and BglII restriction sites into a pTARGET vector (Promega, Madison, WI) using KpnI and BamHI cloning sites.
Preparation of Cytosol and Microsomal Membranes-Rat liver cytosol was prepared as described (19). McA-RH7777 cells were grown to confluency on 100-mm tissue culture dishes and labeled with [ 35 S] protein labeling mix (PerkinElmer Life Sciences) for 3 h (100 Ci/ml/ dish). Microsomal membranes were prepared as described (19).
In Vitro Formation and Isolation of ER-derived Vesicles-In vitro vesicle-formation assay was performed mainly as described (20) with minor modifications (21). The budding reaction was carried out in total volume of 80 l and contained microsomal membranes (50 mg total protein) from McA-RH7777 cells, 600 mg of rat liver cytosol, and an ATP regenerating mix (1.5 mM ATP, 0.5 mM GTP, 10 mM creatine phosphate, 5 units/ml of creatine kinase) in BBV buffer (25 mM HEPES, pH 7.4, 250 mM sorbitol, 70 mM potassium acetate, 1.5 mM magnesium acetate, 5 mM EGTA). For control reactions either the ATP regenerating mix and rat liver cytosol were omitted, or the ATP regenerating mix was substituted with 10 units/ml of the ATP hydrolyzing enzyme, apyrase. The reaction was incubated at 37°C for 30 min. In pilot studies, recovery of apoB100 or albumin failed to significantly increase after this time (data not shown). The reaction was then placed on ice for 2 min. Membranes were collected by centrifugation at 135,000 ϫ g for 10 min at 4°C in Beckman TLA100 rotor (60,000 rpm). Pellets were resuspended in 50 l of ice-cold resuspension buffer (20 mM HEPES, 250 mM sucrose, pH 7.4) by repeated pipetting for 10 min on ice, and potassium acetate and magnesium acetate were added to obtain final concentrations of 150 and 2.5 mM, respectively. After the mixture was centrifuged at 16,000 ϫ g for 20 min at 4°C, 80% of the total of the medium-speed supernatant (MSS) was retained, and the rest of the supernatant was aspirated from the medium-speed pellet. The medium-speed pellet and MSS were resuspended in NET buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS), and apoB was immunoprecipitated with rabbit-anti-rat apoB antibody. For Western blot assays, MSS was centrifuged at 135,000 ϫ g for 10 min at 4°C in a Beckman TLA100 rotor (60,000 rpm) to recover the high-speed pellet containing transport vesicles. For protease protection assays, MSS was treated with trypsin at a final concentration 0.1 mg/ml as described (22), and the indicated proteins were immunoprecipitated with rabbit-anti-rat apoB antibody or ␣ 2 -macroglobulin antiserum.
Immunoisolation of Vesicles Containing apoB-In vitro budding reactions were performed, and vesicles and membranes were separated as above. MSS was treated under non-denaturing conditions with rabbit anti-rat apoB antibody. The supernatant was reserved, the beads were washed, and the immunoprecipitated vesicles, containing apoB, were released with NET buffer containing 2% SDS and 1% Triton X-100. ApoB and albumin were immunoprecipitated from the reserved super-natant, and the material that was released from protein A vesicles with rabbit-anti-rat apoB and rabbit-anti-albumin antibodies.
Density Gradient Fractionation and Immunoanalysis of Vesicles-McA-RH7777 cells were incubated with [ 35 S] protein labeling mix, and vesicles were generated from the membranes purified from these cells. The vesicles were separated on sucrose gradients as described previously (23). To determine the distribution of apoB among the fractions, an aliquot of each fraction was taken, and apoB was immunoprecipitated in NET buffer with anti-apoB antibody. The material released from protein A was separated by SDS-PAGE, and the recovered apoB was detected by autoradiography. To determine the distributions of COPII, albumin, and transferrin, the remainder of each fraction was centrifuged at 60,000 ϫ g for 1 h at 4°C in a Beckman TLA100 rotor, then pellets were resuspended in sample buffer (125 mM Tris-HCl pH 6.8, 20% ␤-mercaptoethanol, 2.5% bromphenol blue), and Western blotting was performed using the appropriate antibodies.
Subcellular Fractionation-McA-RH7777 cells were grown on four 100-mm tissue culture dishes and pre-incubated in a low serum medium in 5% CO 2 at 37°C for 16 h (Dulbecco's modified Eagle's medium containing 0.5% fetal bovine serum, 0.5% horse serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin). Two dishes of cells were incubated with BSA (0.16 mM), or to promote lipid loading of apoB, another two were incubated with OA/BSA (0.6 mM of OA; OA:BSA molar ratio ϭ 5:1) for 2 h in the same medium. Cells were labeled with [ 35 S] protein labeling mix for 3 h (100 Ci/ml) in the presence of BSA or OA/BSA in cysteine/methionine-free medium. Cells were washed with phosphate-buffered saline and harvested in 2.5 ml of homogenization buffer (10 mM Hepes, pH 7.4, 250 mM sucrose, 0.5 mM dithiothreitol, 1ϫ EDTA-free protease inhibitors mixture, 20 units/ml RNase inhibitor). Cells were homogenized by 10 passages through a 25-gauge needle and then 10 passages through a 30-gauge needle. Homogenates were centrifuged at 1900 ϫ g for 10 min at 4°C in a Beckman JS-4.3 rotor. Supernatants (2.3 ml) were loaded onto a sucrose density gradient containing the following layers (from the bottom): 56% sucrose (0.46 ml), 50% (0.92 ml), 45% (1.38 ml), 40% (2.3 ml), 35% (2.3 ml), 30% (1.38 ml), 20% (0.46 ml). After ultracentrifugation in a Beckman SW41 rotor at 39,000 rpm for 18 h at 4°C, 15 fractions were collected from each centrifuge tube (0.75 ml each) from the top to the bottom. The distribution patterns of the subcellular compartment markers were achieved by either enzymatic assay (for mannosidase II, Golgi marker) (24) or Western blot (for calnexin, ER marker; TGN38, trans-Golgi network marker).
Isolation of Microsomal Lumenal Contents-Lumenal contents were released from microsomes in the supernatant by treatment with 0.1 M sodium carbonate (pH 11) and deoxycholic acid (0.025%) for 25 min at room temperature, after which BSA was added to a final concentration of 5 mg/ml. These conditions have been shown previously (25) to quantitatively remove apoB from microsomal membranes but not to delipidate apoB-lipoproteins. The samples were centrifuged at 60,000 rpm in a Beckman TLA 100.4 rotor for 1 h at 4°C to separate microsomal membranes (pellet) from lumenal contents (supernatant). Lumenal contents were adjusted to pH 7.4 by the addition of 10% acetic acid.
Sucrose Gradient Separation of ApoB-containing Lipoproteins-Microsomal membranes were prepared from cells treated or mock-treated with OA as above, and in vitro budding reactions were performed with these microsomal membranes in a final volume of 500 l. Vesicles (high-speed pellet) were treated with sodium carbonate/deoxycholic acid to extract lipoproteins. The supernatants containing the extracted lipoproteins were recovered by centrifugation at 60,000 rpm for 1 h at 4°C in a Beckman TLA100.4 rotor. Lipoproteins were separated according to density by ultracentrifugation on a sucrose gradient (25) after adjusting 4.8 ml of the supernatant to a sucrose concentration of 12.5% (w/v). The adjusted supernatants were then placed on the top of a step gradient consisting of 1.9 ml of 49% sucrose and 1.9 ml of 20% sucrose. Phosphate-buffered saline (2.8 ml) was layered over the gradient. All solutions contained a protease inhibitor mixture. The gradients were centrifuged in a Beckman SW41Ti rotor at 35,000 rpm for 65 h at 10°C. Twelve fractions (0.95 ml each), with densities ranging from 1.005 to 1.21 g/ml, were collected from the top of the tube, and apoB was immunoprecipitated from each fraction with anti-apoB antibody. The density of each fraction was determined as the ratio of the weight of a 1-ml aliquot to the weight of a 1-ml aliquot of water.
Transfection Studies-LipofectAMINE PLUS reagent (Invitrogen) was used to transfect McA-RH7777 cells with either the pTARGET-Sar1 WT-HA plasmid (control) or the pTARGET-Sar1 T39N-HA plasmid according to the protocol of the manufacturer. Five h later, cells were permeabilized, fixed, and treated as described previously (26,27).
Immunoprecipitation and Western Blotting-Samples for immunoprecipitation were diluted in buffer containing 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS and incubated with appropriate antibodies at 4°C for 2 h followed by incubation with protein A-Sepharose at 4°C for 2 h. The beads were washed three times with the same buffer, and immunoprecipitated material was released by heating at 100°C in SDS-PAGE sample buffer. For Western blots, proteins were first resolved by SDS-PAGE and then transferred to a polyvinylidene difluoride membrane (PerkinElmer Life Sciences). Typically, the primary and secondary antibodies were used at 1:1000 and 1:10,000 dilutions, respectively, with final detection of signal by the Western blot Chemiluminescence Reagent Plus (PerkinElmer Life Sciences).

Cell-free Reconstitution of ApoB100 ER Export-An in vitro
budding assay (20,28) was used to study ER export of apoB100. Microsomes isolated from McA-RH7777 cells were incubated in the presence or absence of rat hepatic cytosol and an ATP regenerating system. There are two forms of apoB synthesized in rat hepatoma McA-RH7777 cells, apo48 and apoB100. Here, we monitored the results for apoB100 not only because these cells primarily synthesize this form, but also because only apoB100 is made by human liver, where it serves as the major structural protein for VLDL (8).
ApoB100 appeared in a vesicle-containing (v) fraction in a cytosol and energy-dependent manner (Fig. 1). As expected, the typical hepatic secretory protein albumin was also present in the vesicle fraction, and its appearance was also dependent on cytosol and energy (Fig. 1). The budding efficiencies of the complete reactions for apoB100 and albumin were 16% Ϯ 3 (n ϭ 4) and 23% Ϯ 5 (n ϭ 4), respectively, comparable with that of other proteins studied in similar in vitro budding assays (21,23). Budding was inefficient (less than 3%) if the reaction was performed on ice. The Sec23 protein (Fig. 1, COPII panel) was found in both membrane and vesicle fractions, suggesting that the vesicles were coated with COPII.
The Western blot analyses displayed in Fig. 1 showed that the resident ER proteins calnexin, Sec61␤, and PDI remained in the membrane fraction, making it unlikely that the vesicle fraction was created by nonspecific fragmentation of or leakage from the donor membranes. The absence of TGN38, a TGN marker, in the membrane fraction could not be explained by lack of immunoreactivity of the antibody (see Fig. 8) and is consistent with no or extremely low contamination of the microsomes by Golgi membranes, as predicted from the microsomal isolation protocol we employed (19). Further supporting the low level of Golgi contamination was an immunoblot analysis/comparison between our microsomal membrane preparation and purified rat liver Golgi membranes (kindly provided by David Sabatini, New York University School of Medicine). The Western blot signal for the Golgi proteins syntaxin 6 or Vti1a in the membrane preparation was Ͻ1% of the signal in an equivalent amount of Golgi protein (29,30). Overall, the data in Fig.  1 establish that we were successful in reconstituting apoB100 vesicle-mediated export from very highly enriched, ER-derived microsomes.
Domain Exposure of apoB100 in ER-derived Vesicles-Though it does not possess a classic transmembrane domain, we and others have shown that apoB100 has domains exposed to the cytosol when it is present in the ER and Golgi (31-33). We therefore wondered whether there is also domain exposure of apoB100 in the ER-derived vesicles. Thus, apoB from the vesicle fraction was immunoprecipitated with a polyclonal antibody under either denaturing or non-denaturing conditions. Fig. 2 (panel A) shows that the recovery of apoB100 was comparable using either condition, implying that the majority of vesicles carrying apoB100 have a domain or domains exposed to the cytosol.
Additional evidence for apoB100 domain exposure was obtained by incubating vesicles with exogenous trypsin. Fig. 2 (panel B) shows that apoB100, but not ␣ 2 -macroglobulin (a soluble protein known to be fully translocated (34)), was sensitive to trypsin. This result also demonstrated vesicle integrity. Albumin topology was not similarly examined, because it was shown previously (34) that albumin, presumably because of its highly globular structure, is trypsin-resistant in the absence or presence of Triton X-100.
ApoB100 Exits the ER in Vesicles Different from Those Containing the Typical Hepatic Secretory Protein, Albumin-To determine whether apoB100 resides in the same ER-derived vesicles as other hepatic secretory proteins, we performed the following sequential immunoprecipitation analysis. An in vitro ER budding reaction was performed, and vesicles and membranes were separated by low speed centrifugation. ApoB in the supernatant fraction (vesicles) was first immunoprecipitated with a specific antibody under non-denaturing conditions to isolate apoB-containing vesicles and to maintain their integrity. Aliquots of the supernatant and of the resuspended immunoprecipitate were then subjected to additional immunoprecipitations under denaturing conditions with either anti-apoB or anti-albumin antibody. That the vesicles maintained their integrity was confirmed by incubating them with trypsin after the non-denaturing immunoprecipitation with anti-apoB antibody. This resulted in protease sensitivity data identical to that presented in Fig. 2B, left panel (data not shown). Fig. 3A shows that after the first non-denaturing immunoprecipitation, apoB100 was depleted from the supernatant, in contrast to albumin (compare the two lanes 1). In addition, the vesicles containing apoB100 (lane 2, left) lack albumin (lane 2, right). These data suggest that apoB100 is packaged into vesicles distinct from those harboring a more typical secreted protein.
The Sucrose Gradient Density Distribution of Vesicles Containing apoB100 Is Distinct from Those of Other Hepatic Secretory Proteins-The finding that apoB100 and albumin might reside in distinct vesicle populations prompted us to explore whether vesicles containing apoB100 were distinguishable on sucrose gradients from vesicles containing other cargo. By comparison, it was shown previously (23) that glycosylphosphatidylinositol-anchored proteins reside in distinct vesicle populations from other cargo proteins. To this end, the ER-derived vesicles from an in vitro budding reaction were separated on sucrose gradients, and the distributions of apoB100, albumin, transferrin, and COPII were determined by either immunoprecipitation (apoB100) or Western blot analysis (all others).
As shown in Fig. 3B, apoB100 was found in the bottom (dense) fractions of the gradient, whereas albumin and transferrin co-migrated in the top (lighter) fractions. COPII was found in all fractions as expected for a protein that is vesicleassociated, because a lipid-free protein would not have floated in the sucrose gradient. These results further support our hypothesis that apoB100-lipoproteins are exported from the ER in vesicles distinct from those containing other secreted proteins.
The Export of ApoB100 from the ER Is Sar1-dependent-As shown in Fig. 1 and Fig. 3B, fractions containing apoB100 also contain Sec23, a major COPII component, but this does not directly establish COPII dependence for the ER exit of apoB100. To address this issue, indirect immunofluorescence studies were performed on rat hepatoma cells transfected with plasmids containing WT or the GDP-restricted mutant (T39N) of the Sar1 protein (5). The rationale for this experiment is based on the fact that the small GTPase, Sar1, recruits COPII components to begin the cargo selection/vesicle budding processes for secretory proteins (35). Sar1 activity is regulated by cycling between the GTP-and GDP-bound states, and expression of dominant-negative mutants in which Sar1 exists exclusively in either state prevents the ER-exit of typical cargo (5).
Our initial indirect immunofluorescence analysis using a polyclonal antibody that recognizes both apoB48 and apoB100 confirmed that both forms of apoB were predominately associated with the Golgi at steady state, based on co-localization with mannosidase II (panel A), TGN38 (panel B), and the lack of co-localization with the ER resident protein, PDI (panel C) (Fig. 4). We note, however, that the majority (Ͼ75%) of the total pool of apoB in this clone of rat hepatoma cells is apoB100 (36). That the fluorescent signals represented bona fide protein detection was supported by control experiments in which either the primary or secondary antibody was omitted, resulting in no detectable signal above background (data not shown).
Next, to determine whether apoB transport from the ER to A cell-free budding reaction was incubated at 37°C for 30 min, and vesicles and membranes were separated. ApoB was immunoprecipitated from the vesicular fraction with anti-rat apoB under nondenaturing (i.e. non-solubilizing) conditions. The supernatant was reserved, the beads were washed, and the immunoprecipitated vesicles were released with NET buffer containing 2% SDS and 1% Triton X-100. Proteins in the reserved supernatant (lane 1) and vesicle contents (lane 2) were immunoprecipitated with anti-rat apoB or antialbumin antibody, and the immunoprecipitates were separated by SDS-PAGE and assessed by fluorography. B, density gradient analysis of ER-derived vesicles. A cell-free budding reaction was incubated at 37°C for 30 min, and membranes and vesicles were separated. Vesicles were floated up from a Nycodenz cushion into a sucrose gradient as described (23). Fractions were collected and treated with anti-rat apoB antibody, and the protein was visualized by SDS-PAGE/fluorography. Alternatively, the vesicles were centrifuged at 60,000 ϫ g for 1 h at 4°C in a Beckman TLA100 to recover the pellets, and a Western blot was performed with anti-albumin, anti-transferrin, or anti-COPII antibody. Results shown are representative of three (A) or four (B) independent experiments.
the Golgi is Sar1-dependent, apoB localization studies were repeated in cells transfected with either Sar1 WT or the T39N mutant. As shown in Fig. 5, and consistent with the data presented in Fig. 4, apoB co-localized with the Golgi marker TGN38 in cells transfected with the Sar1 WT plasmid (panel A); in contrast, in cells transfected with the T39N-Sar1 mutant, there was now substantial co-localization with PDI ( Fig. 5B versus Fig. 4C).
As noted earlier, depending on the amount of lipid associated with a hepatic apoB-containing lipoprotein, its size can vary greatly from ϳ20 to upwards of 200 nm. To determine whether the ability of mutant Sar1p to restrict apoB to the ER depended on the size of the cargo, rat hepatoma cells were maintained in basal medium or in medium supplemented with OA before transfection with HA-tagged versions of either WT or mutant Sar1 plasmids. OA supplementation increases triglyceride synthesis and results in a sharp increase in the assembly and secretion of large (VLDL) apoB-lipoproteins (see below and in Fig. 8) (8).
Cells were fixed and stained with primary antibodies to apoB and HA (to confirm successful transfection). As shown in Fig.  6A, transfection of McA-RH7777 cells with the Sar1 WT plasmid resulted in a perinuclear, eccentric apoB signal, which, based on the data in Figs. 4 and 5, is most likely the Golgi. Importantly, transfection with mutant Sar1 T39N was equally effective in control and OA-treated cells in redistributing apoB to a diffuse location, which, based on the data in Fig. 5, is most likely the ER. Similar results for the effects of Sar1 T39N on the distribution of albumin were also observed (data not shown).
Taken together, the results shown in Figs. 4 -6 indicate that transport of apoB from the ER to the Golgi is Sar1-dependent. Furthermore, we propose that independent of their lipidation state, hepatic apoB-lipoproteins exit the ER in COPII vesicles.
Full Lipidation of ApoB-lipoproteins to Form VLDL Occurs Post-ER-The assembly of VLDL is a complex process that includes two major lipidation steps (8). Initial lipidation occurs in the ER and results in the formation of a dense apoB-lipoprotein, a "primordial" particle, which has a density in the LDL-HDL range. The site of further lipidation (the "second step") in which VLDL particles form remains unclear. Some investigators have concluded that VLDLs are completely assembled within the rough ER (12,13), whereas others have provided evidence of a requirement for post-ER events (16,17). Our data in the previous sections suggest that independent of the final lipidation state of an apoB-lipoprotein, the particle is exported from the ER in COPII vesicles. These data, however, do not address the state of particle lipidation upon ER export under conditions that favor VLDL assembly and secretion. Addressing this issue would not only determine whether the transport of lipoprotein cargos of considerably different sizes and characteristics are accomplished by similar mechanisms, but would establish the site, ER or post-ER, where additional lipids required for VLDL formation are added. To these ends, McA-RH7777 cells were incubated overnight with low serum medium and were then treated for 5 h with either BSA or with a complex of OA and BSA (OA/BSA) to induce the secretion of apoB100 as VLDL (36). After the first 2 h, [ 35 S]methionine/ cysteine was added to metabolically label total protein. Conditioned medium samples were reserved for density gradient analysis, and ER membranes were isolated (see "Experimental Procedures"). Cell-free budding assays were performed for 30 min, membranes and vesicles were separated by differential centrifugation, and the vesicle contents were released with sodium carbonate and deoxycholic acid (25,37). The contents of the vesicles, as well as aliquots from conditioned medium, were subjected to density sucrose gradient separation. ApoB was immunoprecipitated from the fractions collected after centrifugation, and the immunoprecipitates were resolved by SDS-PAGE.
As shown in Fig. 7A, ER-derived vesicles carried apoB100containing lipoproteins primarily in the LDL or HDL density range, independent of OA treatment. At the same time, apoB100 was enriched in the VLDL fraction in the medium from cells treated with OA/BSA (compare Fig. 7B, lane 1, lower row, to the other rows in the figure). Thus, although the amount of apoB100 that was secreted in a highly lipidated state varied significantly between the BSA-and OA/BSAtreated cells, the ER-derived vesicles in both cases carried a relatively dense, lipid-poor particle with the same density as a primordial lipoprotein that had undergone only the first step of lipidation.
The result presented in Fig. 7 suggested that the final lipidation of apoB100-lipoproteins to form VLDL particles occurs post-ER. To test this hypothesis directly, rat hepatoma cells were treated as above. They were then lysed and homogenized, and the post-nuclear supernatants were subjected to subcellular fractionation. Distributions of the ER, Golgi, and TGN markers are shown in Fig. 8A and indicate successful enrichment of each compartment. Next, the fractions corresponding to the TGN (3-4), Golgi (6), and ER (8 -12) were treated with sodium carbonate/deoxycholic acid to release the lumenal contents, which were then subjected to sucrose density gradient ultracentrifugation to determine the apoB100 contents in the fractions corresponding to VLDL and VLDL remnants (intermediate density lipoprotein). ApoB was immunoprecipitated from the fractions collected after centrifugation, and the immunoprecipitates were resolved by SDS-PAGE.
As shown in Fig. 8, VLDL density particles appeared more predominantly in the Golgi and TGN fractions from cells treated with OA/BSA (panels C and D, respectively, lanes 1). In contrast, the density distributions of apoB100-lipoproteins in the ER microsomes were similar after either treatment and resembled those found in the ER-derived vesicles (compare Fig.  7A and Fig. 8B). These results add further evidence that apoB100 is exported from the ER as part of a relatively dense particle, independent of the final form of the lipoprotein. Under conditions favorable for VLDL assembly, additional lipid appears to be loaded post-ER, either before or after apoB100lipoproteins reach the Golgi. DISCUSSION It is now generally accepted that for the majority of proteins, COPII-coated vesicles of ϳ60 -80 nm mediate protein export from the ER (1). Although early data to support this model came primarily from studies in yeast, cell-free systems were developed subsequently that allowed for detailed mechanistic investigations of the formation of ER-derived transport vesicles (3,6,7,20,28). To characterize how apoB100-lipoproteins exit the ER, we used a cell-free system consisting of rat hepatoma microsomal membranes and rat hepatic cytosol. Hepatic apoB100-containing lipoproteins, when fully lipidated, are VLDL particles, which are larger than canonical COPII vesicles. VLDL serves an important physiological role by transporting endogenously synthesized lipids, particularly cholesterol and triglycerides, from the liver to peripheral tissues. ApoB100-lipoproteins also have pathophysiological properties as the bearers of the cholesterol that accumulates in coronary arteries during the formation of atherosclerotic plaques (8).
Because of their size, it was speculated previously (9) that the larger apoB100-lipoproteins particles might require special factors for export from the ER or utilize COPII-independent transport mechanisms. However, we present evidence under "Results" suggesting COPII-dependent export of hepatic apoB100-lipoproteins, independent of the final size of the particles. Nonetheless, despite this similarity to the export of other secreted proteins, the vesicles containing apoB100 have a number of distinct properties. 1) They lack abundant, constitutively secreted proteins. 2) Their densities on sucrose gradients are different from vesicles containing albumin and transferrin. 3) Although apoB100 is not a transmembrane protein, a portion of the apoB100 in the particles was exposed to the cytosol, consistent with data obtained in primary hepatocytes and hepatoma cells (31,32,38). 4) The density distribution of the apoB100-lipoproteins in the ER-derived vesicles was relatively constant and lacked the most lipidated species, VLDL, even though secreted particles varied significantly in buoyant density size. Consistent with these data, subcellular fractionation suggested post-ER lipidation when VLDL secretion was favored (see below). 5) Electron microscopy studies indicated that, in addition to the expected 60 -80-nm coated vesicular structures observed in the supernatant of the in vitro budding reaction, larger, ϳ100-nm vesicles were present (data not shown). Therefore, we suggest that the export, subsequent trafficking, and perhaps site of final lipidation for apoB100containing lipoproteins is specialized in the liver relative to other "bulk" secreted proteins. This conclusion is supported by the inability to reconstitute efficient assembly or secretion of apoB100-lipoproteins, especially VLDL, in non-hepatic cells. Moreover, because the formation of these vesicles appears to be COPII-dependent, our data suggest that COPII vesicles can adjust to the demands related to particular cargoes.
Recently, the existence of distinct vesicles involved in the ER export of other proteins has been reported. Perhaps most relevant to the present study is the ER export of the intestinal form of apoB (apoB48). Intestinal apoB48-lipoproteins (chylomi- crons) are considerably larger than hepatic lipoproteins (they can reach 1000 nm in diameter). In fractionation studies, the vesicular structures containing chylomicrons were not in the denser portion of the gradient, as we found for hepatic apoB100-containing vesicles (Fig. 3B), but in the lighter portion (39). Also in contrast to our data, apoB48 recovery in the supernatant was not inhibited but actually increased ϳ10-fold when Sar1p-immunodepleted cytosol was used in cell-free budding reactions. Although budding was Sar1-indepedent, these vesicles still somehow acquired COPII components. The relationship between these findings to the very recent report (40) that mutations in a Sar1 family member are associated with defects in chylomicron secretion is not clear. Nevertheless, in some aspects, the ER export, subsequent processing, and phys-ical characteristics of chylomicron-containing vesicles may be quite different from those for hepatic apoB100-lipoproteins, though in other aspects, there may be common sorting/trafficking mechanisms for apoB-lipoproteins in both cell types.
Another demonstration of the existence of distinct ER-derived vesicles was from Riezman and co-workers (23), who found in a yeast cell-free system that glycosylphosphatidylinositol-anchored proteins, such as Gas1p, are transported in vesicles with different properties than those harboring an integral membrane and soluble cargo protein, known as Gap1p and gp ␣ F, respectively. However, the requirement for COPII function was not directly examined in that study. The authors concluded that the existence of vesicles with distinct properties represented protein sorting that occurred at an early stage in FIG. 7. ER-derived vesicles do not contain VLDL density lipoproteins. A, proteins were labeled by incubation of cells with [ 35 S] protein labeling mix, and vesicles were generated from microsomes that had been isolated from untreated cells or cells treated with OA. The ER-derived vesicles were then incubated with sodium carbonate and deoxycholic acid to release their contents, which were subjected to sucrose density gradient centrifugation to separate apoB-containing lipoproteins. ApoB was immunoprecipitated from each fraction, and the immunoprecipitates were separated by SDS-PAGE and assessed by fluorography. B, conditioned media from the cells treated in panel A were subjected to the same sucrose density gradient, and apoB was analyzed as in A. By densitometry, the percent of apoB100 in density fraction 1 was as follows: panel A, 7% Ϯ 2 (ϪOA) and 3.5% Each fraction was treated with sodium carbonate and deoxycholic acid to release the contents of the microsomes, which were then subjected to sucrose density gradient fractionation to separate apoB-containing lipoproteins. ApoB100 content was assessed by immunoprecipitation/SDS-PAGE/fluorography. The first three fractions (from 12) from the top of the sucrose density gradients are shown in the figure and are from one complete experiment. IDL, intermediate density lipoprotein. the secretory pathway. Consistent with this view, in a yeast cell-free system, Schekman and co-workers (41) reported the packaging of the plasma membrane ATPase subunit, Pma1p, into vesicles that were somewhat larger (ϳ87 nm) than standard COPII vesicles; in this case, Lst1p, a homologue of the Sec24p COPII component was required for transport.
Our data also suggest post-ER lipidation of dense apoB100lipoproteins, consistent with studies in cells of hepatic origin (14 -17). We believe that selective sorting represents an efficient means to direct vesicles containing dense apoB100-lipoproteins to the site where final lipidation to VLDL occurs, and the exposure of a portion of apoB100 to the cytoplasm (see Fig. 2B and Refs. 31-33) might target the cargo to this site.
A specialized sorting process for apoB100-lipoproteins would be consistent with a number of recent reports. First, vesicles containing procollagen exit the ER in a COPII-dependent manner, but they rapidly undergo a COPI-dependent, pre-Golgi sorting step; dominant-negative Arf1p, which blocks COPI-coat assembly, prevented this sorting step (42). Second, and particularly relevant to the first example, Olofsson and co-workers (16) have shown that dominant-negative Arf1p blocks VLDL formation. Third, Balch and co-workers (43) have reported that the cystic fibrosis transmembrane conductance regulator might utilize a non-conventional pathway during its journey from the ER and ultimately to the plasma membrane, one that takes place in COPII vesicles, but that occurs independent of Arf1, Rab1a/Rab2, and syntaxin 5.
In summary, this report represents the first characterization of the ER packaging and exit of hepatic apoB100-containing lipoproteins. Because of the distinctive properties of the vesicles, coupled with recent results on protein sorting in the secretory pathway, it is likely that a highly specialized machinery catalyzes VLDL assembly and secretion. We suggest that the continued characterization of this process will illuminate not only the fundamental features of lipoprotein assembly and secretion but might explain why this process is limited to cells of hepatic and intestinal origin.