| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Review Article |
Correspondence to: Jere P. Segrest, To whom correspondence should be addressed., segrest{at}uab.edu (E-mail)
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONApolipoproteinsLDLLIPID-ASSOCIATING DOMAINS OF...MECHANISM OF ASSEMBLY OF...RECEPTOR-BINDING DOMAINS OF...LDL STRUCTURECRITICAL UNRESOLVED ISSUES...DEVELOPMENT OF A CONSENSUS...REFERENCES |
|---|
There is general consensus that amphipathic 
-helices and ß sheets represent the major lipid-associating motifs of apolipoprotein (apo)B-100. In this review, we examine the existing experimental and computational evidence for the pentapartite domain structure of apoB. In the pentapartite nomenclature presented in this review (NH2-ß1-ß1-2-ß2-3-COOH), the original 1 globular domain (Segrest, J. P. et al. 1994. Arterioscler. Thromb. 14: 1674;–1685) is expanded to include residues 1;–1,000 and renamed the ß1 domain. This change reflects the likelihood that the ß1 domain, like lamprey lipovitellin, is a globular composite of -helical and ß-sheet secondary structures that participates in lipid accumulation in the co-translationally assembled prenascent triglyceride-rich lipoprotein particles. Evidence is presented that the hydrophobic faces of the amphipathic ß sheets of the ß1 and ß2 domains of apoB-100 are in direct contact with the neutral lipid core of apoB-containing lipoproteins and play a role in core lipid organization. Evidence is also presented that these ß sheets largely determine LDL particle diameter. Analysis of published data shows that with a reduction in particle size, there is an increase in the number of amphipathic helices of the 2 and 3 domains associated with the surface lipids of the LDL particle; these increases modulate the surface pressure decreases caused by a reduction in radius of curvature. The properties of the LDL receptor-binding region within the overall domain structure of apoB-100 are also discussed.
Finally, recent three-dimensional models of LDL obtained by cryoelectron microscopy and X-ray crystallography are discussed. These models show three common features: a semidiscoidal shape, a surface knob with the dimensions of the ßC globular domain of lipovitellin, and planar multilayers in the lipid core that are approximately 35 Å apart; the multilayers are thought to represent cholesteryl ester in the smectic phase. These models present a conundrum: are LDL particles circulating at 37°C spheroidal in shape, as generally assumed, or are they semidiscoidal in shape, as suggested by the models? The limited evidence available supports a spheroidal shape. — Segrest, J. P., M. K. Jones, H. De Loof, and N. Dashti. Structure of apolipoprotein B-100 in low density lipoproteins. J. Lipid Res. 2001. 42: 1346;–1367.
Supplementary key words:
amphipathic ß sheets, amphipathic -helices, cryoelectron microscopy, X-ray crystallography, boundary lipid, smectic phase, lipid phase transition, LDL receptor-binding domain
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONApolipoproteinsLDLLIPID-ASSOCIATING DOMAINS OF...MECHANISM OF ASSEMBLY OF...RECEPTOR-BINDING DOMAINS OF...LDL STRUCTURECRITICAL UNRESOLVED ISSUES...DEVELOPMENT OF A CONSENSUS...REFERENCES |
|---|
Lipoproteins are submicroscopic particles composed of lipid and protein held together by noncovalent forces. Their general structure is that of a putative spheroidal microemulsion formed from an outer layer of phospholipids, unesterified cholesterol, and proteins, with a core of neutral lipids, predominately cholesteryl ester and triacylglycerols (TAG). Although the microemulsion is the basic structural motif of lipoproteins, several different lipoprotein classes exist that differ in relative amount of lipids, in the protein/lipid ratio, and in the protein species present, resulting in differences in size, density, and electrophoretic mobility. Lipoproteins are generally classified by density, size, and/or protein composition.
|
Apolipoproteins |
|---|
|
TOPABSTRACTINTRODUCTIONApolipoproteinsLDLLIPID-ASSOCIATING DOMAINS OF...MECHANISM OF ASSEMBLY OF...RECEPTOR-BINDING DOMAINS OF...LDL STRUCTURECRITICAL UNRESOLVED ISSUES...DEVELOPMENT OF A CONSENSUS...REFERENCES |
|---|
Apolipoproteins are amphipathic in nature, in that they have both hydrophobic and hydrophilic regions, and can therefore interact both with the lipids of the lipoprotein and with the aqueous environment (1). Because of the nature of these amphipathic regions, apolipoproteins act as detergents, and have a major role in determining and stabilizing the size and structure of the lipoprotein particle.
Plasma apolipoproteins can be grouped into two classes, the nonexchangeable apolipoproteins [apolipoprotein (apo)B-100 and apoB-48], and the exchangeable apolipoproteins (apoA-I, apoA-II, apoA-IV, apoC-I, apoC-II, apoC-III, and apoE) (2). ApoB-100 is highly insoluble in aqueous solution and is one of the largest monomeric proteins known. On the other hand, the exchangeable apolipoproteins are soluble in aqueous solutions, and the secondary structural motif responsible for their lipid association, the amphipathic -helix, has been extensively studied (1) (2) (3).
|
LDL |
|---|
|
TOPABSTRACTINTRODUCTIONApolipoproteinsLDLLIPID-ASSOCIATING DOMAINS OF...MECHANISM OF ASSEMBLY OF...RECEPTOR-BINDING DOMAINS OF...LDL STRUCTURECRITICAL UNRESOLVED ISSUES...DEVELOPMENT OF A CONSENSUS...REFERENCES |
|---|
The only protein component of LDL is a single molecule of apoB-100 per particle (4) (5) (6). LDL, about 200 Å in diameter, is much smaller in size than the originally secreted VLDL that ranges from 600 to 800 Å. Analytical and structural studies of LDL (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) suggest a range of particle sizes (180;–250 Å). Subfractions of LDL, characterized by variations in density, size, and chemical composition, are assuming important clinical significance; a predominance of small dense LDL particles is associated with an increased risk of coronary heart disease (19) (22) (23).
|
LIPID-ASSOCIATING DOMAINS OF APOB-100 |
|---|
|
TOPABSTRACTINTRODUCTIONApolipoproteinsLDLLIPID-ASSOCIATING DOMAINS OF...MECHANISM OF ASSEMBLY OF...RECEPTOR-BINDING DOMAINS OF...LDL STRUCTURECRITICAL UNRESOLVED ISSUES...DEVELOPMENT OF A CONSENSUS...REFERENCES |
|---|
Amphipathic -helical and ß-strand motifs
The B apolipoproteins are highly insoluble in aqueous solutions and, thus, remain with the lipoprotein particle throughout its metabolism (24). Because of the size and insoluble nature of apoB, it has been difficult to confirm the structural motifs responsible for the lipid-associating properties of this nonexchangeable apolipoprotein (25) (26).
Circular dichroic (CD) spectroscopy of LDL suggested that apoB-100 has an -helical content of 25% or greater (27) (28) (29) (30). Amphipathic -helices, the ubiquitous lipid-associating motifs in the exchangeable apolipoproteins, were detected in the sequence of apoB-100 by helical wheel analysis (5) (31). Using computer analysis, De Loof et al. (32) noted two clusters of potential 22-mer amphipathic helical repeats between residues 2,079;–2,428 and 4,150;–4,484. Further, they showed, using comparison matrix analysis, that the regions between residues 2,035;–2,506 and 4,002;–4,527 contained sequence similarity to the exchangeable apolipoproteins.
Yang et al. (33) (34) used nondissociability of peptides from trypsin-treated intact LDL to develop a map of the lipid-associating regions of apoB-100 ( Fig 1). They determined the regions of apoB on the LDL particle that were trypsin releasable, those that were not, and those there were mixed. Based on this criteria, five broad domains of apoB were identified. Their map defined two major apoB-100 lipid-associating domains between residues 1,701;–3,070 and 4,101;–4,536.
|
Analyzing the data of Yang et al. (33) (34) in another way, two continuous stretches of apoB-100, residues 2,100;–2,700 and 4,100;–4,500, contain less than 10 trypsin-releasable residues per 100 amino acid residues (35). These two regions correspond closely to the amphipathic helical repeats identified by De Loof et al. (32) between residues 2,079;–2,428 and 4,150;–4,484.
As first suggested by Gotto, Levy, and Fredrickson (28), apoB-100 differs from the exchangeable apolipoproteins in that it contains ß-sheet structure, estimated using CD spectra of isolated LDL to be 15;–25% (28) (30) (36) (37). Infrared spectroscopy, considered a better method for determining the content of ß structure than CD spectroscopy, suggested as much as 41;–50% ß sheet in apoB-100 [ (37) and V. K. Mishra et al., unpublished data].
The results of infrared spectroscopy also suggested that the ß sheets of apoB-100 are oriented parallel to the phospholipid monolayer of LDL (37). Relevant to this possibility, several different research groups have postulated that amphipathic ß strands contribute to the high affinity of apoB-100 for the lipid surface of VLDL and LDL (5) (35) (37) (38) (39) (40) (41) (42). Significantly, Fourier analysis of amphipathic structures in apoB-100 by Nolte (42) indicated that two large regions of apoB-100 are enriched in amphipathic ß strands. Supporting the concept that amphipathic ß strands contribute to the lipid affinity of apoB-100, a model amphipathic ß-strand peptide was synthesized and shown to have properties similar to apoB-100 (43). The computer analysis of amphipathic helical repeats by De Loof et al. (32) combined with the Fourier analysis of amphipathic ß strands by Nolte (42) suggested that apoB-100 might contain four consecutive domains alternately enriched in amphipathic -helices and amphipathic ß strands.
Pentapartite model
A computer program called LOCATE was developed by Segrest et al. (35) to examine the lipid-associating domains of apoB-100 in more detail. Analysis with this program confirmed the presence of two regions of amphipathic ß strands alternating with two regions of amphipathic -helices (32) (42), and suggested the presence of a third N-terminal amphipathic -helical domain, giving apoB-100 a pentapartite structure: NH3-1-ß1-2-ß2-3-COOH ( Fig 2). As the amphipathic -helices of the 1 domain were class G* (1) (2) (3), these authors proposed that the 1 domain of apoB represented a globular region. Because the amphipathic -helices of the 2 and 3 domains were mostly class A and Y (the types found in the exchangeable apolipoproteins), they proposed that these two regions of apoB-100 represented flexible domains with reversible lipid affinity. These authors further suggested that the two amphipathic ß-strand domains in apoB-100 (ß1 and ß2) represented the irreversibly lipid-associated regions of this irreversibly associated apolipoprotein.
|
In a second publication, Segrest et al. (44) compared the complete sequence of human apoB-100 with partial sequences from eight additional species of vertebrates. They showed that class A lipid-associating amphipathic -helices cluster in two domains in all species for which those regions have been sequenced, but with little conservation of individual helices: 2 between residues 2,075 ± 25 and 2,575 ± 25, and 3 between residues 4,100 ± 100 and 4,550 ± 50. These authors further showed that amphipathic ß strands cluster in two domains in all species for which these regions have been sequenced, with apparent conservation of several individual amphipathic ß strands: ß1 (approximately residues 827;–2,000) and ß2 (approximately residue 2,571 to residue 4,000 ± 50). Finally, they found that hydrophobic segments were present in apoB-100 sequences of all nine species, but the frequency of occurrence was no greater than generally found in ß sheet-containing proteins. Analysis of the overall conservation of the 1 domain was difficult because of limited amino acid sequence data on this domain of apoB from species other than human. It was concluded that the four alternating lipid-associating domains, -ß1-2-ß2-3-COOH, are common supramolecular features of apoB-100 in all vertebrate species (44).
Effects of apoB truncation on lipid affinity
Several laboratories have reported a number of naturally occurring apoB gene mutations that produce truncated forms of apoB that range in length from 9% (apoB-9) to 89% (apoB-89) of full-length apoB-100 (45) (46). Because many domains throughout the entire length of apoB appear to be involved in lipid binding, truncated apoB is expected to form denser lipid-poor particles. For example, Parhofer et al. (47) demonstrated that apoB-89, apoB-75, and apoB-54.8 were secreted into the VLDL fraction, whereas apoB-31 was secreted into HDL. In general, the shorter the mutant apoB protein, the denser and more lipid poor the particle (48). In fact, studies on the expression of truncated forms of apoB in cultured cells have indicated that the size of the secreted particles and the amount of lipid per particle are proportional to the length of the truncated apoB (49) (50) (51) (52).
Studies in mouse mammary carcinoma-derived cells by Carraway et al. (52) demonstrated that apoB-29, apoB-32.5, and apoB-37 had peak densities of 1.25, 1.22, and 1.16 g/ml, had lipid weight percentages of 30%, 37%, and 49%, and had calculated anhydrous particle diameters of 81, 88, and 101 Å, respectively. The core TAG of the apoB-29, apoB-32.5, apoB-37, and apoB-41 particles accounted for 11%, 28%, 58%, and 68%, respectively, of the total lipid mass. These investigators (52) concluded that the sequences in the C-terminus of apoB-29 bind phospholipids and diacylglycerols, sequences between apoB-29 and apoB-32.5 augment TAG binding, and sequences between apoB-32.5 and apoB-41 account for the marked incorporation of TAG at the rate of 
1 TAG per two amino acids.
Morphology of detergent-solubilized apoB-100
A recent publication by Gantz, Walsh, and Small (53) reported the shape of sodium deoxycholate-solubilized apoB-100 using cryoelectron microscopy. The majority of molecules prepared in negative stain and vitreous ice were curved and had alternating thin and thick regions. In negative stain, the apoB molecules lay on the grid perpendicular to the electron beam and had a mean length of 650 Å. In vitreous ice, the molecules measured up to 650 Å in length and showed one or two beaded regions. Similar regions were also observed in negative stain. Some vitrified molecules contained ribbon-like portions. These images suggested a long flexible beaded-thread morphology for apoB-100 (53).
Conformation of the ß1 and ß2 domains of apoB-100 on LDL particles
Although the amphipathic ß strand is generally accepted as a major lipid-associating motif in apoB, little direct evidence for its mode of association with phospholipid monolayers exists. Small and Atkinson [ (54) and personal communication] analyzed the amphipathic ß strands in the N-terminal half of apoB. For this analysis, they combined the Fourier power spectrum approach of Nolte and Atkinson (55) and the algorithm of White, Stultz, and Smith (56). Their analysis located 57 amphipathic ß strands of 11 residues or more in B41; 41 were between residues 968 and 1,882 (B21;–B41) and 47 were between B18 and B41. Most of the strands were 11-mers but several were longer, up to 15 residues in length. The 41 amphipathic ß strands in B21;–B41 contained over 58% of the residues. These investigators proposed (54) that this region of apoB forms a nearly continuous amphipathic ß sheet. When they modeled the 41 amphipathic ß strands in B21;–B41 as a continuous sheet, they calculated the total 
G of lipid-association for the sheet to be on the order of 500 kcals/mol, an extremely high lipid affinity (54). This proposal is similar to the one by Segrest et al. (35) that the two amphipathic ß-strand domains (ß1 and ß2) with, for all practical purposes, infinite lipid affinity (44), represented the irreversibly lipid-associated regions of apoB-100.
Amphipathic ß strands are found in a few classes of proteins in addition to apoB. One example is the egg yolk lipoprotein, lipovitellin. Fig 3 illustrates the molecular features of a five-stranded antiparallel amphipathic ß-sheet segment, residues 1,374;–1,440, from the lipid-associated nine-stranded amphipathic ßD-sheet domain of lamprey lipovitellin (see Fig 4). The structure of this ß sheet was determined by X-ray crystallography (57) (58); the sheet shown in Fig 3 has an average of 8.5 residues per strand (to give an average sheet width of approximately 30 Å) and 6 residues per turn. In Fig 3, the ß sheet is oriented to view the hydrophobic and hydrophilic faces in the left- and right-hand columns, respectively, and is displayed as cartoon, stick, and space-filling models in the upper, middle, and lower rows, respectively. The space-filling models highlight hydrophobic residues in gold, positively charged residues in blue, and negatively charged residues in red. In the left-hand stick model, one set of interstrand hydrogen bonds is represented by dashed blue lines.
|
|
Local alignment of sequence blocks performed by the computer program MACAW (Multiple Alignment Construction and Analysis Workbench), searching by segment-pair overlap, suggested that the sequence of the ß-sheet structure of lipovitellin shown in Fig 3 possesses a weak similarity to a sequence at the N-terminal end of the ß1 domain of apoB near residue 1,150 (59). This ß sheet thus represents a good working model for the proposed amphipathic ß sheets of the ß1 and ß2 domains of apoB-100. A continuous amphipathic ß-sheet model for the ß1 and ß2 domains of apoB-100 can explain the findings of Gantz, Walsh, and Small (53) that detergent-solubilized apoB-100 is a long thin flexible thread of about 650 Å in length that contains several ribbon-like portions.
|
MECHANISM OF ASSEMBLY OF APOB-CONTAINING LIPOPROTEIN PARTICLES |
|---|
|
TOPABSTRACTINTRODUCTIONApolipoproteinsLDLLIPID-ASSOCIATING DOMAINS OF...MECHANISM OF ASSEMBLY OF...RECEPTOR-BINDING DOMAINS OF...LDL STRUCTURECRITICAL UNRESOLVED ISSUES...DEVELOPMENT OF A CONSENSUS...REFERENCES |
|---|
Background
The assembly of apoB-containing lipoprotein particles occurs co-translationally (60); that is, the C-terminal portion is synthesized on the ribosome of the endoplasmic reticulum (ER) as the N-terminal portion is assembling a small lipoprotein particle. Disulfide-dependent folding of the N-terminal domain of apoB is required for assembly (61) (62) (63). One often-quoted mechanism for the physical assembly of lipid particles containing apoB is the budding oil droplet (60). In this model, the N-terminal portion of apoB is embedded in the inner monolayer of the ER membrane where it nucleates an oil droplet from the supersaturated rough RE membranes. Upon completion of apoB synthesis, this oil droplet is detached from the bilayer to form the nascent lipoprotein. One weakness of this model is that an extensive search by electron microscopy for inner rough ER membrane blebs in liver microsomal preparations has failed (60). Further, thermodynamic considerations make it unlikely that lipoproteins assemble through the wholesale remodeling or dismantling of membrane bilayers as suggested by the budding oil droplet model. It seems more likely that apolipoproteins accrete the lipid for their corresponding lipoprotein particles gradually (64).
The autosomal recessive disorder, abetalipoproteinemia, results in a virtual absence of apoB-containing lipoproteins, and microsomal triglyceride-transfer protein (MTP) is also not detectable (65). Although disputed by some (66), this observation has led to the suggestion that MTP is necessary for the formation of apoB-containing lipoproteins (67) (68) (69). Recent studies in cell lines that do not express apoB and MTP (i.e., HeLa and COS-1 cells) have clearly demonstrated that co-transfection of these cells with these two proteins resulted in secretion of apoB-containing lipoproteins (67) (70). Further, an inhibitor of the MTP has been shown to inhibit apoB secretion from HepG2 cells (68). Evidence to date supports the notion that initial lipidation of the nascent apoB polypeptide may occur through direct association with MTP, and that this step may be required for proper folding of the polypeptide in order to escape ubiquitination and subsequent degradation (71).
Lipid pocket model
In their comparison of apoB-100 from nine species, Segrest et al. (44) reported that a number of the amphipathic -helices in the 1 domain, unlike the 2 and 3 domains, appeared to be conserved. In a follow-up article, Segrest, Jones, and Dashti (59) reported that the first 1,000 residues of human apoB-100 (the 1 domain plus the first 200 residues of the ß1 domain, the ß1 domain) have sequence and amphipathic motif homologies to lamprey lipovitellin. The X-ray crystal structure of lamprey lipovitellin, an egg yolk lipoprotein shown in Fig 4 (57) (58) (72) (73), is known to contain a "lipid pocket" lined by three antiparallel amphipathic ß sheets designated ßB, ßA, and ßD. The top two amphipathic ß sheets, ßB and ßA, are joined together by an antiparallel double-layered bundle consisting of 17 -helices designated the domain. The fourth ß-sheet domain of lamprey lipovitellin, ßC, forms a globular ß barrel structure at the apex of the triangular lipid pocket.
Lipovitellin was compared to apoB-100 by Segrest, Jones, and Dashti (59) because a database search for protein sequences that contained amphipathic ß strands similar to those found in apoB-100 turned up four vitellogenins, the precursor form of lipovitellin, from chicken, frog, lamprey, and C. elegans. Segrest, Jones, and Dashti (59) also showed that most of the 1 domain of human apoB-100 has sequence and amphipathic motif homologies to human MTP, an observation also made by others (74) (75) (76).
Based upon their results, Segrest, Jones, and Dashti (59) suggested that a lipovitellin-like "proteolipid" intermediate containing a lipid pocket is formed by the N-terminal portion of apoB. They suggested that this intermediate produces a lipid nidus required for assembly of apoB-containing lipoprotein particles. Pocket expansion through the addition of amphipathic ß strands from the ß1 domain of apoB then results in the formation of HDL, and then in VLDL-like spheroidal particles ( Fig 5).
|
The apparent absolute requirement of MTP for assembly of apoB-containing lipoproteins could simply mean that MTP plays a purely nonstructural role as a shuttle to fill the lipid pocket of the proposed lipovitellin-like apoB intermediate. However, because MTP binds specifically to the unlipidated ß1 domain of apoB (75) (77) (78), MTP also may play a more central structural role in assembly of apoB-containing lipoproteins. Segrest, Jones, and Dashti (59) hypothesized that in the absence of the MTP, the lipovitellin-like apoB lipid pocket intermediate is incomplete, and no apoB-containing lipoprotein can be assembled (Fig 5).
Particularly relevant in the development of a model is a study by Hussain et al. (77). They showed that positively charged amino acid residues, presumably on amphipathic -helices located between residues 430;–570 of apoB, are critical for MTP binding to apoB. Further, they found that 40% and 70% of this binding activity is abolished by truncation of the apoB:570 construct to residues 509 and 502, respectively. LOCATE identifies three positively charged amphipathic -helices between residues 477;–491, 492;–508, and 527;–541 in the N-terminus of the domain of apoB that are homologous to the C-terminal half of the -helical domain lipovitellin; the locations of these three amphipathic helices, particularly the latter two, correlate extremely well with the Hussain et al. (77) data.
In another recent article, Mann et al. (75) used molecular modeling to suggest a homology of the N-terminal region of lamprey lipovitellin to the N-terminal portions of both apoB and MTP. Based upon the results of site-directed mutagenesis, these authors proposed that initial apoB binding to MTP occurs via their respective homologue domains (75). These results suggest that the ß barrel structure proposed by Mann et al. (75) for the N-terminal region of apoB may be the TAG-binding domain of apoB that accepts "shuttled" monomeric TAG from the ß barrel region of the N-terminal domain of MTP.
As noted earlier in this review, expression of progressively smaller C-terminal truncated forms of apoB-100 in cell culture demonstrated that near the N-terminal end of the ß1 domain, the percentage of the truncated apoB-100 associated with lipid approaches zero (48) (49) (66) (79) (80). There appears to be a threshold in apoB size (somewhere between apoB-23 and apoB-28) under which the polypeptide cannot form a lipoprotein particle. Further, McLeod et al. (80) showed that B-29 forms a HDL-like particle when expressed in McA-RH7777, but VLDL particle formation requires inclusion of multiple amphipathic ß strands from the ß1 domain beyond apoB-29. Other studies have shown that the diameters of secreted lipoprotein particles are linearly proportional to the length of C-terminal truncated apoB-100 fragments until a minimal apoB fragment length is reached (49) (60); the linearity falls off near the end of the ß1 domain, between residues 1,100 and 1,300. In another study, B-17 was recovered from the media free of lipid, but when combined with phospholipid, it formed discoidal complexes, suggesting the presence of amphipathic motifs within the B-17 sequence (66). Taken together, truncation studies suggest a rather abrupt change in the nature of the interactions between apoB and bulk lipid at the ß1-ß1 boundary. They support the suggestions (35) (44) that a) a major portion of the ß1 domain is globular and not an integral part of the bulk lipid of the apoB-containing lipoprotein particles, and b) the ß1 domain is extended and is integrated into the bulk lipid of these same particles.
|
RECEPTOR-BINDING DOMAINS OF APOB-100 |
|---|
|
TOPABSTRACTINTRODUCTIONApolipoproteinsLDLLIPID-ASSOCIATING DOMAINS OF...MECHANISM OF ASSEMBLY OF...RECEPTOR-BINDING DOMAINS OF...LDL STRUCTURECRITICAL UNRESOLVED ISSUES...DEVELOPMENT OF A CONSENSUS...REFERENCES |
|---|
Published results
As with so many other aspects of apoB-100, characterizing its interaction with the LDL receptor at a molecular level turned out to be a difficult task. When the sequence became available, it was not clear whether predictions and early experiments supported one (38), two (4), or more (81) domains interacting independently or in concert with the LDL receptor. Indeed, Cladaras et al. (5) performed hydrophobic moment calculations on apoB, as previously done by DeLoof et al. (82) on apoE, the other well-characterized ligand of the LDL receptor (83), and this pointed to the presence of several possible interacting regions enriched in positively charged residues. The top candidate of this analysis, sometimes called site B, had much in common with the receptor-binding domain of apoE, and experiments with a synthetic peptide (apoB 3,345;–3,381) (38) showed the importance of this region in receptor binding and internalization of LDL particles. This did not, however, exclude the importance and/or participation of other domains.
Additional evidence came from experiments with monoclonal antibodies. Unlike the region between residues 2,980 and 3,780, the rest of the protein remained accessible for antibodies while LDL was bound to its receptor (84). Subsequently, Law and Scott (85) compared the sequences of apoB in seven species, and concluded that site B (3,359;–3,367) is the primary site involved in receptor interaction.
Truncation mutants of apoB also provide information, as apoB-67 (the amino-terminal 67% of the protein) did not interact with the LDL receptor (86), but apoB-75 (the protein up to residue 3,387) was fully active (87). One important recent report further validates this and provides additional insights into the modulation of the receptor-binding domain as it becomes available for interaction during metabolic conversion of VLDL to LDL. Boren et al. (88) convincingly showed through site-directed mutagenesis that site B is indeed the domain interacting with the receptor. In addition, these experiments showed that the arginine in position 3,500, when mutated to a glutamine, changes the conformation of the C-terminal tail, reducing receptor binding and explaining the molecular basis of the genetic disorder, familial defective apoB-100 (89). The same mutation in apoB devoid of its 20% C-terminal domain did not disrupt receptor binding, but dramatically increased the binding of VLDL, thereby initiating our understanding of the conformational changes in apoB that modulate its metabolic fate.
Although the preeminence of site B in receptor binding with the LDL receptor is now clear, a recent report suggested that charged clusters might be important in the catabolism of truncated apoB mutants. ApoB-70.5 (truncating at residue 3,196) contains site A (3,129;–3,151), but not site B; it does not interact with the LDL receptor, but it interacts with LRP-2, the megalin receptor (90). The latter observation seems to be an exception, as unmodified apoB has not been described as the regular mediator of the interaction with other members of the LDL-receptor gene family including the chylomicron remnant receptor or the VLDL receptor [for a review, see ref. (91)].
In addition to LDL-receptor binding, apoB is known to interact with proteoglycans, a process also relevant in the pathogenesis of atherosclerosis (92). Heparin has been used as a model for this interaction, and several heparin-binding domains of apoB have been localized (93) (94) Synthetic peptide work in an in vivo atherosclerosis model demonstrated the importance of a charged cluster other than site B (i.e., residues 1,000;–1,016) (95), and more recent work (96) showed the potential for other, also positively charged peptides. In addition, Goldberg et al. (97) showed that apoB-48 lacking site B was capable of binding proteoglycans, explaining why apoB-48-containing lipoproteins may be equally atherogenic. The work of Boren et al. (98) on apoB-100, however, points strongly toward site B as the primary binding site for these interactions with full-length apoB. In addition, they showed that mutants defective in proteoglycan binding could retain their LDL-receptor affinity. Together, these experiments again suggest that apoB is capable of major conformational changes with important metabolic consequences.
The repercussion of these interactions with proteoglycans is the retention of lipoproteins in the arterial wall, making it possible for the apoB-containing lipoprotein particles to become modified [for a review of the different mechanisms, see ref. (92)]. These modified lipoproteins are avidly taken up by the cell though different receptors, resulting in foam cell formation. The quantitative importance of these different mechanisms, the molecular detail of these receptor interactions, and the apoB domains involved remain largely unexplored.
New computer analyses
In this section, we analyze the sequence and structural homology of the putative LDL-receptor binding region within the amphipathic ß-strand domain, ß2. These analyses provide insights into mechanisms of receptor binding that both complement and supplement previously published results.
Fig 6 is a LOCATE analysis (35) (44) of residues 2,501;–4,100 from human apoB-100, showing the predicted distribution of amphipathic ß strands, positively charged amphipathic -helices, prolines, prolines plus valines, and cysteines. Three long well-defined positively charged amphipathic -helices, residues 3,129;–3,151, 3,170;–3,208, and 3,347;–4,479, are identified. The first and last of these helices correspond approximately to site A and site B, respectively, as previously described (4) (5) (38) (81) (82). The middle helix, termed site C, has not been described previously.
|
Three proline-rich domains (prd) previously described (32) (99) are clearly delineated in Fig 6 between residues 2,551;–2,740 (prd1), 3,216;–3,291 (prd2), and 3,687;–3,865 (prd3). These domains, possibly representing partial gene duplications (32), are also rich in valine (Fig 6). As previously noted (34), prd2 separates site A from site B; this domain also separates site C from site B.
Three lines of evidence suggest that the prd2 domain forms a condensed protein structure unassociated with the lipid surface of LDL (condensed domain). First, as defined by the disulfide bond between the cysteines at residues 3,167 and 3,297 (Fig 6), prd2 (as well as site C) is part of a local loop in the structure of apoB-100 on the LDL particle. Because this disulfide bond is absent in apoB from all other species in which sequence information is available (44), the loop encompassing prd2 and site C seems to be independent of disulfide bond formation. Second, a study of apoB peptides released from the surface of native LDL (99) concluded that the prd2 domain is surface exposed and free of interaction with lipid; 4 of the 11 peptides isolated in this study were from the prd2 domain. Finally, the pdr2 domain, site C, and both cysteine residue positions are missing in the sequence of salmon apoB-100 [ref. (100), Fig 6, and Fig 7], yet fish LDL binds to the human fibroblast LDL receptor with comparable affinity to human LDL (101).
|
The other two prd (prd1 and prd3) also appear to form condensed domains on the surface of LDL. The study by Forgez et al. (99) found that peptides from both prd1 and prd3 were also released by trypsin treatment of native LDL. Further, the prd3 (and possibly the prd1) domain, like the prd2 domain, is missing in the sequence of salmon apoB-100 (Fig 7).
Fig 8 (left panel) is a plot of the positively charged amphipathic -helices identified by LOCATE between residues 3,001 and 3,500 in apoB-100 from 10 species of vertebrates (human, monkey, pig, mouse, rat, hamster, rabbit, chicken, frog, and salmon). As can be seen, site A is present as a well-conserved amphipathic -helix located between residues 3,129 and 3,151 in mammals, but is missing in birds, amphibians, and fish. These results support the view that site A is not the primary LDL-receptor binding site. The presence of site A in mammals only is an intriguing observation with uncertain significance.
|
On the other hand, site B is present as one or two considerably less well-conserved amphipathic -helices between residues 3,345 and 3,377;–3,404 in all species examined including fish. These results are compatible with site B being the primary LDL-receptor binding site, although its interspecies conservation is not as strong as previously suggested by Law and Scott (85).
Site C is present and fairly well conserved in all species except salmon in which that portion of the sequence is deleted (Fig 8, black bar, right panel), and in hamster in which there is an additional positively charged amphipathic -helix located between site B and the prd2 domain that is not present in other species. The possible function of site C is unknown.
Fig 8 (right panel) is a LOCATE plot of all proline residues between residues 3,001 and 3,500 in apoB-100 from the 10 species of vertebrates whose sequences are available. The location of the disulfide bond present in human apoB-100 is also shown. As can be seen, a striking clustering of mostly conserved proline residues is present between residues 3,216 and 3,291 in mammals, birds, and amphibians; a cluster that is deleted in salmon (Fig 8, black bar, right panel). Interestingly, the homologous residue positions of the two cysteines forming the bond are framed on their N- and C-terminal sides by isolated prolines that just fall within the salmon deletion (Fig 8, black bar, right panel).
We postulate that all three prd (prd1, prd2, prd3) form condensed domains, defined as protein folds on the surface of LDL not associated with lipid. Further, we speculate that these regions may be related to apoB-100 structural changes during the conversion from VLDL to LDL.
|
LDL STRUCTURE |
|---|
|
TOPABSTRACTINTRODUCTIONApolipoproteinsLDLLIPID-ASSOCIATING DOMAINS OF...MECHANISM OF ASSEMBLY OF...RECEPTOR-BINDING DOMAINS OF...LDL STRUCTURECRITICAL UNRESOLVED ISSUES...DEVELOPMENT OF A CONSENSUS...REFERENCES |
|---|
Published LDL models
It has been generally assumed that LDL is a spheroidal particle approximately 200 Å in diameter that contains a cholesteryl ester-rich core surrounded by a phospholipid-rich shell (60). The issue addressed in this section is the molecular organization of the apoB-100 molecule on the phospholipid-rich surface of the LDL particle and the implications of this organization for overall particle structure and function.
Schumaker and colleagues have painstakingly mapped the positions of 11 anti-apoB monoclonal antibodies onto the surface of human LDL by electron microscopy (102). The first 89% of apoB-100 was modeled as a thick ribbon that wraps once around the LDL, completing the encirclement by approximately amino acid residue 4,050 (the junction of the ß2 and the 3 domains). The thickness of the ribbon was proposed to be approximately 20 Å, sufficient to penetrate the monolayer and to make contact with the core. They proposed a kink in the ribbon beginning almost halfway along its length at approximately apoB-48 (the start of the 2 domain). The C-terminal 11% of apoB (the 3 domain) was termed a "bow," an elongated structure of about 480 residues beginning at residue 4,050, stretching back into one hemisphere, and then crossing the ribbon between residues 3,000 and 3,500 into the other hemisphere.
The most complete modeling of the lipid and protein components of LDL subclasses is an elegant study by McNamara et al. (21). These authors combined the use of nondenaturing gradient gel electrophoresis and ultracentrifugation with calculations of the number of molecules of each lipid and protein component of eight distinct LDL subclasses from 66 subjects to derive detailed compositional models for each subclass. Their conclusion was that differences in LDL subclasses involve both changes in lipid composition and conformational changes in apoB-100. In particular, they concluded that apoB-100 progressively unfolds to cover an increasing area of the LDL particle surface as LDL subclasses decrease in size.
Several models for LDL have been suggested from studies using low angle X-ray scattering (13) (14) (103). Most of the models were based upon an oil droplet or microemulsion model containing a predominantly cholesteryl ester lipid core (104). Relevant to the cholesteryl ester core model, a broad phase transition in LDL was observed by differential scanning calorimetry and X-ray and neutron solution scattering at a temperature in the range of 19;–32°C (12) (104) (105) (106) (107) (108) (109) (110) (111). The transition was associated with a smectic-to-disorder phase change of cholesterol esters within the particles. Below the transition temperature, in the smectic phase, a 1/36 Å-1 reflection was observed in X-ray solution-scattering experiments (12).
van Antwerpen and collaborators (112) (113) (114) reported images of LDL analyzed in a vitrified frozen-hydrated condition without chemical fixation or any form of staining (cryoelectron microscopy). Based upon these studies, it was concluded that the overall shape of human LDL was discoidal ( Fig 9a). The observed LDL discs were reported to have a diameter of 214 ± 13 Å and a height of 121 ± 11 Å. These authors concluded that apoB-100 appeared to form two ring-shaped structures around the perimeter of the LDL disk.
|
Spin and Atkinson (115) also reported images of LDL in vitreous ice using electron cryomicroscopy (at approximately 30-Å resolution). As shown in Fig 9b, LDL appeared in their study to be a quasi-spherical particle approximately 220;–240 Å in diameter, with a region of low density surrounded by a ring (in projection) of high density believed to represent apoB-100. The ring was noted to be composed of four or five large regions of high density material, presumably protein superdomains, connected by regions of lower density. Some images were egg shaped and contained a pointed end; the pointed end (arrow in Fig 9b) was postulated to represent the N-terminal globular region (the ß1 domain) of apoB. These authors also observed disc-shaped structures similar to those seen by van Antwerpen and collaborators (112) (113) (114), but considered these objects artifacts.
In a third article using electron cryomicroscopy, a three-dimensional structure of LDL particles at 27-Å resolution was reported by Orlova et al. (116). Multivariate statistical and cluster analyses were used to identify a subset of structurally homogeneous LDL particles from a larger pool of particles with heterogeneous sizes and features. Angular reconstitution was then used to develop a three-dimensional model from a subset of 2,600 particles in 129 classes. Fig 9c is a three-dimensional map of the ellipsoid LDL structure with dimensions of 250 x 210 x 175 Å (116). Two features of this model are especially noteworthy. First, a knob-shaped electron-dense object appears on the surface of the model (red arrow), with dimensions (35;–45 Å) that approximate the ßC globular domain of lipovitellin. Further, the knob contains a central cavity (10;–20 Å in diameter) that is similar to the cavity found in the ßC domain (red arrow). Second, the core of the model contains three major higher density planar layers approximately 35 Å apart (Fig 9c, black arrow). The authors suggested that these layers might represent cholesteryl ester in the smectic phase. Because LDL undergoes a smectic-to-disordered phase transition in the range of 19;–32°C (12) (104) (105) (106) (107) (108) (109) (110) (111), both the flattening of the LDL model and its core organization may be artifacts of the cryogenic temperatures. This model presumably is related to the discoidal structures reported by van Antwerpen and collaborators (112) (113) (114).
Two groups have been able to crystallize LDL (117) (118). The second of these groups (119) recently determined a preliminary X-ray crystallographic model for LDL-2 at 27-Å resolution. Fig 9d shows a stereo view of their model (M. Baumstark, personal communication). Interestingly, this model appears to show the two principal features of the Orlova et al. (116) model: a surface knob with the dimensions of the ßC globular domain of lipovitellin (red arrow, Fig 9d) and a planar multilamellar core (black arrow, Fig 9d). The diffraction data was collected at cryogenic temperatures.
New analysis of published data
The constraint in both the Chatterton et al. (102) and the McNamara et al. (21) studies was a lack of knowledge at that time of the pentapartite organization of apoB-100 (35) (44). We have used the information contained in the pentapartite domain structure of apoB-100 (35) (44) shown in Fig 2 in a re-analysis of the organization of apoB-100 on the surface of the eight so-well-characterized LDL subclasses of McNamara et al. (21). Their model used the following two assumptions:
First, McNamara et al. (21) assumed that both phospholipid and free cholesterol are in contact with the core lipid, and they applied surface pressure assumptions directly to the core surface in order to determine the fraction of the core surface covered by these polar lipids. In our calculations, we assumed that phospholipid and free cholesterol affect the packing of the polar lipids at the aqueous surface of the LDL particles, but that only the fatty acyl chains of the phospholipid make contact with the lipid core.
Second, they assumed that the core lipids of LDL are arranged in a regular spheroidal shape that is in direct contact over a fraction of its total surface with a 20-Å thick section of apoB-100 (21). Our model assumed that only that portion of apoB-100 that forms amphipathic ß sheets (35) (44) is in direct contact with the lipid core. Further, because no protein structural motif is known to form a stable structure that extends 20 Å into a lipid monolayer, we assumed that the core lipid in contact with the amphipathic ß sheets forms ridges that extend to within 7 Å of the LDL aqueous surface ( Fig 10). We further assumed that amphipathic helices from the domains of apoB-100 (35) (44) act much like free cholesterol to increase the packing density of the polar lipids at the aqueous surface of the LDL particles.
|
Six additional assumptions were used to develop our model: first, we assumed, as first suggested by Small and Atkinson [ (54) and personal communication], that the individual ß strands of the ß1 and ß2 domains of apoB-100 form continuous ß sheets. Further, the number (Nß) and the mean residue length (Lß) of lipid-associating amphipathic ß strands [predominantly located in the ß domains of apoB-100 as defined previously (35) (44)] can be determined using the computer program LOCATE (35) (44) (120) (121). From this, we calculated that for the ß1 sheet, Nß = 81 and Lß = 12.2 and, assuming condensed loops at the three prd of ß2, Nß = 63 and Lß = 12.0 for ß2.
Second, each amphipathic ß sheet has a mean thickness of approximately 7 Å (determined by molecular modeling; results not shown) and, thus, exerts its maximal effect on surface area at a depth of 3.5 Å from the aqueous surface of each LDL subclass.
Third, the surface area (SA) per polar lipid (i.e., phospholipid + free cholesterol) has been calculated for each of the individual LDL subclasses by McNamara et al. (21), and varies from 51;–55 Å2.
Fourth, amphipathic helices of apoB-100 are postulated to be evolutionarily selected for a wide range of lipid affinities (
) from a large number with low lipid affinity or greater to only a few with extremely high . The number and mean residue length of lipid-associating amphipathic -helices for a given lower limit of can be determined by the computer program; the higher the cutoff, the fewer the amphipathic helices.
Fifth, each lipid-associated amphipathic -helix exerts its maximal effect on surface pressure at a depth of 7 Å from the aqueous surface of each LDL subclass (122) (see footnote to Table 1).
|
Sixth, the radius (rCi) and surface area (SACi) of the effective core (core minus the volume of the ß core ridges) for each LDL subclass, i, were calculated assuming that phospholipid monolayers containing amphipathic helices are 23 Å thick (122). Thus rCi = (rLDLi - 23) and SACi = [4
x (rLDLi - 23)2].
Table 1 contains calculations of the fractional surface area coverage by polar lipid, amphipathic ß sheets, and amphipathic -helices as a function of LDL subclass. The equations used are included in a footnote to the table. As first pointed out by McNamara et al. (21), these results indicate that the fractional area occupied by surface lipids decreases with decreasing LDL subclass size, whereas the fractional surface estimated to be covered by amphipathic -helices increases with decreasing size.
The total surface area in Å2 covered by amphipathic -helices can then be calculated as a function of LDL subclass; the results are shown in Table 1. Compatible with the conclusion by McNamara et al. (21), these calculations indicate that the total surface area covered by amphipathic -helices increases as LDL subclasses decrease in size.
Determination of the approximate number of amphipathic -helices associated with each LDL subclass can be made from the surface area data. These calculations involve using the computer program LOCATE to construct a table ( Table 2) of the number of lipid-associating amphipathic -helices and the LDL surface area that would be occupied by these helices (see footnote to Table 2) as a function of lipid affinity less than a given . Then, the LDL subclass surface area data is interpolated onto the table. The number of lipid-associated amphipathic helices (AH) and the number of phospholipid (PL) molecules per lipid-associated amphipathic helix (PL/AH) calculated for each LDL subclass by this interpolation are shown in Table 1. The smallest subclass, LDL-8, has the lowest PL/AH ratio (6.8), with a ±10% range of 6.3 to 8.9. This ratio suggests that the amphipathic helices of LDL-8 are separated, on average, by one layer of phospholipid (see later discussion), although extensive helix-helix interaction is another possibility.
|
Another
6 residues, hydrophobic moment 
