apoE3[K146N/R147W] acts as a dominant negative apoE form that prevents remnant clearance and inhibits the biogenesis of HDL.

The K146N/R147W substitutions in apoE3 were described in patients with a dominant form of type III hyperlipoproteinemia. The effects of these mutations on the in vivo functions of apoE were studied by adenovirus-mediated gene transfer in different mouse models. Expression of the apoE3[K146N/R147W] mutant in apoE-deficient (apoE(-/-)) or apoA-I-deficient (apoA-I(-/-))×apoE(-/-) mice exacerbated the hypercholesterolemia and increased plasma apoE and triglyceride levels. In apoE(-/-) mice, the apoE3[K146N/R147W] mutant displaced apoA-I from the VLDL/LDL/HDL region and caused the accumulation of discoidal apoE-containing HDL. The WT apoE3 cleared the cholesterol of apoE(-/-) mice without induction of hypertriglyceridemia and promoted formation of spherical HDL. A unique property of the truncated apoE3[K146N/R147W]202 mutant, compared with similarly truncated apoE forms, is that it did not correct the hypercholesterolemia. The contribution of LPL and LCAT in the induction of the dyslipidemia was studied. Treatment of apoE(-/-) mice with apoE3[K146N/R147W] and LPL corrected the hypertriglyceridemia, but did not prevent the formation of discoidal HDL. Treatment with LCAT corrected hypertriglyceridemia and generated spherical HDL. The combined data indicate that the K146N/R147W substitutions convert the full-length and the truncated apoE3[K146N/R147W] mutant into a dominant negative ligand that prevents receptor-mediated remnant clearance, exacerbates the dyslipidemia, and inhibits the biogenesis of HDL.

R147W]202 was generated using the pGEM7-[K146N/R147W] vector as a template and the mutagenic primers apoE3[L203Stop] F and apoE3[L203Stop]R (supplementary Table IA). The apoE gene containing the mutations was excised with Hin dIII and XbaI and subcloned to the pAdTrack-CMV vector. This strategy for generation of pAdTrack-CMV constructs is presented schematically in Fig. 1 of ( 38 ). The LPL cDNA cloned in the pCMV-Sport6 vector (Open Biosystems, catalog number MHS1010-73520) was excised using the KpnI and Hin dIII restriction enzymes and subcloned into the corresponding sites of the pAdTrack-CMV vector. The LCAT adenovirus construct was generated as described ( 39 ). The recombinant adenoviruses were generated as described using the Ad-Easy-1 system where the adenovirus construct is generated in BJ-5183 bacterial cells ( 35,40 ). Following large-scale infection of HEK 293 cell cultures, the recombinant adenoviruses were purifi ed by two consecutive CsCl 2 ultracentrifugation steps, dialyzed, and titrated. Titers of approximately 2-5 × 10 7 plaque forming units (pfu)/µl were obtained.

Animal studies, RNA isolation, and apoE mRNA quantitation
Male and female apoE Ϫ / Ϫ or apoA-I weeks old were used in these studies. The apoE Ϫ / Ϫ mice were purchased from Jackson Laboratories (Bar Harbor, ME), mice deficient for apoA-I and apoE were a gift of Dr. Fayanne Thorngate and Dr. David Williams ( 41 ). The mice were maintained on a 12 h light/dark cycle and standard rodent chow. All procedures performed on the mice were in accordance with National Institutes of Health guidelines and following a protocol (AN-14219.2012.10) approved by the Institutional Animal Care and Use Committee. Groups of four or more mice were injected intravenously through the tail vein with doses of 5 × 10 8 to 2 × 10 9 pfu of adenoviruses expressing WT and truncated human apoE forms. The high (2 × 10 9 ) adenovirus dose was used in gene transfer experiments involving the truncated (1-202) apoE forms in order to establish that overexpression of truncated apoE does not induce hypertriglyceridemia. Blood was obtained from the tail vein after a 4 h fast preceding adenoviral injection and 1, 2, 3, and 4 days postinfection. The plasma was isolated, aliquoted, and stored at 4°C for immediate use or at Ϫ 20°C for subsequent use. Mice were euthanized 4 days postinfection, hepatic RNA was isolated by the triazole method, and the apoE mRNA levels were determined by quantitative PCR using the standard TaqMan thermal cycling program and Applied Biosystems probe sets for human apoE and 18s rRNA (catalog numbers HJ00171168_m1 and 4319413E, respectively).

Fast protein liquid chromatography fractionation and plasma lipid and apoE determinations
For fast protein liquid chromatography (FPLC) analysis of plasma samples, 20 l of plasma were diluted 1:3 with PBS, loaded onto a Superose 6 column in a SMART micro FPLC system (Pharmacia), and eluted with PBS. A total of 25 fractions of 50 l volume each were collected for further analysis. Triglycerides, total cholesterol (TC), and free cholesterol of plasma and the FPLC fractions were determined spectrophotometrically at 540 nm and 490 nm, respectively, as described previously ( 35 ), using the INFINITY triglycerides reagent (Thermo Fisher Scientifi c, Waltham, MA), cholesterol E, and free cholesterol C (Wako Chemicals USA, Inc.) according to the manufacturer's instructions. Plasma apoE concentrations were determined by sandwich ELISA using apoE standards supplied by Wako Chemicals (Richmond, VA) ( 35 ).

Density gradient ultracentrifugation and electron microscopy analysis
To assess the ability of WT and truncated apoE forms to associate with different lipoproteins, an aliquot of 0.24 ml of plasma receptor-related protein 1 (LRP1) and heparin sulfate proteoglycans (HSPGs), respectively ( 19,28,29 ).
Certain apoE phenotypes and genotypes are associated with recessive or dominant forms of type III hyperlipoproteinemia (HLP). The recessive form of type III HLP that is found in subjects with the E2/2 phenotype results from the substitution of Cys for Arg-158 ( 30 ). This mutation, combined with other genetic or environmental factors, affects the catabolism of apoE-containing lipoproteins and results in the accumulation in plasma of remnants of lipoprotein catabolism ( 2,31 ).
A variety of rare mutations in the 136-147 region of apoE are associated with a dominant mode of inheritance of type III HLP, which is expressed at an early age ( 2,12,32 ). The degree of penetrance of the disease varies depending on the mutation ( 2 ). Of particular interest is the apoE1-Hammersmith phenotype that carries the K146N/ R147W substitutions in apoE3. The carriers of this phenotype expressed the clinical characteristics of type III HLP at the age of 3 years ( 33 ).
In the present study, we have investigated the molecular etiology of the dominant form of type III HLP that is associated with the apoE3[K146N/R147W] mutations using adenovirus-mediated gene transfer in apoE-defi cient (apoE Ϫ / Ϫ ) and apoA-I and apoE double deficient (apoA-I Ϫ / Ϫ × apoE Ϫ / Ϫ ) mice. We found that the K146N/ R147W substitutions in apoE3 exacerbated the hypercholesterolemia that is associated with apoE defi ciency ( 4 ) and induced severe hypertriglyceridemia in the mouse models tested that could be corrected by excess LPL. These mutations also affected the functions of LCAT and caused accumulation of discoidal HDL particles in plasma and this defect could be corrected by excess LCAT.
Hypertriglyceridemia was prevented but hypercholesterolemia persisted when the C-terminal 203-299 residues of the apoE3[K146N/R147W] mutant were deleted. The inability of the truncated apoE3[K146N/R147W]202 mutant to clear plasma cholesterol is unique to this mutant and was not observed in previous studies using apoE-202 truncations of other normal apoE isoforms, as well as apoE4[R142C] that is associated with a dominant form of type III HLP ( 34,35 ). The fi ndings suggest that the apoE3 [K146N/R147W] mutant acts as a dominant negative ligand that blocks clearance of the lipoprotein remnants and inhibits the maturation of apoA-I-containing HDL particles.

Materials
Materials not mentioned in the Materials and Methods were obtained from sources described previously ( 36,37 ).

Construction of recombinant adenoviruses
apoE mutants were generated using the pGEM7-apoE3 vector containing exons II, III, and IV of the human gene as a template ( 35 ) We have found that a low dose (5 × 10 8 pfu) of an adenovirus expressing apoE3[K146N/R147W] increased plasma cholesterol in apoE Ϫ / Ϫ mice over a 4 day period and induced severe hypertriglyceridemia ( Fig. 1A, B ). In contrast, infection of apoE Ϫ / Ϫ mice with a similar dose (5 × 10 8 pfu) of adenovirus expressing WT apoE3 decreased plasma cholesterol, did not affect plasma triglyceride levels, and caused an increase in plasma human apoE levels ( Fig. 1A-C ). Despite the fact that the plasma levels of WT human apoE3 are beyond the physiological levels of mouse apoE (6.8-9.1 mg/dl) ( 43 ), the low dose (5 × 10 8 pfu) of adenovirus expressing WT apoE3 used did not cause hypertriglyceridemia in apoE Ϫ / Ϫ mice. There was a great increase in plasma triglycerides and apoE levels in the apoE3[K146N/ R147W] mutant and this increase was correlated with the increase in plasma apoE levels ( Fig. 1B, C ). In the present study, we found that even a very high dose (2 × 10 9 pfu) of the truncated apoE3[K146N/R147W]202 did not clear cholesterol but also did not induce hypertriglyceridemia ( Fig. 1A-C ). The inability of this truncated apoE3[K146N/ R147W]202 form to clear plasma cholesterol in apoE Ϫ / Ϫ mice, differentiates this mutant from all the apoE truncated forms studied previously, which cleared plasma cholesterol but did not induce hypertriglyceridemia ( 34,35,44 ). In contrast to the inability of the truncated forms to induce hypertriglyceridemia, even when they are expressed at high levels, previous studies showed that overexpression of full-length apoE forms induced severe hypertriglyceridemia and did not clear the plasma cholesterol in apoE Ϫ / Ϫ mice ( 35,44 ). This was attributed to the folding of the C-terminal portion of the full-length protein bound to lipoprotein particles in a way that masks the LDL receptorbinding domain of apoE ( 44 ).
The increased plasma levels of the full-length and the truncated apoE3[K146N/R147W] form ( Fig. 1C) are consistent with defective recognition by the LDL receptor due to the K146N/R147W mutations. The low levels of triglycerides when the truncated apoE3[K146N/R147W]202 form is overexpressed (Fig. 1B) supports the concept that induction of hypertriglyceridemia requires the C-terminal domain of apoE, as shown previously ( 34,35 ). The decrease in plasma apoE of the truncated relative to the fulllength apoE mutant on day 4 (Fig. 1C), which is also associated with a similar decrease in plasma cholesterol levels, may be associated with the faster removal from plasma of discoidal apoE particles containing the truncated form of apoE. Previous studies indicated increased catabolism by the liver of apoA-I containing discoidal particles ( 45,46 ). Real-time quantitative RT-PCR analysis showed that the mRNA levels of the WT apoE3 and the apoE3[K146N/ R147W] mutant were comparable , whereas the mRNA levels of the truncated apoE3[K146N/R147W]202 mutant were much higher, refl ecting the higher dose of the recombinant adenovirus selected for this experiment (supplementary Table IB). The mouse apoA-I mRNA levels were not altered 4 days postinfection with the adenoviruses expressing the WT apoE or the full-length or the truncated apoE3[K146N/R147W] form (data not shown). Two other truncated apoE forms studied previously that were used as was fractionated by KBr density gradient ultracentrifugation. Ten fractions were collected and analyzed by SDS-PAGE as described previously ( 42 ). Aliquots of fractions 6 and 7 obtained by density gradient ultracentrifugation that contained most of the apoE were analyzed by electron microscopy (EM). The photomicrographs were taken at ×75,000 magnifi cation and enlarged three times as described previously ( 42 ).

Non-denaturing two-dimensional electrophoresis
The distribution of HDL subfractions in plasma was analyzed by two-dimensional gel electrophoresis as described ( 22 ). apoE was detected by probing the membrane with a goat polyclonal anti-human apoE antibody (K74190G; Meridian Biosciences, Saco, ME).

ABCA1-dependent cholesterol effl ux assay
ABCA1-dependent effl ux of cholesterol was measured in cultures of J774 macrophages in which expression of ABCA1 was induced by a cAMP analog using WT and mutant apoE forms as cholesterol acceptors. The J774 mouse macrophages were labeled with 0.3 Ci/ml [

Statistical analysis
Data are presented as the mean ± SD. Statistically signifi cant differences between two groups of mice were determined using Student's t -test with unequal variance.

Lipids and lipoprotein profi le of apoE ؊ / ؊ mice expressing the apoE3[K146N/R147W] mutant
To elucidate how the K146N/R147W substitutions in human apoE3 induce dyslipidemia, we have assessed the ability of WT apoE3 as well as the full-length apoE3[K146N/ R147W] and various truncated apoE forms extending from residues 1 to 202, to clear the high plasma cholesterol levels of the apoE Ϫ / Ϫ mice and to affect the plasma triglycerides following gene transfer. The rationale of the experiments with the truncated apoE forms was to assess the contribution of the C-terminal 203-229 domain of apoE3[K146N/R147W] in the clearance of plasma cholesterol without the induction of hypertriglyceridemia. Previous studies showed that apoE-202 truncations of other normal apoE isoforms, as well as of apoE4[R142C] that is associated with a dominant form of type III HLP, cleared cholesterol and did not induce hypertriglyceridemia ( 34,35 ). A different behavior of this mutant would differentiate it from the previously studied apoE forms and might explain the dominant nature of this mutation in the induction of type III HLP.  whereas the discoidal particles observed in Fig. 2G represent apoE-containing HDL.
Two dimensional gel electrophoresis of plasma followed by probing of the blots with a human apoE antibody showed similar, but not identical, broad patterns of particles both for the WT apoE3 and the apoE3[K146N/R147W] mutant (

Studies in apoA-I
The effects of the apoE3[K146N/R147W] mutations were also studied by gene transfer in apoA-I Ϫ / Ϫ ×apoE Ϫ / Ϫ mice in order to obtain a clearer picture of the mechanisms by which the apoE mutations affect HDL particle formation. This information could not be unambiguously obtained by the experiments described in domain in the truncated apoE3[K146N/R147W]202 appears to improve, to some extent, its functional interaction with LCAT.
FPLC fractionation of plasma showed that mice expressing WT apoE3 had an HDL cholesterol peak as described previously ( 3 ). Mice expressing the apoE3[K146N/ R147W]202 mutant likewise contained a smaller HDL cholesterol peak along with a cholesterol shoulder extending in the LDL and IDL region. Mice expressing the apoE[K146N/R147W] mutant had a negligible HDL cholesterol peak, but also contained a shoulder extending in the VLDL and LDL region ( Fig. 2A ). The majority of the triglycerides in mice expressing the apoE3[K146N/R147W] mutant were found in the VLDL region and, to a much lesser extent, in the IDL and LDL region ( Fig. 2B ). The FPLC profi les are compatible with the distribution of the apoA-I and apoE in HDL and other lipoprotein fractions as determined by density gradient ultracentrifugation of plasma ( Fig. 2C-E ).

Density gradient ultracentrifugation, two-dimensional gel electrophoresis of the plasma, and EM analysis of HDL
The proteomic composition of HDL and other lipoprotein fractions were assessed by density gradient ultracentrifugation of plasma obtained 4 days post-adenovirus infection and analysis of the fractions by SDS-PAGE. When the apoE Ϫ / Ϫ mice were infected with a low dose (5 × 10 8 pfu) of apoE3-expressing adenovirus, the majority of the apoE was found in the HDL3 region where mouse apoA-I also resides. This apoprotein distribution indicated that the HDL fraction of mice expressing apoE3 described in Fig. 2A , contains mostly apoA-I-, and to a lesser extent, apoE-containing HDL. Smaller amounts of apoE along with apoB-48 were found in the VLDL/IDL/LDL region ( Fig. 2C ).
When the apoE Ϫ / Ϫ mice were infected with a similar dose (5 × 10 8 pfu) of adenovirus expressing apoE3[K146N/ R147W], approximately 60% of apoE along with apoB-48 was found in the VLDL/IDL/LDL region and the remainder in the HDL2/HDL3 region. Unexpectedly, the expression of mutant apoE was associated with complete absence of the apoA-I in the HDL region and the other lipoprotein fractions ( Fig. 2D ). In apoA-I transgenic mice that express high levels of human apoA-I, expression of the apoE [K146N/R147W] mutant decreased but did not eliminate the amount of apoA-I that fl oats in the HDL region (data not shown).
Analysis of fractions 6 and 7, obtained by density gradient ultracentrifugation by EM, showed that the WT apoE3 and the truncated apoE3[K146N/R147W]202 mutant formed spherical particles ( Fig. 2F, H ) and the apoE3[K146N/R147W] mutant formed discoidal HDL and larger size particles corresponding in size to IDL/LDL ( Fig. 2G ). Based on the apoprotein composition observed in Fig. 2C-E , the spherical particles observed in Fig. 2F, H represent apoA-I-containing and to a lesser extent apoE-containing HDL particles, to those observed in the plasma of apoE Ϫ / Ϫ mice expressing the same apoE3 forms (presented in Fig. 2I-K ). The experiments described in Fig. 3H, I made it clear that the expression of both the full-length and the truncated apoE3[K146N/R147W] form promoted the formation of discoidal apoE-containing HDL particles, whereas the WT apoE3 promoted the formation of spherical apoEcontaining HDL ( Fig. 3G ).

The role of LPL and LCAT in the induction of dyslipidemia in apoE ؊ / ؊ mice expressing apoE3[K146N/ R147W]
The accumulation of triglycerides and apoE in apoE Ϫ / Ϫ mice that received the adenovirus expressing apoE3[K146N/ R147W] ( Fig. 1A-C )  mice expressing apoE3, the majority of apoE was in the HDL2/HDL3 fraction, but a considerable portion was also found in the VLDL/IDL/LDL fraction ( Fig. 3D ). In mice expressing the apoE3[K146N/R147W] mutant, the great majority of apoE was shifted toward the VLDL/IDL/LDL region and a smaller amount remained in the HDL region ( Fig. 3E ). A similar distribution in apoE was observed in apoA-I Ϫ / Ϫ ×apoE Ϫ / Ϫ mice expressing the truncated apoE3[K146N/R147W]202 mutant ( Fig. 3F ). EM analysis of fractions 6 and 7, obtained by density ultracentrifugation of the plasma, showed that WT apoE3 generated spherical particles and the full-length and truncated apoE3[K146N/R147W] forms generated discoidal particles ( Fig. 3G-I ). Small amounts of discoidal particles, along with larger particles corresponding in size to LDL/IDL, were also found in fractions 4 and 5 in mice expressing the fulllength and truncated apoE3[K146N/R147W] forms (data not shown). Because the double defi cient mice do not contain mouse apoE or apoA-I, the spherical and discoidal particles observed represent apoE-containing HDL. Twodimensional gel electrophoresis showed similar patterns apoE was distributed primarily in the HDL2/HDL3 region ( Fig. 5A ). The increased apoE levels in the HDL region were associated with the loss of plasma apoA-I that otherwise fl oats in the same region (compare Figs. 2D and 5A ). EM analysis of HDL corresponding to fractions 6 and 7 of Fig. 5A showed the presence of predominantly discoidal particles ( Fig. 5C ), indicating that correction of hypertriglyceridemia did not correct the defective maturation of HDL. Few discoidal particles, along with larger size particles corresponding in size to LDL/IDL particles, were also observed in fractions 4 and 5 (supplementary Fig. IC). The persistence of discoidal HDL particles following LPL treatment is consistent with the observed decrease of the CE/ TC ratio in plasma (supplementary Table II). The presence of discoidal particles ( Fig. 2G ) and the decrease of the CE/TC ratio in the plasma of apoE Ϫ / Ϫ mice expressing the apoE3[K146N/R147W] mutant adenovirus expressing human LPL, gradually increased the plasma cholesterol levels 2-4 days postinfection ( Fig. 4A ). The increase in plasma cholesterol was not statistically different to that observed following infection of apoE Ϫ / Ϫ mice with adenovirus expressing apoE3[K146N/R147W] alone. Importantly, coinfection of mice with the two viruses caused only a moderate increase of plasma triglycerides 3-4 days postinfection ( Fig. 4B ). Furthermore, the LPL overexpression increased slightly the plasma apoE levels 2-4 days postinfection ( Fig. 4C ). FPLC analysis of plasma from mice expressing the apoE3[K146N/R147W] mutant and LPL showed the presence of a small HDL cholesterol peak along with a large shoulder in the LDL/IDL region ( Fig. 4D ). Residual triglycerides were distributed in the VLDL region ( Fig. 4E ).
Density gradient ultracentrifugation of the plasma also supported the increase in plasma apoE and showed that ( Fig. 5D ). Formation of spherical particles is consistent with the normalization of the CE/TC ratio of plasma and the reappearance of apoA-I in the HDL region following LCAT treatment ( Fig. 5B ) (supplementary Table II).

Ability of the WT apoE3 and the full-length and truncated apoE3[K146N/R147W] mutant to promote ABCA1mediated cholesterol effl ux
Cholesterol effl ux studies showed that the ability of WT apoE3, as well as the full-length and truncated apoE3[K146N/ R147W] mutants, were similar ( Fig. 6 ).

Physicochemical studies
We evaluated the impact of K146N/R147W substitutions on the structure of apoE3. We found that compared with WT apoE3, these mutations decreased the thermal stability of the full-length apoE3[K146N/R147W] mutant and increased the binding of ANS, indicating exposure of hydrophobic residues in the surface of apoE3[K146N/R147W] (  Fig. IIIB). There was no significant difference in ␣ -helical content and the other physicochemical parameters described in Table 1

Background
A variety of rare apoE mutations between residues 136-147 have been described in humans that are associated (supplementary Table II ( Fig. 4A, B ). The increase in plasma cholesterol was not statistically different to that observed by treatment of apoE Ϫ / Ϫ mice with adenovirus expressing apoE3[K146N/R147W] mutant alone or in the presence of LPL. Furthermore, the LCAT treatment increased plasma apoE 2-4 days postinfection to levels that were slightly lower than those observed in apoE Ϫ / Ϫ mice treated with apoE3[K146N/R147W] alone ( Fig. 4C ). The increase in plasma apoE levels is also supported by density gradient ultracentrifugation of the plasma and SDS-PAGE analysis of the resulting fractions, as described below. FPLC analysis of plasma from mice expressing the apoE[K146N/ R147W] mutant and LCAT showed the presence of a large HDL cholesterol peak along with a shoulder extended in the LDL/IDL region ( Fig. 4D ). Residual triglycerides were distributed in the VLDL region ( Fig. 4E ).
Density gradient ultracentrifugation of the plasma and SDS-PAGE analysis of the resulting fractions showed that the LCAT treatment shifted apoE to the VLDL/IDL/LDL region and restored apoA-I in the HDL region ( Fig. 5B ). EM analysis of the HDL fractions 6 and 7 of Fig. 5B showed that apoE Ϫ / Ϫ mice infected with adenovirus expressing apoE3[K146N/R147W] and LCAT formed spherical particles The K146N/R147W mutations caused small changes in the physicochemical properties of apoE3[K146N/ R147W] mutant but did not affect its ability to promote

ABCA1-mediated cholesterol effl ux
The physicochemical studies showed that the apoE3 [K146N/R147W] mutant had decreased thermal stability and increased affi nity for ANS, indicating that the mutations increase the exposure of hydrophobic residues on the surface of the mutant apoE. The ability of the apoE3 [K146N/R147W] mutant to promote ABCA1-mediated cholesterol effl ux was similar to that of WT apoE3. This is consistent with the ability of the mutant to form apoE-containing HDL particles in vivo.

Lipid and lipoprotein phenotype of apoE ؊ / ؊ mice expressing the apoE3[K146N/R147W] mutant
Initially, we used adenovirus-mediated gene transfer in apoE Ϫ / Ϫ mice to study the lipid and lipoprotein phenotype that results from the expression of the apoE3 [K146N/R147W] mutant. The advantage of the adenovirus system is that when the mutant protein is expressed in apoE Ϫ / Ϫ mice, it creates the phenotype expected in subjects with homozygous defect in the corresponding allele within a few days. Human subjects or experimental animals homozygous for the apoE3[K146N/R147W] mutations have not been studied previously. The phenotype that was generated following the gene transfer is characterized by severe hypercholesterolemia and hypertriglyceridemia, reduced CE/TC ratio in plasma, formation of discoidal apoE-containing HDL, very high plasma apoE levels, and accumulation of apoE in the VLDL/ IDL/LDL region. It is expected that a similar phenotype in humans homozygous for the mutations will have severe clinical consequences that may lead to cardiovascular disease. Abnormal lipid and lipoprotein phenotypes have been observed in mice as a result of overexpression of the endogenous mouse apoE ( 53 ). Thus the observed dyslipidemia in this study cannot be attributed to abnormal interactions of the human apoE with the mouse apolipoproteins. The most likely interpretation with dominant inheritance of type III HLP that is manifested at an early age ( 2 ).
The importance of the 136-152 region of apoE for binding to the LDL receptor was assessed by in vitro mutagenesis ( 12,47,48 ). It has been recognized that the binding affi nity of reconstituted HDL containing these apoE mutants for the LDL receptor determined in vitro did not always correlate with the severity of the dyslipidemia observed in vivo. Thus, it was suggested that the occurrence of the disease can be infl uenced by other secondary factors, including the apoE phenotype or the affinity of the specific mutants for HSPGs ( 2,12,47,(49)(50)(51)(52).
The apoE3[K146N/R147W] variant studied here was initially designated apoE1-Hammersmith ( 33 ). This apoE variant contains two additional negative charges as compared with apoE3 due to substitution of two positively charged residues by neutral residues. Two family members heterozygous for the apoE1 allele exhibited the clinical markers of type III HLP at a very early age, including tuberoeruptive xanthomas, high plasma cholesterol and triglyceride levels, and the appearance of a broad ␤ band following electrophoresis of plasma ( 33 ).  Values are mean ± SD of three experiments . Parameters obtained from the indicated measurements are as follows: " ␣ -helix 222nm " is the percentage of ␣ -helical content of the protein as calculated from the molecular ellipticity of the protein sample at 222 nm; "T m " is middle point of the thermal denaturation transition (melting temperature); "n" is an indicator of the cooperativity of the thermal unfolding transition and is calculated using the Hill equation n = (log81)/log(T 0.9 /T 0.1 ), where T 0.9 and T 0.1 are the temperatures where the unfolding transition reaches a fractional completion of 0.9 and 0.1, respectively; " ⌬ H app " is the relative enthalpy change during the thermal denaturation; and "fold-increase" is the increase in ANS fl uorescence in the presence of 1.8 M protein relative to free ANS in the same buffer . a P = 0.0054. b P = 0.0244. c P < 0.0001. d P = 0.0040.
It is unclear whether the defective maturation of discoidal into spherical particles is a direct consequence of the K146N/R147W mutation and its impact on the maturation of HDL or is the result of the increased plasma apoE levels that may affect LCAT activity. Accumulation of discoidal HDL particles was observed previously in mice expressing apoA-I mutants ( 42 ). The experiments in mouse models that lack CETP activity may not fully refl ect potential changes in lipid and lipoprotein profi les that can occur in human patients. However, in mice expressing the apoE3[K146N/R147W] mutant, HDL is virtually absent and the cholesterol is distributed to the VLDL, and to a much lesser extent to the shoulder that appears in the IDL/LDL region. It is possible that in the presence of CETP, the area of this shoulder might be decreased due to the transfer of its CE to the VLDL region.

Deletion of the C-terminal residues 203-299 of apoE3[K146N/R147W] prevented induction of hypertriglyceridemia but exacerbated the hypercholesterolemia
Previous studies showed that proteoliposomes containing truncated apoE-184 or apoE-191 that lack the C-terminal apoE region can bind to the LDL receptor ( 48 ). The combined data established that the amino terminal 1-202 or 1-184 domains of apoE are effi cient ligands for the LDL receptor ( 34,48,61 ).
apoE contains an N-terminal and a C-terminal domain which fold independently ( 62 ). It has been proposed that apoE inter-domain interactions affect the structure of apoE isoforms (63)(64)(65) and their different affinity for VLDL (64)(65)(66), as well as the contribution of apoE4 in Alzheimer's disease ( 67 ).
In the present study, the truncated apoE3[K146N/ R147W]202 mutant prevented the induction of hypertriglyceridemia in apoE Ϫ / Ϫ or apoA-I Ϫ / Ϫ ×apoE Ϫ / Ϫ mice. This fi nding is consistent with previous studies involving other truncated apoE forms, including the truncated form of apoE4[R142C] that is also associated with a dominant form of type III HLP ( 34 ).

Potential mechanism of hypercholesterolemia induced by the full-length and the truncated apoE3[K146N/R147W] mutant
apoE3[K146N/R147W] induced an increase in plasma cholesterol levels in apoE Ϫ / Ϫ or apoA-I mice that could be the result of altered recognition of the receptor binding domain of the mutant protein by the LDL receptor. Alternatively, it could be the result of unfavorable interactions between the N-and C-terminal domains that shield the receptor binding domain of apoE, as suggested previously ( 44 ), or a combination of both factors. It has been proposed that apoE inter-domain interactions may be at least partially responsible for the subtle differences in the structural and the functional properties of different apoE isoforms (63)(64)(65)(66)(67). These two possibilities were tested by adenovirus gene transfer experiments using the truncated apoE3[K146N/R147W]202 form that extends from residue 1 to residue 202. The expectation was is that the accumulation of the mutant apoE3[K146N/ R147W] cannot be cleared effi ciently by the LDL receptor and this is manifested by the concomitant increase of plasma apoE, plasma cholesterol, and plasma triglyceride levels.

Potential etiology of the apoE3[K146N/R147W]-induced hypertriglyceridemia
The accumulation of apoE in the VLDL/IDL/LDL region appears to create insuffi ciency for both LPL and LCAT activity. As reported previously ( 54,55 ) the insufficiency of LPL may result from inhibition of this enzyme by the apoE which accumulates in VLDL/IDL/LDL size lipoprotein particles. However, one cannot exclude negative interaction with other components of the triglyceride rich lipoprotein-hydrolyzing system . The LPL treatment corrected the hypertriglyceridemia but did not affect the hypercholesterolemia or the plasma CE/TC ratio and did not reverse the formation of discoidal HDL particles. The increased levels of LPL in the circulation following gene transfer could, in principle, promote lipoprotein clearance both by increasing the rate of lipolysis and exposing the receptor-binding domain of apoE, as well as by direct receptor-mediated clearance mechanisms involving LPL ( 56,57 ). Other studies have suggested that excess of secreted apoE may partially displace the LPL and/or apoCII and thus reduce the rate of lipoprotein triglyceride lipolysis ( 58 ). The ability of excess LPL to correct the hypertriglyceridemia induced by apoE3[K146N/R147W] observed in this study, suggests that the activity of LPL, rather than apoCII, may be rate-limiting for the clearance of VLDL/ IDL/LDL triglycerides. Consistently with our fi nding, a previous study also showed that defective recognition of apoE-containing lipoproteins, either due to mutations in apoE or defi ciency in the LDL receptor, may increase the sensitivity and severity of hypertriglyceridemia ( 1 ). The same study also showed that the LDL receptor alone can account for the clearance of apoE-containing lipoprotein remnants in mice, and that the contribution of the other members of the LDL-receptor family and HSPGs may be limited ( 1 ).

Potential etiology of the apoE3[K146N/R147W]induced defective maturation of HDL that results in the accumulation of discoidal HDL particles in plasma
A common property of the full-length as well as of the truncated apoE3[K146N/R147W]202 mutant is that they inhibit the esterifi cation of cholesterol, as evidenced by the decrease of the CE/TC ratio of plasma, and promote the formation of discoidal HDL particles. The LCAT treatment corrected the hypertriglyceridemia, normalized the plasma CE/TC ratio, restored the apoA-I in the HDL fractions, and generated spherical HDL particles, but did not correct the hypercholesterolemia. The insuffi ciency in LCAT may be the result of fast catabolism of LCAT bound to discoidal HDL particles that are formed in the plasma of mice expressing apoE3[K146N/R147W] by the kidney. Accelerated catabolism of discoidal particles and reduction in LCAT activity has been observed due to apoA-I mutations ( 59,60 ). that if the cholesterol clearance defect was the result of unfavorable N-and C-terminal domain interactions that shield the receptor binding domain, then deletion of the C-terminal domain would be expected to correct the cholesterol clearance defect.
The present study showed that similar to the full-length apoE3[K146N/R147W] mutant protein, high levels of expression of the truncated apoE3[K146N/R147W]202 form further increased the high cholesterol levels of apoE Ϫ / Ϫ mice, despite the fact that it did not induce hypertriglyceridemia. This property of the truncated apoE3[K146N/ R147W]202 differentiated this mutant from all previously studied truncated apoE forms, which cleared cholesterol without induction of hypertriglyceridemia ( 34,35,68 ). These fi ndings indicate that the amino terminal 1-202 domain of the mutant protein does not inhibit lipolysis, but is not capable of clearing the high cholesterol levels of the apoE Ϫ / Ϫ mice .
The fact that gene transfer of the full-length apoE3[K146N/R147W] and the truncated apoE3[K146N/ R147W]202 form increased the plasma cholesterol levels of the apoE Ϫ / Ϫ or apoA-I Ϫ / Ϫ ×apoE Ϫ / Ϫ mice, indicates that both proteins act as dominant negative ligands that prevent receptor-mediated remnant clearance. The fact that the defective clearance of lipoprotein remnants persists following treatment with LPL or LCAT indicates that the apoE mutations are the primary cause of the dyslipidemia and the insuffi ciency of LPL and LCAT .
Taken together with previous studies ( 34,35,68 ), our data indicate that the dyslipidemia induced by the apoE3 [K146N/R147W] in mice, and possibly human patients carrying these mutations, most likely is brought about due to the ability of this mutant to act as a dominant negative ligand that, upon binding, blocks the LDL receptor and prevents the catabolism of triglyceride-rich apoEcontaining lipoprotein particles. The accumulation of apoE in VLDL/LDL/IDL appears to inhibit the functions of LPL and causes hypertriglyceridemia. It also inhibits the functions of LCAT and causes accumulation of discoidal HDL particles in plasma. The hypertriglyceridemia can be corrected by coexpression of apoE3[K146N/ R147W] with LPL or LCAT. However, the HDL maturation defect, which is demonstrated by the accumulation of discoidal particles, persists following LPL treatment and can only be corrected by excess LCAT. This indicates that the apoE3[K146N/R147W] has a direct effect on the activity of LCAT that is independent of the lipid abnormalities caused by this mutant.
Lipid-free and lipid bound apoE can interact with various receptors and membrane proteins that may trigger signaling cascades (19)(20)(21). It is possible that the severe proatherogenic phenotype generated by the apoE3[K146N/ R147W] mutant may result partially from an aberrant signaling mechanism following binding of this mutant to different apoE receptors. Future studies are required to address this question.