Binding of Microsomal Triglyceride Transfer Protein to Lipids Results in Increased Affinity for Apolipoprotein B EVIDENCE FOR STABLE MICROSOMAL MTP-LIPID COMPLEXES*

Apolipoprotein B (apoB) and microsomal triglyceride transfer protein (MTP) are known to interact with each other. We evaluated the effect of different lipids on the protein-protein interactions between MTP and apoB100 or its C-terminally truncated forms. Negatively charged lipids decreased protein-protein interactions between apoB and MTP. In contrast, zwitterionic phospholipids enhanced (2–4-fold) the binding of apoB100 to MTP by increasing affinity (1.5–3-fold) between these proteins without affecting the number of binding sites. Similarly, phospholipids augmented (1.5–4-fold) the binding of various C-terminally truncated apoB peptides to MTP. The increased binding was greater for apoB peptides containing lipid-binding domains, such as apoB28 and apoB42. Surpris-ingly, preincubation of apoB28 with lipid vesicles had no effect on MTP binding. In contrast, incubation of MTP with lipid vesicles resulted in a stable association of MTP with vesicles, and MTP-lipid vesicles bound better (5-fold increase) to LDL than did lipid-free MTP. To determine whether MTP exists stably associated with lipids in cells, microsomal contents from COS cells expressing MTP, HepG2 cells, and mouse liver were visualized by enhanced chemiluminescence followed by fluoro- grahy. containing MTP, was not precipitated with apoB, used for the precipitation of MTP. For this l of anti-97-kDa antibodies 20 of protein G-Sepharose added supernatants,

Apolipoprotein B (apoB) 1 is essential for the transport of lipids packaged into lipid-protein complexes called lipoproteins. Assembly of these lipoproteins requires, in addition to apoB, microsomal triglyceride transfer protein (MTP). In vitro kinetic studies suggest that MTP may catalyze transfer of lipids between vesicles by a shuttle mechanism (1). In this process, each MTP molecule is proposed to interact transiently with donor vesicles, extract up to 2 mol of PC and 0.25 mol of triglycerides, dissociate from these vesicles, interact transiently with acceptor vesicles, and rapidly deliver lipid molecules to these vesicles (1,2). The importance of this lipid transfer activity in lipoprotein assembly was established by identifying mutations in MTP gene that result in loss of lipid transfer activity in patients that lack apoB-containing lipoproteins in their plasma (3)(4)(5), by identifying compounds that inhibit in vitro lipid transfer activity and decrease apoB secretion (6 -9), and by manipulating gene expression and correlating it to changes in apoB secretion (10 -14).
Lipoprotein assembly occurs in different steps (15)(16)(17)(18)(19)(20)(21)(22)(23)(24). MTP plays a crucial role in the first step of lipoprotein assembly in which "primordial lipoproteins" are formed. This event involves minimal lipidation that renders nascent apoB secretion-competent. Subsequently, lipids are added in bulk resulting in the "core expansion" of primordial lipoproteins and formation of "nascent lipoproteins" (22)(23)(24). A critical event in core expansion is perhaps the formation of lipid droplets. Raabe et al. (10) have observed that in the absence of MTP there is paucity of lipid droplets in the lumen of the endoplasmic reticulum and Golgi apparatus. These studies suggest that MTP may play a critical role in the formation of lipid droplets in the endoplasmic reticulum. However, this role has not yet been elucidated. Moreover, it is not known whether MTP exists stably associated with lipid droplets in microsomes.
In addition to its lipid transfer activity, MTP has also been suggested to act as a chaperone in the assembly of apoBcontaining lipoproteins. The evidence for its chaperone activity is largely based on the observations that MTP physically interacts with apoB. Protein-protein interactions between apoB and MTP were first demonstrated by co-immunoprecipitation experiments (25,26). Subsequently, in vitro studies involving solid-liquid interphase binding assays (27)(28)(29)(30) and two-hybrid binding analyses (31,32) showed that these proteins may interact at multiple binding sites. There is some evidence indicating that protein-protein interactions between apoB and MTP may be important for lipoprotein assembly. An antagonist that inhibits the binding of MTP to apoB but does not inhibit its lipid transfer activity has been shown to decrease apoB secretion from HepG2 cells (30).
Lipids modulate protein-protein interactions between apoB and MTP. We have shown that these interactions are affected by the degree of apoB lipidation (27). For example, very low density lipoprotein binds poorly to MTP, and this binding is increased by partial delipidation. Similarly, binding of LDL to MTP is also increased after partial delipidation with taurocholate. Wu et al. (25) have characterized two different kinds of interactions between apoB and MTP. At basal level, apoB-MTP binding occurs transiently in HepG2 cells resulting in the association of MTP with ϳ10% of total intracellular nascent apoB. The duration of these interactions is prolonged when the intracellular degradation of apoB is inhibited. They also observed that apoB-MTP binding and duration of the binding was increased 3-5-fold after the induction of lipid synthesis by oleic acid supplementation. Thus, it is known that lipids affect apoB-MTP binding; however, the molecular mechanisms by which lipids modulate apoB-MTP binding or other protein-protein interactions are not known. In this study, we show that neutral lipids increase affinity between apoB and MTP by binding to MTP and provide evidence for the presence of stable microsomal MTP-lipid complexes.

EXPERIMENTAL PROCEDURES
Materials-All of the assays were performed with the purified heterodimeric MTP complex of 97 and 55 kDa subunits (6,33). Antibodies used for ELISA have been described (34,35).
Preparation of Lipid Vesicles-Lipid vesicles were prepared as described before (6,33). The required amounts of lipids in chloroform were added to a glass vial, and the solvent was evaporated under N 2 . A calculated amount of the buffer (15 mM Tris-Cl, 40 mM NaCl, 1 mM EDTA, 0.02% sodium azide, pH 7.4) was added, and the contents were sonicated (Bronson Sonifier; setting, 5; output, 50 -60%) on ice until the solution was mostly clear. The vesicles were then centrifuged (Beckman, 35,000 rpm, 15 min, 4°C), and the supernatant was used to assay their effect on the binding of LDL to immobilized MTP (27,29).
Human Recombinant apoB Polypeptides-McA-RH7777 cells stably transfected with human recombinant apoB17, apoB28, or apoB42 cDNA have been described (34,36,37). For experiments, 80% confluent monolayers were incubated with serum-free medium containing 0.2% bovine serum albumin for 48 h. The concentrated conditioned medium was used to determine the amount of apoB polypeptides present and to study the binding of these peptides to the immobilized MTP in the presence and the absence of phospholipid vesicles. The apoB polypeptides that interacted with MTP were quantified by ELISA (27,34,35). COS-7 cells were transiently transfected with plasmids expressing apoB17 and apoB:270 -570 using Fugene-6 reagent according to the manufacturer's protocol.
Detection of MTP in the Lumen of Microsomes-Confluent monolayers of cells were washed three or four times with ice-cold phosphate buffered saline, and cells were scraped into 2 ml of homogenization buffer (50 mM Tris-HCl, 50 mM KCl, 5 mM EDTA, 250 mM sucrose, pH 7.4) containing protease inhibitors. The cells were homogenized by 20 passes through a ball bearing homogenizer and centrifuged (SW 60Ti, 60,000 rpm, 4°C, 2 h). The microsomal pellets were suspended in 10 mM Tris, 150 mM KCl, 250 mM sucrose, pH 7.4, adjusted to 100 mM sodium carbonate, pH 11.5 (38), incubated overnight at 4°C with gentle tumbling to release microsomal contents, and centrifuged (SW 60 Ti, 60,000 rpm, 4°C, 2 h). The supernatants were adjusted to a density of 1.24 g/ml by the addition of KBr, overlaid sequentially with 2 ml each of 1.21, 1.063, 1.019, and 1.006 g/ml density solutions and centrifuged (SW41 rotor, 40,000 rpm, 12°C, 18 h), and 1-ml fractions were collected from the top.
To immunoprecipitate apoB, each fraction was incubated overnight at 4°C with 5 l of sheep anti-human apoB antibodies and 20 l of protein G-Sepharose and centrifuged (10,000 rpm, 1 min, 4°C). Pellets and supernatants were processed separately. The pellets that contain proteins associated with apoB were washed and treated with sample buffer. Eluted proteins were separated on 10% polyacrylamide gels, transferred to nitrocellulose, treated sequentially with 1:2000 diluted anti-97-kDa antibodies (kind gift of Dr. Haris Jamil of Bristol-Myers Squibb) and 1:5000 diluted peroxidase-conjugated protein A. The bands were visualized by enhanced chemiluminescence followed by fluorograhy. Supernatants containing MTP, which was not precipitated with apoB, were used for the precipitation of MTP. For this purpose, 5 l of anti-97-kDa antibodies and 20 l of protein G-Sepharose were added to supernatants, and the pellets were processed as described above.
Other Analyses-Protein was determined using the Coomassie Plus reagent (Pierce) with bovine serum albumin as a standard (39). Optical density in ELISA plates was measured using a Dynatech MRX microplate reader (Dynatech Labs, Chantilly, VA). The data were plotted as the means Ϯ S.D. The molecular masses used for apoA-I, apoB100, apoB28, and MTP were 23.3, 512, 143.4, and 146 kDa, respectively.

Effect of Phospholipids on apoB-MTP Interactions-MTP is a lipid transfer protein. apoB is a lipid carrier protein.
Thus, we hypothesized that lipids would modulate interactions between these proteins. To test this hypothesis, the effect of phospholipid vesicles on these interactions was studied. PC and sphingomyelin increased the interactions 1.5-2-fold (Fig. 1A). As a control, we studied the effect of phospholipids on the binding of LDL to a monoclonal antibody, 1D1. 1D1 binds to an epitope in the N terminus (amino acids 474 -539) of human apoB (40,41), a site that overlaps with amino acids 270 -570 and that constitutes a putative MTP-binding site (28). PC slightly enhanced (Ϸ25% increase) the binding of LDL to 1D1, whereas sphingomyelin had an inhibitory effect (50% inhibition at 2.4 mM; Fig.  1B). These studies indicated that zwitterionic phospholipid vesicles increase interactions between LDL and MTP.
MTP transfers triglycerides (TG), whereas apoB forms TGrich lipoproteins. To study the effect of TG on apoB-MTP interactions, PC vesicles or emulsions containing two different concentrations of TG were prepared, and their effect on LDL binding to MTP was compared with those of PC vesicles with no TG (Fig. 1C). PC vesicles enhanced (Ͼ4-fold) the LDL binding in a dose-dependent manner. Inclusion of TG had no additional effect on PC-mediated enhanced binding of LDL to MTP.
The effect of different lipids was studied in more detail using PC-based vesicles (Table I). Again, PC vesicles increased the binding of LDL to MTP Ϸ3-fold. The addition of sphingomyelin or cholesterol to PC vesicles had no additional effect. However, inclusion of negatively charged phospholipids, phosphatidylinositol (PI) and phosphatidylserine, in PC vesicles significantly decreased the PC-stimulated binding of LDL to MTP. These studies suggested that zwitterionic phospholipids promote apoB-MTP interactions, whereas negatively charged phospholipids decrease them.
Interactions between C-terminally Truncated apoB Polypeptides and MTP-We then studied the effect of phospholipids on the binding of MTP to different C-terminally truncated apoB polypeptides that are secreted with different degree of lipidation or secreted as lipid-poor polypeptides (Table II). We used conditioned medium from McA-RH7777 cells stably expressing apoB42, apoB28, and apoB17. apoB42 is mainly secreted as high density lipoprotein-size particle, whereas apoB28 is secreted as both lipid-free and in lipidated state. On the other hand, apoB17 is mainly secreted as lipid-free polypeptide by the stably transfected cells. PC vesicles increased the binding of apoB42 and apoB28 by 400%, whereas apoB17 binding was increased by only 60% (Table II). These studies indicated that phospholipids increased interactions between C-terminally truncated apoB polypeptides and that the effect of PC was more pronounced with polypeptides that contain ␤1 lipid-binding domains (42)(43)(44) compared with apoB17 that lacks this domain.
Next, we compared the effect of different concentrations of PC and PI vesicles on the binding of C-terminally truncated apoB peptides to MTP (Fig. 2). As seen before (Fig. 1A), increasing concentrations of PC resulted in enhanced binding of LDL to MTP ( Fig. 2A). However, increasing concentrations of PI decreased the binding of LDL to MTP, and these results are in agreement with those in Table I. PC increased the binding of apoB42 (Fig. 2B), apoB28 (Fig. 2C), and apoB17 ( Fig. 2D), whereas PI decreased the binding of these peptides to MTP. The increase caused by PC ranged from 50 to 400% and paralleled with increases in the length of apoB. By contrast, decreases caused by PI were ϳ50% in all cases and were independent of increases in apoB length. These studies indicate that PC and PI have differential effect on the binding of different apoB polypeptides to MTP.
Effect of Phospholipids on the Kinetics of apoB-MTP Binding-Subsequently, we determined the effect of phospholipids on the kinetic parameters of apoB-MTP binding (Fig. 3). The effect of phospholipid vesicles on the binding of LDL to immobilized MTP was more pronounced at lower concentrations. In  fact, no significant binding of LDL to MTP could be observed at low concentrations (Յ3 nM) in the absence of vesicles (Fig. 3A). The effect of vesicles was significant at Յ12.5 nM of LDL. At higher concentrations the binding of LDL to immobilized MTP in the presence and the absence of vesicles was not significantly different. Nonlinear regression analysis revealed that phospholipids decreased K d (23 Ϯ 8 nM versus 69 Ϯ 6 nM) by 3-fold without significantly affecting the B max (13 Ϯ 0.5 fmol versus 10 Ϯ 1.2 fmol). In four other experiments, the decrease in K d ranged between 1.5-and 3-fold. Next, we studied the effect of phospholipid vesicles on the binding of 125 I-MTP to immobilized LDL (Fig. 3B). In this experiment, the binding of 125 I-MTP in the absence of phospholipids did not reach saturation, and thus the K d and B max values were not determined. Nonetheless, the fold increase in the binding of 125 I-MTP was significantly higher at lower concentrations (3-fold at 2-8 nM) than at higher concentrations (2-fold at 17 and 34 nM). Furthermore, at the highest concentration (68 nM) the difference in binding was no longer statistically significant (Fig. 3B), indicating that the effect of PC vesicles was probably on the K d and not on the B max . These experiments indicated that phospholipid vesicles increase affinity between apoB and MTP without altering the number of binding sites. How Do Phospholipids Increase Affinity between apoB and MTP?-Consideration was given to the possibility that interactions between apoB and lipids may be important. To evaluate this possibility, we studied the interactions between apoB28 with PC vesicles. Conditioned medium obtained from McA-RH7777 cells stably transfected with human apoB28 was incubated with lipid vesicles. As shown in Fig. 4A, preincubation of the conditioned medium with PC vesicles resulted in a slight shift of apoB28 peak toward lower density. What is more important, apoB28 (fractions 13-19; Fig. 4A) could be separated from lipid vesicles (fractions 1-4; Fig. 4B). Fractions (fractions 14 -18; Fig. 4A) containing apoB28 were pooled, dialyzed, and used for MTP binding (Table III). The binding of apoB28 preincubated with phospholipid vesicles was slightly lower than the binding of apoB28 that was not incubated with vesicles. These studies indicate that interactions between lipids and apoB28 are not important for the enhanced interactions between apoB and MTP.
Next, we hypothesized that the association of MTP with lipids may be more important in the enhanced apoB-MTP binding. To test this hypothesis, we first determined whether PC vesicles bind to MTP by incubating labeled vesicles with MTP and then determining the flotation properties of the vesicles (Fig. 4B). Incubation of vesicles with MTP resulted in a diminished vesicle peak (fractions 1-4; Fig. 4B) with a concomitant appearance of a hump at the trailing end of the vesicle peak (fractions 5-8; Fig. 4B), indicating that MTP had some influence on the flotation properties of the PC vesicles, most likely by interacting with them. To determine whether MTP interacted with PC vesicles, 125 I-MTP was incubated with or without PC vesicles and subjected to density gradient ultracentrifugation (Fig. 4C). MTP, incubated without lipid vesicles, was recovered in the bottom fractions (fractions 21-24) corresponding to a density Ͼ1.21 g/ml (Fig. 4C, peak A). In contrast, MTP incubated with vesicles was distributed in two fractions. One fraction was recovered as lipid-poor MTP (fractions 21-24; Fig.  4C, peak B) similar to that observed for MTP alone (Fig. 4C,  peak A). This fraction may represent MTP with 2 mol of PC (1). In addition, MTP (Ϸ50%) was also recovered associated with lipid vesicles (fractions 3-9; Fig. 4C, peak C). The profile of MTP associated with vesicles was similar to that observed for vesicles incubated with unlabeled MTP (Fig. 4B). These studies show that MTP stably associates with phospholipid vesicles.
To study the binding of vesicle-associated MTP to LDL, peaks A, B, and C were pooled, dialyzed, and then used to study their binding to immobilized LDL. As shown in Table III, the binding of lipid-poor MTP (Fig. 4C, peak B) to LDL was increased (72%, p Ͻ 0.005) compared with the binding of MTP not incubated with lipids (Fig. 4C, peak A). More important, the binding of vesicle-associated MTP (Fig. 4C, peak C) to LDL was 5-fold higher than MTP alone. These studies indicate that MTP associated with vesicles binds better to LDL compared with lipidpoor and lipid-free MTP.
Consideration was given to the possibility that the enhanced interactions between LDL and vesicles-associated MTP were due to interactions between lipids. To test this hypothesis, we designed experiments to study the binding of apoA-I and apoA-I-lipid complexes to immobilized LDL. apoA-I incubated with or without lipid vesicles was subjected to ultracentrifugation (Fig.  4D). In contrast to MTP (Fig. 4C), all of the apoA-I bound to phospholipid vesicles (Fig. 4D). Next the binding of apoA-I and apoA-I-lipid complexes to immobilized LDL was performed (Table III). In contrast to the binding of MTP-lipid vesicles to LDL, binding of apoA-I-lipid complexes to LDL was significantly less (Ϫ84%) than the binding of free apoA-I (Table III). Thus, these studies indicate that MTP stably associates with lipids, and this association results in its increased affinity for apoB. Furthermore, increased binding of vesicle-associated MTP to LDL is dependent on protein-protein interactions between apoB and MTP.
Presence of MTP Associated with Lipids in the Lumen of the Endoplasmic Reticulum-To determine whether MTP exists stably bound to lipids in the lumen of the endoplasmic reticulum, COS-7 cells (monkey kidney cells that neither express apoB nor MTP) were transfected with plasmids expressing human MTP (45). After 3 days, microsomal lumenal contents were subjected to ultracentrifugation, and the presence of MTP associated or not associated with lipids was determined by Western blotting (Fig. 5A). MTP was mainly present in fractions corresponding to d Յ 1.21 g/ml. These data indicate that MTP associates and forms stable complexes with lipids in these cells.
To study the effect of apoB on the association of MTP with lipids, COS-7 cells were co-transfected with plasmids expressing human MTP and apoB17 (Fig. 5B). Three days post-transfection, the microsomes were prepared, and lumenal contents were ultracentrifuged. apoB was immunoprecipitated under nonreducing conditions (25) from each fraction and transferred to nitrocellulose, and a co-immunoprecipitated 97-kDa MTP subunit was visualized using specific antibodies and chemiluminescence detection (Fig. 5B, panel i). MTP was present in fractions corresponding to d Ͻ 1.21 and d Ͼ 1.21 g/ml. These studies indicate that MTP co-immunoprecipitates with apoB17 and are in agreement with published data (25,26).
We then looked for MTP present unassociated with apoB17. The supernatants from apoB immunoprecipitates (Fig. 5B, panel i) were further subjected to immunoprecipitation with polyclonal anti-apoB antibodies, and the presence of apoB was determined by Western blotting (data not shown). In these experiments, no apoB could be detected, indicating that all of the apoB was quantitatively immunoprecipitated in the first immunoprecipitation. Subsequently, the supernatants were immunoprecipitated with anti-MTP antibodies, and the presence of the 97-kDa subunit was visualized by Western blotting (Fig. 5B, panel ii). MTP that was unassociated with apoB was present in fractions corresponding to a density of d Յ 1.21 g/ml, indicating that it was associated with lipids. In addition, MTP was also present in lipid-free (d Ն 1.21 g/ml) fractions. Therefore, these studies indicate that MTP unassociated with apoB exists associated and unassociated with lipids.
As a control, we studied the flotation properties of apoB in COS cells transiently transfected with plasmids expressing apoB17 (Fig. 5C) or apoB:270 -570 (Fig. 5D). apoB17 represents the N-terminal 17% of apoB100. apoB:270 -570 is a chimeric protein expressing amino acids 270 -570 of apoB attached to a FLAG epitope and has been shown to bind MTP (28). Both apoB17 and apoB:270 -570 were mainly present in the bottom fractions corresponding to d Ն 1.21 g/ml, indicating that these apoB polypeptides mainly exist unassociated with lipids in the lumen of the endoplasmic reticulum.
Consideration was given to the possibility that MTP might have associated with lipids as a consequence of limiting PDI levels in COS cells. Thus, we determined whether MTP is present associated with lipids in cells of hepatic origin where PDI is not expected to be limiting. For this purpose, lumenal proteins from the microsomes of HepG2 cells were subjected to density gradient ultracentrifugation, and different fractions were collected from the top (Fig. 5E). apoB was immunoprecipitated from each fraction under nondenaturing conditions, and co-immunoprecipitated MTP was visualized with anti-MTP antibodies (Fig. 5E, panel iii). MTP was present in all of the fractions; however, the majority was in fractions corresponding to d Յ 1.21 g/ml, suggesting its association with apoB in various intracellular lipoproteins. To determine the presence of MTP unassociated with apoB, the supernatants were subjected to immunoprecipitation with anti-MTP antibodies (Fig.  5E, panel iv). The majority of MTP unassociated with apoB was present in d Յ 1.21 g/ml fractions, indicating again that this MTP exists associated with various amounts of lipids. The presence of lipid-associated MTP was further studied in microsomes isolated from mouse and bovine liver. As shown in Fig.  5F, majority of MTP was present in d Յ 1.21 g/ml fractions in the mouse liver. However, some MTP was also present unas- sociated with lipids (d Ն 1.21 g/ml fractions). Similar results were observed for bovine liver (data not shown). Therefore, we conclude that MTP exists associated and unassociated with lipids in hepatocytes.

Modulation of Protein-Protein Interactions by Lipids-MTP
and apoB are lipid transfer and lipid carrier proteins, respectively. Both proteins are essential for the assembly of triglyceride-rich lipoproteins. They are known to physically associate with each other in the lumen of the endoplasmic reticulum. In this study, we provide evidence that lipids modulate protein-protein interactions between apoB and MTP. Negatively charged phospholipids decrease apoB-MTP binding (Table I and Fig. 2). The inhibition by negatively charged lipids was similar for the binding of apoB100 and its various C-terminally truncated forms, indicating that the inhibition was not related to the length of apoB polypeptide. We have previously shown that ionic interactions play an important role in apoB-MTP binding (27,29). It is likely that negatively charged phospholipids inhibited the ionic interactions between these proteins.

TABLE III
Binding of MTP/lipid complexes to immobilized LDL MTP, apoB28, or apoA-I were incubated with PC vesicles, and protein-lipid complexes were isolated by ultracentrifugation (Fig. 4), dialyzed, and used for studying their binding to different immobilized proteins as indicated below. nificantly higher (2-4-fold) for longer apoB polypeptides containing ␤1 lipid-binding domains. The smaller increase for the binding of apoB17 may be because it binds better than other polypeptides (27) and may represent optimal binding. Alternatively, sequences beyond the N-terminal 17% of apoB may be required for the enhanced affinity. To examine the role of ␤1 lipid-binding domains beyond apoB17, we studied the binding of apoB28 incubated with lipid vesicles. Surprisingly, preincubation of apoB28 with lipids did not result in increased binding to MTP. Instead, we observed that lipids stably associated with MTP (Figs. 4 and 5 and Table III) and increased its affinity for apoB (Fig. 3).
How does stable association of lipids with MTP increase its affinity for apoB? First, binding of MTP to lipids may result in conformational changes that result in the expression of new binding sites for apoB. Our observation (Fig. 3) that there is no increase in the number of binding sites excludes this possibility. Second, binding of MTP with lipids may cause a localized conformational change in MTP that results in higher affinity without increasing the number of binding sites. Structural changes caused by lipid binding have been observed for apoE (46,47). It is known that free apoE does not bind LDL receptors. However, after association with lipids it undergoes a conformational change that favors interactions with LDL receptors. Third, binding of MTP-lipid complexes with apoB may involve secondary interactions leading to increased affinity constants. The secondary interactions would be dependent on primary protein-protein interactions and may not manifest as additional binding sites. For example, protein-protein interactions between apoB and MTP may juxtapose lipid-binding domains of apoB with lipids associated with MTP and promote additional hydrophobic interactions, resulting in increased affinity constants. We favor the third mechanism because the affinity of MTP-lipid complexes is higher for the apoB peptides that contain lipid-binding ␤-sheets.
The regions in MTP that form stable complexes with lipids are not known. These regions are probably different from the domains that interact with apoB because zwitterionic lipids increase affinity between proteins. Binding of lipids at the apoB-binding domain would have decreased affinity for apoB. Furthermore, regions in MTP that form stable lipid complexes are probably different from the cavity involved in lipid transfer. Based on homology with lipovitellin, Read et al. (48) have suggested that MTP contains a lipid transfer cavity that can accommodate only few lipid molecules and that for this to occur MTP associates with lipid vesicles. Atzel and Wetterau (1) have shown that there are two binding sites for lipids. One site is involved in rapid shuttling of lipids and may represent the lipid transfer domain. This site binds triglycerides and phospholipids. However, they also identified an additional phospholipidbinding site that is not involved in rapid lipid transfer. It is possible that this site may be involved in stable association with lipids. It remains to be determined whether stable association of MTP with lipids affects its ability to transfer lipids between vesicles.
Importance of MTP-Lipid Complexes in Lipoprotein Assembly and Secretion-At this time, the physiologic significance of the stable association of MTP with zwitterionic phospholipids is not known. It is interesting to note that an absolute requirement for newly synthesized PC in lipoprotein assembly has been observed in hepatocytes obtained from choline-deficient rats (49). There is evidence to suggest that phospholipids are added to nascent apoB during early biogenesis of intestinal lipoproteins (23,24,50). Furthermore, studies from the laboratory of Steve Young indicate that MTP may be important for the presence of lipid droplets in the lumen of the endoplasmic FIG. 5. Association of MTP with lipids in the lumen of microsomes. A, COS cells expressing MTP. COS-7 cells were transfected with plasmids expressing human MTP. After 3 days, cells were homogenized, and microsomal membranes were isolated as pellets after 100,000 ϫ g centrifugation. Microsomes were treated with sodium carbonate, and lumenal contents were subjected to ultracentrifugation. MTP was immunoprecipitated from each fraction using anti-MTP antibodies, transferred to nitrocellulose, and detected using anti-bovine 97-kDa subunit antibodies. The data are representative of three independent experiments. B, COS cells expressing MTP and apoB17. COS-7 cells were transfected with plasmids expressing apoB17 and MTP. After 3 days, lumenal contents were subjected to ultracentrifugation. Panel i, individual fractions were first subjected to immunoprecipitation with anti-apoB antibodies, and the precipitated proteins were separated on SDS-polyacrylamide gels. The MTP co-immunoprecipitated with apoB was visualized using anti-MTP antibodies and chemiluminescence detection. Panel ii, supernatants from panel i were subjected to immunoprecipitation with anti-MTP antibodies. Immunoprecipitated proteins were separated on polyacrylamide gels and visualized after treatment with anti-MTP antibodies. The data are representative of three independent experiments. C, COS cells expressing apoB17. COS-7 cells were transfected with plasmids expressing human recombinant apoB17. Microsomal lumenal contents, prepared as described for A, were subjected to ultracentrifugation. apoB17 was immunoprecipitated using polyclonal anti-apoB antibodies from each fraction and subsequently blotted with a monoclonal antibody, 1D1. D, COS cells expressing B:270 -570. For this experiment, COS-7 cells were transfected with plasmids expressing apoB-FLAG chimeras expressing amino acids 270 -570 of apoB. Lumenal contents were ultracentrifuged, and apoB was immunoprecipitated with polyclonal antibodies. The precipitated proteins were separated on polyacrylamide gels and probed with M2, a monoclonal antibody that recognized FLAG epitope. E, HepG2 cells. Confluent monolayers of HepG2 cells were homogenized, microsomes were prepared by centrifugation, and lumenal contents were released by sodium bicarbonate treatment and subjected to ultracentrifugation. Panel iii, each fraction was immunoprecipitated with anti-apoB antibodies, separated on polyacrylamide gels, transferred to nitrocellulose membranes, and treated with anti-MTP antibodies. Panel iv, supernatants from apoB precipitation were subjected to immunoprecipitation with anti-MTP antibodies. The precipitated proteins were resolved on SDS-polyacrylamide gels and used for MTP detection. The data are representative of three independent experiments. F, mouse liver. Microsomal contents were subjected to ultracentrifugation. MTP in each fraction was first precipitated using goat anti-bovine MTP antibodies, separated on polyacrylamide gels, transferred to nitrocellulose membranes, treated with purified anti-97-kDa antibodies, and visualized with chemiluminescence reactions. Lane a contains purified bovine MTP. Lane b contains a portion of the lumenal contents obtained from HepG2 cells. This was present only in panel A. Lanes 1-12 represent different fractions obtained from the top of the gradient. The approximate densities in these fractions have been indicated. reticulum. They showed that liver-specific knockout of MTP gene expression in mice resulted in the absence of lipid staining particles in the endoplasmic reticulum and Golgi apparatus (10). At this time, it is not clear why the absence of MTP should result in decreased lipid droplets in the lumen of the endoplasmic reticulum. It is possible that MTP may transfer lipids to another acceptor, other than apoB, and may assist in the formation of lipid droplets. In the present study, we observed that MTP exists associated with lipids in the lumen of the endoplasmic reticulum (Fig. 5) and suggest that MTP may itself be involved in the formation and stabilization of lipid droplets.
Our studies may provide a possible explanation for the increased apoB-MTP binding observed in oleic acid-treated HepG2 cells (25). Wu et al. (25) showed that oleic acid treatment increased the physical association between apoB and MTP by 3-5-fold, and this enhanced association was dependent on lipid synthesis. Although we did not study the effect of oleic acid supplementation on MTP-lipid complexes, it can be anticipated that oleic acid supplementation would result in increased lipid synthesis and result in greater association of MTP with lipids. This would enhance the affinity of more MTP molecules for apoB and would result in increased association between these proteins for longer duration.
Lipid-associated MTP may play a role in lipoprotein assembly. Lipoprotein assembly may occur in discrete steps. First, nascent apoB is lipidated, resulting in the formation of primordial lipoproteins. It is possible that MTP-associated with lipids may serve as a nucleus for apoB to wrap around it, and this may lead to the assembly of primordial lipoprotein particles. Second, lipids droplets are formed. Based on the data presented in this study, MTP may play a role in the formation of lipid droplets. Third, bulk lipids are added, probably because of the fusion of lipid droplets, to primordial lipoproteins, causing core expansion of these particles and formation of nascent lipoproteins. MTP-lipid complexes may be involved in the bulk transfer of lipids to apoB during the core expansion.
In summary, we have demonstrated that: 1) lipids modulate the binding between LDL and MTP; 2) PC increases the affinity between MTP and apoB (Fig. 3); 3) PC increases interactions between C-terminally truncated forms of apoB and MTP, and this effect is more pronounced with polypeptides that contain lipid-binding domains; 4) MTP forms stable association with lipid vesicles; and 5) the lipid-associated MTP binds better to LDL than the lipid-poor and lipid-free MTP. The MTP-lipid complexes bind better to apoB most likely because proteinprotein interactions between apoB and MTP juxtapose and facilitate interactions between lipids associated with MTP and lipid-binding domains of apoB. Furthermore, we have shown that MTP exists associated with lipids in microsomes and suggest that MTP-lipid complexes play a positive role in lipoprotein biogenesis. Further investigations are needed to clarify the role of MTP-lipid complexes in the formation of lipid droplets and their role in lipoprotein assembly.