Biophysical Properties of Apolipoprotein E4 Variants: Implications in Molecular Mechanisms of Correction of Hypertriglyceridemia

Abstract

In humans and animal models, high plasma concentrations of apolipoprotein (apo) E are associated with hypertriglyceridemia. It has been shown that overexpression of human wild-type (WT) apoE4 in apoE-deficient mice induces hypertriglyceridemia. In contrast, overexpression of an apoE4 variant, apoE4-mut1 (apoE4(L261A, W264A, F265A, L268A, V269A)), does not induce hypertriglyceridemia and corrects hypercholesterolemia. Furthermore, overexpression of another variant, apoE4-mut2 (apoE4(W276A, L279A, V280A, V283A)), induces mild hypertriglyceridemia and does not correct hypercholesterolemia. To better understand how these mutations improve the function of apoE4, we investigated the conformation and stability of apoE4-mut1 and apoE4-mut2 and their binding to dimyristoyl phosphatidylcholine (DMPC) vesicles and to triglyceride (TG)-rich emulsion particles. We found that the mutations introduced in apoE4-mut1 lead to a more stable and compactly folded conformation of apoE4. These structural changes are associated with a slower rate of solubilization of DMPC vesicles by apoE4-mut1 and reduced binding of the protein to emulsion particles as compared to WT apoE4. Under conditions of apoE4 overexpression, the reduced binding of apoE4-mut1 to TG-rich lipoprotein particles may facilitate the lipolysis of these particles and may alter the conformation of the lipoprotein-bound apoE in a way that favors the efficient clearance of the lipoprotein remnants. Mutations introduced in apoE4-mut2 result in smaller structural alterations compared to those observed in apoE4-mut1. The slightly altered structural properties of apoE4-mut2 are associated with slightly reduced binding of this protein to TG-rich lipoprotein particles and milder hypertriglyceridemia as compared to WT apoE4.

Human apolipoprotein E (apoE) is a key component of the lipoprotein transport system and is required for the maintenance of lipid homeostasis in the circulation and the brain (1–3). In humans, there are three natural apoE isoforms that differ from each other by amino acid substitutions at positions 112 and 158. ApoE3 (Cys-112, Arg-158) is the most common isoform; apoE2 (Cys-112, Cys-158) is associated with type III hyperlipoproteinemia, while apoE4 (Arg-112, Arg-158) is associated with high plasma cholesterol level and an increased risk for both coronary heart disease and Alzheimer’s diseases (4–7). ApoE is one of the major protein constituents of triglyceride (TG)-rich chylomicrons and very low density lipoproteins (VLDL). In the bloodstream, these TGrich lipoproteins are converted into remnants through hydrolysis of their core TG by lipoprotein lipase (LPL). At physiological concentrations, apoE mediates the hepatic uptake of the lipoprotein remnants via the low density lipoprotein (LDL) receptor and thereby, regulates plasma lipid levels and contributes to atheroprotection (1–3, 7). However, elevated plasma apoE levels in humans and in animal models have been positively correlated with hypertriglyceridemia (8–10). It has been shown that overexpression of wild-type (WT) apoE in apoE-deficient (apoE^−/−) mice triggers hypertriglyceridemia due to increased VLDL TG secretion and impaired VLDL lipolysis (9–10).

Induction of hypertriglyceridemia in mice overexpressing apoE was prevented by Cterminal truncations of apoE, suggesting that the 260–299 region of apoE is essential for induction of hypertriglyceridemia (10, 11). In an attempt to further identify specific residues and regions of the C-terminal segment of apoE involved in the induction of hypertriglyceridemia, two apoE4 variants, apoE4-mut1 (apoE4(L261A, W264A, F265A, L268A, V269A)) and apoE4-mut2 (apoE4(W276A, L279A, V280A, V283A)) (Figure 1), were generated and studied in vivo in apoE^−/− mice using adenovirus-mediated gene transfer (17). Each of these variants has alanine substitutions for hydrophobic residues in one of two hydrophobic stretches of amino acids, 261–269 or 276–283, that are thought to be involved in lipid and lipoprotein binding of apoE (16, 18–20). These studies showed that overexpression of apoE4-mut1 in apoE^−/− mice corrected hypercholesterolemia and did not induce hypertriglyceridemia. In contrast, comparable levels of expression of apoE4-mut2 did not correct hypercholesterolemia and induced mild hypertriglyceridemia (Figure 1, inset). The molecular basis for the improved function of these apoE4 variants is not fully understood. The results of the animal studies suggested that mutations introduced in apoE4-mut1, and to a lesser extent in apoE4-mut2, may decrease the ability of apoE4 to be associated with TG-rich lipoproteins in plasma (17). The decreased affinity of the apoE4 variants to TG-rich particles could explain, in part, the improved function of these variants. However, no direct experimental evidences of the altered ability of apoE4-mut1 and apoE4-mut2 to bind to TG-rich particles have been obtained.

Figure 1.

Mutations introduced in apoE4-mut1 and apoE4-mut2. In the schematic representation of apoE4 (the top), cylinders represent helical structures in the N-terminal domain (amino acids 1–191) based on the X-ray crystallography structure of the isolated domain (12). The rectangle and the wavy line represent the C-terminal domain (amino acids 222–299) and linker, respectively. Their structures are unknown; the C-terminal domain is thought to be mostly helical and the linker is disordered, on the basis of structural prediction and CD spectroscopy (13–16). The shaded areas M1 and M2 show sites of the mutations introduced in apoE4-mut1 and apoE4-mut2, respectively. The shaded blocks below the schematic representation of apoE4 show the amino acid sequences of the corresponding segments (amino acids 261–269 and 276–283) for WT apoE4, apoE4-mut1, and apoE4-mut2; the mutated residues are shown in bold. The dotted line indicates a putative salt bridge between Arg61 and Glu255 that stabilizes a closer contact between the N- and C-terminal domains in the tertiary structure. Inset: Effects of high level expression of the apoE4 variants on plasma lipid levels in apoE^−/−mice. Plasma TG (black columns) and cholesterol (gray columns) concentrations in apoE^−/− mice are shown before (apoE^−/−) and four days after infection with recombinant adenoviruses expressing WT apoE4, apoE4-mut1, or apoE4-mut2. Adopted from ref. 17.

Here, we studid in-vitro the ability of apoE4-mut1and apoE4-mut2 to bind to synthetic VLDL-like particles and to solubilize dimyristoylphosphatidylcholine (DMPC) vesicles. VLDL-like emulsion particles prepared from triolein and phosphatidylcholine (PC) were utilized as a model for TG-rich lipoproteins (21, 22). To better understand how the mutations introduced in apoE4-mut1and apoE4-mut2 alter the function of apoE4, we investigated the conformation and stability of the apoE4 variants. Previous studies have shown that the structural and biophysical properties of apoE dictate the function of the protein (15, 16, 19). For example, apoE4 has increased domain interaction and more loose tertiary structure, as compared to apoE2 and apoE3, which is thought to lead to the preferential binding of apoE4 to large VLDL and the preferential binding of apoE3 and apoE2 to smaller high density lipoprotein in plasma (23, 24). The conformational stability of the common isoforms of apoE correlates inversely with the rate of their association with phospholipid vesicles and the affinity of the isoforms to TG-rich emulsion particles (25–27). We found that the E4-mutlvariant that prevents the induction of hypertriglyceridemia has a more compactly folded and stable structure and a reduced ability to solubilize DMPC vesicles and to bind to TG-rich particles as compared to WT apoE4. In contrast, the E4-mut2 variant that induces milder hypertriglyceridemia, compared to WT apoE4, has milder changes in the conformation and stability and the ability to solubilize DMPC vesicles and to bind to TG-rich particles.

EXPERIMENTAL PROCEDURES

Materials

Materials not mentioned in the Experimental procedures have been obtained from sources described previously (17, 28, 29).

Construction of Recombinant Adenoviruses, Expression, Purification and Preparation of Proteins

The generation of recombinant adenoviruses carrying the genomic sequence for the WT, apoE4-mut1 (apoE4(L261A, W264A, F265A, L268A, V269A)), and apoE4-mut2 (apoE4(W276A, L279A, V280A, V283A)) has been described previously (17). For apoE4 production, human HTB13 cells in roller bottles were infected with adenoviruses expressing the WT or the mutant apoE4 forms and the medium was harvested as described previously (30). Purification of apoE from the culture medium was performed using dextran sulfate-Sepharose column as described before (30). The purified proteins were analyzed by 12% SDS-PAGE and by Western blotting. Fractions containing apoE4 of >95% purity were dialyzed against 5 mM ammonium bicarbonate, lyophilized, and stored at −80°C. Before use in the experiments, lyophilized proteins were dissolved in a solution containing 6 M guanidine hydrochloride (GdnHCl) and 1% (v/v) β-mercaptoethanol and dialyzed extensively against an appropriate buffer.

Circular Dichroism (CD) Spectroscopy

CD-measurements were performed on an AVIV 62DS or AVIV 215 spectropolarimeter (AVIV Associates, Inc.) equipped with a thermoelectric temperature control and calibrated with d-10-camforsulfonic acid in 2, 5, and 10 mm path length quartz cuvettes as previously described (29, 31). Proteins were in 10 mM sodium phosphate buffer (pH 7.4), freshly dialyzed from 6 M GndHCl and 1% β-mercaptoethanol solution. Far-UV CD spectra were recorded from 260 to 185 nm at 25°C; for each sample, three to five scans were performed and averaged. The thermally induced unfolding of the proteins was monitored by ellipticity at 222 nm over the temperature range 5 to 90°C at two heating rates, 0.25°C/min and 0.65°C/min. Because of the high tendency of full size apoE to aggregate (32), the experiments were performed at relatively low protein concentrations, 10–30 µg/mL, including the lowest possible concentration for each particular experiment. For each apoE4 form, spectra and melting curves were recorded at several protein concentrations within the indicated range and then, after the baseline subtraction, were normalized to molar residue ellipticity [Θ]. The α-helix content was estimated from the molar residue ellipticity at 222 nm, [Θ₂₂₂] (33).

For analysis of thermal unfolding, the fraction of unfolded protein at a given temperature was determined from the equation:

$$\text{Unfolded Fraction}=({[{\mathrm{\Theta }}_{222}]}_{\mathrm{F}}-[{\mathrm{\Theta }}_{222}])/({[{\mathrm{\Theta }}_{222}]}_{\mathrm{F}}-{[{\mathrm{\Theta }}_{222}]}_{\mathrm{U}}),$$

where [Θ₂₂₂] is a molar residue ellipticity at a given temperature and [Θ₂₂₂]_F and [Θ₂₂₂]_U are molar residue ellipticity for the protein in folded (native) and unfolded (thermally denaturated) states, respectively. The values for the unfolded fraction were plotted against temperature and the data were smoothed over a five degree window. The data [Θ₂₂₂ ] were algebraically differentiated with respect to temperature and the derivative function d[(Θ₂₂₂)]/dT was plotted against temperature. For each thermal unfolding curve, the apparent melting temperature was determined as the midpoint of thermal unfolding at which unfolding of apoE was half complete, T_1/2, and as the maximum of the first derivative function, T^d_m (34). The width of the first derivative function d[Θ₂₂₂]/dT at the half maximal height, Δ(d[Θ₂₂₂]/dT), was determined for each protein to compare cooperativity of unfolding of the proteins (34).

ANS (8-Anilino-2-naphthalene-sulfonate) Fluorescence Measurements

ANS fluorescence emission spectra were collected on a FluoroMax-2 fluorescence spectrometer (Instruments S.A., Inc.) as previously described (28, 29). Fluorescence of ANS was recorded in Tris buffer (10 mM Tris-HCl, 150mM NaCl, 0.02% NaN3, 1 mM EDTA, pH.7.4) at ANS concentration 125 µM in the presence of 25 µg/ml (0.74 µM) apoE4 or in the absence of any protein. To control possible effects of apoE4 aggregation, the experiments were repeated at ANS concentration 250 µM in the presence of 50 µg/ml (1.48 µM) apoE4. This increase in apoE4 concentration had no effect on the results of the experiment that is consistent with the absence of the protein self-association at the indicated concentrations. In each experiment, ANS fluorescence was also recorded in the presence of the “reference” proteins, carbonic anhydrase or bovine serum albumin, at the corresponding concentrations. Wavelength of maximum fluorescence (WMF) and intensity of fluorescence emission at WMF were determined from each spectrum.

DMPC Turbidity Clearance Studies

The solubilization of DMPC multilammelar vesicles by apoE was studied according to modified procedures (25, 35, 36). To prepare the vesicles, 1 mg of DMPC was dissolved in 1 ml of 2:1 (v/v) chloroform/ methanol solution, dried under nitrogen to form a thin lipid film, and left overnight under vacuum to get rid of any traces of the organic solvent. The lipids were then resuspended in 1 ml of Tris buffer containing 0.1M GndHCl and vortexed vigorously for 1–2 min to become milky turbid suspension. The 0.1 M GndHCl was added to the buffer to ensure the absence of apoE self-association during the experiment; this low denaturant concentration has been shown to have no effect on the protein conformation and the turbidity clearance data (25). For the clearance experiments, ~ 0.95 ml of the stock suspension diluted to the lipid concentration 60 µg/ml was placed in 1 cm path length quartz cuvett and preincubated for 5 min at 24°C within a holder of a spectrophotometer (Varian Cary-300) with thermoelectric temperature control. Monitoring of absorbance at 325 nm indicated no significant changes in turbidity of the lipid suspension during the preincubation. A small volume of apoE in Tris buffer (freshly dialyzed from 6 M GdnHCl and 1% (v/v) β-mercaptoethanol) was added to the lipid suspension in the cuvette to give a protein concentration of 7.5 µg/ml (0.22 µM) and gently mixed by repeated pipetting within 10 s before resuming the recording of absorbance at 325 nm. In the control experiments, buffer containing no protein was added to the cuvette with the lipid suspension. After recording turbidity for 2 to 4 hrs, the apoE/DMPC mixtures were left overnight in the cuvettes within the spectrophotometer holders maintaining the controlled temperature of 24°C to be observed under electron microscope the following morning, after 20 h of incubation.

Preparation of TG-rich Emulsion Particles

TG-rich emulsion particles were prepared by sonication followed by ultracentrifugation according to modified procedures (22, 37) originally developed by Miller and Small (21). Chloroform solutions of 100 mg triolein and 25 mg egg yolk PC were mixed together, dried under nitrogen and vacuum desiccated overnight at 4 °C. The dried lipids were suspended in 20 ml of Tris buffer and sonicated at 80–90 watts continuous power using a Branson sonifier, model 350, with a flat tip. Sonication was performed for 20 min under a stream of nitrogen; the temperature of the sample was maintained at 40–50°C by immersion in an ice-water bath. The sonicated mixture was transferred into centrifuge tubes, overlayered with warm Tris buffer and centrifuged in an SW41 rotor (Beckman, CA) at 20°C for 15 min at 18,000 rpm. A top creamy layer and an under-laying cloudy layer containing large chylomicronlike particles were removed by tube slicing. The remaining sample was transferred into a new tube, overlayered with Tris buffer and centrifuged in the same rotor at 20°C for 1 h at 25,000 rpm. The top layer was recovered as TG-rich emulsion and analyzed by phospholipids and triglyceride assays to determine composition and by electron microscopy to examine morphology and determine particle size. The emulsions were used for the binding assays within 2 days after isolation; during this period no changes were detected on the electron microscopy photographs.

Binding of ApoE4 to Emulsion Particles

Binding of apoE4 to emulsion was studied according to modified procedures (22, 37). To minimize the required amounts of nonlabeled apoE4 forms, a fixed amount of the proteins was incubated with various amounts of emulsion. Specifically, 80 µg of WT apoE4, apoE4-mut1, or apoE4-mut2 (freshly dialyzed from 6 M GdnCHl and 1% β-mercaptoethanol) were incubated with increasing amounts of emulsion in 1.4 ml of Tris buffer to give a PC to protein molar ratio ranging from 50 to 1020. Incubation was performed for 1 h at 27°C in a water bath with gentle shaking. Then each mixture was overlayered with NaCl solution of the density 1.006 g/cm³ and spun in an SW60 rotor for 80 min at 40,000 rpm at 25 °C. The emulsions floating to the top of the tubes were recovered by tube slicing; they contained emulsionbound apoE4. Lipid-free apoE4 was in the bottom fractions. The concentration of bound apoE was calculated by subtracting the background free apoE concentration in the top fractions as described (37). Emulsions from four different preparations were used for the binding assays. With each isolated emulsion, four to six emulsion/apoE4 mixtures with various PC to protein ratios were studied for each protein.

Electron Microscopy

Analysis of the morphology and estimation of the size of isolated TG-rich emulsion particles and apoE/DMPC complexes were performed using negative staining electron microscopy as described (36). Emulsions were appropriately diluted; apoE/DMPC complexes were processed without dilution. Images of random fields of particles were recorded on Kodak SO-163 film (Eastman Kodak Co, Rochester, NY) using a CM12 electron microscope (Philips Electron Optics, Eindhoven, The Netherlands).

Analytical Procedures

Protein concentrations were determined by the Lowry procedure (38) and by absorbance at 280 nm with the extinction coefficient 1.34 ml/ (mg cm). TG concentration was determined by Infinity^™ TG Kit (Thermo Electron, Australia). PC concentration was determined by Bartlett phosphorus assay (39).

RESULTS

CD Analysis of the α-Helical Content and Thermal Unfolding

Far-UV CD data were recorded at protein concentrations 10–30 µg/mL in 10 mM sodium phosphate buffer, pH 7.4. Because of the high tendency of apoE to aggregate, spectra and thermal unfolding for each apoE4 form were recorded at several protein concentrations within the indicated range. No changes in the normalized spectra or normalized melting curves were observed with the variations in the protein concentration, which is consistent with the absence of protein self-association within the range of the concentrations used in our experiments. This agrees with previous data (40, 41) showing that apoE4 is largely monomeric at the low concentrations used in the CD studies. The α-helical content of the apoE4 forms (Table 1) was estimated from the normalized far-UV spectra. The α-helical content of WT apoE4 was ~62% that corresponds to ~189 residues in the helical conformation. The value for the α-helical content of WT apoE4 is within the range of the previously published estimations, 46 to 64% (16, 42, 43). There was no significant difference between the α-helical content of apoE4-mut2 and WT apoE4, although in all the experiments, apoE4-mut2 consistently exhibited slightly lower values for α-helicity. In contrast, apoE4-mut1 had ~11% lower α-helical content compared to WT apoE4 that corresponds to ~34 fewer residues in the helical conformation of this variant.

Table 1.

α-helical Content and Parameters of Thermal Denaturation for WT and the Variant Forms of apoE4 Determined by Far-UV CD^a

apoE4	α-helixb (%)	Number of residues in α-helix c	T1/2d °C	Tmd, °C	Δ(d[Θ222]/dT )d, °C
WT apoE4	62 ± 3	~189	46 ± 0.5	47 ± 0.5	33 ± 1.0
apoE4-mut2	60 ± 2	~183 (−6)	49 ± 1.3 e	52 ± 1.3 f	27 ± 1.5 f
apoE4-mut1	51 ± 2f,h	~155 (−34)	51 ± 0.5 g	53 ± 1 g	25 ± 0.6 g

^aValues are means ± SD from three to five experiments.

^bEstimated from the value [Θ₂₂₂] at 25 °C according to ref. 33 ; systematic error is ±3%, statistical error is within ±2%.

^cValues in parenthesis show changes in the number of residues compared to WT apoE4.

^dParameters determined from the thermal unfolding curves; systematic error is within ±1°C. T_1/2 is the midpoint of thermal unfolding. T_m is a maximum of the first derivative function d[Θ₂₂₂]/dT; Δ(d[Θ₂₂₂]/dT) is the width of the first derivative function d[Θ₂₂₂]/dT at the half maximal height.

^eSignificance of differences from the value for WT apoE4: p<0.05.

^fSignificance of differences from the value for WT apoE4: p<0.01.

^gSignificance of differences from the value for WT apoE4: p<0.005.

^hSignificance of differences from the value for apoE4-mut2: p<0.01.

To investigate stability of the apoE4 forms, thermally induced unfolding of the proteins was monitored using CD. The thermal unfolding curves plotted as a fraction of unfolded protein against temperature are shown in Figure 2A. The values of the midpoint of thermal unfolding at which the protein is half unfolded, T_1/2, are shown in Table 1. The shifts of the unfolding curves for the mutants to higher temperatures suggest a trend in stability (WT apoE4 < apoE4-mut2 < apoE4-mut1), consistent with the changes in T_1/2 in the same order. However, the differences in T_1/2 between apoE4-mut2 and the other proteins are not statistically significant. Figure 2B shows the first derivative functions d[Θ₂₂₂]/dT of the thermal unfolding curves that emphasize differences in slope for the unfolding curves. Parts of the derivative function at low and high temperatures that correspond to less cooperative structural changes in each protein are not shown because of the high proportion of noise in these parts of the derivative function. The temperature of the maximum of the derivatives curves, T^d_m, are in agreement with the midpoint of thermal unfolding T_1/2 (Table 1), which is consistent with reversibility of the thermal unfolding of the proteins. Similar to the values for T_1/2, the values for the apparent melting temperature T^d_m indicate an increase in the stability of the apoE4 variants compared to WT apoE4.

Figure 2.

Thermal unfolding of the apoE4 variants. (A) Thermal unfolding curves for the apoE4 forms are derived from a fractional change in CD ellipticity at wavelength of 222 nm, (Θ₂₂₂), at increasing temperature. Protein concentration was 15 µg/mL. (B) First derivative functions d(Θ₂₂₂)/dT of the thermal unfolding curves. The derivative curves are shown in the temperature region ~27 to ~70°C that corresponds to more cooperative structural changes. The peak position that determines the apparent melting temperature, T^d _m, and the width of the first derivative curve at the half of it maximal height, Δ(dΘ₂₂₂/dT), are indicated for WT apoE4. WT apoE4 (black stars), apoE4-mut1 (open black circles), and apoE4-mut2 (open gray triangles).

The width of the first derivative function d[Θ₂₂₂]/dT at the half maximal height, Δ(d[Θ₂₂₂]/dT), which is inversely proportional to the apparent van’t Hoff enthalpy of the transition (34), decreases in the order WT apoE4 > apoE4-mut2 ≥ apoE4-mut1 (Table 1). This is consistent with the higher cooperativity of unfolding of the variant proteins, especially apoE4-mut1, compared to WT apoE4, in the temperature region ~27 to ~70°C. The heights of the peaks for the derivative functions d[Θ₂₂₂]/dT, which is greatest for apoE4-mut1, slightly lower for apoE4-mut2, and significantly lower for apoE4 (Figure 2B), agrees with the trend in cooperativity of unfolding: WT apoE4 < apoE4-mut2 ≤ apoE4-mut1.

ANS Binding

To explore changes in the tertiary structure of the variant apoE4 forms, we monitored intrinsic fluorescence of the amphipathic fluorescent dye ANS in the presence of WT apoE4, apoE4-mut1, and apoE4-mut2 (Figure 3). The intrinsic fluorescence of ANS is known to be increased and red-shifted when the dye is bound to protein hydrophobic surfaces or cavities, while the water-phase dye does not contribute to the emission. Table 2 shows parameters of ANS fluorescence in the presence of the apoE4 forms and the “reference” proteins, carbonic anhydrase as a typical globular watersoluble protein and bovine serum albumin as a protein exposing hydrophobic binding pockets (data for bovine serum albumin not shown in Figure 3). Of all three apoE4 forms, WT apoE4 results in the largest intensity and red shift of ANS fluorescence suggesting a loosely folded conformation with the greatest exposure of hydrophobic surfaces. For apoE4-mut2, compared to WT apoE4, there is a ~10% decrease in the intensity and a ~6 nm red shift of ANS fluorescence, that suggests a reduction of exposed hydrophobic surfaces and a more compact folding of apoE4-mut2. In the presence of apoE4-mut1, the ANS fluorescence spectrum changes most dramatically compared to the fluorescence in the presence of WT apoE4: 50% decrease in the intensity and ~9 nm red shift that indicates a significant reduction in exposed hydrophobic sites. The data suggest that of the three apoE4 forms, apoE4-mut1 has a most compactly folded structure and WT apoE4 has a most loosely folded structure.

Figure 3.

ANS fluorescence recorded in the presence of the apoE4 variants. Fluorescence of ANS (125 µM) was recorded in the presence of 25 µg/ml WT apoE4 (black stars), apoE4-mut1 (open black circles), apoE4-mut2 (open gray triangles), carbonic anhydrase (bars), or in buffer alone (solid line). Two latter spectra are superimposed.

Table 2.

ANS Fluorescence in the Presence of WT and the Variant Forms of apoE4 and Reference Proteins.

protein	I (relative units)	WMF (nm)
apoE WT	1.0	478
apoE-mut1	0.5	487
apoE-mut2	0.9	484
Bovine serum albumin	2.9	475
Carbonic anhydrase	0.2	516
ANS in buffer alone	0.2	516

Fluorescence of ANS (125 µM) was recorded in the presence of 0.025 mg/ml apoE4 WT, apoE4-mut1, apoE4-mut2, carbonic anhydrase, or bovine serum albumin or in Tris buffer alone. I is ANS fluorescence intensity in relative units compared to the fluorescence in Tris buffer alone.

DMPC Turbidity Clearance

Electron microscopy of apoE4/DMPC mixtures. To analyze the kinetics of solubilization of lipid vesicles by the apoE4 forms, we used a standard apolipoprotein DMPC-binding assay that examines the clearance of DMPC multilamellar vesicles by the apolipoproteins. The clearance of DMPC vesicles was triggered by adding a protein and monitored by turbidity at 325 nm. In order to avoid oligomerization of the full-size apoE forms, the clearance experiments were performed at a very low protein concentration (7.5 µg/ml). This resulted in a lower apolipoprotein:lipid weight ratio (1:8) in our experiments compared to the ratio 1:2.5 or 1:2 in similar experiments described elsewhere (25, 36).

The time-courses for clearance of DMPC vesicles by apoE4 forms (Figure 4) show that the mutations introduced in apoE4-mut2 do not affect significantly the rate of solubilization of the lipid vesicles by E4. In contrast, apoE4-mut1 clears the turbidity significantly slower than WT apoE4 and apoE4-mut2 indicating a lower rate of solubilization of DMPC vesicles by apoE4-mut1. Residual turbidity was significantly higher for apoE4-mut1 than for WT apoE4 suggesting the presence of larger particles in the apoE4-mut1/DMPC mixture. To confirm this, we performed electron microscopy analysis of the mixtures WT apoE4/DMPC and apoE4-mut1/DMPC after 20 h preincubation at 24°C. The micrographs show the presence of discoidal particles in both mixtures (Figure 5). The discoidal particles are apoE4:DMPC complexes formed during the incubation; they are seen “face-up” or stacked on edge in short rouleaux produced during the negative staining process. The average thickness of the discoidal particles measured from the periodicity of discs stacked in rouleaux is (5.7±0.5) nm that is consistent with the thickness of DMPC bilayer discs determined previously (44). The estimated average diameter of the discoidal particles was similar (~20.5 ± 2.5 nm) for particles formed by WT apoE4 and by apoE4-mut1. However, the apoE4-mut1/DMPC mixtures (Figure 5B) contain fewer discoidal complexes and shorter rouleaux than the WT apoE4/DMPC mixtures (Figure 5A) indicating significantly lower production of the discoidal protein-lipid particles by apoE4-mut1. Furthermore, the micrographs of the apoE4-mut1/DMPC mixtures show the presence of large lipid vesicles that are the remaining DMPC vesicles not solubilized by the proteins, consistent with the lower lipidbinding ability of apoE4-mut1.

Figure 4.

The time course of DMPC turbidity clearance by the apoE variants. A small amount of protein or buffer was added to DMPC multilamellar vesicles (60 µg/ml) preincubated at 24°C to get the final protein concentration 7.5 µg/ml. The turbidity clearance was monitored by absorbance at 325 nm at the controlled temperature 24°C. WT apoE4 (black stars), apoE4-mut1 (open black circles), apoE4-mut2 (open gray triangles), and buffer alone (black dots).

Figure 5.

Electron micrographs of negatively stained apoE4/DMPC mixtures. Mixtures of apoE4 (7.5 µg/ml) and DMPC multilamellar vesicles (60 µg/ml) were incubated for 20 h at 24°C before imaging by electron microscopy. The representative fields are shown for WT apoE4/DMPC (A) and apoE4-mut1/DMPC (B). The scale bar corresponds to 25 nm.

Binding of apoE4 Forms to Emulsion Particles

Emulsion particles were made from triolein and egg yolk PC and used as a model for TG-rich lipoproteins. The triolein to PC weight ratio in the emulsions was 3.9 ± 0.3 (mean ± SD of six isolated emulsion). Thus, triolein was the major component of emulsion particles accounting for almost 80% of the total mass. The average size of particles in a typical isolated emulsion calculated from the electron micrographs was 55 ± 25 nm (mean ±SD, n=198). The large standard deviations reflect a broad size distribution that resembles a size distribution of plasma VLDL. Importantly, variations in the particle composition and size between different isolated emulsions used in our studies were insignificant.

To investigate if the mutations made in apoE4-mut1 and apoE4-mut2 affect the ability of apoE4 to bind to TG-rich emulsion particles, WT apoE4 and each of the variants were incubated with emulsions at various PC to protein molar ratios ranging from 50 to 1020. The emulsion-bound protein and unbound protein were then separated by ultracentrifugation of the mixtures. For each apoE4 form, a portion of the emulsionbound protein was plotted versus a PC to protein molar ratio in the incubation mixtures (Figure 6). This presentation of results of the binding assays was chosen over plotting a concentration of the emulsion-bound protein versus a concentration of the unbound protein (24, 37), as our binding assays were performed using the fixed amount of protein and the increasing amounts of emulsion. The plots in Figure 6 indicate that at each PC/protein ratio, the portion of the bound protein is the highest for WT apoE4, slightly lower for apoE4-mut2, and significantly lower for apoE4-mut1. The plots in Figure 6 represent results of binding assays performed with one of four isolated emulsions. Similar results were obtained in binding assays performed with each other isolated emulsion. However, because binding assays performed with emulsions from different preparations might be carried out at slightly different PC/protein ratios, the standard deviations for the portion of bound protein in Figure 6 are shown only for the PC/protein molar ratio of 255 and 600, as three to five binding assays for each protein were performed at the PC/protein ratio of 255 ± 5 and 600 ± 7 (mean ± SD). At these both values of the PC/protein ratio, the average portion of the bound protein decreases in the order: WT apoE4 ≥ apoE4-mut2 > apoE4-mut1 (Table 3).

Figure 6.

Binding of the apoE4 variants to emulsion particles. Emulsion/apoE4 mixtures were incubated for 1 hr and then spun to separate bound and unbound protein. A portion of bound apoE4 was determined for various PC to protein molar ratios in the incubation mixtures. WT apoE4 (black stars), apoE4-mut1 (open black circles), or apoE4-mut2 (open gray triangles).

Table 3.

Binding of WT and the Variant Forms of apoE4 to TG-Rich Emulsion Particles^a.

apoE4	PC to protein molar ratio in the incubation mixtures
	255 ± 5		600 ± 7
	Portion of the bound protein b	Protein on the emulsion particles c (amino acids/mol PC)	Portion of the bound protein b	Protein on the emulsion particles c (amino acids/mol PC)
WT apoE4	0.39 ± 0.04	0.71 ± 0.08	0.53 ± 0.02	0.41 ± 0.03
apoE4-mut2	0.32 ± 0.04	0.59 ± 0.10	0.47 ± 0.02	0.37 ± 0.07
apoE4-mut1	0.19 ± 0.02 d, g	0.38 ± 0.007 e, h	0.28 ± 0.03 d, f	0.23 ± 0.02 d, g

^aValues are means ± SD from three to five experiments.

^bPortion of the bound protein of the total protein added to emulsion.

^cParameter derived from composition of apoE4-emulsion complexes recovered by ultracentrifugation following the incubation of apoE4 with emulsion.

^dSignificance of difference from the value for WT apoE4: p<0.005.

^eSignificance of difference from the value for WT apoE4: p<0.01.

^fSignificance of difference from the value for apoE4-mut2: p<0.005.

^gSignificance of difference from the value for apoE4-mut2: p<0.05.

^hSignificance of difference from the value for apoE4-mut2: 0.1<p<0.05.

To analyze the relative content of apoE4 on the resultant apoE4-emulsion complexes formed during the incubation, the bound protein to PC weight ratio was determined in the top fractions recovered after ultracentrifugation following incubation. The values for this ratio in terms of a number of amino acids of bound protein per molecule of PC for the incubations with the initial PC/protein molar ratio of 255 ± 5 and 600 ± 7 are listed in Table 3. Regardless of the initial proportion between apoE4 and emulsion during the incubation, the number of amino acids per PC molecule in the protein-emulsion complexes is lower for apoE4-mut1 by almost 50% compared to WT apoE4 and by almost 40% compared to apoE4-mut2. Figure 7 shows the numbers of bound apoE4 molecules per emulsion particle at various initial PC to protein molar ratios in the incubation mixtures containing WT apoE4 or apoE4-mut1.The number of bound apoE4 molecules per emulsion particle was calculated from the molar ratio of the bound protein to PC in the top fractions recovered after ultracentrifugation, the average emulsion particle diameter, and the assumptions that the emulsion surface area per amino acid residue in a helix is 0.15 nm² (45, 46) and the emulsion surface area covered by one PC molecule is 0.7 nm² (47). The data in Figure 7 show that at each initial PC to protein ratio, less apoE4-mut1 molecules, than WT apoE4 molecules, are bound to one emulsion particle. At each initial PC to protein ratio, there was a trend for the numbers of apoE4 molecules per particle: WT apoE4 ≥ apoE4-mut2 > apoE4-mut1 (data for apoE4-mut2 not shown in Figure 7).

Figure 7.

Bound apoE4 molecules per emulsion particle. Numbers of the bound protein molecules per particle were estimated at various PC to protein molar ratios in the incubation mixtures. See “Results” for details. WT apoE4 (stars), apoE4-mut1 (open circles).

DISCUSSION

It has been shown that apoE^−/− mice overexpressing human WT apoE develop severe hypertriglyceridemia (9, 10). Development of hypertriglyceridemia may be prevented by specific mutations of apoE (10, 11, 17). In this work, we studied the properties of two apoE4 variants, apoE4-mut1 (apoE4(L261A, W264A, F265A, L268A, V269A)) that leads to normal plasma triglyceride and cholesterol levels and apoE4-mut2 (apoE4(W276A, L279A, V280A, V283A)) that leads to mild hypertriglyceridemia and hypercholesterolemia when these variants are overexpressed in apoE^−/− mice. We were trying to identify alterations in the structural, stability and lipid and lipoprotein-binding properties of these apoE4 variants to better understand how the mutations introduced in these apoE4 forms favorably change the protein function. Particularly, we investigated binding of the apoE4 variants to synthetic VLDL-like emulsion particles as a model of plasma large TG-rich lipoproteins and their remnants. It has been established that apoE bound to TG-rich lipoproteins and their remnants modulates the catabolism of plasma triglycerides (9, 17, 48, 49).

According to the CD analysis, apoE4-mut1 has significantly fewer number of residues in the α-helical conformation than apoE4-mut2 and WT apoE4 (~155 vs. ~183 and ~189, correspondingly). This indicates that mutation of five hydrophobic and bulky residues in the 8-residue segment 261–269 in apoE4-mut1 results in a loss of ~34 residue in the helical conformation of apoE4. In contrast, mutation of four hydrophobic and bulky residues in the 8- residue segment 276–283 in apoE4-mut2 results in a smaller loss of helical residues (~6 residues). This suggests that residues within the segment 261–269 may be involved in tertiary (inter-helical) interactions. Disruption of these interactions by the amino acid substitutions in this segment results in unfolding of ~34-residue helical segment(s), including likely the 261–269 segment and ~ 26-residue segment(s) in other part(s) of the molecule, perhaps in the N-terminal domain. In the amino acid sequence of apoE4, residues 261–269 are close to Glu255 that forms a putative salt bridge with Arg61 that is thought to stabilize a close contact between the protein N- and C-terminal domains (19, 50). Mutations made in apoE4-mut1 may change the protein conformation in the vicinity of Glu255 and thus obstruct the salt-bridge formation with Arg61. This may result in disruption of intra-molecular interactions involved in stabilizing the α-helical structure of those apoE4 segments (consisting of ~26-residues) that are unfolded in apoE4-mut1 and may be located in the N-terminal domain. In contrast, the mutated residues in apoE4-mut2 located within the segment 276–283 are more remote from Glu255 in the amino acid sequence. Therefore, mutations made in apoE4-mut2 have likely only a local effect on the protein conformation (a loss of ~6 helical residues in the mutated region) and a less significant effect on the salt bridge between Glu255 and Arg61 and domain interaction.

According to the ANS studies, mutation of the hydrophobic and bulky residues between amino acids 261 and 269 in apoE4-mut1 results in significantly fewer exposed hydrophobic surfaces in apoE4. This agrees with the existence of a largely exposed hydrophobic site created by residues 261–272 in apoE4 (16, 40). It appears that the compactness of the protein structure is increasing in the order: WT apoE4 ≤ apoE4-mut2 < apoE4-mut1, while the α-helical content is decreasing in the same order. These data suggest that the salt bridge between Glu255 and Arg61 in apoE4 may create spatial constraints that hinder compact tertiary folding of the protein. Eliminating (weakening) this salt bridge in apoE4-mut1 (and in a much lesser extent in apoE4-mut2) is associated with unfolding of some helical regions and eliminating (weakening) the spatial constraints that allows more compact protein folding.

Not withstanding the reduced helical secondary conformation, the more compact structure of apoE4-mut1 appears to be more stabilized and to unfold more cooperatively than a relatively loosely folded structure of WT apoE4 (Figure 2, Table 1). The increased stability of apoE4-mut1, in spite of the fewer number of residues in the helical conformation of the protein (Table 1), suggests that tertiary interactions stabilizing the compactly folded conformation of apoE4-mut1 are more resistant to heat than interactions stabilizing the conformation of WT apoE4, including domain interaction and interactions stabilizing the additional 34-residue helical segment(s). Interestingly, the conformational stability of apoE3 isoform is higher than stability of apoE4 isoform (41, 42), also in spite of the absence of domain interaction (16) and a lower α-helical content in apoE3 (by ~5–10%, our data). Similarly, apoE3 is more compactly folded than apoE4 (24). More cooperative thermal unfolding of apoE4-mut1 suggests that this variant is less prone to form a stable folding intermediate than more loosely folded WT apoE4. This may have a beneficial effect on the function of apoE4-mut1-containing lipoproteins, because the increased propensity of apoE to form partially folded stable intermediates is thought to be implicated in abnormal lipoprotein metabolism (51, 52).

The more stabilized conformation of apoE4-mut1 that is likely less prone to form stable intermediate is associated with a significant reduction in lipid binding, in agreement with the observed inverse correlation between stability and DMPC binding for full-size apoE (25, 26, 52). ApoE4-mut1shows a significant reduction in both the rate and ability to associate with DMPC vesicles (Figure 4 and Figure 5) and the ability to bind to TG-rich emulsion particles (Figure 6). The reduced total length of α-helices in apoE4-mut1 (~34 helical residues less then in WT apoE4) may contribute to the reduced lipid-binding of this protein, because the total length of helices in an apolipoprotein determines its association with lipids (53). Fewer exposed hydrophobic sites and more compact tertiary conformation of apoE4-mut1 can also contribute to the slower rate and ability of this mutant to associate with phospholipid vesicles and the ability to bind to TG-rich particles.

The reduced ability of apoE4 to solubilize multilamellar phospholipid vesicles and bind to TG-rich lipoproteins may have a crucial effect on lipoprotein metabolism and plasma lipid levels. ApoE bound to TG-rich lipoproteins directly inhibits the LPL-mediated lipolysis of TG by promoting substrate dissociation from LPL (48). Thus, reduced binding of apoE to TG-rich particles will promote LPL-mediated hydrolysis of the particles. Also, studies of animal models overexpressing apoE suggested that hypertriglyceridemia developed in these animals results partially from displacement of apoC-II from TG-rich lipoproteins by apoE (9, 54). Apo C-II is required for the activation of LPL (55); thus, the association of apoC-II with the TG-rich lipoproteins enhances the lipolysis of their TG. Mutations made in apoE4-mut1 result in the conformational and stability changes that reduce significantly the binding of apoE4 to TG-rich lipoproteins and thus favor the increased binding of apoC-II that promotes LPL-mediated lipolysis and clearance of lipoprotein remnants.

The conformational and stability characteristics and ability to bind to TG-rich particles for apoE4-mut2 appear to be between the characteristics for WT apoE4 and apoE4-mut1. The α-helical content, ANS fluorescence intensity that correlates with the exposure of hydrophobic surfaces, the protein destabilization and the ability to bind to TG-rich model particles decrease in the order: WT apoE4 ≥ apoE4-mut2 > apoE4-mut1. The in vivo studies showed that under conditions of overexpression of apoE4-mut2, these slightly altered properties of the protein are associated with milder hypertriglyceridemia (Figure 1, inset).

The data in Figure 7 show that one emulsion particle binds less apoE4-mut1 molecules than WT apoE4-molecules throughout the entire range of the initial PC/protein ratios in the incubation mixtures (from 1025 to 40). The extrapolation of these data to the conditions of protein saturation (the initial ratio PC/protein ⟶ 0) suggests that under these conditions, when a maximal possible amount of protein is bound to emulsion, the emulsion surface will also accommodate fewer molecules of apoE4-mut1 than WT apoE4. This suggests that one molecule of WT apoE4 occupies a smaller area on emulsion particles than one molecule of apoE4-mut1. Compared to apoE4-mut1, WT apoE4 molecule may be more deeply immersed into the lipid layer or/and have more compact conformation on the emulsion surface. This may imply that under conditions of apoE overexpression, when the protein concentration in plasma and on TG-rich lipoproteins is high, the receptor binding region of WT apoE4 may be buried or masked and thereby unable to mediate clearance of the lipoproteins. It has been suggested that at high surface concentrations of apoE4 on TG-rich particles, the whole N-terminal domain of the protein may even be displaced from the lipid surface and adopt a “closed” fourhelix bundle conformation (37, 56) that precludes the interaction of the receptor-binding region with the LDL-receptor and impairs clearance of remnant lipoproteins (57). In contrast, under conditions of overexpression of apoE4-mut1, each protein molecule occupies a larger area on the emulsion surface that may allow the “open” lipid-bound conformation of the N-terminal domain. In this conformation, the receptor-binding region of apoE is recognized by the LDL receptor and thus may promote the efficient clearance of the apoE-containing lipoproteins. This may also explain why, in contrast to WT apoE4, overexpression of apoE4-mut1 corrects hypercholesterolemia and does not induce hypertriglyceridemia. The amounts of apoE4-mut2 bound to emulsion are not significantly different from the corresponding amounts of WT apoE4 (Table 3). This may imply that under conditions of overexpression of apoE4-mut2, the topology and conformation on the emulsion surface for this variant are similar to those of WT apoE4, so that the receptor-binding region of apoE4-mut2 may be buried or masked, or the whole N-terminal domain of this mutant may be displaced from the particle surface. This may explain why overexpression of apoE4-mut2, similar to WT apoE4, is associated with hypercholesterolemia and hypertriglyceridemia.

It appears that the mutations made in apoE4-mut1 lead to changes in the biophysical characteristics of apoE4 that make it resemble apoE3. The interaction between the N- and C-terminal domains is a characteristic of apoE4 and does not exist in apoE3 (15, 19). Our studies suggest that the domain interaction may be also reduced in apoE4-mut1 compared to WT apoE4. ApoE3, similar to apoE4-mut1, has a more compactly folded conformation as shown by ANS fluorescence (24), is more stable (41, 42), solubilizes DMPC vesicles slowly (25, 40), and has a reduced ability to bind to TG-rich emulsion particles as compared to WT apoE4 (24). This “conversion” of apoE4 into apoE3-like protein by the (L261A, W264A, F265A, L268A, V269A) mutation is associated with normalization of plasma lipid levels of apoE^−/− mice without induction of hypertriglyceridemia. It has been established that in humans, the most common isoform apoE3, in contrast to apoE4, is not associated with high plasma cholesterol level and an increased risk of development of atherosclerosis (1–3, 7). However, similar to apoE4, apoE3 at high plasma concentrations is associated with hypertriglyceridemia (8–11). In contrast, overexpression of apoE4-mut1 in apoE^−/− mice does not induce hypertriglyceridemia. The bigger differences in some properties between WT apoE4 and apoE4-mut1, compared to the differences in these properties between WT apoE4 and apoE3, may be a key to the beneficial effects of the mutations made in apoE4-mut1. For example, both apoE3 and apoE4-mut1 have less exposed hydrophobic surfaces than WT apoE4 as shown by ANS fluorescence data; however, ANS intensity in the presence of WT apoE4 is only ~1.2 times larger than the intensity in the presence of apoE3 (58) but two times larger than in the presence of apoE4-mut1 (Table 2). This suggests that apoE4- mut1 has much more exposed hydrophobic surfaces than apoE3. Both apoE3 and apoE4-mut1 are more stable than WT apoE4; however the difference in the midpoint of thermal denaturation T_1/2 between apoE3 and WT apoE4 is only ~2°C (40), in contrast to ~5°C difference between apoE4-mut1 and WT apoE4 (Table 1). This indicates significantly higher stability of apoE4-mut1 compared to apoE3.

Remarkably, we found earlier that two mutants of human apoA-I, the apoAI[E110A/E111A] and apoA-I[Δ(61–78)], that induce severe hypertriglyceridemia in mice had a more loosely folded structure and decreased stability compared to WT apoA-I, which does not induce hypertriglyceridemia (28, 29). These structural changes in the apoA-I mutants facilitate their binding to TG-rich large lipoproteins (28, 59). Thus, the association of a destabilized, loosely folded structure and the increased affinity to TGrich lipoproteins with hypertriglyceridemia is observed for both apoE4 and apoA-I. In this connection, it will be interesting to study other variant forms of the apolipoproteins that induce or correct hypertriglyceridemia.

In summary, our data show that the apoE4-mut1 variant that prevents the induction of hypertriglyceridemia has a more stable and compactly folded conformation, a decreased rate and ability to solubilize phospholipid vesicles, and a reduced ability to bind to large TG-rich particles. These properties of apoE4-mut1 may lead to enhanced lipolysis and receptor-mediated clearance of the apoE4-mut1-containing lipoprotein particles. The mutation apoE4-mut2 that results in milder hypertriglyceridemia leads to smaller or insignificant changes in the biophysical properties of apoE4. Overall, these data provide new biophysical insights into structure-function relationships of apoE4 that may underlie prevention of hypertriglyceridemia. The molecular etiology of common forms of hypertriglyceridemia in humans remains unknown. It is possible that structural mutations that destabilize apoE, apoA-I, or other apolipoproteins may contribute to hypertriglyceridemia in humans (60). Understanding the molecular mechanisms that underlie the improved biological properties of the apoE variant forms may facilitate the development of new therapeutic approaches for correction of lipid disorders in humans. For example, using small molecules that potentially can bind to apoE4 in vivo and modulate the protein biophysical properties (reviewed in ref. 61) may be one of the therapeutic strategies.

Abbreviations

ANS

8-Anilino-2-naphthalene-sulfonate

apo

apolipoprotein

apoE^−/−

apoE-deficient

CD

circular dichroism

DMPC

dimyristoylphosphatidylcholine

GdnHCL

guanidine hydrochloride

LDL

low density lipoprotein(s)

LPL

lipoprotein lipase

PC

phosphatidylcholine

TG

triglyceride

VLDL

very low density lipoprotein(s)

WMF

wavelength of maximum fluorescence

WT

wild type