Lipid Binding-induced Conformational Change in Human Apolipoprotein E EVIDENCE FOR TWO LIPID-BOUND STATES ON SPHERICAL PARTICLES*

Apolipoprotein (apo) E contains two structural domains, a 22-kDa (amino acids 1–191) N-terminal domain and a 10-kDa (amino acids 223–299) C-terminal domain. To better understand apoE-lipid interactions on lipoprotein surfaces, we determined the thermodynamic parameters for binding of apoE4 and its 22- and 10-kDa fragments to triolein-egg phosphatidylcholine emulsions using a centrifugation assay and titration calorimetry. In both large (120 nm) and small (35 nm) emulsion particles, the binding affinities decreased in the order 10-kDa fragment (cid:1) 34-kDa intact apoE4 > 22-kDa fragment, whereas the maximal binding capacity of intact apoE4 was much larger than those of the 22- and 10-kDa fragments. These results suggest that at maximal binding, the binding be-havior of intact apoE4 is different from that of each fragment and that the N-terminal domain of intact apoE4 does not contact lipid. Isothermal titration calorimetry measurements showed that apoE binding to emulsions was an exothermic process. Binding dilution were determined in control experiments by injecting either protein solution or emulsion suspension into buffer, and these heats were subtracted from the heats determined in the corresponding pro- tein-emulsion binding experiments to give the enthalpy of binding. Analytical Procedures— Protein concentrations were determined by the procedure of Lowry et al. (32). Lipid concentrations were determined with a phospholipid (Wako Chemicals, Richmond, VA) and tri- glyceride (Sigma) assay kits. 14 C radioactivity was assessed by standard liquid scintillation procedures. Polyacrylamide gel electrophoresis (8– 25% gradient) in the presence of SDS was performed with an Amer-sham Pharmacia Biotech Phast electrophoresis system to monitor the purity of the proteins.

tein, plays a key role in lipoprotein metabolism, serving as a high affinity ligand for the low density lipoprotein (LDL) receptor family and cell surface heparan sulfate proteoglycans (1,2). Defective binding of apoE to receptors causes cholesterolrich lipoprotein particles to accumulate in the plasma and is the mechanism of type III hyperlipoproteinemia, a genetic disorder characterized by elevated plasma cholesterol and triglyceride levels and accelerated coronary artery disease (3). Association of apoE with lipid is required for its high affinity binding to the LDL receptor (4). However, the molecular details of the apoE-lipid interaction remain unclear.
ApoE is composed of two major structural and functional domains. The 22-kDa N-terminal domain contains the receptor-binding region, whereas the 10-kDa C-terminal domain has a high affinity for lipid and is responsible for lipoprotein binding (5,6). The three-dimensional structure of the N-terminal domain has been shown by x-ray crystallographic studies to be an elongated globular four-helix bundle (7,8). Molecular area measurements at an air-water interface suggest that the fourhelix bundle can undergo a conformational change exposing the hydrophobic faces to interact with lipid (9). Indeed, recent studies with infrared spectroscopy (10), fluorescence resonance energy transfer (11), and interhelical disulfide mutants of apoE N-terminal domain (12) indicate that the four-helix bundle undergoes conformational opening on phospholipid discs. Although these studies show ␣-helical reorganization upon lipid binding, the final organization of the helices with respect to one another is unknown (12,13).
The conformational reorganization of the N-terminal domain upon interaction with lipid is associated with enhanced receptor binding activity. NMR measurements showed that, when apoE was complexed with dimyristoyl phosphatidylcholine (DMPC), lysines 143 and 146 in the N-terminal domain (both in the LDL receptor-binding region) have unusually low pK a values, reflecting local increases in positive electrostatic potential caused by reorganization of the ␣-helices (14). The increased electrostatic potential, coupled with enhanced exposure to the aqueous phase of the polar face of the amphipathic-helix containing residues 136 -150, seems to explain why lipid association is required for the high affinity binding of apoE for the LDL receptor (14,15). However, certain apoE-containing lipoprotein particles, such as intact chylomicrons, seem to be receptor-inactive (16). In studies with perfused rat liver (17,18) or phospholipase A 2 -treated chylomicrons (19), it has been proposed that the conformation of lipoprotein-bound apoE can be modulated by lipoprotein lipid composition (20,21) or other apolipoproteins (e.g. C apolipoproteins) (22,23). Based on the observation that lipoprotein-associated apoE displays variable receptor binding ability, Narayanaswami and Ryan (24) have proposed two lipid-bound states model of apoE. The C-terminal domain anchors apoE to the lipoprotein surfaces, whereas the N-terminal domain undergoes reversible conformational changes that modulate receptor binding activity.
In the present study, we determined the quantitative contributions of the N-terminal and the C-terminal domains of apoE to lipid binding on lipoprotein-like emulsion particles with engineered apoE4 and its 22-kDa (residues 1-191) and 10-kDa (residues 223-299) fragments.

EXPERIMENTAL PROCEDURES
Materials-Egg yolk phosphatidylcholine (PC) and triolein were purchased from Sigma, and stock solutions were stored in chloroform/ methanol (2:1) under nitrogen at Ϫ20°C. [ 14 C]Formaldehyde (40 -60 Ci/mol) in distilled water was purchased from PerkinElmer Life Sciences. NaCNBH 3 (Aldrich) was recrystallized from methylene chloride before use (25). All other salts and reagents were analytical grade.
Bacteriological media were obtained from Fisher. The prokaryotic expression vector pET32a was from Novagen (Madison, WI), and the competent Escherichia coli strain BL21 star (DE3) was from Invitrogen (Carlsbad, CA). The competent E. coli strain DH5␣ was from Life Technologies, Inc. Polymerase chain reaction supplies were from Qiagen (Chatsworth, CA). Restriction enzymes were purchased from Promega (Madison, WI). Isopropyl-␤-D-galactopyranoside, ␤-mercaptoethanol, aprotinin, and ampicillin were from Sigma. Ultrapure guanidine HCl was from ICN Pharmaceuticals (Costa Mesa, CA). Oligonucleotides were from Integrated DNA Technologies (Coralville, IA), and DNA purification kits were from Qiagen.
Expression and Purification of Proteins-The full-length human apoE4 and its 22-and 10-kDa fragments were expressed and purified as described (26) with some modification. The cDNA for full-length human apoE4, the 22-kDa fragment, or the 10-kDa fragment were ligated into a thioredoxin fusion expression vector pET32a and transformed into the E. coli strain BL21 star (DE3). The transformed E. coli were cultured in LB medium at 37°C, and thioredoxin-apoE expression was induced with isopropyl-␤-D-galactopyranoside for 3 h. After the bacterial pellet was sonicated and the lysate was centrifuged to remove debris, the fusion protein was cleaved with thrombin to remove thioredoxin from full-length apoE4 or the 22-or 10-kDa fragment. For the full-length apoE4, the fusion protein was complexed with DMPC before it was cleaved with thrombin to protect the protease susceptible internal hinge region (this step was not necessary for the 22-and 10-kDa fragments). After inactivation of the thrombin with ␤-mercaptoethanol, the mixture was lyophilized and delipidated, and the apoE pellet was dissolved in 6 M guanidine HCl, pH 7.4, containing 1% ␤-mercaptoethanol. The apoE was isolated by gel filtration chromatography on a Sephacryl S-300 column. If further purification (Ͼ95%) was needed, the proteins were subjected to gel filtration with a Superdex 75 column or anion exchange chromatography with a HiTrap Q column.
Preparation of Large and Small Emulsion Particles-Emulsion particles were prepared by sonication and purified by ultracentrifugation as described (27,28). Appropriate aliquots of stock solutions of triolein and egg PC (weight ratios of triolein/PC were 3.5:1 for large emulsions and 1:1 for small emulsions) were mixed in a glass tube and dried under a stream of nitrogen, and the tube was placed into a vacuum desiccator overnight. For large emulsions, the dry lipids were suspended in Tris buffer (10 mM Tris-HCl, 150 mM NaCl, 0.02% NaN 3 , 1 mM EDTA, pH 7.4) and sonicated at 80 watts with a Branson sonifier model 350 and flat tip for 30 min at 50 -60°C under a stream of nitrogen. The crude emulsions were centrifuged at 3,000 rpm in a Beckman GPR centrifuge for 15 min to remove titanium. The emulsions were then centrifuged in a Beckman L7 ultracentrifuge with a SW40Ti rotor at 15,000 rpm for 15 min at 20°C to remove large particles (creamy layer) and then washed twice with the buffer by resuspending the emulsion and spinning at 17,000 rpm for 1 h at 20°C to remove any contaminating liposomes. The resulting creamy top layer was collected as large emulsion particles. For small emulsions, the dry lipids were suspended in Tris buffer containing 10% (w/v) sucrose. After sonication under the conditions described above, the emulsions were centrifuged as described above to remove titanium and dialyzed against buffer overnight to remove free sucrose. The emulsions were then centrifuged at 40,000 rpm at 20°C for 20 min to separate large particles (top layer) and small particles (middle layer). The cloudy middle fraction was collected and respun at 28,000 rpm, 20°C, for 12 h. The resulting creamy top layer was collected as small emulsion particles. The weight ratio of triolein to PC in the emulsions was 6:1-7:1 for large emulsions or 2:1-2.5:1 for small emulsions. The average particle diameter determined by quasi-elastic light scattering measurements was 120 Ϯ 10 nm for large emulsions or 35 Ϯ 5 nm for small emulsions, respectively.
Binding of ApoE to Emulsion Particles-Binding of apoE4 and its 22and 10-kDa fragments to emulsion particles was assayed with a centrifugation method (29,30). The 14 C label was introduced into the proteins to a specific activity of ϳ1 Ci/mg protein by reductive methylation of lysines with [ 14 C]formaldehyde as described (14,25,31). This trace labeling of apoE leads to modification of less than one lysine residue/apoE molecule, and there is no detectable change in the physical properties of the protein. Previously, we have established that reductive methylation of all lysine residues has no effect on the interaction of apoE with lipid (31). 14 C-Labeled apoE4 or the 22-or 10-kDa fragment (freshly dialyzed from 1% ␤-mercaptoethanol and 6 M guanidine HCl solution into Tris buffer, pH 7.4) and emulsions (0.3 mg phospholipid/ml) were incubated for 1 h at room temperature with gentle shaking in 1.4 ml of Tris buffer (pH 7.4) containing 0.25 M sucrose. After incubation, the mixtures were centrifuged in a Beckman L7 ultracentrifuge with a 50Ti rotor at 30,000 rpm (for large emulsions) or 50,000 rpm (for small emulsions) for 30 min. All of the lipids were found in the top fraction, and the radioactivity in 200-l aliquots of the top and bottom fractions was quantitated in a liquid scintillation counter. The bound apoE concentration was calculated by subtracting the background free apoE concentration in the top fraction; the latter was obtained from the results of centrifugation of lipid-free apoE solutions. Binding data were fitted by nonlinear regression to a one binding site model with the GraphPad Prizm program.
Isothermal Titration Calorimetry (ITC) Measurements-Heats of apoE binding to emulsions were measured with a MicroCal MCS isothermal titration calorimeter (MicroCal Inc., Northampton, MA) at 25°C. All solutions were degassed under vacuum before use. The reactant was placed in the sample cell (1.33 ml) and titrated with 8 -10-l aliquots of the injectant with continual stirring at 400 rpm. Heats of

FIG. 1. Binding isotherms of full-length apoE4 and its 22-and 10-kDa fragments to small (A) and large (B) emulsion particles.
Various concentrations of full-length apoE4 (ⅷ), the 22-kDa fragment (E), or the 10-kDa fragment (‚) were incubated with emulsion particles at a PC to protein molar ratio in the range of 50 -1000 at room temperature. Each point represents the mean Ϯ S.D. from two independent experiments each done in duplicate. The amount of bound protein is expressed in terms of the molar ratio of amino acid/PC in the emulsion surface. The binding curves were obtained by nonlinear regression fitting to a one-binding site model. dilution were determined in control experiments by injecting either protein solution or emulsion suspension into buffer, and these heats were subtracted from the heats determined in the corresponding protein-emulsion binding experiments to give the enthalpy of binding.
Analytical Procedures-Protein concentrations were determined by the procedure of Lowry et al. (32). Lipid concentrations were determined with a phospholipid (Wako Chemicals, Richmond, VA) and triglyceride (Sigma) assay kits. 14 C radioactivity was assessed by standard liquid scintillation procedures. Polyacrylamide gel electrophoresis (8 -25% gradient) in the presence of SDS was performed with an Amersham Pharmacia Biotech Phast electrophoresis system to monitor the purity of the proteins.

RESULTS
Binding of ApoE to Emulsion Particles-To assess the lipid binding properties of full-length apoE4 and its 22-and 10-kDa fragments, the 14 C-labeled proteins were incubated with small (average diameter of 35 nm) and large (120 nm) emulsion particles at various protein concentrations at room temperature. Emulsion-bound protein was separated from unbound protein by ultracentrifugation. All proteins appeared to bind to the emulsion surface in a saturable manner regardless of particle size (Fig. 1). The dissociation constant (K d ) and the maximal binding capacity (B max ) are listed in Table I. Full-length apoE4 and the 10-kDa fragment bound with greater affinity than the 22-kDa fragment to both small and large emulsions, indicating that the C-terminal domain has a dominant effect on the lipid binding affinity of full-length apoE4. It should be noted that although a previous study of the distribution of the 22-kDa fragments of apoE among plasma lipoproteins demonstrated that the 22-kDa fragment does not bind to any triglyceride-rich lipoprotein particles (33), our results clearly showed that the apoE4 22-kDa fragment can bind to some extent to the surface of emulsion particles.
The four-helix bundle of the 22-kDa fragment is thought to open on lipid binding (9 -12). However, a recent study using a disulfide bond engineering approach demonstrated that the initial interaction of the apoE4 22-kDa fragment with DMPC vesicles does not require the complete opening of the four-helix bundle (12). To confirm the bundle opening of the 22-kDa fragment on the emulsion surface, we tested the emulsion binding of the triple interhelical disulfide-linked apoE4 22-kDa mutant (12) in which the opening of the four-helix bundle is completely restricted. This mutant in fact did not bind to either of the emulsion particles (data not shown), indicating that the 22-kDa fragment binds to the emulsion surface with the fourhelix bundle in an open conformation, as occurs in the formation of apoE-phospholipid discoidal particles (10 -14).
The binding capacity of full-length apoE4 (around 0.8 amino acid/PC molecule) was comparable with the previous data of human apoE3 binding to emulsion particles (34,35). Assuming that all of the amphipathic ␣-helices in the apoE molecule interact similarly with lipid, the B max values in terms of amino acids/PC molecule of all proteins should be identical. However, the B max of full-length apoE4 was much larger than the B max values of the 22-and 10-kDa fragments, suggesting that the protein-lipid interaction of full-length apoE4 on the emulsion surface is different from that of each fragment. These results, together with the finding that the emulsion binding affinity of the C-terminal fragment is much higher than the N-terminal fragment, suggest that, at maximal binding, the N-terminal four-helix bundle in full-length apoE4 does not have an open conformation on the emulsion surface. This concept has been proposed by Narayanaswami and Ryan (24) for the situation in which other apolipoproteins, such as apoCs, induce alteration in the conformation of the N-terminal domain of apoE on spherical lipoprotein particles.
ITC Measurements-To further compare the interactions of both domains with lipid, we used ITC to measure the heats of apoE binding to emulsions. The heat change on apolipoprotein binding to lipid contains contributions from the protein-lipid interaction and from any resulting changes in the ␣-helix content of the protein (36,37). Thus, the heat used or released during the binding process reflects the overall process of protein-lipid interaction. We determined the enthalpies of emulsion binding of full-length apoE4 and its 22-and 10-kDa fragments by injecting small amounts of protein solution into an emulsion suspension of defined lipid concentration (38). Fig. 2 shows the results of injecting 10-l aliquots of fulllength apoE4 solution into small and large emulsions, together with a control injection in which apoE4 was injected into buffer only. Because of the large lipid-to-protein ratio, the injected apoE4 is likely to be almost completely bound to the emulsion surface, and in fact the heats of consecutive injections were virtually identical. Subtracting the heats of dilution yielded the binding enthalpies. Using the binding constants given in Table   FIG  a Percentages of emulsion surface area covered with protein assuming that area/amino acid residue in an ␣-helix is 0.15 nm 2 . b Number of protein molecules/particle was calculated from particle diameter assuming that the cross-sectional area of a PC molecule is 0.7 nm 2 . c p Ͻ 0.05 compared to small emulsions (35 nm).
I, a comprehensive set of thermodynamic parameters in apoE4 binding to emulsions was obtained. Table II summarizes the enthalpies, free energies, and entropies of binding of full-length apoE4 and its fragments to small and large emulsions. Although the binding to both emulsions was found to be an exothermic process, the binding to large particles was distinctly more exothermic than binding to small particles for all proteins. As a result, the binding to large particles is enthalpically driven, whereas that to small particles is entropically driven. Interestingly, such an effect of emulsion particle size on the thermodynamic properties of apoE-lipid interaction is quite opposite to those found in apolipoprotein A-I model peptidevesicle (39) and antibacterial peptide-vesicle (40) interactions.
In contrast to the ITC measurement conditions in which there are few apoE molecules packed on the emulsion surface, the B max values in the 14 C-labeled apoE binding experiments (Table I) reflect the saturated condition of apoE binding to emulsions. To obtain the thermotropic information for the apoE-lipid interaction under saturating binding conditions, we performed the reverse experiment by injecting the emulsions into the apoE solution. Under this condition, the emulsion surface is always saturated with apoE molecules. Fig. 3 shows the injections of large emulsion particles into full-length apoE4 or buffer solution. Because the binding enthalpies directly obtained from this ITC measurement were expressed in terms of PC molecules, we calculated the binding enthalpies per protein using the B max values listed in Table I. For example, we obtained the binding enthalpy of full-length apoE4 of Ϫ39 kcal/ mol protein (Fig. 4B) using the experimental value of Ϫ99 cal/mol PC (Fig. 3) and the B max value of 2.6 mmol protein/mol PC (Table I). Fig. 4 summarizes the binding enthalpies of full-length apoE4 and its 22-and 10-kDa fragments to large emulsion particles under the two different conditions. If the protein-lipid interaction is independent of the surface concentration of protein, the binding enthalpies in the two limiting conditions should be identical. In fact, the binding enthalpies of the 22-and 10-kDa fragments were similar in the two conditions, although the binding enthalpies in a saturated condition (Fig. 4B) tended to be slightly smaller than those in a diluted condition (Fig. 4A). In contrast, the binding enthalpy of fulllength apoE4 in a saturated condition was much less exothermic than that in a diluted condition, suggesting that full-length apoE4 binds to emulsions differently in the two conditions. In addition, comparison of binding enthalpies among full-length apoE4 and its fragments in a diluted condition demonstrated that the binding enthalpy of full-length apoE4 is similar to the sum of the enthalpies of the 22-and 10-kDa fragments, indicating that both the 22-and the 10-kDa domains in full-length apoE4 interact with lipid on the emulsion surface. It should be noted that the enthalpy of binding of full-length apoE4 is about 10% less than the sum of the enthalpies for the two domains (Table II); this small difference may be due to a slight alteration in the lipid interaction of one domain induced by the other domain.

DISCUSSION
For high affinity binding to the LDL receptor, apoE must be associated with lipid. It has been hypothesized that the fourhelix bundle of the N-terminal domain undergoes a lipid-induced reorganization of the helices that exposes the hydrophobic faces to interact with lipid. Although this reorganization of apoE bound to phospholipid discs appears to involve opening of the four-helix bundle structure, it is not clear whether apoE undergoes such a conformational change on spherical particles.  FIG. 4. Binding enthalpies of full-length apoE4 and its 22-and 10-kDa fragments to large emulsions obtained under two limiting conditions. A, protein solutions were injected into excess emulsion at a PC-to-protein molar ratio of Ͼ10,000. B, emulsion was injected into excess protein at a PC-to-protein molar ratio of Ͻ40. a All of the enthalpies of binding to small emulsions (35 nm) are not significantly different, but those to large emulsions (120 nm) are significantly different (p Ͻ 0.05). b Free energy was calculated according to ⌬G ϭ ϪRT ln 55.5(1/K d ) using the binding constants given in Table I. c The entropy of binding was calculated from ⌬G ϭ ⌬H Ϫ T⌬S.
from the particle surface by the C-terminal domain that has a strong lipid affinity.
Effects of Particle Size on ApoE-Lipid Interactions-It has been shown that apoE4 has a greater preference for binding to very low density lipoprotein (VLDL) than high density lipoprotein and that interaction between N-and C-terminal domains is responsible for this VLDL preference (41,42). Although the molecular mechanism of the VLDL preference of apoE4 is not known, the proposed explanation is that the domain interaction may stabilize an extended helical structure in the Cterminal domain that is better accommodated on a less curved VLDL surface (42). Direct binding experiments to small and large emulsion particles showed no significant differences in the binding parameters for full-length apoE4 and its two domains, except that full-length apoE4 has a slightly higher affinity for small particles than large particles (Table I). Therefore, apoE4 seems to display no binding preference within this size range of emulsion particles, which approximates the VLDL-to-chylomicron size spectrum.
In marked contrast, the thermodynamic binding parameters of apoE were found to be significantly different between small and large emulsion particles (Table II). Although the binding free energies of full-length apoE4 and its fragments are similar for small and large emulsions, large differences exist in the binding enthalpies and entropies. A previous ITC study of apolipoprotein A-I model peptide-phospholipid vesicle interactions demonstrated that the binding of the model peptides to small vesicles was enthalpically driven with a small entropy change, whereas that to large vesicles was entropically driven (39). Thermodynamic binding parameters of apoE to phospholipid vesicles also appear to have the same particle size dependence. 2 Such an enthalpy-entropy compensation mechanism can be explained by differences in the lipid packing in which insertion of ␣-helices into tightly packed large vesicles leads to a greater increase in the lipid fluidity than occurs with the more disordered small vesicles (40). However, in our study, the apoE binding to large particles was found to be enthalpically driven, whereas that to small particles was entropically driven. It is not clear why emulsion particle size exerts opposite effects on thermodynamic binding parameters of apoE. One possible explanation concerns the different surface structure in emulsions and phospholipid vesicles. A recent NMR study comparing the surface structures of large emulsion particles and vesicles in-dicated that PC polar headgroups are more separated and exposed to water molecules in emulsions than in vesicles and that this plays a determinant role in apolipoprotein binding to the lipid surface (43). In addition, comparison of surface lipid fluidity using fluorescent probes revealed that small emulsions have lower surface fluidity than large emulsions (44). Therefore, the thermodynamic binding parameters in Table II suggest that the binding of apoE to a relatively well ordered small emulsion surface causes a greater disordering of surface structure compared with the situation with the more fluid large emulsion surface.
ApoE Conformation on Emulsion Particles-Full-length apoE4 and the 10-kDa fragment bind emulsion particles with much higher affinity than does the 22-kDa fragment. These observations are consistent with the previous study demonstrating that these proteins have different abilities to form discoidal complexes with DMPC: intact apoE3 ϳ 10-kDa fragment Ͼ 22-kDa fragment (45). These results indicate that the C-terminal domain dominantly regulates the lipid binding of full-length apoE as proposed for the lipoprotein association of this protein (46). Comparison of the binding capacities among full-length apoE4 and its fragments demonstrated that the binding capacity of full-length apoE4 is greater than those of the 22-and 10-kDa fragments. By assuming a surface area of 0.15 nm 2 /residue for ␣-helical proteins (47) and 0.7 nm 2 /molecule for PC (48) in the emulsion surface monolayer, the fraction of emulsion surface covered by protein is 17-18% for full-length apoE4 and only 6 -7% for the 22-and 10-kDa fragments in both small and large emulsion particles. This difference in protein surface area on the emulsion surface between full-length apoE4 and its fragments, together with the fact that the lipid binding affinity of the C-terminal fragment is much higher than the N-terminal fragment, suggests that the N-terminal domain in full-length apoE4 does not interact with lipid because it is anchored to the emulsion surface by the C-terminal domain.
To test this hypothesis, we measured binding enthalpies of apoE to large emulsion particles in the two limiting conditions that enable us to distinguish the two possible conformations on the particle surface (Fig. 5). The binding enthalpy of full-length apoE4 was approximately equal to a sum of those of the 22-and 10-kDa fragments at a low surface concentration of protein (Fig. 4A). Together with the fact that the triple interhelical disulfide-linked apoE4 22-kDa mutant did not bind to emulsion particles, these results indicate that full-length apoE4 can bind to the spherical surface with the four-helix bundle in an open 2 P. Acharya and J. Snow, unpublished data. conformation. However, when the protein binding on the emulsion surface became saturated, the binding enthalpy of fulllength apoE4 was less exothermic and rather similar to that of each fragment alone (Fig. 4B). This suggests that the N-terminal domain in full-length apoE4 swings into the aqueous phase away from lipid contact probably because of the binding competition with the C-terminal domain, which has a strong lipid affinity. This four-helix bundle closed conformation would allow more apoE molecules to bind to the emulsion surface, resulting in the greater binding capacity of full-length apoE4 than each fragment, as observed in Table I.
Recently, Narayanaswami and Ryan (24) proposed that apoE undergoes a conformational change upon binding to lipid. The C-terminal domain anchors the protein to the lipoprotein surface, modulating the receptor binding properties through alteration of the conformation of the N-terminal domain between an open lipid-bound receptor-active state and the globular fourhelix bundle, receptor-inactive state. They hypothesized that other apolipoproteins, such as apoCs, mediate this conformational change by displacement of the N-terminal domain of apoE from lipoprotein surface. Our data suggest that such a conformational change can occur with apoE alone depending upon its surface concentration, because the C-terminal domain has greatly higher lipid binding affinity compared with the N-terminal domain. In fact, in a fibroblast LDL receptor competitive binding assay, the concentrations of apoE3 required for 50% replacement of 125 I-labeled LDL were 0.01-0.02 and 1.0 g/ml for apoE-DMPC discs and apoE-bound VLDL-like emulsion particles, respectively (49). This indicates that the LDL receptor binding activity of apoE bound to emulsion particles is much lower than that of apoE-DMPC discs. In addition, Kypreos et al. (50) demonstrated that overexpression of fulllength apoE induces hyperlipidemia in apoE-deficient mice, whereas the N-terminal 1-185 residues of apoE are sufficient for the clearance of apoE-containing lipoprotein remnants by the liver. This suggests that full-length apoE on lipoprotein particles might be in a conformation that hinders receptor recognition.
In summary, the current study presents the first complete description of the thermodynamic binding parameters of fulllength apoE4 and the 22-and 10-kDa fragments to lipid particles, allowing us to elucidate the contributions of each domain of apoE to its interaction with lipid. Our results suggest that the apoE molecule has two distinct lipid-bound states involving the N-terminal four-helix bundle open and closed conformations on the particle surface. This finding may explain in part why lipoprotein-associated apoE displays variable receptor binding activity during lipoprotein metabolism.