Identification of domains in apolipoprotein B100 that confer a high requirement for the microsomal triglyceride transfer protein.

The microsomal triglyceride transfer protein (MTP) is required for the assembly and secretion of apoB-containing lipoproteins. To investigate the role of MTP in lipoprotein assembly, we determined the ability of carboxyl-terminally truncated forms of apoB to be secreted from cells treated with the MTP inhibitor 4'-bromo-3'-methylmetaqualone (Benoist, F., Nicodeme, E., and Grand-Perret, T. (1996) Eur. J. Biochem. 240, 713-720). In Caco-2 and mhAT3F cells that produce apoB100 and apoB48, the inhibitor preferentially blocked apoB100 secretion. When the inhibitor was tested on McA-RH7777 cells stably transfected with cDNAs encoding human apoB100, apoB72, apoB53, apoB29, and apoB18, the secretion of apoB100, apoB72, and apoB53 was preferentially impaired relative to apoB48 and shorter forms. To delineate the region between apoB48 and apoB53 that has a high requirement for MTP, we used puromycin to generate a range of truncated forms of apoB in HepG2 cells. The secretion of apoB53 and longer forms of apoB was markedly affected by low concentrations of the MTP inhibitor (approximately 1 microM), whereas apoB51 and smaller forms of apoB were only affected at higher concentrations (> 10 microM). The size-related sensitivity to MTP inhibitor was not due to late processing or retention, since the same result was observed when nascent lipoproteins were isolated from the endoplasmic reticulum. The MTP inhibitor did not alter the density of the secreted lipoproteins, indicating that each apoB polypeptide requires a minimally defined amount of lipid to attain a secretable conformation. Our results suggest that the folding of the domain between apoB51 and apoB53 has a high requirement for lipid. This domain is predicted to form amphipathic alpha-helices and to bind lipid reversibly. It proceeds and is followed by rigid amphipathic beta-sheets that are predicted to associate with lipid irreversibly. We speculate that these domains enable apoB to switch from a stable lipid-poor conformation in apoB48 to another lipid-rich conformation in apoB100 during lipoprotein assembly.

Apolipoprotein B (apoB) 1 mediates the formation of triacylglycerol-rich lipoproteins in the liver and intestine. In humans, the liver secretes a large form of 4536 residues called apoB100, whereas the intestine secretes apoB48 corresponding to the 48% amino-terminal part of apoB100 (reviewed in Ref. 1). The two forms derive from the same gene by the process of sitespecific cytidine deamination of nucleotide 6666 of apoB100 mRNA (2). This generates a stop codon and defines the carboxyl terminus of apoB48 (3,4). Rodent hepatocytes produce both apoB100 and apoB48 (5). A variety of nonsense or frameshift mutations of the human apoB gene have been described in familial hypobetalipoproteinemia (6,7). Affected individuals typically produce carboxyl-terminally truncated forms of apoB. The lipid:protein ratio and the size of the lipoprotein are reduced according to the size of apoB truncated forms, leading to hypolipidemia. This suggests that the length of apoB determines the amount of lipid that can assemble into a lipoprotein particle.
The intracellular assembly of apoB with lipid in hepatocytes and enterocytes absolutely requires a microsomal triglyceride transfer protein (MTP) complex (8 -15). Carboxyl-terminally truncated forms of apoB longer than apoB23 are only secreted as lipoprotein particles, whereas smaller forms of apoB are secreted with little or no associated lipid and do not require MTP to attain a secretion-competent conformation (8 -10).
The mechanism of triglyceride-rich lipoprotein assembly is not a simple size-related lipidation of apoB. ApoB48 has been demonstrated to be lipidated through two discrete steps in rat liver cells (16,17). The second lipidation step of apoB48 is not governed by the size of the apoB polypeptide but seems to be triggered by short hydrophobic sequences within apoB48 (18). This could explain why intestinal chylomicrons are more lipidrich than hepatic very low density lipoproteins despite the fact that they contain apoB48 rather than apoB100. Whether or not the second lipidation step requires MTP activity is still under debate (19,20). Partial inhibition of MTP has been reported to affect the secretion of apoB100 but not of apoB48 from intestinal Caco-2 cells (21,22). By contrast, the data of Jamil et al. (23) indicate that the secretion of apoB48 and apoB100 from McRH7777 is similarly impaired.
Little is known about amino acid sequences within apoB100 that are involved in the lipoprotein assembly process. Using computer modeling, Segrest et al. (24) have suggested that apoB100 is composed of a pentapartite structure of three ␣-he-lixes alternated with two ␤-sheets involved in lipid binding. The aim of the present study was to determine whether the obligatory role of MTP for the assembly and secretion of very low density lipoproteins is related to a defined domain of apoB100. To assess this, we compared the secretion from various cells of a series of carboxyl-terminally truncated forms of apoB100 in the presence of up to 40 M of a specific MTP inhibitor (25). We were able to localize a structural domain within apoB100 that confers a high requirement for MTP activity, suggesting that this particular domain plays a key role in the lipoprotein assembly process.

EXPERIMENTAL PROCEDURES
Materials-L-[ 35 S]Methionine (Redivue TM , 37 TBq/mmol) was purchased from Amersham Pharmacia Biotech. Culture media, additives and fetal calf serum were obtained from Life Technologies, Inc. The MTP inhibitor 4Ј-bromo-3Ј-methylmetaqualone was synthesized by GlaxoWellcome. Molecular weight markers for electrophoresis were Rainbow TM (Amersham Pharmacia Biotech). All other chemicals were from Sigma.
Cell Culture-HepG2 and Caco-2 cell lines were obtained from the American Type Culture Collection. The MhAT3F mouse hepatocyte-like cell line, derived from mice transgenic for SV40 genes under the antithrombin III promoter, was provided by B. Antoine (26). McA-RH7777 stably transfected with human apoB truncated variants were maintained in culture as described previously (27,28). Cells were seeded into 24-well plates (200,000 cells/1.7 cm 2 ) containing basal Eagle's medium supplemented with penicillin and streptomycin (100 units/ml each) and 10% heat-inactivated fetal calf serum in a humidified incubator (5% CO 2 ) at 37°C. All cells were used after 4 days of culture, except for Caco-2 cells, which were used after 10 days of culture to allow partial differentiation.
Metabolic Labeling-To avoid problems of fatty acid availability (29), all experiments were performed in the presence of oleic acid (0.5 mM) complexed to albumin prepared as described previously (25). The MTP inhibitor was dissolved in Me 2 SO/ethanol (1:9, v/v) at 4 mM and diluted in ethanol down to 0.01 mM before the addition to the culture medium. Cells were incubated for 30 min in methionine-free RPMI 1640 medium before the pulse labeling with L-[ 35 S]methionine (0.4 -2 MBq/well, 15-30 min of pulse). The chase was performed by replacement of the medium with RPMI 1640 medium. In experiments using mhAT3F or Caco-2 cells, the chase duration was 2 h. For HepG2 cells, apoB truncated forms were generated by performing a 150-min chase in the presence of 10 M cycloheximide and 150 M of puromycin as described by Boren et al. (30). Stably transfected McA-RH7777 cells expressing human truncated apoB forms were labeled for 4 h.
ApoB Immunoprecipitation-Secreted apoB forms were quantified by immunoprecipitation from culture medium after the addition of 1 ml of 60 mM Tris buffer, pH 7, 2 mM ETDA, 1% (v/v) Nonidet P-40, 1 M NaCl, 1 mg/ml bovine serum albumin, 36 g/ml aprotinin, 1 g/ml antipain, 50 g/ml 4-(2-aminoethyl)benzenesulfonyl fluoride. Immunoprecipitations were performed using goat anti-human apoB antiserum (Sigma catalog no. 357-25) after preclearing with gelatin-agarose beads and protein G-Sepharose beads. The gelatin-agarose beads were added to remove soluble fibronectin. This was important, since fibronectin could be easily precipitated and has an apparent molecular mass of 250 kDa. In some experiments, apoAI was immunoprecipitated by using sheep polyclonal antibodies (Boehringer Mannheim catalog no. 726478). All of the samples were analyzed by denaturing polyacrylamide gel electrophoresis (SDS-PAGE) (31) on a 5-12% acrylamide sigmoid gradient gel under reducing conditions. After drying the gel, the radioactivity was detected using a PhosphorImager TM screen (Molecular Dynamics). All pulse-chase experiments were reproduced at least three times.
Lipoprotein Isolation by Sequential Flotation Ultracentrifugation-In some experiments, lipoproteins were analyzed by sequential ultracentrifugation. The density was adjusted to 1.040 by the addition of KBr. After 2 h of ultracentrifugation at 120,000 rpm in Beckman TLA 120.2 rotor (480,000 ϫ g), the top fraction was collected for immunoprecipitation, whereas the bottom fraction was readjusted to 1.070 before another ultracentrifugation step. The density was thus successively adjusted to 1.040, 1.070, 1.110, and 1.21, and the four lipoprotein fractions plus the protein fraction (bottom of 1.21 ultracentrifugation) were analyzed by immunoprecipitation as before.
Isolation of Lipoproteins from the ER Lumen-In some experiments, lipoproteins present in the lumen of the ER were extracted using NaCO 3 . HepG2 cells were incubated for 30 min in the presence of oleic acid (0.5 mM) before the pulse labeling with L-[ 35 S]methionine (2 MBq/ well, 10 min of pulse). ApoB truncated forms were generated by performing a 10-min chase in the presence of 10 M cycloheximide and 150 M of puromycin as before. Cells were washed and incubated for 30 min on ice with 200 mM of NaCO 3 and a protease inhibitor mixture (Complete TM ; Boehringer Mannheim). After 30 min of ultracentrifugation at 100,000 rpm in Beckman TLA 120.2 rotor (330,000 ϫ g), the supernatant containing the free lipoproteins was used either for immunoprecipitation or density determination as described above.

RESULTS
The Secretion of ApoB48 Is Less Sensitive to MTP Inhibition than ApoB100 -We have previously shown that the MTP inhibitor 4Ј-bromo-3Ј-methylmetaqualone decreases apoB100 secretion by human hepatic cells either in primary culture or in the HepG2 cell line (25). To determine whether or not the secretion of apoB48 was also affected by an MTP inhibitor, we used cells that naturally produce simultaneously apoB100 and apoB48 due to partial mRNA editing.
MhAT3F is a well differentiated mouse hepatocyte-like cell line (26,32) that has a protein secretion profile very similar to primary hepatocytes. Cells were pretreated for 15 min with the MTP inhibitor prior to pulse labeling with L-[ 35 S]methionine for 30 min. After a 120-min chase, secreted apoB were immunoprecipitated. As shown in Fig. 1, apoB100 secretion by mhAT3F cells was inhibited by submicromolar concentrations of MTP inhibitor, whereas apoB48 required higher MTP inhibitor concentrations to be affected. At a 1.5 M concentration of the MTP inhibitor, apoB100 secretion was almost abolished (18% of control), whereas apoB48 secretion was marginally affected (71% of control). Albumin secretion remained insensi-FIG. 1. The effect of MTP inhibition on apoB100 and apoB48 secretion by mhAT3F and Caco-2 cell lines. The mouse hepatocyte-like cell line mhAT3F and human intestinal Caco-2 cell line were incubated in the presence of oleic acid (0.5 mM) and labeled with L-[ 35 S]methionine for 30 min followed by 120 min of chase. The MTP inhibitor was added at various concentrations 15 min before the pulse. ApoB was immunoprecipitated before SDS-PAGE, whereas albumin was directly analyzed by SDS-PAGE. Radioactivity was determined by PhosphorImager TM screen autoradiography.
tive to the compound, thus confirming the lack of toxicity of the MTP inhibitor.
The Caco-2 cell line derives from a human intestinal tumor and produces both apoB forms because of partial differentiation (33). Caco-2 cells were treated and labeled as mhAT3F. Although the ratio of apoB100 versus apoB48 was higher in Caco-2 cells compared with mhAT3F cells (5 and 0.2, respectively), the same type of selectivity for apoB100 was observed upon treatment with the MTP inhibitor. At 10 M, apoB100 secretion was reduced to 21% of control, whereas apoB48 secretion was similar to control (82%).
Such a selectivity suggests that the requirement for MTPmediated lipid assembly is not the same for apoB100 and apoB48 in two cell lines from different species (mouse or human) and tissues (liver or intestine). Furthermore, the relative sensitivity of both apoB forms toward MTP inhibition is not correlated with the proportion of apoB100 versus apoB48 secreted by cells. These results confirm that the observed selectivity to MTP inhibition is due to different intrinsic properties of apoB100 and apoB48.
The Secretion of ApoB48 from Transfected McA-RH7777 Cells Is Less Sensitive to MTP Inhibition Than ApoB53-To confirm the result that the MTP inhibitor is more deleterious for the assembly and secretion of apoB100-containing lipoproteins than for apoB48 lipoproteins, we examined the effect of the MTP inhibitor on the secretion of a series of carboxylterminally truncated forms of human apoB (apoB18, apoB29, apoB53, apoB72, and apoB100) from stably transfected rat hepatoma McA-RH7777 cell lines (27,28). The secretion experiments were performed in the presence of oleic acid (0.5 mM) to avoid possible lipid deficiency due to apoB overexpression. Cells transfected with full-length human apoB100 secrete simultaneously apoB100 and apoB48. These cells were compared with cells expressing human apoB72, apoB53, apoB29, or apoB18. Cells were labeled with L-[ 35 S]methionine for 4 h in the presence of the MTP inhibitor (10 M). This compound dramatically decreased the secretion of apoB100, apoB72, and apoB53 (less than 20% of control) but had no effect on the secretion of apoB48, apoB29, and apoB18 ( Fig. 2A). As shown in Fig. 2B, the dose response to the MTP inhibitor confirmed that apoB truncated forms can be distinguished in two groups: forms resistant to and forms sensitive to MTP inhibition. The transition between the two groups can be localized roughly between apoB48 and apoB53.
MTP Inhibition Differently Affects the Secretion of Carboxylterminally Truncated Forms of ApoB-We have previously demonstrated the correlation between the concentration of this MTP inhibitor and the extent of the inhibition of MTP-mediated lipid transfer activity both in vitro and in cells (25). To delineate more precisely the relationship between apoB size, MTP activity, and apoB secretion, we evaluated the ability of a large series of carboxyl-terminally truncated forms of apoB100 to be secreted by HepG2 cells. To this end, we used the method described by Boren et al. (30). This method takes advantage of the capacity of puromycin to induce premature termination of polypeptide elongation (34) and allows simultaneous expression of a wide range of apoB truncated forms. As shown in Fig.  3 (lane a), HepG2 cells pulse-labeled with L-[ 35 S]methionine for 15 min and chased for 150 min secreted only full-length apoB100. By contrast, if the chase was performed in the presence of puromycin and cycloheximide, several apoB truncated forms were secreted (Fig. 3, lane b). As described by others (30,35), a relatively discontinuous pattern of truncated forms was observed. Thus, the quantification of 26 truncated forms ranging from apoB17 to full-length apoB100 can be accurately performed (Fig. 3, lane c) and designated as percentage of full- length apoB100 using human apoB100, apoB72, apoB53, apoB48, apoB29, and apoB18 as molecular mass standards (27,28).
We next examined the effect of the MTP inhibitor on the secretion of these 26 truncated forms of apoB. As shown in Fig.  4, low concentrations of the inhibitor markedly reduced the secretion of apoB polypeptides longer than apoB53. At higher concentrations, intermediate sized forms of apoB began to be affected. Smaller truncated forms (apoB17 to apoB26) were insensitive to high concentrations of the MTP inhibitor.
Importantly, we found that the effect of MTP inhibition was not simply a linear function of apoB length (Figs. 4 and 5). A clear transition was observed between apoB51 and apoB53. The concentration of the MTP inhibitor required to decrease the secretion of apoB51 by 50% (IC 50 ) was 11.3 M compared with 1.2 M for apoB53 (Fig. 5). Even in the absence of oleic acid, the IC 50 values for apoB51 and apoB53 still remained different, being 3.2 and 0.46 M, respectively (data not shown).
These results on HepG2 cells are consistent with our previous findings using mhAT3F, Caco-2, and transfected McA-RH7777 cell lines and allow us to define more precisely the critical domain within apoB that confers a high requirement for MTP activity. This domain is localized between apoB51 and apoB53, which represents a sequence of less than 100 amino acids.
Inhibition of MTP Does Not Modify the Density of ApoBcontaining Lipoproteins but Decreases the Number of Lipoprotein Particles Secreted by HepG2 Cells-Previous studies have established that there is a direct relationship between apoB size and the density of lipoproteins secreted from HepG2 cells (30,35). We studied the effect of the MTP inhibitor on the density of lipoproteins formed with truncated apoB generated by puromycin treatment. Lipoproteins secreted by HepG2 cells incubated with or without the MTP inhibitor (1.5 M) were analyzed by sequential flotation ultracentrifugation. For each fraction, apoB was immunoprecipitated and analyzed as before. ApoB truncated forms smaller than apoB29 were found mainly in the density Ͼ1.21 fraction (Fig. 6, lane i). This fraction contained also albumin-and lipid-free proteins (data not shown). ApoB32 to apoB44, apoB46 to apoB63, and apoB68 to apoB83 were found within densities of 1.21-1.11, 1.11-1.07, and 1.07-1.04, respectively (lanes g, e, and c). ApoB100 and apoB90 were mainly found at density less than 1.04 (lane a). This correlation between apoB sizes and densities has been described using various approaches (6,7,27,28,30,(35)(36)(37).
In our hands, inhibition of MTP did not modify the density at which truncated forms are found, although it decreased the amount of large truncated forms of apoB (Fig. 6, lanes a-j). Large forms such as apoB100 or apoB90 were still present at a density less than 1.040 (Fig. 6, lane b versus lane a). Similarly, smaller forms such as apoB34 or apoB35 remained in the 1.210 -1.110 density range (lane h versus lane g), and none of them shifted to the lipid-free fraction in the presence of the MTP inhibitor (lane j). These results suggest that the inhibition of the lipid assembly mediated by MTP does not significantly modify the lipid:protein ratio but rather decreases the number of secreted particles. Taken together, these results indicate first that each carboxyl-terminally apoB truncated form requires a defined amount of lipid to attain a secretable conformation and second that the inhibition of MTP-mediated assembly with lipid reduces the number of particles that attain this state. The sequence between apoB51 and apoB53 constitutes a structural domain that is highly dependent on MTP-mediated lipid transfer for proper folding and secretion.
The Size-related Sensitivity to MTP Inhibition Is Observed in Nascent Lipoproteins-Next we examined whether the sensitivity of truncated forms of apoB longer than apoB53 to MTP inhibition occurred at the early stages of the lipoprotein assembly process. Following a 10-min chase in the presence of puromycin and cycloheximide, nascent lipoproteins were extracted from the ER of HepG2 cells with NaCO 3 . The MTP inhibitor (10 M) substantially decreased the number of apoB polypeptides longer than apoB53 in the lumen of the ER (Fig. 7, lane b versus  lane a). The sharp transition in behavior between apoB51 and apoB53 was similar to that observed in our secretion experiment (Figs. 4, 5, and 7, lanes c and d). Likewise, the densities of the lipoproteins in the lumen of the ER were comparable with those secreted following a 2-h chase (Fig. 8). These results indicate that the size-related sensitivity of apoB to MTP inhibition is due to very early events in the lipoprotein assembly process and not related to differences in the downstream processing or secretion events. In addition, we conclude that all of the lipid loaded onto apoB in HepG2 cells takes place within the first 10 min of assembly.

DISCUSSION
ApoB100 contains a series of structural domains that are predicted to have different functional roles during the synthesis and assembly of apoB100 with lipids to form a lipoprotein. The amino terminus of apoB is rich in disulfide linkages (4, 38 -40, 42) and is predicted to have a compact globular struc-ture that is highly homologous to the ancient lipid storage and transport protein lipovitellin. 2 The remainder of apoB forms a belt-like structure wrapped around the surface of the lipoprotein particle (4,24,40,(43)(44)(45). The lipid binding structures of apoB100 are predicted to form two extensive clusters of amphipathic ␣-helices, which are predicted to be flexible and bind lipid reversibly, and two extensive ␤-sheets, which are pre-  04 (lanes a and b), between 1.04 and 1.07 (lanes c and d), between 1.07 and 1.11 (lanes e and f), between 1.11 and 1.21 (lanes g and h), or above 1.21 (lanes i and j) were analyzed by immunoprecipitation. ApoB secretion in control condition (lanes a, c, e, g, and i) was compared with apoB secretion in the presence of a 1.5 M concentration of the MTP inhibitor (lanes b, d, f, h, and j). 7. The effect of MTP inhibition on truncated forms of apoB present in the ER lumen. HepG2 cells were treated and pulselabeled as in Fig. 4 but with a 10-min pulse. The duration of the chase period in the presence of puromycin and cycloheximide was 10 min for intracellular lipoproteins (lanes a and b) that were released from the ER lumen using NaCO 3 (see "Experimental Procedures"). The duration of the chase period was 120 min for secreted lipoproteins (lanes c and d).
All samples were analyzed after apoB immunoprecipitation as in Fig. 4.  Control cells (lanes a and c) were compared with cells treated with a 10 M concentration of the MTP inhibitor added 15 min before the pulse (lanes b and d). Six truncated forms of apoB with estimated sizes of apoB26, apoB41, apoB50, apoB53, apoB77, and apoB100 are indicated with dotted arrows (see Fig. 8).
dicted to be rigid and associate with lipid irreversibly (24). The portions of apoB100 that form the amphipathic ␣-helical domains are not susceptible to trypsin digestion (46,47), indicating that they either are firmly anchored to the core of the lipoprotein particle or that they possess a compact globular structure.
The structure and the function of the domains of apoB100 can be assessed by studying the early stages of apoB100 production. Two major phenomenon seem to be rate-limiting in the production of apoB100-containing lipoproteins: (i) the translocation of apoB nascent polypeptide through the ER membrane and (ii) the assembly of translocated apoB with sufficient lipid to form a secretion-competent lipoprotein (30,48). Untranslocated apoB remains bound to the ER membrane and is targeted for degradation (49,50), although it can be rescued in certain circumstances (51). It is now clear that the lipid transfer protein MTP plays an obligatory role in the formation of apoBcontaining lipoproteins (8 -11, 19, 52). We have previously shown that a specific inhibitor of the lipid transfer activity of MTP, 4Ј-bromo-3Ј-methylmetaqualone, inhibits apoB100 secre-tion from hepatic cells and that this is associated with early presecretory degradation of apoB (25). Recently, we have shown that apoB degradation can be observed before the termination of polypeptide elongation in MTP inhibitor-treated cells, and that this degradation is mediated by the proteasome (53). This co-translational degradation takes place after the polypeptide has reached 65% of full length, suggesting that a sequence in this region absolutely requires high MTP activity to escape the quality control degradation pathway.
To delineate more precisely the domain of apoB requiring MTP-dependent lipidation, we studied the relationship between the size of truncated forms of apoB and the efficacy of the MTP inhibitor to decrease their secretion. First, we found that apoB48 secretion was resistant to the inhibition of MTP compared with full-length apoB100 both in Caco-2 and mhAT3F cells. Using McA-RH7777 cells transfected with cDNA encoding carboxyl-terminally truncated forms of human apoB, we showed that two groups of apoB truncated forms could be distinguished. ApoB48 and all smaller forms were not affected by the MTP inhibitor, whereas the secretion of apoB53 and larger forms was strongly decreased.
To map more precisely the domain of apoB involved, we have generated a set of apoB truncated forms ranging from apoB17 to full-length apoB100, using the property of puromycin to induce premature release of nascent polypeptides from the ribosomes (34). These apoB truncated forms can be processed by the cells and secreted as lipoproteins (30,35). The density and the particle circumference have been shown to be a function of the sizes of truncated forms of apoB (35). The same relationship has been demonstrated using cells transfected with cDNA coding for apoB truncated forms (27,28,(35)(36)(37). Nevertheless, the role of the MTP in apoB-containing lipoprotein assembly has never been studied other than by comparing results obtained in cells that do or do not express MTP. Here, we progressively reduced MTP activity by increasing concentrations of a potent and specific MTP inhibitor: 4Ј-bromo-3Јmethylmetaqualone (25). This approach allows the detection of a discontinuous pattern of sensitivity to MTP inhibition. Truncated forms smaller than apoB26 were not affected by high concentrations of the MTP inhibitor. These small forms were found in densities greater than the 1.21 fraction. This suggests that these forms contain almost no lipid and is consistent with the fact that forms smaller than apoB23 have been reported to be secreted even in MTP-deficient cells (27). Truncated forms between apoB29 and apoB51 were only affected by high concentrations (10 -40 M) of the MTP inhibitor, and the smaller they are the less sensitive they are. Thus, a relatively low residual MTP activity is sufficient to ensure proper lipidation and secretion of apoB29 to apoB51 forms. These forms were found at the same density range as apoAI (data not shown). Surprisingly, a huge gap in sensitivity to MTP inhibition was observed between forms smaller than apoB51 and all larger forms. Forms ranging from apoB53 to apoB100 were very sensitive to submicromolar concentrations of MTP inhibitor, suggesting that all of them required highly active MTP to be secreted. The same relationship between size and sensitivity to the MTP inhibitor was observed for the nascent lipoproteins present in ER lumen after 10 min of chase, indicating that the differential effect of the MTP inhibitor on lipoprotein secretion occurs during the very early stages of lipoprotein assembly.
HepG2 cells are deficient in mobilizing lipid stores (54) and produce apoB100 almost exclusively at a low density lipoprotein rather than a very low density lipoprotein density (55,56). Thus, HepG2 cells represent a model for studying minimally lipidated apoB in which the second lipidation step responsible for bulk lipid addition is almost absent. In the present study, FIG. 8. Comparison between the density of lipoproteins assembled around truncated forms of apoB present in the ER or secreted. HepG2 cells were treated, pulse-labeled, and chased, and lipoproteins were extracted as described in the legend to Fig. 7. Lipoprotein fractions were obtained by sequential flotation ultracentrifugation as described in the legend to Fig. 6 and analyzed by immunoprecipitation. The data for six representative apoBs (see Fig. 7) are shown. The densities of NaCO 3 -extracted lipoproteins from the ER, after a 10-min chase (closed symbols) were compared with the densities of secreted lipoproteins after a 120-min chase (open symbols). Control cells (circles, dashed lines) were compared with cells treated with a 2.5 M concentration of the MTP inhibitor (triangles, solid lines). The densities of fractions A, B, C, D, and E are less than 1.04, between 1.04 and 1.07, between 1.07 and 1.11, between 1.11 and 1.21, and above 1.21, respectively. Results were expressed as percentage of the total amount of each truncated form found in all fractions at 10 min of chase (ER lumen) in control cells.
we have used this cell system to establish that MTP inhibition does not modify the density of secreted apoB-containing lipoproteins, despite decreased production. Thus, a reduction of MTP-mediated lipid supply has no effect on the lipid to protein ratio but rather decreases the number of lipoproteins secreted, as previously suggested (57). This observation indicates that each apoB polypeptide requires a minimally defined amount of lipid to attain a secretion-competent conformation.
The size of the domain determining the high sensitivity to the MTP inhibitor seems very small compared with apoB100 sequence. Experiments performed on transfected McA-RH7777 cells indicate that this domain is located between apoB48 and apoB53. Using the puromycin method, we further mapped this domain to between apoB51 and apoB53, which encompasses less than 2% of full-length apoB100. The region of apoB between apoB48 and apoB55 corresponds to an extensive ␣-helical domain (4,24,40). Computer predictions and lipid and monoclonal antibody (41) binding studies suggest that this region is flexible and that it binds lipid reversibly. Upstream and downstream of this domain are the two extensive amphipathic ␤-sheets, which are predicted to bind lipid irreversibly and to have a much more rigid structure. The precise role of these domains remains to be defined, but we can speculate that lipidation by MTP could induce conformational changes in nascent apoB that are absolutely required for further lipidation and for lipoprotein assembly. The localization of the ␣-helical domain immediately after the end of apoB48 suggests that this domain, possibly in conjunction with the succeeding ␤-sheets, allows apoB to switch from one stable conformation (lipid-poor apoB48) to another stable conformation (lipid-rich apoB100) during polypeptide elongation. The presence in apoB100 of the ␣-helical domain and ␤-sheet beyond the carboxyl terminus of apoB48 must presumably confer on apoB100 the high level of MTP-mediated lipid transfer activity needed for the lipoprotein assembly process. The fact that only large apoB nascent polypeptides undergo co-translational degradation by the proteasome (53) is consistent with the misfolding of apoB sequences downstream apoB53 when MTP is inhibited. An alternative explanation for the inverse relationship between apoB size and sensitivity to MTP inhibition would be that mild inhibition of MTP reduces the amount of lipid associated with all forms of apoB at the point they are released from the ribosome, and if so that underlipidated forms of large apoBcontaining lipoproteins are more susceptible to proteolysis as they proceed along the secretory pathway than their smaller apoB polypeptides. The observation that the same relationship exists between apoB size and sensitivity to the MTP inhibitor for both nascent lipoproteins, in the ER, and secreted lipoproteins rules out this possibility and places the effects we observe proximate to co-translational lipoprotein assembly.
The results we describe in the present paper are consistent with results published by others and have two major implications for the understanding of apoB100 lipoprotein assembly. First, they suggest that special sequences (consecutive amphipathic ␣-helical and ␤-sheet domains) in apoB100 trigger the lipidation mediated by MTP, which is required for proper folding. When lipid addition at the site of apoB assembly is reduced through MTP inhibition, apoB is diverted to a degradation pathway. Second, they indicate that during early steps of lipoprotein assembly, the lipid:protein ratio is only determined by the length of apoB.