Functional analysis of disulfide linkages clustered within the amino terminus of human apolipoprotein B.

We tested the involvement of N-terminal six disulfide bonds (Cys-1 through Cys-12) of human apolipoprotein (apo) B in the assembly and secretion of lipoproteins using two C-terminal-truncated apoB variants, namely B50 and B18. In transfected rat hepatoma McA-RH7777 cells, B50 could assemble very low density lipoproteins (VLDL), and B18 was secreted as high density lipoproteins. When all 12 cysteine residues were substituted with alanines in B50, the mutant protein (B50C1-12) lost its ability to assemble lipid and was degraded intracellularly. However, mutation had no effect on B50C1-12 translation or translocation across the microsomal membrane. Post-translational degradation of B50C1-12 was partially inhibited by the proteasome inhibitor MG132. To determine which cysteines were critical in VLDL assembly and secretion, we prepared three additional mutant B50s, each containing four selected Cys-to-Ala substitutions in tandem (i.e. Cys-1 to Cys-4, Cys-5 to Cys-8, and Cys-9 to Cys-12). Expression of these mutants showed that disruption of disulfide bond formation within Cys-5 to Cys-8 diminished apoB secretion, whereas within Cys-1 to Cys-4 or Cys-9 to Cys-12 had lesser or no effect. In another two mutants in which only one disulfide bond (i.e. between Cys-5 and Cys-6 or between Cys-7 and Cys-8) was eliminated, only secretion of B50 with mutations at Cys-7 and Cys-8 was decreased. Thus, the disulfide bond involving Cys-7 and Cys-8 is most important for VLDL assembly and secretion. In addition, assembly and secretion of VLDL containing endogenous B100 or B48 were impaired in cells transfected with B50s containing Cys-7 and Cys-8 mutation. The Cys-to-Ala substitution abolished recognition of B50 by MB19, a conformational antibody with an epitope at the N terminus of human apoB. The Cys-to-Ala substitution also attenuated secretion of B18, but the effect of the mutation on B18 secretion was less evident than on B50.

tide consisting of 4536 amino acids (1,2). In humans, apoB100 serves as a backbone for the assembly of triacylglycerol-enriched very low density lipoproteins (VLDL) (3). Assembly of VLDL requires coordinate biosynthesis of apoB100 and lipid components (4,5). The association with lipid may occur during apoB100 translation and its translocation across the endoplasmic reticulum (ER), a process facilitated by the microsomal triglyceride transfer protein (MTP) (6). The sequence elements within apoB that are responsible for lipid recruitment have not been precisely defined. Mutational studies with C-terminaltruncated apoB variants have shown that the amount of core lipid in the resulting lipoprotein particles is positively correlated with the apoB polypeptide length (7)(8)(9)(10)(11), suggesting that the lipid recruiting ability is a function of the number of hydrophobic sequences. However, when abundant lipid is available, assembly of VLDL can be achieved by utilizing C-terminal-truncated apoB forms (e.g. apoB48) (12)(13)(14). Thus, both the length of apoB polypeptide and sufficient lipid supply are important determinants in the assembly and secretion of VLDL.
Despite the absence of information concerning specific sequence elements involved in VLDL assembly, several experimental observations suggest that the N-terminal 17% of apoB100 (i.e. apoB17) is critical in lipoprotein formation. Transfection studies have shown that although the C-terminaltruncated apoB forms that contain the autologous N terminus of apoB100 are usually secretion-competent (7,11,13), recombinant apoB variants that lack the N-terminal sequences of apoB are often secreted poorly (13,15). Several studies have demonstrated that apoB17 has relatively low ability to associate with neutral lipids (7,16). Secondary structure analysis has predicted that apoB17 is composed mainly of amphipathic ␣-helices (2,16) and may form a globular structure owing to the existence of concentrated disulfide bonds (17). It is not clear why the region of apoB with low lipid binding ability is essential for efficient assembly of lipid into a lipoprotein.
Of eight disulfide linkages within human plasma apoB, six are located within the N-terminal 500 amino acids (17). Recently, involvement of the concentrated disulfide bonds within the N terminus of apoB in lipoprotein assembly and secretion has been examined. Using dithiothreitol to disrupt disulfide bonding, Shelness and Thornburg (18) have observed that formation of the N-terminal disulfide bonds, which occurs within 1 min of translation, is required for the initiation of MTP-dependent lipid recruitment and the secretion of lipoproteins. It has been postulated that folding of the N-terminal region, presumably mediated through disulfide bonding, is essential for lipoprotein assembly (19). Considering the intrinsic nature of protein folding and the role of disulfide bonds in stabilizing native protein conformation, we hypothesized that the N-terminal six disulfide bonds play a crucial role in VLDL assembly and secretion.
In the current studies, we have directly examined the function of six disulfide bonds using two truncated forms of apoB, apoB50 that has the ability to form VLDL and apoB18 that does not associate with neutral lipids. In these proteins, the cysteine residues were selectively substituted with alanine. Our results demonstrate that the fourth disulfide bond (between Cys-7 and Cys-8) is most important for apoB50 to assemble VLDL. However, the requirement for disulfide bonds is less manifest in apoB18. The present results suggest that an interaction between the disulfide-bonded domain and the downstream lipid binding sequences may play an important role in the post-translational stability of apoB and in VLDL assembly and secretion. Preparation of ApoB Expression Plasmids Containing Cys-to-Ala Substitution-Selective Cys-to-Ala substitutions were introduced into cDNA constructs encoding apoB18 (7) or apoB50. We first created an expression plasmid for B50 (pB50L-L) in which the 3Ј end of the apoB50 coding sequences contained an MluI site at nucleotide 7011 of the apoB cDNA that was introduced by site-specific mutagenesis as described previously (20). To create Cys-to-Ala substitutions, a 2.45-kb EcoRI-HindIII fragment was excised from pB100L-L (prepared in pCMV5) (11) by RecA-assisted restriction endonuclease (RARE) cleavage using two oligonucleotides, RARE-1 (5Ј-ccgcggccgcataggccactagtgaattcgggcgggctgagtgccctcggttgctgc-3Ј) and RARE-2 (5Ј-gcaaggatttttcccagacagtgtcaacaaagctttgtactgggttaatggtcaagttcc-3Ј). RARE-1 protects an EcoRI site within the polylinker of pCMV5 that flanked the 5Ј end of the apoB cDNA (11,20), whereas RARE-2 protects the HindIII site at nucleotide 2279 of the apoB cDNA. The RARE cleavage was performed as described in detail (21), except that both EcoRI methylase and AluI methylase were used together. After RARE cleavage, the 2.45-kb EcoRI-HindIII fragment was inserted into the pZero-1 vector (Invitrogen) to generate a subclone designated pZero-2.45kb that was then subjected to site-directed mutagenesis using the Morph System (5 Prime 3 3 Prime, Inc.). To create pZeroC1-4, pZeroC5-8, and pZeroC9 -12 in which 4 cysteines (i.e. Cys-1 to Cys-4, Cys-5 to Cys-8, and Cys-9 to Cys-12, respectively) were changed to alanines, oligonucleotides designated Cys-1 to Cys-12 shown in Table I were used. To create pZeroC1-12 in which all 12 cysteines (i.e. Cys-1 to Cys-12) were changed to alanines, a fragment encompassing Cys-1 to Cys-4 mutations was excised from pZeroC1-4 by digestion with EcoRI (in the polylinker of pCMV5) and Bsu36I (nucleotide 449 of the apoB cDNA) and ligated to pZeroC5-8 that had been digested with the same enzymes to generate pZeroC1-8. Then a XmnI-HindIII fragment (nucleotides 1547-2279 of the apoB cDNA) containing Cys-9 to Cys-12 mutations was excised from pZeroC9 -12 and used to replace the identical sequence in pZeroC1-8 to generate the resulting pZeroC1-12. To create pZeroC5-6 and pZeroC7-8 in which Cys-5 to Cys-6 and Cys-7 to Cys-8, respectively, were changed to alanines, an EcoRI-BclI fragment containing Cys-5 and Cys-6 mutations from pZeroC5-8 and an EcoRI-BclI fragment containing control sequences from the subclone pZero-2.45 kb were swapped. The mutated 2.45-kb EcoRI-HindIII fragments containing various Cys-to-Ala substitutions were inserted into pB50L-L after RARE cleavage with oligomers RARE-1 and RARE-2 to create apoB50 or into pCMV5 to create apoB18 expression plasmids. Sequences en-compassing the Cys-to-Ala substitutions were verified in each plasmid by DNA sequencing with an ABI 373A DNA sequencer (Perkin-Elmer) using oligonucleotides shown in Table I. The resulting plasmids were purified by centrifugation twice in a a CsCl gradient before transfection experiments.

Materials-Restriction
Cell Culture and Transfection-McA-RH7777 cells were cultured in Dulbecco's modified Eagle's medium containing 20% fetal bovine serum. Stable transformants expressing recombinant apoB were generated using McA-RH7777 cells according to the previously described method (7).
Density Ultracentrifugation of Medium ApoBs-Cells were cultured in Dulbecco's modified Eagle's medium containing 20% serum and 0.4 mM oleate for 10 h. The conditioned media were fractionated into 12 fractions in a sucrose density gradient (12). Total apoBs in each fraction were recovered by immunoprecipitation using a polyclonal anti-apoB antibody (Boehringer Mannheim) that recognized both human and rat apoBs. After separation by electrophoresis on a 5% polyacrylamide gel containing 0.1% SDS (SDS-PAGE), human apoBs were visualized by immunoblotting using antibody 1D1 and rat apoB using antibody LRB220.
Metabolic Labeling of ApoBs-Cells were pulse-labeled with [ 35 S]methionine/cysteine for 1 h and chased for up to 4.5 h in the presence of 0.4 mM oleate and 20% serum (13). Cell and medium apoBs were immunoprecipitated using the polyclonal antibody (Boehringer Mannheim) and were analyzed by SDS-PAGE. The recovery of apoB with the polyclonal antibody was 80 -90% for human apoB and 70 -80% for rat apoB.
Northern Blot Analysis-Northern blot analysis was performed as described previously (22) using a SacI-SacI fragment (nucleotides 3194 -3835) of the human apoB cDNA as a probe.
Microsomal Membrane Isolation, Protease Digestion, and Characterization of Lumenal Lipoprotein Particles-Cells were cultured in Dulbecco's modified Eagle's medium (20% serum Ϯ 0.4 mM oleate) for 6 h. The cells (combined from three 10-cm dishes) were disrupted by ballbearing homogenization in a lysis buffer (10 mM Tris, pH 7.4, 250 mM sucrose, 0.1 mM leupeptin, 0.1 mM phenylmethylsulfonyl fluoride, 2 g/ml aprotinin, 20 g/ml N-acetyl-leucyl-leucyl-norleucinal) containing 20 mM N-ethylmaleimide. The intact microsomes isolated from the post-nuclear supernatant were treated with 0.1 M sodium carbonate, pH 11.3, for 30 min in the presence of protease inhibitors and 20 mM N-ethylmaleimide. The microsomal membranes and lumenal contents were separated by centrifugation at 400,000 ϫ g for 30 min (23). The membrane pellets were dissolved in sample buffer and analyzed by SDS-PAGE under reducing or nonreducing conditions. The lumenal contents (0.8 ml) were diluted to 5 ml with phosphate-buffered saline containing 12.5% (w/v) sucrose and subjected to sucrose gradient ultracentrifugation (12).
Trypsinization of ApoB in Permeabilized Cells-Permeabilization and trypsinization of transfected cells were performed as described (24).
Endoglycosidase H Digestion-Endoglycosidase H (New England Biolabs) digestion was performed according to the manufacturer's instructions. The reaction products were separated by SDS-PAGE and visualized by immunoblot analysis using antibody 1D1 (25). cagaagaagccaagcasa Cys-7-8 gcttcaagcccatccgcacagg Cys-9-10 gtccagccccatcactttacaa Cys-11-12 ggagaagcatcatcaaggaaag a All sequences are shown 5Ј to 3Ј. Mutations are shown in bold in the mutagenic oligonucleotides Cys-1 to Cys-12.
Other Assays-Protein was quantified by the method of Lowry (26) using bovine serum albumin as a standard.
Substituting 12 Cysteines with Alanines Impairs ApoB50 Secretion-We first determined the effect of 12 Cys-to-Ala substitutions on apoB secretion. We used B50 as a model since it has the ability to form VLDL. The secreted control B50 from stable cells (cultured in exogenous oleate) was associated with particles of d Յ 1.01 g/ml (corresponding to VLDL) and of d ϭ 1.09 -1.13 g/ml (corresponding to high density lipoproteins) ( Fig. 2A). Like the oleate-induced assembly and secretion of B48-VLDL (12)(13)(14), secretion of B50 as VLDL was observed only when the culture medium was supplemented with oleate (data not shown).
In contrast to control B50, mutant B50C1-12 was poorly secreted from three stable cell lines examined. One representative B50C1-12 cell line was used for the subsequent experiments. The majority of B50C1-12 was associated with d ϭ 1.13-1.18 g/ml fractions, whereas a trace amount was found in d Յ 1.01 g/ml fractions ( Fig. 2A). The diminished secretion of B50C1-12 (Fig. 2C, Medium) was not attributable to low levels of protein expression, since Northern blot (Fig. 2B) and pulsechase analyses (Fig. 2C, Cell) showed similar results between control and mutant B50s. In all three stable cell lines, expression of B50C1-12 affected the secretion of endogenous rat B100 and B48 as VLDL (data from one representative cell line are shown in Fig. 2A). Similarly, pulse-chase analysis revealed that the secretion of endogenous B100 and B48 was reduced by 80% in B50C1-12 cells as compared with control B50 cells (Fig.  2C, Medium), although the cell B100 or B48 levels were unaffected (Fig. 2C, Cell).
We generated two additional constructs (Cys-5-6 and Cys-7-8) to determine if elimination of one disulfide bond would affect secretion of B50. Immunoblot analysis showed that B50C5-6 and B50C7-8 were both secreted from stable cells with different efficiencies (Fig. 4A). Although accumulation of B50C5-6 in the medium was comparable to that of control B50, accumulation of B50C7-8 in the medium was only ϳ30% that of control B50 (determined by scanning densitometry of the immunoblots). These results suggest that Cys-7 and Cys-8 are probably most crucial for the assembly and secretion of lipoproteins. Similar to other mutants with selected Cys-to-Ala substitution (e.g. C1-4 and C9 -12 in Fig. 3B and 3C), B50C5-6 and B50C7-8 also retained the ability to associate with both high density lipoproteins and VLDL particles (Fig. 4B). However, the level of secreted VLDL was markedly reduced in B50C7-8. Notably, secretion of endogenous rat B100 or B48 was also decreased by ϳ80% in B50C7-8-transfected cells (Fig.  4C). Therefore, expression of all mutants lacking Cys-7 and Cys-8 (i.e. B50C1-12, B50C5-8, and B50C7-8) exerted an inhibitory effect on endogenous VLDL secretion.
Secretion efficiency of mutant B50 was quantified by pulsechase analysis. Data from a typical experiment using cells transfected with control B50 or B50C5-8 are shown in Fig. 5A. Synthesis of B50C5-8 was comparable to that of control B50, whereas secretion of B50C5-8 was markedly decreased. Data of pulse-chase experiments with all B50 variants are summarized in Fig. 5B. Variants that contained a mutation at Cys-7 and Cys-8 (e.g. B50C1-12, B50C5-8, and B50C7-8) all exhibited markedly diminished secretion efficiency, whereas other variants exhibited near normal (e.g. B50C9 -12 and B50C5-6) or moderately decreased (e.g. B50C1-4) secretion efficiency.

Requirement of Disulfide Bonds Is Manifest Less in ApoB18 -
The role of disulfide bonds in apoB secretion was also tested using B18. When only two pairs of cysteine residues or either pair within Cys-5-8 (i.e. Cys-5-6 or Cys-7-8) were mutated, there was no appreciable difference in B18 secretion (Fig. 6A). In contrast, secretion of B50C5-8 was totally impaired, and the level of B50C7-8 was markedly reduced as compared with B50C5-6 ( Fig. 6A). Therefore, formation of disulfide bonds within Cys-5-8 is essential for B50 to assemble into lipoproteins, but it is not as important for B18. However, secretion of B18 was abolished by mutation of all six disulfide bonds (C1-12, Fig. 6A), suggesting that severe alteration of the N-terminal folding impedes apoB secretion regardless of its lipid association.
There are eight potential N-glycosylation sites within B50, of which five are glycosylated in LDL-apoB at asparagine (N) residues 158, 956, 1341, 1350, and 1496 (17). Since Cys-5 (residue 159) is adjacent to asparagine 158, it is possible that NC 159 S to NAS substitution may affect glycosylation (Fig. 6B,  top). We subjected control B50, B50C1-12, control B18, and B18C1-12 (which contains one N-glycosylated asparagine) to endoglycosidase H digestion and found that all of the apoB proteins were sensitive to endoglycosidase H (Fig. 6B). Thus, normal glycosylation at the N terminus of apoB occurred and The atypical bands below B50 in B50C1-4 were only present in that cell line and were not observed in others. C, density profiles of secreted endogenous rat B100 and B48 from indicated B50s transfected cells. The experiments were performed essentially as described in Fig. 2A. was not altered by Cys-to-Ala substitution.
Translocation of apoB50 Is Not Affected by Cys-to-Ala Substitution-Protease protection assay using isolated microsomes was performed to determine if Cys-to-Ala substitution has any effect on B50 translocation. In Fig. 7A, top shows quantification of the proportion of total apoB that is protected from limited proteolysis in the absence or presence of trypsin, whereas the bottom shows representative immunoblots. Recombinant B50, either control or mutant (C1-12), was almost fully protected from trypsin (i.e. 83% of control B50 and 87% of B50C1-12 were protected), suggesting that disruption of disulfide bonding does not affect apoB translocation. Endogenous B48 was also insensitive to trypsin, but B100 was only partially protected (ϳ50% of total). The complete resistance of protein disulfide isomerase to trypsin confirmed the integrity of the microsomes (Fig. 7A). Translocation of apoB was also examined using permeabilized cells in which trypsin digestion was performed in situ. There was no difference in the degree of protection from trypsin digestion of control or mutant B50s (data not shown).
Cytosolic  35 S-B50 was quantified by scintillation counting (bottom) as described in Fig. 2C. B, stable cell lines expressing individual B50 variants were pulse-labeled for 1 h and chased for 2 h. Results are expressed as the percent of the initial radiolabeled B50s recovered in medium and in cells at the end of the 2-h chase. The dashed line indicates the secretion efficiency of control B50. Total, the sum of radioactivity associated with cell and medium B50s.
within the ER (29). We treated the cells with MG132 (Fig. 7B) and found that MG132 had no effect on post-translational degradation of endogenous B100 (Fig. 7B, top) or control B50 (Fig. 7B, middle). Scanning densitometry revealed that irrespective of MG132 treatment, ϳ65% (in control B50 cells) or ϳ35% (in B50C1-12 cells) of B100 and ϳ40% of control B50 were recovered after incubation with MG132. Thus, post-trans-lational degradation of B100 or control B50 is not catalyzed by the proteasome. The recovery of B50C1-12 increased from 9% without MG132 to 17% with MG132 (Fig. 7B, bottom), suggesting that a fraction of B50C1-12 was probably degraded posttranslationally by the proteasome, and the majority of B50C1-12 was degraded in the ER. The enhanced degradation of B100 and B50 in B50C1-12 cells compared with that in  7. Translocation and post-translational degradation of apoB. A, trypsin digestion of microsome-associated apoB. Microsomes prepared from cells expressing control B50 or B50C1-12 were subjected to limited proteolysis with trypsin (50 g/ml) in the presence or absence of 0.5% Triton X-100 (TX-100). Top, percent of human B50, endogenous rat B100, B48, and protein disulfide isomerase (PDI) that were resistant to trypsin digestion. The results were analyzed by scanning densitometry of the immunoblots shown in the bottom. B, effect of MG132 on apoB degradation. Cells expressing control B50 or B50C1-12 were permeabilized by digitonin for 10 min and incubated with (ϩ) or without (Ϫ) MG132 in a cytoskeletal buffer (24) for 2 h. Cells were solubilized, and cellular apoBs were analyzed by immunoblotting. Top, endogenous rat apoBs. Middle, control B50. Bottom, B50C1-12. control B50 cells will be discussed below.
Mutant ApoB50 C1-12 Is Associated with Lipid-poor Particles within the Microsomes-We next examined whether the mutant B50s were membrane-bound or lumenal in nature. Immunoblotting under reducing condition (Fig. 8A, top) showed that at steady state, control B50 was associated with both membrane and lumenal fractions (control lanes). In contrast, B50C1-12 was associated almost exclusively with membrane fraction (C1-12 lanes). Treatment with oleate had little effect on the distribution of B50 (either control or mutant) between membrane and lumen. The absence of B50C1-12 in the microsomal lumen was also apparent when the lumenal content was fractionated by density ultracentrifugation (Fig. 8B). Some studies have suggested that the membrane-bound apoB may serve as the precursor of lumenal lipoprotein-associated apoB (12,29). We assessed possible protein-protein interaction of membrane-associated B50 using nonreducing SDS-PAGE (Fig.  8A, top). Mutant B50C1-12 migrated slower than control B50 during electrophoresis (compare control lanes with C1-12 lanes), suggesting a less restrained configuration of the mutant protein. Notably, several bands of high molecular mass representing covalent complexes that reacted with antibody 1D1 were observed in microsomal membranes of control B50 cells. These bands were intensified when the cells were treated with exogenous oleate. However, the covalent complexes were not present in B50C1-12 cells.
Finally we assessed the effect of Cys-to-Ala substitution on apoB conformation using monoclonal antibody MB19 (30). It was originally suggested that MB19 recognized a polymorphism (Thr 3 Ile substitution at amino acid 71) of B100 (31). Later studies have found that MB19 probably recognizes a conformation of the N terminus of B100 (32). Immunoblots shown in Fig. 8C demonstrate that MB19 did not react with B50 under reducing conditions. However, control B50 could be detected with MB19 under nonreducing conditions. Mutant B50C1-12 reacted with MB19 poorly under nonreducing conditions, suggesting a conformational change within the MB19 epitope. The poor reactivity of MB19 toward apoB under nonreducing condition was also observed with other mutant B50s, including C1-4, C5-8, C9 -12, C5-6, and C7-8 (Fig. 8D). Thus, all six disulfide bonds are likely involved in the presentation of the MB19 epitope. DISCUSSION Protein folding and conformational stabilization by disulfide bonds during translation and translocation into the ER is regarded as a crucial step for the sorting of many secretory proteins. Using site-directed mutagenesis to selectively eliminate disulfide bond formation within B50, we have observed that not all, but only the fourth pair of cysteines (Cys-7 and Cys-8) is essential for the formation and secretion of B50-VLDL (Fig. 4B) and for efficient B50 secretion (Fig. 5B). Mutation at other disulfide bonds has less or no effect, whereas elimination of all six disulfide bonds (B50C1-12) almost abolishes B50 secretion as lipoproteins. The decreased secretion of mutant B50 is not caused by reduced protein expression nor is it attributable to impaired translocation across the ER membrane. Rather, the failure to secrete mutant B50 results from a defect at the post-translational level, leading to degradation of the secretion-incompetent proteins. It is most likely that the inefficiency of lipid recruitment impairs secretion of the mutant proteins. The specific requirement for the fourth disulfide bond in B50 is striking, which suggests that the failed assembly and secretion of mutant B50-VLDL may not be simply the consequence of misfolding of the N terminus of apoB. In fact, mutation at cysteine residues other than Cys-7 and Cys-8 equally impairs proper folding of apoB (as determined by the reduced affinity to MB19 shown in Fig. 8D) but has less effect on B50 secretion. Therefore, it is possible that the fourth disulfide bond plays an important functional role in lipid assembly other than a structural role in folding. However, the possibility that mutation at Cys-7 and Cys-8 may affect the correct formation of other disulfide bonds within apoB has not been ruled out.
Unexpectedly, mutation at Cys-7 and Cys-8 also impairs assembly and secretion of endogenous VLDL (Figs. 2A, 3C, and 4C). Intracellular degradation of endogenous B100 and B48 are markedly increased in cells transfected with mutant B50. Expression of misfolded proteins inside the ER has been shown to induce expression of some molecular chaperones that assist both the folding and degradation processes (33,34). Considering that lipid assembly is a general process for both endogenous and exogenous apoBs, expression of mutant B50 lacking the functional disulfide bond at Cys-7 and Cys-8 may trigger a stress response that typically affects lipid assembly. This would then generate unstable apoB molecules that are more susceptible to degradation. The nature of this stress response is not known. Identification of molecular chaperones or protease systems that are induced in cells transfected with mutant B50 may provide an explanation for the enhanced degradation of endogenous apoBs.
Another important observation made in the current study is the oleate-induced covalent complex formation of apoB in the microsomal membrane (Fig. 8). Direct interaction of apoB with MTP and other ER resident proteins, including BiP/grp78 and calnexin, has been recently demonstrated (35,36). Our results are the first demonstration that complex formation is sensitive to mutations at the N-terminal cysteines of apoB. The concept of a high-order association of ER resident proteins forming a matrix or a gel inside the ER lumen led to a new paradigm that newly synthesized proteins form complexes of heterogeneous sizes with each other or with ER molecular chaperones to acquire more intrachain disulfide bonds until they are fully oxidized and released from the aggregates (37,38). Vesicular stomatitis virus G proteins and procollagen chains are a few of the examples demonstrating complex formation with protein folding catalysts such as BiP/grp78 and protein disulfide isomerase before their maturation (38,39). Similar biochemical events may also occur during apoB-lipoprotein assembly, involving transient protein-protein interactions until the apoB polypeptide achieves a certain degree of lipidation to become secretion-competent.
The current studies have emphasized that interfering with folding of the N terminus of apoB in general has less effect on the secretion of B18 than on B50. Although secretion of apoB18 is abolished when the six pairs of cysteines are mutated, its secretion is relatively resistant to one-or two-pair cysteine mutations. Thus, the stringent requirement for the disulfide bonding appears to be by and large dependent on MTP-mediated lipid assembly. On the other hand, gross disruption of disulfide bonding affects both lipid recruitment and protein secretion. On the basis of these observations, we postulate that correct folding and lipid association may represent two interdependent biochemical events that, in concert with the ER quality control system, regulate the assembly and secretion of VLDL. (a) Proper folding of a critical region involving the fourth disulfide bond might signal a lipid recruitment sensor to initiate VLDL assembly. This forward-signaling event (that is, appropriate folding of the N terminus precedes lipid assembly by acting as an acceptor for MTP) has been suggested by studies with dithiothreitol-treated HepG2 cells (19). (b) The downstream lipid binding sequences, after acquiring adequate lipid load might in turn signal the ER quality control to release the secretion-competent particle. This reversesignaling event has been suggested by studies with oleatetreated Sf-21 cells transfected with apoB, MTP, and protein disulfide isomerase (40).