In vitro reconstitution of assembly of apolipoprotein B48-containing lipoproteins.

Human apolipoprotein B48 (apoB48) and apoB15 (the NH2-terminal 48 and 15% of apoB100, respectively) were translated in vitro from their respective mRNAs using a rabbit reticulocyte lysate and microsomes derived from rat liver or dog pancreas. Synthesis of phosphatidylcholine and triacylglycerols was reconstituted in freshly isolated microsomes by the addition of precursors of these glycerolipids (acylcoenzyme A, glycerol 3-phosphate, and CDP-choline) before, during, or after translation. Assembly of apoB15 and apoB48 with newly synthesized phospholipids and triacylglycerols was favored by active, co-translational lipid synthesis. Moreover, translocation of apoB48 but not B15 into the microsomal lumen was increased in the presence of co-translational lipid synthesis. When apoB48 was translated in vitro, approximately 50% of apoB48 was buoyant at a density of <1.10 g/ml in the lumen of liver microsomes only when lipid synthesis was reconstituted during translation. Microsomal triacylglycerol transfer protein has been proposed to be essential for lipidation and/or translocation of apoB48. However, apoB48 was translocated into the lumen of dog pancreas microsomes in which the activity of the microsomal triacylglycerol transfer protein was not detectable. These data indicate that (i) apoB15 and apoB48 bind newly synthesized phosphatidylcholine during translocation; (ii) apoB48 but not apoB15 associates co-translationally with triacylglycerols; (iii) translocation of apoB48 but not apoB15 is stimulated by lipid synthesis; (iv) assembly of buoyant apoB48-containing lipoproteins can be reconstituted in vitro in the presence of active lipid synthesis; and (v) even in microsomes lacking microsomal triacylglycerol transfer protein activity, apoB48 is translocated into the lumen.

Human apolipoprotein (apo) 1 B100 is an extremely large (4,536 amino acids) hydrophobic protein that is essential for assembly and secretion of VLDL from liver. One unique property of apoB100 is that it is secreted only when assembled with lipids, a process that apparently occurs in the ER (1,2). ApoB secretion is modulated mainly post-transcriptionally (3), most likely by lipid supply (5; for review, see Refs. 4 and 6). For example, the addition of oleic acid to HepG2 human hepatoma cells stimulates both lipid synthesis and apoB secretion (5,7,8). ApoB secretion also requires active synthesis of phosphatidylcholine (9) and is decreased when the synthesis of cholesterol (10) or cholesteryl esters (11) is inhibited. The intracellular site(s), and mechanism, of the addition of lipids (cholesterol, cholesteryl esters, phospholipids, and triacylglycerols) to apoB are not yet clear.
ApoB secretion requires microsomal triacylglycerol transfer protein (MTP), which is expressed only in liver and intestine (12)(13)(14)(15). In vitro, MTP transfers lipids, particularly neutral lipids, between membranes (16 -18). The critical requirement of MTP for apoB secretion was demonstrated in the human disease abetalipoproteinemia, in which apoB-containing lipoproteins are almost completely absent from serum (12,19) because the gene for MTP is mutated. Moreover, when nonhepatic, non-intestinal cells express certain recombinant truncated apoB variants, apoB is secreted only when MTP is coexpressed (13,14). MTP is a luminal ER protein associated with protein disulfide isomerase (20,21) and with apoB, in the microsomal lumen (22). These findings have suggested that MTP is involved in supplying lipids to apoB for assembly of lipoprotein particles in the lumen of the ER.
Transport of apoB out of the ER appears to be the ratelimiting step in apoB secretion (23), and it has been suggested that translocation of apoB across the ER membrane is a key regulatory step in VLDL secretion (1,6,7,24). One unique feature of apoB secretion is that, in contrast to the widely accepted model for secretion of typical secretory proteins, newly synthesized apoB appears to exist in two intracellular populations: a lipoprotein-associated form wholly within the ER lumen, and a membrane-associated form in which portions of apoB are exposed to the cytosol and are susceptible to degradation (7,23,25,26). One hypothesis is that lipid supply determines how much apoB is translocated into the ER lumen, and apoB that is not translocated is subsequently degraded (27,28). Evidence in support of exposure of some apoB to the cytosol is: (i) in intact microsomes some apoB is degraded by exogenously added proteases (1,25,29,30), and (ii) in HepG2 cells apoB can be co-immunoprecipitated with the cytosolic heatshock protein, hsp70 (31). However, the existence of cytosolic domains of apoB has been questioned (32,33).
Recent studies by Lingappa and colleagues (34 -37) on in vitro translation/translocation of carboxyl-truncated apoB variants have suggested that apoB undergoes an unusual translocation process in which the protein pauses transiently at distinct sites along the nascent chain during translocation. During translocational pausing, translation continues, and domains of apoB become exposed to the cytosol. ApoB chains that pause but fail to restart during translocation are proposed to remain * This research was supported by a grant from the Heart and Stroke Foundation of Alberta (to J. E. V.) and a postdoctoral fellowship (to A. E. R.) from the Alberta Heritage Foundation for Medical Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  spanning the microsomal membrane, generating segments of apoB which are exposed to the cytosol and are degraded upon proteolysis of intact microsomes. A new class of topogenic sequences, pause transfer sequences, which are rich in charged amino acids, has been shown to mediate translocational pausing (35,37). The reason why apoB but not other secretory proteins contains pause transfer sequences is not known. However, we and others (34) speculate that translocational pausing might provide time for apoB to associate co-translocationally with lipids and/or to become folded into a secretion-competent form.
In the present study we have reconstituted in a cell-free system the assembly of apoB with triacylglycerols and phosphatidylcholine and have generated buoyant lipoprotein particles in the microsomal lumen. Translocation of apoB48 but not the truncated apoB variant apoB15 is stimulated by active lipid synthesis. In addition, the ability of apoB48 to translocate in vitro into the lumen of dog pancreas microsomes, which lack MTP activity, suggests that MTP is not absolutely required for translocation of apoB48 into the ER lumen.  (38). Polyvinylidene difluoride membranes were obtained from Millipore. The cDNA encoding ␤-lactamase and reagents used for transcription of cDNAs were from Promega as was the rabbit reticulocyte lysate used for in vitro translation. Triton X-100, tosylphenylalanyl chloromethyl ketone-treated trypsin, soybean trypsin inhibitor, proteinase K, protein A-Sepharose, and micrococcal nuclease were purchased from Sigma. Sepharose 4B was from Pharmacia. The sheep anti-human apoB100 polyclonal antibody was purchased from Boehringer Mannheim. The polyclonal antibody directed against ␤-lactamase was as described previously (35). All reagents used for polyacrylamide gel electrophoresis were from Bio-Rad. The BCA protein assay kit was from Pierce. All other chemicals were purchased from either Sigma or Fisher Scientific.

Materials
Construction of Plasmids and Cell-free Transcription-Plasmids encoding apoB15, apoB48, and SLSTgG were generous gifts from Dr. V. R. Lingappa, University of California, San Francisco, and were constructed using a modified pSP64 vector (Promega). The plasmid SLSTgG encodes a chimeric protein consisting of the signal sequence of bovine prolactin linked to 100 residues of ␤-lactamase, 49 residues of the stop transfer sequence of the heavy chain, and 111 residues of chimpanzee ␣-globin (35). All plasmids were linearized downstream of the termination codons of their coding regions (35,36) and transcribed in vitro for 120 min using SP6 RNA polymerase Riboprobe Core System II (Promega).
Preparation of Microsomes-Male Sprague-Dawley rats (120 g) were fed a diet of normal rat chow ad libitum. Liver microsomes were prepared as described for isolation of dog pancreas microsomes (39). The majority of ribosomes were stripped from microsomes by treatment first with 100 mM EDTA (pH 7.4) for 30 min at 0°C, then with micrococcal nuclease to reduce background translation products due to endogenous RNA (39). Dog pancreas microsomes were purchased from Promega and in addition were prepared according to the method of Walter and Blobel (39). Microsomes were either used fresh or stored at Ϫ70°C.
Cell-free Translation and Translocation-Purified mRNA transcripts for apoB15, apoB48, ␤-lactamase, and SLSTgG were translated in a rabbit reticulocyte lysate (Promega) for 1 h at 30°C in the absence or presence of 4.0 A 280 units of microsomes and in the presence of 1 Ci of Tran 35 S-label/10 l of translation mixture (39). A protease protection assay was used to assess the degree of translocation of proteins into the microsomal lumen (40). Briefly, aliquots of translation products were digested with 0.4 mg/ml proteinase K for 30 min at 0°C in the presence or absence of 1.0% Triton X-100. Proteolysis was terminated with 5 mM phenylmethylsulfonyl fluoride. Translated proteins were subjected to electrophoresis on 7% polyacrylamide gels containing 0.1% SDS and visualized by autoradiography (1-16 h at Ϫ70°C).
Isolation of Luminal Proteins-In some instances, microsomal vesicles were reisolated after translation by ultracentrifugation through a 100-l cushion of 0.5 M sucrose for 4 min at 110,000 ϫ g (Beckman 100.1 rotor). Reisolated membranes were suspended in 100 l of 0.25 M sucrose containing either Tris-HCl (pH 7.5) or 100 mM sodium carbon-ate (pH 11.5), which releases both luminal contents and loosely bound membrane proteins from microsomes (41). Throughout this paper sodium carbonate-extractable proteins are defined as both luminal proteins and proteins that are loosely bound to microsomal membranes. The sample was incubated at 0°C for 30 min, and luminal contents were separated from membranes by centrifugation for 15 min at 110,000 ϫ g (Beckman 100.1 rotor). Membranes were resuspended in buffer containing 10 mM Tris-HCl (pH 8.3), 2% SDS, and 8 M urea. Luminal proteins were concentrated by precipitation with 12% trichloroacetic acid. The precipitate was dissolved in buffer containing 10 mM Tris-HCl (pH 8.3), 2% SDS, and 8 M urea. In some experiments, the density of luminal contents was adjusted to 1.10 g/ml with solid NaBr, and the sample was centrifuged for 2.5 h at 420,000 ϫ g (Beckman 100.2 rotor). ApoB translation products were immunoprecipitated from the top third and bottom two-thirds of the tube contents using anti-human apoB polyclonal antibody. Translation products were boiled for 2 min in buffer A (1% SDS, 150 mM Tris-HCl (pH 8.0), 150 mM sodium chloride, 10 mM EDTA, 1 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride). The concentration of SDS was adjusted to 0.15% with buffer A lacking SDS, and 1-2 l of antiserum was added. The samples were incubated overnight at 4°C on a rotatory shaker, after which 15 l of protein A-Sepharose was added, and samples were rotated for 3 h at room temperature. The Sepharose beads were washed three times with buffer A and twice with distilled water. Immunoprecipitated proteins were released from the Sepharose by boiling the beads for 2 min in buffer containing 8 M urea and 2% SDS.
Reconstitution of Lipid Synthesis in Rat Liver Microsomes-Freshly isolated microsomes were included in the translation mixture (39) with precursors of glycerolipid biosynthesis: oleoyl-CoA (50 M), glycerol 3-phosphate (100 M), and CDP-choline (1.5 M). Lipid precursors were added either for 1 h at 30°C at initiation of translation or were added for 1 h at 30°C before the addition of mRNA. In some instances, lipid precursors were added post-translationally for 1 h at 30°C after translation had been terminated by the addition of emetin (0.1 M) for 15 min at 25°C (35). In some translation experiments, microsomes were omitted. To determine the extent of co-translocational association of triacylglycerols or phosphatidylcholine with newly translated proteins, 1 Ci of [ 3 H]oleoyl-CoA or 1 Ci of [ 14 C]CDP-choline, respectively, was included in the translation reaction. After translation, microsomes were reisolated and luminal contents released with sodium carbonate as above. Newly synthesized apoB and ␤-lactamase were isolated from luminal contents by immunoaffinity chromatography under nondenaturing conditions using Sepharose 4B linked covalently to polyclonal antibodies directed against either human apoB100 or ␤-lactamase (1). Lipids were extracted from the beads (42), and phosphatidylcholine and triacylglycerols were isolated by thin layer chromatography in a twosolvent system consisting of chloroform:methanol:acetic acid:formic acid:water, 70:30:12:4:2 (v/v) followed by hexane:isopropyl ether:acetic acid, 65:35:2 (v/v). Proteins were subsequently extracted from the beads (1). Radiolabeling of proteins with 35 S and of lipids (phosphatidylcholine with 14 C and triacylglycerols with 3 H) associated with newly translated proteins, was determined.
Electrophoresis and Autoradiography-Proteins were separated by electrophoresis on 7% polyacrylamide minigels containing 0.1% SDS and then transferred to polyvinylidene difluoride membranes for 60 min at 100 volts, and used for autoradiography (1). In some experiments, minigels were impregnated with Amplify and dried. Membranes or dried gels were autoradiographed for 1-16 h at Ϫ70°C.
Assay of MTP Activity-Lipid transfer from donor membranes to acceptor membranes was measured in an assay similar to that described previously by Wetterau et al. (19). Donor and acceptor small unilamellar vesicles were prepared by bath sonication in 15 mM Tris-HCl (pH 7.5) containing 1 mM EDTA, 40 mM NaCl, and 0.02% sodium azide. The lipid transfer assay mixture contained donor membranes (40 nmol of egg phosphatidylcholine, 7.5 mol % cardiolipin, and 0.25 mol % glycerol tri-[1-14 C]oleate), acceptor membranes (240 nmol of egg phosphatidylcholine), and 5.0 mg of bovine serum albumin in a total volume of 0.7 ml. The reaction was started by the addition of a source of MTP. After 60 min the reaction was terminated by the addition of 0.5 ml of DE52 cellulose pre-equilibrated with 15 mM Tris-HCl (pH 7.6), 1.0 mM EDTA, and 0.02% sodium azide. The mixture was agitated for 5 min and then centrifuged at maximum speed in a Biofuge B centrifuge (Baxter Scientific, McGraw Park, IL) for 3 min to pellet the DE52 containing bound donor vesicles. Radioactivity was measured in 0.5 ml of supernatant. Background transfer in the absence of MTP was subtracted to calculate MTP-mediated transfer of triacylglycerol from donor vesicles.

ApoB15 and ApoB48
Co-translationally Bind Newly Synthesized Phosphatidylcholine-As a prerequisite for secretion, apoB100 and apoB48 are assembled into lipoprotein particles consisting of a hydrophobic core of neutral lipids (primarily triacylglycerols with some cholesteryl esters) covered by a surface monolayer of phospholipids, unesterified cholesterol, and apoproteins. Unlike large native apoB, apoB15 (the recombinant NH 2 -terminal 15% of human apoB100) has a density of Ͼ1.2 g/ml when secreted, implying that it is associated with little neutral core lipid. However, several hydrophobic domains, representative of those occurring throughout the apoB100 molecule, are present within apoB15. These regions are assumed to bind lipids in the assembled lipoprotein particle (43)(44)(45). The possibility exists, therefore, that some lipids, particularly phospholipids, are associated with secreted apoB15. Indeed, apoB17 is secreted in association with some triacylglycerols and cholesteryl esters (46), and the co-translational assembly of phospholipids with apoB has been suggested previously (1,47).
The co-translational association of phospholipid with apoB15 was investigated in in vitro translation experiments in which apoB15 mRNA was translated in the presence of [ 35 S]methionine and in the absence or presence of rat liver microsomes. Precursors of phosphatidylcholine biosynthesis (oleoyl-CoA, glycerol 3-phosphate, and [ 14 C]CDP-choline) were included either before or during translation. Alternatively, lipid precursors were added post-translationally, after the addition of emetin, which terminates translation (35). After the reaction was complete, microsomes were reisolated, and luminal contents and loosely bound membrane proteins were released by sodium carbonate treatment. ApoB15 was isolated from the sodium carbonate extract by immunoaffinity chromatography under nondenaturing conditions using anti-human apoB100 antibody linked to Sepharose 4B (1). ApoB15-bound lipids were extracted from the beads, and incorporation of 14 C into phosphatidylcholine was measured. Subsequently, 35 S-proteins were extracted from the Sepharose. When microsomes were omitted from the translation mixture, very little [ 14 C]phosphatidylcholine was bound to apoB15 (Table I). However, when microsomes were present and precursors of phosphatidylcholine biosynthesis were added either before or during translation, luminal apoB15 readily associated with [ 14 C]phosphati-dylcholine. In contrast, when lipid precursors were added posttranslationally, the amount of [ 14 C]phosphatidylcholine bound to apoB15 was less than 14% of that when the protein was translated in the presence of lipid precursors. The 14 C/ 35 S ratio in immunoprecipitated apoB15 (Table I) is indicative of how much newly synthesized phosphatidylcholine is co-translationally associated with newly translated apoB15. This ratio was 9 -11-fold higher when precursors of glycerolipid biosynthesis were added during (ratio ϭ 1.14 Ϯ 0.12) or before (ratio ϭ 1.42 Ϯ 0.03) translation than when added post-translationally (ratio ϭ 0.13 Ϯ 0.06) ( Table I). These data imply that once apoB15 is translocated into the lumen relatively little phosphatidylcholine subsequently binds to apoB15. The amount of luminal [ 35 S]apoB15 was similar whether lipid precursors were present or absent during translation.
In parallel experiments, lipid precursors were added during translation of the typical secretory protein ␤-lactamase in the presence of microsomes. Negligible amounts of [ 14 C]phosphatidylcholine were associated with immunoprecipitated ␤-lactamase ( 14 C/ 35 S ratio ϭ 0.004 Ϯ 0.002, Table I). These observations demonstrate that apoB15 but not ␤-lactamase cotranslationally associates with phosphatidylcholine.
ApoB48 was also successfully translated in vitro in both the presence and absence of microsomes (Table II). As was the case for apoB15, more [ 14 C]phosphatidylcholine was associated with apoB48 when lipid precursors were added pre-or co-translationally ( 14 C/ 35 S ratio ϭ 1.77 Ϯ 0.61 and 1.75 Ϯ 0.49, respectively), than post-translationally ( 14 C/ 35 S ratio ϭ 1.19 Ϯ 0.72) (Table II). These experiments also reveal that relatively more phosphatidylcholine associates post-translationally with apoB48 ( 14 C/ 35 S ratio ϭ 1.19 Ϯ 0.72, Table II) than with apoB15 ( 14 C/ 35 S ratio ϭ 0.13 Ϯ 0.06, Table I). Although we cannot fully explain this phenomenon, the greater post-translational association of phospholipid with apoB48 than with apoB15 might be related to a less efficient translation/translocation of apoB48 (because of its large size of ϳ225,000 kDa) compared with apoB15 (ϳ70 kDa). Alternatively, the greater association of lipids post-translationally with B48 than with B15 might reflect the presence of more lipid binding domains in apoB48 than in apoB15 (43,35), allowing more post-translational assembly of phosphatidylcholine with apoB48. Nevertheless, these experiments demonstrate that both apoB15 and apoB48 bind phosphatidylcholine during translation/ translocation.
The data in Table II also show that when active lipid synthesis was reconstituted during translation, approximately  5 M), 50 M oleoyl-CoA, and 100 M glycerol 3-phosphate before (pre-trans), during (co-trans), or after (post-trans) translation. For the last, translation was terminated by the addition of emetin. After translation, microsomes were reisolated and treated with sodium carbonate to release luminal contents. ApoB15 and ␤-lactamase were immunoprecipitated from luminal contents under nondenaturing conditions, phosphatidylcholine associated with immunoprecipitated proteins was extracted and isolated, and radioactive incorporation was measured. Immunoprecipitated proteins were extracted, and the incorporation of 35 Table I, by the addition of lipid precursors before (pre-trans), during (co-trans), or after (posttrans) translation. Microsomes were reisolated and treated with sodium carbonate to release luminal contents. ApoB48 and ␤-lactamase were immunoprecipitated under nondenaturing conditions, phosphatidylcholine associated with immunoprecipitated protein was extracted and isolated, and radioactivity was determined. Incorporation of 35 Table III and Fig. 1 (see below). ApoB48 Co-translationally Binds Triacylglycerols More Efficiently Than Does ApoB15-Since in vivo apoB48 but not apoB15 is secreted in association with a neutral lipid core, we next investigated whether or not these apoproteins bound newly synthesized triacylglycerols during in vitro translation/ translocation. Precursors of triacylglycerol biosynthesis ([ 3 H] oleoyl-CoA and glycerol 3-phosphate) were added to the translation mixture before, during, or after translation. ApoB48 co-translationally associated with newly synthesized triacylglycerols (Table III). When lipid precursors were added after, rather than before or during, translation, markedly less [ 3 H]triacylglycerol bound to apoB48. When lipid precursors were added co-translationally or pre-translationally, the 3 H/ 35 S ratio in immunoprecipitated apoB48 was 0.82 Ϯ 0.30 and 0.79 Ϯ 0.26, respectively, whereas this ratio was ϳ45% less when lipid precursors were added post-translationally ( 3 H/ 35 S ratio ϭ 0.43 Ϯ 0.15). A negligible amount of [ 3 H]triacylglycerol co-translationally associated with ␤-lactamase (data not shown). Moreover, as is the situation with apoB15 secreted from intact cells, far less newly synthesized triacylglycerol co-translationally associated with apoB15 ( 3 H/ 35 S ratio ϭ 0.09 Ϯ 0.01) than with apoB48 (ratio ϭ 0.82 Ϯ 0.30) (Table III).
These experiments indicate that apoB48 binds newly synthesized phosphatidylcholine and triacylglycerols during translation/translocation, whereas apoB15 associates co-translationally with newly synthesized phosphatidylcholine but not triacylglycerols.
Active Lipid Synthesis Stimulates ApoB48 Translocation-Tables II and III indicate that translocation of newly synthesized apoB48 into the ER lumen was stimulated ϳ2-fold when the translation mixture was supplemented with precursors of glycerolipid synthesis pre-translationally or co-translationally, rather than post-translationally. In contrast, translocation of apoB15 was not enhanced when lipid biosynthetic precursors were present during translation (Table I).
The impact of active lipid synthesis on apoB48 translocation was also assessed by analysis of the mass of apoB48 present in the microsomal lumen. Lipid precursors were either absent or present during translation. After translation was complete, microsomes were reisolated, luminal contents and loosely bound membrane proteins were separated from membranes by sodium carbonate treatment, and proteins were separated by polyacrylamide gel electrophoresis. The distribution of apoB48 between microsomal membranes and luminal contents was determined by autoradiography and densitometric scanning of the films. As shown in Fig. 1, when lipid precursors were added post-translationally, the majority (85.7%) of newly translated apoB48 was tightly associated with microsomal membranes, with only a small amount (14.3%) being extractable by sodium carbonate and therefore being either in the lumen or loosely membrane-associated. In contrast, when lipid synthesis was active during translation, the majority of apoB48 (74.5%) was translocated into the microsomal lumen, with a smaller amount (25.5%) being membrane-associated. These experiments demonstrate that active glycerolipid synthesis promotes apoB48 translocation into the microsomal lumen.
MTP Is Not Required for Translocation of ApoB48 into the Lumen of Dog Pancreas Microsomes-One factor that might be expected to control the extent of apoB lipidation and the rate of apoB translocation is the ER luminal protein, MTP. This protein is expressed in liver and intestine but not in pancreas (12). Therefore, the dependence of translocation of apoB48 on the presence of MTP was determined by in vitro translation of apoB48 in the presence of lipid precursors and either dog pancreas microsomes or rat liver microsomes. After translation, the extent of translocation of apoB48 into the microsomal lumen was assessed by a protease protection assay in which proteins of intact microsomes were assumed to be luminal if they were resistant to proteolysis by exogenously added protease (48). Fig. 2 shows that translation of apoB48 occurred approximately equally efficiently in the presence or absence of rat liver microsomes or dog pancreas microsomes. In the in vitro translation system, endogenous mRNAs are stripped from membranes before translation is initiated; therefore only species of mRNA that are added to the translation mixture (in this case for apoB48 and SLSTgG) are translated. Fig. 2 shows that in the incubation without microsomes, or in incubations with rat liver microsomes, a translation product corresponding to apoB48 was detected on the gel. However, with dog pancreas microsomes a doublet was evident in the region of apoB48. The upper band of the doublet clearly corresponds to full-length apoB48 as synthesized in incubations without microsomes or with liver microsomes. The lower band of the doublet is probably an alternative translation product derived from apoB48 mRNA, since apoB48 mRNA and SLSTgG mRNA were the only mRNAs present. One possible explanation is that the smaller apoB48 mRNA translation product was generated by translation from an alternative start site. Another possible origin of the truncated species of apoB48 is from translation of a partial degradation product of full-length apoB48 mRNA.
When rat liver microsomes were treated with proteinase K after translation, nearly all apoB48 was resistant to digestion and was therefore luminal. The lack of digestion of apoB48 by proteinase K was not due either to apoB48 being inherently resistant to proteolysis by proteinase K or to the protease being inactive, because in the presence of the detergent Triton X-100 all apoB48 was degraded by proteinase K. In dog pancreas microsomes, the upper band of the apoB48 doublet (corresponding to full-length apoB48) was almost completely resistant to digestion by exogenously added proteinase K. In contrast, the protein corresponding to the lower band of the doublet was not protected from exogenous protease and was therefore exposed on the outside of the microsomes. These data indicate that the truncated apoB48 species might not contain the NH 2 -terminal region of the full-length apoB48 and might therefore lack the signal sequence. Such a protein would be translated but not translocated across the membrane.
As a control to confirm the protease susceptibility of cytosolic portions of a transmembrane protein in our assay, the chimeric protein SLSTgG was co-translated with apoB48 in the same incubation mixtures. SLSTgG contains a stop transfer sequence and therefore has a transmembrane orientation (35). Almost all of the SLSTgG was cleaved when microsomes from dog pancreas and rat liver were treated with proteinase K, and a ϳ14-kDa fragment of the protein was generated which was resistant to proteolysis and was therefore luminal (Fig. 2). In the presence of detergent, the luminal fragment of SLSTgG was completely degraded by proteinase K. These data confirm the transmembrane disposition of this chimera and demonstrate that proteinase K had access to cytosolic domains of newly synthesized polypeptide chains. These experiments indicate that apoB48 is translocated efficiently into the microsomal lumen, even in pancreatic microsomes that have been reported to lack MTP (12). In vitro translated apoB15 was also translocated into the microsomal lumen equally efficiently in the presence of microsomes derived from either dog pancreas or rat liver (data not shown).
The mRNA for MTP is not detectable in human pancreas (12). As confirmation that dog pancreas microsomes also lack MTP activity, we compared the ability of microsomes from rat liver and dog pancreas to transfer triacylglycerols between membranes in vitro in a standard assay for MTP activity (19). Table IV shows that dog pancreas microsomes, both those purchased from Promega and those prepared in our laboratory, possessed only ϳ0.1% of the triacylglycerol transfer activity of rat liver microsomes.
These experiments suggest that in the in vitro translation system apoB48 translocates across microsomal membranes equally efficiently in the presence or absence of MTP activity.
Reconstitution of Assembly of ApoB48-containing Lipoproteins-We next investigated whether or not luminal apoB48 generated during in vitro translation was present in buoyant lipoprotein particles. ApoB48 mRNA was translated in the presence of rat liver microsomes and precursors of glycerolipid synthesis. After translation, microsomes were reisolated, and luminal contents and loosely bound membrane proteins were released with sodium carbonate. The density of the luminal contents was adjusted to 1.10 g/ml, and the samples were subjected to ultracentrifugation. The top third and the bottom two-thirds of the contents of the tube were collected individually. Proteins in each fraction were separated by polyacrylamide gel electrophoresis and analyzed by autoradiograpy. Fig.  3 shows that approximately half of the newly synthesized luminal apoB48 floated at a density of 1.10 g/ml and was therefore in the form of lipoprotein particles.

Role of Glycerolipid Synthesis in Translocation of ApoB15
and ApoB48 -We have examined the involvement of synthesis of triacylglycerols and phosphatidylcholine in translocation of apoB across microsomal membranes using an in vitro translation system. Human apoB48 (a component of triacylglycerolrich lipoprotein particles) and apoB15 (which is secreted from cells in association with very little lipid at a density of Ͼ1.2 g/ml) were translated in vitro in the presence of rat liver microsomes. Translocation of newly synthesized proteins across the microsomal membrane was assessed by a protease protection assay and by isolation of microsomal luminal contents. Both apoB15 and apoB48 were translated and translocated across the membranes. During translation/translocation, glycerolipid synthesis was reconstituted, and newly synthesized phosphatidylcholine, which is the major phospholipid of all plasma lipoproteins, became co-translationally associated with both apoB15 and apoB48. In contrast, newly synthesized triacylglycerols bound to apoB48 but not to apoB15. Our finding that triacylglycerols assemble with apoB48 co-translationally in vitro agrees with the suggestion of Boström et al. (49) that   FIG. 2. Translocation of apoB48 across microsomes from dog pancreas and rat liver. Purified mRNA transcripts encoding apoB48 and the control transmembrane protein SLSTgG were translated in a rabbit reticulocyte lysate supplemented with lipid precursors as described for Fig. 1. The reaction proceeded for 1 h at 30°C in the presence of Tran 35 S-label and in the presence or absence of 4.0 A 280 units of dog pancreas microsomes or rat liver microsomes/25 l of translation mixture (39). Aliquots of translation products were digested with proteinase K (50 g/ml) for 30 min at 0°C in the presence or absence of 1% Triton X-100. Proteolysis was terminated in the presence of 5 mM phenylmethylsulfonyl fluoride. Products were analyzed by electrophoresis on a 7% polyacrylamide gel containing 0.1% SDS, followed by autoradiography. Arrowheads indicate the positions on the gel of standard apoB48 and SLSTgG. The asterisk indicates the ϳ14-kDa peptide fragment generated by proteolysis of SLSTgG. Data are from one representative experiment of five similar independent ones. association of apoB with newly synthesized triacylglycerols begins before translation is complete. Our data are also consistent with the observation that apoB variants longer than approximately the NH 2 -terminal 28% of apoB100 are secreted in association with a neutral lipid core with density of Ͻ1.17 g/ml, whereas apoB molecules smaller than apoB28 are secreted with a density of Ͼ1.17 g/ml and presumably do not contain significant amounts of triacylglycerols (50,51). However, even the NH 2 -terminal 17% of apoB100 is apparently secreted from mouse C127 mammary-derived cells in association with some triacylglycerols (46). Our studies (Tables II and III and Fig. 1) also indicated that active glycerolipid synthesis during translation enhanced the translocation of newly synthesized apoB48 across the microsomal membrane. Moreover, when precursors of glycerolipid synthesis were added co-translationally, ϳ50% of luminal apoB48 was in the form of lipoprotein particles with a density of Ͻ1.10 g/ml. In contrast, translocation of apoB15 into the microsomal lumen was not stimulated by active glycerolipid synthesis. These data support the hypothesis that active synthesis of triacylglycerols and/or phosphatidylcholine stimulates the translocation of apoB species that form lipoprotein particles containing a neutral lipid core. In contrast, translocation of smaller truncated variants of apoB, which do not assemble into triacylglycerol-rich lipoprotein particles, is not stimulated by ongoing glycerolipid synthesis.
Although both triacylglycerol synthesis and phospholipid synthesis were reconstituted in the in vitro translation reactions, it is more likely that enhanced translocation of apoB48 was due to stimulation of triacylglycerol synthesis, rather than phospholipid synthesis, for the following reasons. First, both apoB15 and apoB48 co-translationally associate with phosphatidylcholine whereas triacylglycerols associate with apoB48 but not significantly with apoB15. Correspondingly, translocation of apoB48 but not apoB15 was stimulated when glycerolipid synthesis was reconstituted. Second, treatment of HepG2 cells with Triacsin D inhibits triacylglycerol synthesis and apoB secretion (5). However, even under conditions for which Triacsin D does not inhibit phosphatidylcholine synthesis, apoB secretion is inhibited. 2 It should be noted, however, that although it is generally believed that oleic acid stimulates apoB secretion in HepG2 cells because of increased synthesis of triacylglycerols (7,49,52), phospholipid synthesis is also enhanced by oleic acid (53). Third, when rat hepatocytes are deprived of choline, the synthesis of phosphatidylcholine but not triacylglycerols is inhibited, and secretion of apoB100 and apoB48 and their associated lipids is reduced (9). However, during choline deficiency, inhibition of apoB secretion does not result from impaired apoB translocation since the same number of apoB-containing particles is present in the lumen of the ER derived from livers of choline-deficient and choline-supplemented rats (30). The choline deficiency-induced reduction of apoB secretion appears to be due to increased degradation of defective apoB-containing particles in a post-ER compartment (54) rather than a defect in translocation. Moreover, we have recently shown directly in in vitro translation experiments that translocation of neither apoB15 nor apoB48 is impaired in microsomes derived from choline-deficient rat livers. 3 These data indicate, therefore, that decreased synthesis of phosphatidylcholine, with a corresponding reduction in the phosphatidylcholine content of microsomal membranes, does not restrict apoB translocation.
The source of triacylglycerols used for assembly of apoBcontaining lipoproteins such as VLDL and the mechanism by which these lipids are delivered to apoB are not yet clear. Gibbons and co-workers (55-57) and Yang et al. (58,59) have proposed that the bulk of triacylglycerols associated with apoB are derived from intracellular stores rather than from de novo synthesis. The proposal is that cytosolic triacylglycerols are hydrolyzed by a triacylglycerol lipase, the lipolysis products are reacylated in the ER, and the resulting triacylglycerols are assembled with apoB. Some triacylglycerols associated with apoB, however, are believed to be produced by de novo synthesis from glycerol 3-phosphate in the ER membrane. In support of this hypothesis, HepG2 cells do not assemble typical VLDL particles but instead secrete apoB-containing particles having densities corresponding to low and high density lipoproteins (60) which are triacylglycerol-poor compared with VLDL. Triacylglycerols are apparently synthesized by the de novo pathway in HepG2 cells, and some associate with apoB (61). In contrast, lipolysis of the cytosolic pool of triacylglycerols and subsequent reacylation of the lipolysis products are thought to occur inefficiently in these cells, resulting in inefficient incorporation of triacylglycerols from this source into apoB-containing lipoproteins (55)(56)(57)(58)(59)61). In our in vitro reconstitution system, triacylglycerols are assembled co-translationally with apoB48. Presumably, the source of triacylglycerols used for assembly with apoB48 in our in vitro system represents primarily triacylglycerols synthesized de novo from glycerol 3-phosphate; we would anticipate that the cytosolic pool of triacylglycerols is not available. The absence of an active pathway for hydrolysis of cytosolic triacylglycerol stores in the in vitro translation experiments might explain why relatively dense apoB48-containing lipoproteins are produced.
Role of MTP in Translocation of ApoB48 -We also examined the requirement of MTP for translocation of apoB48 into the microsomal lumen in the in vitro translation system. Secretion of apoB-containing lipoproteins from the liver, intestine, and cultured cells is known to be dependent upon MTP (12,13,15,62). One exception to this general rule is that when C127 mammary-derived cells are transfected with apoB41 cDNA, apoB41 is secreted in association with triacylglycerols, even  FIG. 3. A portion of translocated apoB48 is in the form of buoyant lipoproteins. ApoB48 mRNA was translated for 1 h at 30°C with Tran 35 S-label in the presence or absence of rat liver microsomes. The biosynthesis of triacylglycerols and phosphatidylcholine was reconstituted as for Fig. 1. After translation, microsomes were reisolated and treated with sodium carbonate, and luminal contents were separated from membranes by centrifugation. The density of luminal contents was adjusted to 1.10 g/ml, then samples were centrifuged for 3 h at 430,000 ϫ g. The top third and bottom two-thirds of the contents of the tubes were collected individually. Proteins were separated by electrophoresis on a 7% polyacrylamide gel containing 0.1% SDS and analyzed by autoradiograpy. Data are from one experiment representative of four similar ones. though MTP protein and activity are apparently absent (63). Although the exact function of MTP in lipoprotein assembly has not been defined, MTP is thought to transfer triacylglycerols to newly synthesized apoB in the ER lumen. Indeed, MTP catalyzes the transfer of neutral lipids (triacylglycerols and cholesteryl esters) (16), and to a lesser extent phospholipids (18), between membranes in vitro. MTP also associates with apoB in HepG2 cells (61). Even though MTP transfers phosphatidylcholine between membranes with only ϳ10% of the efficiency of triacylglycerols (18), the number of molecules of phosphatidylcholine with which MTP would come into contact on the microsomal membrane is ϳ10 times higher than the number of triacylglycerol molecules at this location (64).
In our experiments, apoB48 was translocated approximately equally efficiently into the lumen of microsomes derived from rat liver (which contains robust MTP activity) and dog pancreas (which lacks detectable MTP activity). Approximately 50% of in vitro translated luminal apoB48 in rat liver microsomes was in buoyant lipoprotein particles of density Ͻ1.10 g/ml. However, we could not detect any luminal apoB48 that floated at a density of Ͻ1.10 g/ml in dog pancreas microsomes (data not shown). These observations suggest that although MTP might not be required for translocation of apoB48 across the membrane, MTP might be required for assembly of apoB48 with sufficient lipid for formation of lipoprotein particles that are stable in the aqueous environment of the ER lumen. In previous experiments using cultured cells transfected with truncated forms of apoB100 we have also observed that apoB translocation can be dissociated from assembly of apoB with a neutral lipid core (27). Our experiments do not, however, eliminate the possibility that our in vitro reconstituted system of lipoprotein assembly might be lacking some normal requirements for lipoprotein assembly which are present in an intact animal or that MTP might augment apoB translocation in vivo.