Discrete roles of apoA‐I and apoE in the biogenesis of HDL species: Lessons learned from gene transfer studies in different mouse models

Abstract

Using adenovirus‐mediated gene transfer in apolipoprotein A‐I (apoA‐I)‐deficient mice, we have established that apoA‐I mutations inhibit discrete steps in a pathway that leads to the biogenesis and remodeling of high‐density lipoprotein (HDL).

To this point, five discrete categories of apoA‐I mutants have been characterized that may affect the interactions of apoA‐I with ATP‐binding cassette superfamily A, member 1 (ABCA1) or lecithin:cholesterol acyl transferase (LCAT) or may influence the plasma phospholipid transfer protein activity or may cause various forms of dyslipidemia.

Biogenesis of HDL is not a unique property of apoA‐I. Using adenovirus‐mediated gene transfer of apoE in apoA‐I‐ or ABCA1‐deficient mice, we have established that apolipoprotein E (apoE) also participates in a novel pathway of biogenesis of apoE‐containing HDL particles. This process requires the functions of the ABCA1 lipid transporter and LCAT, and it is promoted by substitution of hydrophobic residues in the 261 to 269 region of apoE by Ala.

The apoE‐containing HDL particles formed in the circulation may have atheroprotective properties. ApoE‐containing HDL may also have important biological functions in the brain that confer protection from Alzheimer's disease.

• ABCA1
• apolipoprotein A‐I
• apolipoprotein E
• HDL biogenesis
• hypoalphalipoproteinemia
• LCAT

Abbreviations
ABCA1	=	ATP‐binding cassette superfamily A, member 1
ABCA1−/−	=	ABCA1‐deficient
apoA‐I	=	apolipoprotein A‐I
apoA‐I−/−	=	apoA‐I‐deficient mice
apoE	=	apolipoprotein E
apoE−/−	=	apoE‐deficient
ATP	=	adenosine triphosphate
cAMP	=	cyclic adenosine monophosphate
CE	=	cholesteryl esters
CETP	=	cholesteryl ester transfer protein
EM	=	electron microscopy
FPLC	=	fast pressure liquid chromatography
GFP	=	green fluorescence protein
HDL	=	high‐density lipoprotein
IDL	=	intermediate‐density lipoprotein
LCAT	=	lecithin:cholesterol acyl transferase
LDL	=	low‐density lipoprotein
pfu	=	plaque‐forming unit
PLTP	=	phospholipid transfer protein
SR‐BI	=	scavenger receptor class B type I
SR‐BI−/−	=	SR‐BI‐deficient
TC	=	total cholesterol
VLDL	=	very‐low‐density lipoprotein
WT	=	wild type

Introduction

ApoA‐I participates in several steps in the biogenesis and catabolism of HDL

High‐density lipoprotein (HDL) is synthesized and catabolized through a complex pathway shown in Figure 1A,B, 1. HDL assembly is initiated by an ATP‐binding cassette superfamily A, member 1 (ABCA1)‐mediated transfer of cellular phospholipids and cholesterol to extracellular lipid‐poor apolipoprotein A‐I (apoA‐I). Through a series of intermediate steps that are not fully understood, lipidated apoA‐I is gradually converted to discoidal particles and subsequently to spherical α‐HDL particles by the action of lecithin:cholesterol acyl transferase (LCAT). Under normal conditions human plasma contains 3%–5% preβ‐HDL particles and 95%–97% α‐HDL particles 2, 3, whereas discoidal HDL is not detectable in normal human plasma. Following synthesis, HDL interacts with the HDL receptor scavenger receptor class B type I (SR‐BI), and this interaction mediates selective uptake of cholesterol esters and other lipids by the cell 4, 5, as well as bidirectional movement of unesterified cholesterol 6–9. Cellular cholesterol may also be effluxed to HDL by the cell surface transporter ABCG1 10. Finally, hydrolysis of lipids of HDL is mediated by various lipases (lipoprotein lipase, hepatic lipase, endothelial lipase), and exchange of lipids by the phospholipid transfer protein (PLTP) and by the cholesteryl ester transfer protein (CETP). Lipid‐free apoA‐I secreted by the liver, intestine, and few other tissues 11 may also interact with ABCA1 present in peripheral cells and tissues and thus contribute to lipid efflux and potentially to HDL biogenesis.

Figure 1 The pathway of biogenesis(A) and catabolism (B) of HDL. Numbers 1–11 indicate key cell membrane or plasma proteins shown to influence HDL levels or composition. They are: 1) apolipoprotein A‐I; 2) ATP binding cassette transporter A1; 3) lecithin:cholesterol acyl transferase; 4) cholesteryl ester transfer protein; 5) hepatic lipase; 6) endothelial lipase; 7) phospholipid transfer protein; 8) lipoprotein lipase; 9) scavenger receptor class B type I; 10) LDL receptor; 11) ATP binding cassette transporter G1 (ABCG1). The figure is modified from reference 1. C: Secondary structure of apoA‐I based on high‐resolution X‐ray crystallography 58. The domains of apoA‐I that affect its various biological functions are shown in colors as follows. Pink and blue colors indicate regions that affect interactions of apoA‐I with ABCA1 and LCAT, respectively. Red indicates the region associated with inhibition of PLTP activity and dyslipidemia. Green (and green asterisks) indicates regions and amino acids associated with hypertriglyceridemia. Gray asterisks indicate amino acids that are involved in interactions of apoA‐I with SR‐BI.

To probe the pathway of biogenesis and catabolism of HDL, we have used adenovirus‐mediated gene transfer of apoA‐I mutants in apoA‐I‐deficient and other mouse models. The expectation was that these apoA‐I mutants will inhibit discrete steps of the HDL pathway and either lead to rapid catabolism or accumulation of intermediate products. The first part of this review will provide examples of characteristic mutations of apoA‐I (Figure 1C) that affect different steps of the HDL biogenesis and catabolism 12–17. The second part will describe a novel pathway of biogenesis of apolipoprotein E (apoE)‐containing HDL and how certain apoE mutations promote formation of spherical apoE‐containing HDL 18–20.

Search for mutations in apoA‐I with defective ABCA1‐mediated cholesterol efflux that may prevent the biogenesis of HDL

To explore the mechanism of ABCA1‐mediated efflux of cellular lipid to apoA‐I, we expressed in vitro wild‐type (WT) apoA‐I and numerous variants carrying point mutations or deletions. The apoA‐I mutants thus generated were used as lipid acceptors either in cultured cells expressing endogenous ABCA1 (J774 macrophages activated by a cyclic adenosine monophosphate (cAMP) analog), or in HEK293 cells transfected with a cDNA encoding ABCA1, and were used for in vitro ABCA1‐mediated lipid efflux and apoA‐I/ABCA1 chemical cross‐linking experiments. The questions addressed were: whether apoA‐I interacts directly or indirectly with ABCA1 to promote cholesterol and phospholipid efflux; which domains of apoA‐I are required for ABCA1‐dependent lipid efflux; and how lipid efflux correlates with apoA‐I/ABCA1 cross‐linking.

These in vitro studies showed that ABCA1‐mediated cholesterol and phospholipid efflux was moderately decreased by amino‐terminal deletions and was diminished by carboxy‐terminal deletions in which residues 220–231 were removed (Figure 2A; Table I) 12. Chemical cross‐linking/immunoprecipitation studies showed that the relative ability of apoA‐I mutants to promote ABCA1‐mediated lipid efflux correlated with the ability of these mutants to be cross‐linked efficiently to ABCA1 21.

Table I. In vivo phenotypes of mice expressing apoA‐I deletion and point mutants and changes in the in vitro functions of these apoA‐I mutants 12–17, 59, 60.

Mutation	In vivo phenotypes ^a						Changes in the in vitro functions ^b			HDL subpopulations
	Plasma total cholesterol (x‐fold)	Plasma triglycerides (x‐fold)	VLDL lipolysis	Plasma PLTP activity (%) ^d	Particles in the HDL region	CE/TC of HDL region (%) ^e	Activation of LCAT ^f	ABCA1‐mediated cholesterol efflux	SR‐BI‐mediated cholesterol efflux
ApoA‐I[Δ(1–41)]	69% of normal	61% of normal	– ^c	–	spherical	–	–	86%	–	–
ApoA‐I[Δ(7–43)] ^g	82% of normal	–	–	–	spherical	76	27%	91%	–	–
ApoA‐I[Δ(7–65)] ^g	66% of normal	–	–	–	discoidal	39	43%	97%	–	–
ApoA‐I[Δ(1–41)Δ(185–243)]	46% of normal	19% of normal	–	–	discoidal	–	–	80%	–	–
ApoA‐I[Asp102Ala/Asp103Ala]	94% of normal	62% of normal	–	–	spherical	103	50%	96%	21%	–
ApoA‐I[Δ(62–78)]	2‐fold increase	9.5‐fold increase	decreased	92	spherical	90	47%	101%	96%	normal
ApoA‐I[Glu110Ala/Glu111Ala]	1.9‐fold increase	6.9‐fold increase	decreased	–	spherical	–	37%	107%	53%	normal
ApoA‐I[Δ(89–99)]	3.2‐fold increase	normal	–	32	discoidal	38	77%	68%	78%	increased preβ1‐/α‐HDL ratio
ApoA‐I[Arg160Val/His162Ala]	55% of normal	72% of normal	–	–	discoidal	44	0.7%	93%	49%	near‐absence of α1, α2, α3 subpopulations
ApoA‐I[Δ(144–165)]	25% of normal	67% of normal	–	–	few spherical	62	0.4%	71%	116%	significant reduction in the α1, α2, α3 subpopulations
ApoA‐I[Leu141Arg]Pisa	16% of normal	17% of normal	–	–	very few spherical	40	0.4%	86%	108%	absence of α1, α2, α3 subpopulations, accumulation of α4 and preβ particles
ApoA‐I[Leu159Arg]Fin	11% of normal	40% of normal	–	–	very few spherical	2.5	2.1%	113%	–	absence of α1, α2, α3 subpopulations, accumulation of α4, and preβ particles
ApoA‐I[Arg151Cys]Paris	39% of normal	13% of normal	–	–	spherical and few discoidal	76	28%	94%	–	abnormal preβ particles with electrophoretic mobility between preβ and α
ApoA‐I[Arg160Leu]Oslo	23% of normal	47% of normal	–	–	discoidal	7	2.7%	105%	–	significant reduction in the α1, α2, α3 subpopulations, accumulation of α4 particles
ApoA‐I[Arg149Ala]	41% of normal	62% of normal	–	–	discoidal	21	5.2%	98%	–	reduction in the α1, α2, α3 subpopulations, accumulation of α4 particles
ApoA‐I[Glu191Ala/His193Ala/Lys195Ala]	97% of normal	89% of normal	–	–	spherical	100	107%	92%	–	normal
ApoA‐I[Δ(143–164)] Transgenic mice ^h	29% of normal	–	–	–	–	59	4%	–	–	–

^a Parameters obtained from mice expressing the apoA‐I deletion and point mutants as compared to mice expressing the WT apoA‐I.

^b Parameters determined by in vitro assays for the apoA‐I mutants compared to the same parameters for the WT apoA‐I.

^c apparent maxium velocity/apparent MichaelisMenten constant

^d PLTP activity relative to the PLTP activity from plasma of mice expressing WT apoA‐I.

^e CE/TC of particles in the HDL region relative to the CE/TC of the HDL from mice expressing WT apoA‐I.

^f Refers to catalytic efficiency, apparent maximum velocity/apparent Michaelis‐Menten constant (Vmax_app/Km_app) of LCAT with rHDL particles containing mutant apoA‐I forms as compared to rHDL particles containing WT apoA‐I.

^g Reference 59.

^h Reference 60.

Figure 2 A: cAMP‐dependent (ABCA1‐mediated) cholesterol efflux in J774 mouse macrophages. The experiment was performed as described 12. B: FPLC profile of total plasma cholesterol of apoA‐I^−/− mice expressing the WT apoA‐I, or the carboxy‐terminal mutants apoA‐I[Δ(185–243)], apoA‐I[Δ(220–243)], apoA‐I[Δ(232–243)], apoA‐I[Glu191Ala/His193Ala/Lys195Ala], or the control protein GFP. Plasma samples were obtained from mice infected with 1×10⁹ pfu of the recombinant adenoviruses expressing the WT or mutant forms of apoA‐I or the control protein GFP 4 days postinfection. The experiment was performed as described 16. C: Electron microscopy pictures of the HDL fractions obtained by density gradient ultracentrifugation from the plasma of apoA‐I^−/− mice expressing the WT or mutant forms of apoA‐I or the control protein GFP. The experiment was performed as described 16. D: Two‐dimensional gel electrophoresis and Western blotting analysis of plasma of apoA‐I^−/− mice expressing the WT, the point mutant or the truncated mutant forms of apoA‐I described above, or the control protein GFP. HDL subpopulations containing apoA‐I were detected using antihuman apoA‐I antibodies 16.

A functional relationship between apoA‐I/ABCA1 cross‐linking and lipid efflux was also seen in the concurrent inhibition of both apoA‐I cross‐linking to ABCA1 and lipid efflux by two small molecule inhibitors: block lipid transport‐4 (BLT4) and (1‐(2‐methoxy‐phenyl)‐3‐naphthalen‐2‐yl‐urea) (glyburide) 22. All these in vitro data are consistent with a lipid efflux mechanism that involves direct interaction between apoA‐I and ABCA1, although other earlier studies suggested that protein‐protein interactions may not be required for lipid efflux 23.

In the in vivo studies, apoA‐I‐deficient (apoA‐I^−/−) mice were infected with 1 to 2×10⁹ plaque‐forming units (pfu) of an adenovirus expressing WT apoA‐I or the various apoA‐I mutants that were previously studied for ABCA1‐mediated lipid efflux and ABCA1 cross‐linking. Four to five days following gene transfer, lipid analyses, gel filtration chromatography, and two‐dimensional gel electrophoretic analyses were performed on the plasma of the apoA‐I^−/− mice expressing the WT or mutant apoA‐I forms 12, 16, 24. Hepatic apoA‐I mRNA levels were also determined to ensure comparable expression of the mutant genes. These experiments showed that the WT apoA‐I, the apoA‐I[Δ(232–243)], and the apoA‐I[Glu191A/His193A/Lys195A] generated HDL in vivo as determined by fast pressure liquid chromatography (FPLC) fractionation (Figure 2B) 16. Electron microscopy (EM) analysis showed that the HDL consisted of spherical particles (Figure 2C I–III; Table I) 16. In contrast, apoA‐I[Δ(185–243)] and apoA‐I[Δ(220–243)] generated few HDL particles that were similar in numbers with the particles found in the plasma of apoA‐I^−/− mice that express the green fluorescence protein (GFP) (Figure 2C IV–VI) 16, 24. Two‐dimensional gel electrophoresis showed that the majority of HDL derived from apoA‐I[Δ(232–243)] and apoA‐I[Glu191Ala/His193Ala/Lys195Ala] consisted of particles with α electrophoretic mobility and a small fraction consisted of particles with β electrophoretic mobility (Figure 2D I–III) 16. In contrast, the carboxy‐terminal deletion mutants apoA‐I[Δ(185–243)] and apoA‐I[Δ(220–243)] formed very little HDL, which in the case of apoA‐I[Δ(185–243)] contained exclusively preβ1‐HDL particles (Figure 2D IV) and in the case of apoA‐I[Δ(220–243)] contained preβ‐HDL and a few α‐HDL particles (Figure 2D V) 16. These findings suggested that carboxy‐terminal deletion mutants of apoA‐I that lacked the 220–231 region prevent the lipidation of apoA‐I, thus blocking the first step in the biogenesis of HDL 25. The combined in vitro and in vivo findings suggest that apoA‐I mutants that lack residues 220–231 are defective in the ABCA1‐mediated efflux of cellular phospholipids and cholesterol, cross‐link poorly to ABCA1, and fail to form discoidal or spherical HDL in vivo but form exclusively or predominantly preβ‐HDL particles. Furthermore, the apoA‐I[Δ(232–243)] mutant that retains the 220–231 region and the point mutant in the 191–195 region tested, promote normally the biogenesis of HDL.

Low HDL syndromes caused by apoA‐I mutations can be corrected by treatment with LCAT

To explain the etiology and find a mode of therapy for some forms of genetically determined low levels of HDL, we have generated recombinant adenoviruses expressing naturally occurring apoA‐I mutations and two bioengineered mutants and studied their properties in vitro and in vivo.

Two of the natural mutants studied, apoA‐I[Leu141Arg]_Pisa26–29 and apoA‐I[Leu159Arg]_Fin30–32, are associated with very low HDL and apoA‐I levels in humans 26, 28, 29, 31 and premature atherosclerosis 26, 32. The other two naturally occurring mutants, apoA‐I[Arg151Cys]_Paris33, 34 and apoA‐I[Arg160Leu]_Oslo35, are also associated with low HDL levels in humans but present milder phenotypes. The bioengineered mutants apoA‐I[Arg149Ala] and apoA‐I[Arg160Val/His162Ala] were selected because the substitution of the positively charged Arg by Ala at positions 149 and 160 reduces the positive electrostatic potential that exists in the 149–160 region of apoA‐I and diminishes the ability of the mutant forms to activate LCAT in vitro36.

Our studies showed that all mutants were secreted efficiently from cells, had diminished capacity to activate LCAT, and near‐normal capacity to promote ABCA1‐mediated cholesterol efflux in vitro (Table I).

Gene transfer of the apoA‐I[Arg160Leu]_Oslo, apoA‐I[Arg149Ala], apoA‐I[Arg151Cys]_Paris, and apoA‐I[Arg160Val/His162Ala] mutants in apoA‐I^−/− mice generated aberrant HDL phenotypes 15, 17. The total plasma cholesterol was significantly reduced (23%–55% as compared to mice expressing WT apoA‐I) (Table I). Plasma apoA‐I and HDL levels were decreased significantly compared to WT apoA‐I (Figure 3A I–IV; Table I) 15, 17. Electron microscopy of the HDL obtained from mice expressing the various apoA‐I mutants showed that apoA‐I[Arg151Cys]_Paris formed mostly spherical and few discoidal particles (Figure 3B III), apoA‐I[Arg160Leu]_Oslo and apoA‐I[Arg149Ala] formed a mixture of discoidal and spherical HDL particles (Figure 3B II and IV), and apoA‐I[Arg160Val/His162Ala] formed discoidal particles (Figure 3B V) 15, 17. The WT apoA‐I formed spherical HDL, whereas apoA‐I^−/− mice expressing GFP formed few particles (Figure 3B I and VI). Two‐dimensional gel electrophoresis of plasma obtained from mice expressing the WT apoA‐I and the various apoA‐I mutants showed that apoA‐I[Arg160Leu]_Oslo, apoA‐I[Arg149Ala] and apoA‐I[Arg160Val/His162Ala] promoted the formation of preβ1 and α4‐HDL subpopulations (Figure 3C II, IV, and V) 15, 17. ApoA‐I[Arg151Cys]_Paris generated subpopulations of different sizes that migrate between preβ and α‐HDL (Figure 3C III) 17. The WT apoA‐I formed predominantly α‐HDL and small amounts of preβ‐HDL particles (Figure 3C I). Simultaneous treatment of mice with adenoviruses expressing any of the four mutants and human LCAT normalized the plasma apoA‐I levels as well as the HDL cholesterol levels (Figure 3A I–IV), promoted the formation of spherical HDL particles (Figure 3B VII–X), and also promoted the formation of mostly α‐HDL subpopulations of larger size (Figure 3C VI–IX) 15, 17. The LCAT treatment also increased the cholesteryl ester (CE)/total cholesterol (TC) ratio of HDL from 0.3 (without treatment) to 0.8 following treatment 15, 17.

Figure 3 A:I‐IV. FPLC profiles of total plasma cholesterol of apoA‐I mice expressing apoA‐I[Arg149Ala], apoA‐I[Arg151Cys]_Paris, apoA‐I[Arg160Leu]_Oslo and apoA‐I[Arg160Val/His162Ala] in the presence and absence of LCAT as indicated in the Figure. The FPLC profile of mice expressing WT apoA‐I is also shown Panels AI‐IV. The experiments were performed as described 15, 17. B: Electron microscopy pictures of the HDL fractions obtained by density gradient ultracentrifugation from the plasma of apoA‐I^−/− mice expressing the WT or mutant forms of apoA‐I or the control protein GFP or the same apoA‐I forms along with LCAT as indicated in Panels BI‐X. Experiments were performed as described 15, 17. C: Two‐dimensional gel electrophoresis and Western blot analysis of plasma of apoA‐I^−/− mice expressing the WT or mutant forms of apoA‐I alone or in the presence of LCAT as indicated in Panels CI‐IX. HDL subpopulations containing apoA‐I were detected using antihuman apoA‐I antibodies. Experiments were performed as described 15, 17.

The in vivo analysis of the apoA‐I[Leu141Arg]_Pisa and apoA‐I[Leu159Arg]_Fin mutants in apoA‐I^−/−mice indicate that apoA‐I[Leu141Arg]_Pisa and apoA‐I[Leu159Arg]_Fin mutations appear to inhibit an early step in the biogenesis of HDL due to inefficient esterification of the cholesterol of the preβ1‐HDL particles by the endogenous LCAT (27,32,37). The LCAT insufficiency appears to result from depletion of the plasma LCAT mass 37. The defects in apoA‐I appear to promote rapid catabolism of newly lipidated apoA‐I as well as plasma LCAT and thus prevent the formation of either discoidal or spherical HDL particles 37. In contrast, the apoA‐I[Arg151Cys]_Paris, apoA‐I[Arg160Leu]_Oslo, and apoA‐I[Arg149Ala] allow formation of discoidal particles but, due to insufficiency of the endogenous LCAT, prevent the conversion of discoidal to spherical HDL 17. The correction of the aberrant HDL phenotypes for all mutants studied here by treatment with LCAT suggests a potential therapeutic intervention to correct HDL abnormalities that result from specific mutations in apoA‐I.

Mutations in apoA‐I may induce hypertriglyceridemia and other dyslipidemias

We have identified three mutations of apoA‐I that have a profound effect on the overall cholesterol and triglyceride homeostasis 13, 14. Two of these, apoA‐I[Δ(62–78)] and apoA‐I[Glu110Ala/Glu111Ala], caused combined hyperlipidemia, characterized by high plasma cholesterol and severe hypertriglyceridemia 13, 14. All the triglycerides and the majority of the excess cholesterol were distributed in very‐low‐density lipoprotein (VLDL)/intermediate‐density lipoprotein (IDL)‐sized lipoproteins. Density gradient ultracentrifugation showed that this VLDL/IDL fraction was enriched in apoA‐I (Figure 4A). Also the VLDL/IDL fraction had decreased levels of apoE and apoCII and increased levels of apoB‐48 (Figure 4B) 13, 14. The presence of apoA‐I and the depletion of apoCII in the VLDL/IDL fraction raised the possibility that the mutations were associated with inhibition of lipolysis in vivo. Coinfection of mice with adenoviruses expressing apoA‐I[Glu110Ala/Glu111Ala] and the human lipoprotein lipase reduced plasma and VLDL triglyceride levels but had a smaller effect on plasma cholesterol levels.

Figure 4 A: Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS‐PAGE) analysis of density gradient ultracentrifugation fractions of plasma obtained from apoA‐I^−/− mice 4 days postinfection with adenoviruses expressing WT apoA‐I or the apoA‐I[Glu110Ala/Glu111Ala] mutant. The densities of the fraction are indicated on the top of the figure. ApoA‐I was detected by Western blotting using polyclonal antibody to human apoA‐I 13. B: Assessment of the apoprotein composition of VLDL/LDL isolated from plasma of mice infected with WT apoA‐I and apoA‐I[Glu110Ala/Glu111Ala] mutant. The apoproteins were detected by Western blotting using polyclonal antibodies to mouse apoB, apoE and apoCII 13. C: SDS‐PAGE analysis of density gradient ultracentrifugation fractions of plasma of apoA‐I^−/− mice expressing the WT apoA‐I or the apoA‐I[Δ(89–99)] mutant. ApoA‐I was detected by staining with Coomassie Brilliant Blue. On the right side of the panel is shown the CE/TC ratio from a pool of lipoprotein fractions that correspond to the HDL region. Experiments were performed as described 14. D, E: Electron microscopy pictures of HDL fractions obtained from apoA‐I^−/− mice expressing the WT apoA‐I (D) or the apoA‐I[Δ(89–99)] mutant (E) following density gradient ultracentrifugation of plasma. The photomicrographs were taken at 75,000× magnification and enlarged three times. The experiment was performed as described 14. F, G: Analysis of plasma obtained from mice expressing the WT apoA‐I (F) or the apoA‐I[Δ(89–99)] mutant (G) following two‐dimensional gel electrophoresis and Western blotting. The experiment was performed as described 14.

The third mutation, apoA‐I[Δ(89–99)] 14, caused (relative to WT apoA‐I) hypercholesterolemia characterized by increased cholesterol and phospholipids in the VLDL/IDL/low‐density lipoprotein (LDL) size lipoproteins, normal triglyceride levels, and a substantially decreased CE/TC ratio in HDL and LDL (Figure 4C). The Δ(89–99) deletion also increased apoA‐I in the LDL, caused accumulation of discoidal HDL (compare Figure 4E with Figure 4D), and increased the levels of preβ1‐ relative to the α‐HDL subpopulation particles (compare Figure 4G with Figure 4F) 14. Strikingly, mice expressing the apoA‐I[Δ(89–99)] mutant exhibited PLTP activity that was only 32% of that of mice expressing the WT apoA‐I 14. It has been proposed that PLTP functions by linking the donor and acceptor lipoprotein particles thus facilitating the net transfer of phospholipids to HDL 38. Previous studies also have suggested that apoA‐I interacts physically with PLTP 39 and that this interaction may underlie the transfer of the phospholipids from the donor molecule to HDL.

Figure 5 shows the normal pathway of biogenesis and catabolism of HDL and various steps where the pathway can be affected and lead to dyslipidemia. This includes: 1) lack of synthesis of HDL due to mutations in ABCA1 or apoA‐I that prevent apoA‐I/ABCA1 interactions; 2) failure to synthesize discoidal or spherical HDL due to fast catabolism of minimally lipidated apoA‐I; 3) induction of hypertriglyceridemia by the apoA‐I[Δ(62–78)] and apoA‐I[Glu110Ala/Glu111Ala] mutants; 4) accumulation of discoidal HDL, inhibition of PLTP, and the induction of hypercholesterolemia by the apoA‐[Δ(89–99)] mutant; and 5) the accumulation of discoidal HDL due to the mutations in the 149–160 region of apoA‐I. The phenotypes produced by targeted mutagenesis of apoA‐I will be valuable for the detection of similar phenotypes in humans and can serve as diagnostic and prognostic markers of dyslipidemia and/or atherosclerosis.

Figure 5 Schematic representation of the defects in the HDL pathway that can account for 1) lack of HDL synthesis due to mutation in ABCA1 or carboxy‐terminal apoA‐I deletions that affect cholesterol efflux; 2) hypertriglyceridemia induced by the apoA‐I[Δ(62–78)] and apoA‐I[Glu110Ala/Glu111Ala] mutants; 3) the high plasma cholesterol levels, inhibition of PLTP activity, and accumulation of discoidal HDL particles that is induced by the apoA‐I[Δ(89–99)] mutant; and 4) accumulation of discoidal HDL due to LCAT deficiency or apoA‐I mutations that cause depletion of plasma LCAT and either prevent formation of discoidal and spherical HDL (apoA‐I[Leu141Arg]_PISA, apoA‐I[Leu159Arg]_Fin) or cause accumulation of discoidal HDL (apoA‐I[Arg149Ala], apoA‐I[Arg151Cys]_Paris, apoA‐I[Arg160Leu]_Oslo and apoA‐I[Arg160Val/His162Ala)]). PL = phospholipids; C = cholesterol; TC = total cholesterol; CE = cholesteryl esters.

Key messages

• Apolipoprotein A‐I (apoA‐I) mutations inhibit discrete steps in the pathway of biogenesis and remodeling of high‐density lipoprotein (HDL).

• The aberrant HDL phenotypes caused by naturally occurring apoA‐I mutations can be corrected by administration of lecithin:cholesterol acyl transferase (LCAT).

• Apolipoprotein E (apoE) along with ATP‐binding cassette superfamily A, member 1 (ABCA1) and LCAT participate in the biogenesis of apoE‐containing HDL.

Novel pathway of biogenesis of apoE‐containing HDL

Using adenovirus‐mediated gene transfer in apoA‐I‐ or ABCA1‐deficient mice, we obtained unequivocal evidence that apoE participates in a novel pathway of biogenesis of apoE‐containing HDL particles that also requires the functions of the ABCA1 lipid transporter and LCAT 18. Infection of apoA‐I^−/− mice with 2×10⁹ pfu of an apoE4‐expressing adenovirus increased both HDL and the triglyceride‐rich VLDL/IDL/LDL fraction and generated discoidal HDL particles (compare Figure 6B with Figure 6A). ABCA1‐deficient (ABCA1^−/−) mice treated similarly failed to form HDL particles (Figure 6C, D), suggesting that ABCA1 is essential for the biogenesis of apoE‐containing HDL. Combined infection of apoA‐I^−/− mice with a mixture of adenoviruses expressing both apoE4 (2×10⁹ pfu) and human LCAT (5×10⁸ pfu) cleared the triglyceride‐rich lipoproteins, increased HDL, and converted the discoidal to spherical HDL (Figure 6E). Similar treatment of ABCA1^−/− mice failed to promote formation of HDL, suggesting that LCAT is essential for the maturation of apoE‐containing HDL 18. Overall, the findings indicate that apoE has a dual functionality. In addition to its documented functions in the clearance of triglyceride‐rich lipoproteins, it participates in the biogenesis of apoE‐containing HDL in a process that is similar to that of apoA‐I (Figure 7).

Figure 6 Electron microscopy pictures of the fractions corresponding to the HDL region obtained from apoA‐I^−/− mice (A), apoA‐I^−/− mice expressing apoE4 (B), ABCA1^−/− mice (C), ABCA1^−/− mice expressing apoE4 (D), and apoA‐I^−/− mice expressing a combination of apoE4 and LCAT (E). Following density gradient ultracentrifugation, the fractions that float to the HDL region were obtained and analyzed by electron microscopy as described 18.

Figure 7 Diagram showing the participation of apoE in the chylomicron pathway(branch I), the participation of apoE in the biogenesis of apoE‐containing HDL (branch II), and the participation of apoA‐I in the biogenesis of apoA‐I‐containing HDL (branch III).

ApoE‐containing HDL may have antioxidant and anti‐inflammatory functions similar to those described for apoA‐I‐containing HDL, which may contribute to the atheroprotective properties of apoE 40–43. ApoE‐containing HDL may also have important biological functions in the brain 44.

ApoE mutants in the 261–269 region promote the formation of spherical apoE‐containing HDL

Overexpression of apolipoprotein E (apoE) induces hypertriglyceridemia in C57BL/6 or apoE‐deficient mice, which is abrogated by deletion of the carboxy‐terminal segment 260–299 1, 19.

We have used adenovirus‐mediated gene transfer in apoA‐I^−/− mice to test the effect of four sets of apoE mutations within the 261–269 region on the formation of spherical or discoidal apoE‐containing HDL as well as the induction of hypertriglyceridemia 19, 20.

A single amino acid substitution (apoE4[Phe265Ala]) induced hypertriglyceridemia in apoA‐I^−/− mice and was associated with moderate HDL cholesterol levels (Figure 8A I, B I). A double substitution (apoE4[Leu261Ala,Trp264Ala]) induced milder hypertriglyceridemia and was associated with increased HDL cholesterol levels (Figure 8A II, B II). A triple substitution (apoE4[Leu261Ala/Trp264Ala/Phe265Ala]) or five residue substitutions (apoE4[Leu261Ala/Trp264Ala/Phe265Ala/Leu268Ala/Val269Ala]) did not induce hypertriglyceridemia and both were associated with greatly increased HDL cholesterol levels (Figure 8A III, B III and A IV, B IV, respectively) 19, 20. EM analysis of the HDL fractions showed that apoE4[Leu261Ala/Trp264Ala/Phe265Ala] and the apoE4[Leu261Ala/Trp264Ala/Phe265Ala/Leu268Ala/Val269Ala] formed spherical HDL (Figure 8C III, IV), apoE4[Phe265Ala] formed discoidal HDL (Figure 8CI), and the apoE4[Leu261Ala/Trp264Ala] formed mostly spherical and few discoidal HDL particles (Figure 8C II) 20. The phenotype of apoE‐mutC carrying three amino acid substitions, in terms of lipid levels and HDL formation was similar in apoE4 as well as in apoE2 background 20.

Figure 8 Plasma cholesterol and FPLC profiles of apoA‐I^−/− mice infected with recombinant adenoviruses expressing WT apoE4, apoE4‐mutA (one amino acid substitution), apoE4‐mutB (two amino acids substitutions), apoE4‐mutC (three amino acids substitutions), or apoE4‐mut1 (five amino acids substitutions) as indicated on the top of the Figure. ApoA‐I^−/− mice were infected with 2×10⁹ pfu of recombinant adenovirus, and serum samples were isolated 4 days postinfection as described 20. A: total plasma cholesterol and triglycerides; B: cholesterol FPLC profiles; C: electron microscopy pictures of HDL.

These studies established that substitutions of Leu261, Trp264 and Phe265 by Ala in either the apoE4 or apoE2 background promoted formation of spherical apoE‐containing HDL and also prevented apoE‐induced hypertriglyceridemia. Bioengineered apoE variants that can promote formation of spherical apoE‐containing HDL may find therapeutic applications in the future for the prevention of atherosclerosis.

Functions of apoE‐containing HDL in the brain

ApoE is the only medium‐size apolipoprotein present in the brain 11, 45 and alone may contribute to lipid homeostasis in the brain 44. The interactions of apoE with ABCA1 in the brain promote cholesterol efflux and lead to the biogenesis of apoE‐containing HDL 25,44. Once formed, apoE‐containing HDL can interact with the HDL receptor SR‐BI, and this interaction leads to the uptake of cholesteryl esters and cholesterol efflux 46. ApoE‐containing HDL may further have antioxidant and anti‐inflammatory functions in the brain similar to those attributed to apoA‐I‐containing HDL 40–43. Existing evidence indicated that apoE‐containing HDL through receptor‐mediated processes may sequester Aβ away from the brain parenchyma and prevent amyloid deposition and plaque formation 47–53. In contrast, lipid‐free apoE that cannot bind to apoE‐recognizing receptors 54 may promote Aβ polymerization and plaque formation (Figure 9) 55–57.

Figure 9 Schematic representation of the pathway of biogenesis of apoE‐containing HDL in the brain. The picture depicts a potential role of apoE‐containing HDL in lipid homeostasis in the brain and the removal of the amyloid peptide β (Aβ). It also depicts a potential role of lipid‐free apoE in the Aβ polymerization and formation of amyloid plaques.

Materials and methods

Materials used and various assays, such as cholesterol efflux assays, cross‐linking of apoA‐I to ABCA1, lipid‐lipoprotein apoA‐I and apoE measurements have been described 12–21.

Acknowledgements

This work was supported by grants from the National Institutes of Health (HL48739 and HL68216), the 6th Framework Programme of the European Union (No. LSHM‐CT‐2006‐0376331), and a grant from the American Heart Association (SDG 0535443T). Alexander Vezeridis is a trainee of the Predoctoral Training Grant in Cardiovascular Biology (T32 HL07969).