Molecular mechanisms for ABCA1-mediated cholesterol efflux

ABSTRACT

The maintenance of cellular cholesterol homeostasis is essential for normal cell function and viability. Excessive cholesterol accumulation is detrimental to cells and serves as the molecular basis of many diseases, such as atherosclerosis, Alzheimer’s disease, and diabetes mellitus. The peripheral cells do not have the ability to degrade cholesterol. Cholesterol efflux is therefore the only pathway to eliminate excessive cholesterol from these cells. This process is predominantly mediated by ATP-binding cassette transporter A1 (ABCA1), an integral membrane protein. ABCA1 is known to transfer intracellular free cholesterol and phospholipids to apolipoprotein A-I (apoA-I) for generating nascent high-density lipoprotein (nHDL) particles. nHDL can accept more free cholesterol from peripheral cells. Free cholesterol is then converted to cholesteryl ester by lecithin:cholesterol acyltransferase to form mature HDL. HDL-bound cholesterol enters the liver for biliary secretion and fecal excretion. Although how cholesterol is transported by ABCA1 to apoA-I remains incompletely understood, nine models have been proposed to explain this effect. In this review, we focus on the current view of the mechanisms underlying ABCA1-mediated cholesterol efflux to provide an important framework for future investigation and lipid-lowering therapy.

KEYWORDS:

• ABCA1
• cholesterol
• phospholipids
• apoA-I
• nHDL

1. Introduction

All mammalian cells require cholesterol that is essential for normal cell function, from maintenance of membrane fluidity to biogenesis of bile acids, steroid hormones and vitamin D. Excessive cholesterol, however, is toxic to the cells. It is well established that cholesterol accumulation in macrophages and vascular smooth muscle cells (VSMCs) leads to foam cell formation and atherogenesis [1]. Atherosclerosis is the pathological basis of most cardiovascular disease (CVD), the leading cause of morbidity and mortality worldwide. A 25-year epidemiological study has demonstrated that CVD accounts for >17 million deaths globally every year, which is predicted to rise to >23 million by 2030 [2]. Excessive cholesterol is also associated with occurrence and development of Alzheimer’s disease and diabetes mellitus [3,4]. Thus, the maintenance of optimal cellular cholesterol concentration is necessary for human healthy.

In mammals, cells obtain cholesterol through endogenous synthesis and exogenous uptake. The liver is the only organ than can convert cholesterol to bile acids for biliary secretion and fecal excretion. However, the peripheral cells, particularly macrophages and VSMCs in the arterial wall, do not possess this capability to metabolize cholesterol. To maintain cholesterol homeostasis and prevent toxicity, these cells must release surplus cholesterol into extracellular acceptors such as high-density lipoprotein (HDL) for removal. ATP-binding cassette transporter A1 (ABCA1), a transmembrane protein, is highly expressed in the liver, intestine, brain, and macrophages. Its major function is to mediate the transport of free cholesterol (FC) and phospholipids from cells to apolipoprotein A-I (apoA-I) for generating nascent HDL (nHDL) particles [5]. Studies from our group and others have revealed that ABCA1-dependent cholesterol efflux plays a central role in blocking the transformation of macrophages and VSMCs into foam cells, the hallmark of early atherosclerotic lesions [6–10]. Despite the detailed molecular mechanisms by which ABCA1 transfers cholesterol to apoA-I are not fully understood, a variety of models have been proposed to explain this effect. In the present review, we summarize the current knowledge about the structure and functions of ABCA1 with an emphasis on how it mediates cholesterol efflux.

2. Structural features of ABCA1

Human ABCA1 gene has been mapped to chromosome 9q31.1, spans 149 kb, and comprises 50 exons and 49 introns [11]. The mature ABCA1 protein is composed of 2261 amino acid residues with a predicted molecular mass of 254 kDa. As an integral membrane protein, ABCA1 has two transmembrane domains (TMDs) linked covalently, each containing six transmembrane α-helices (TMs) [12]. Unique to the ABCA subfamily, ABCA1 has two large extracellular domains (ECDs) and multiple membrane-spanning segments (Figure 1)[13]. ECD1 resides between TM1 and TM2, and ECD2 resides between TM7 and TM8. They are connected by two intramolecular disulfide bonds and serve as the binding sites for apoA-I [14]. There are two nucleotide-binding domains (NBDs) in its intracellular region. Their amino acid sequences are highly conserved among ABC transporters. Each NBD contains two conserved peptide motifs known as Walker A and Walker B, and a Walker C signature unique to ABC proteins [15]. Walker A is responsible for ATP binding while Walker B is associated with subsequent hydrolysis [16]. It is worth noting that ATP can bind equally to both NBDs. The hydrolysis of ATP in these sites can alter ECD conformation, which is essential for apoA-I binding [17]. In the NBD1, the PEST (proline, glutamic acid, serine, and threonine) sequence is responsible for ABCA1 degradation by calpain [18]. However, whether Walker C signature is involved in nHDL biogenesis remains unclear and needs further characterization.

Figure 1. Topological diagram of ABCA1.

ABCA1 is a full transporter that is composed of two ECDs, multiple membrane-spanning segments, 12 TMs, and two NBDs. Each NBD contains Walker A, Walker B, and Walker C signature.

A recent study revealed the single-particle cryoelectron microscopy structure of human ABCA1 at 4.1 Å resolution [19]. The overall structure resembles an elongated torch that is about 200 Å high. Both ECDs are the upward flaring flame, and other domains constitute the handle. There is a shallow pocket enclosed by the intracellular segments of TMs 1/2/5. The amino acid residues within these TMs are mostly polar and charged, which are responsible for binding to the polar heads of phospholipid molecules. Both TMD1 and TMD2 constitute a narrow chamber as an intracellular gate to move lipids from the inner leaflet to the outer leaflet of the cellular membrane. The helical domains in ECD1 and ECD2 together enclose an elongated hydrophobic tunnel that acts as a temporary storage and delivery passage for lipids. These structural features provide an important framework for better understanding ABCA1-mediated cholesterol efflux.

3. Biological roles of ABCA1

Cholesterol efflux is the first and rate-limiting step of reverse cholesterol transport (RCT), a process by which peripheral cholesterol is transferred by HDL to the liver for excretion into the bile and eventually feces [20,21]. Currently, four efflux pathways have been identified in the arterial wall cells [22]. Two passive processes involve simple diffusion via the aqueous phase and facilitated diffusion mediated by scavenger receptor class B type I (SR-BI). Cholesterol flux is bidirectional and driven by FC concentration gradient between the cellular membrane and HDL particles [23,24]. When FC amount in the cellular membrane prevails over that in HDL particles, there is net cholesterol removal from cells. Despite these passive pathways occur ubiquitously, they are relatively inefficient. In contrast, two active processes mediated by ABCA1 and ATP-binding cassette transporter G1 (ABCG1) are thought to play a major role in exporting intracellular cholesterol. ABCA1 mediates initial transport of FC and phospholipid molecules to apoA-I for nHDL biogenesis, while ABCG1 facilitates subsequent continued FC efflux to these nHDL particles for further maturation [25]. In cholesterol-loaded mouse peritoneal macrophages incubated with diluted human serum, ABCA1 and ABCG1 are responsible for 50 and 20% of cholesterol efflux, respectively [26]. Thus, ABCA1 is the most important transporter of intracellular cholesterol efflux. ABCA1 is ubiquitously expressed, and most types of cells in the body do not possess the ability to degrade cholesterol. To maintain cholesterol homeostasis, ABCA1-dependent cholesterol efflux occurs in almost all cell types, particularly macrophages and VSMCs.

Mutations in the ABCA1 gene is known to cause Tangier disease, which is characterized by extremely low plasma levels of HDL cholesterol (HDL-C) and premature atherosclerosis [27]. Macrophage-specific ablation of ABCA1 promotes foam cell formation, inhibits RCT, and increases plaque area in low-density lipoprotein receptor-deficient (Ldlr^–/–) mice [28,29]. We also reported that heat shock protein 70, histone methyltransferase enhancer of zeste homolog 2, miR-19b, and pregnancy-associated plasma protein-A decrease ABCA1 expression and subsequent cholesterol efflux in THP-1 macrophage-derived foam cells and then accelerate the development of atherosclerosis in apolipoprotein E knockout (apoE^–/–) mice [30–33]. Conversely, macrophage ABCA1 overexpression markedly promotes the efflux of cholesterol from mouse peritoneal macrophages to apoA-I and alleviates atherosclerosis in Ldlr^{–/ –} mice [34]. Human ABCA1 transgenic mice lacking apoE develop dramatically smaller, less-complex lesions when compared with their apoE^{–/ –} counterparts [35]. In addition, studies from our laboratory showed that apoA-I binding protein and diosgenin elevate plasma HDL-C levels, promote RCT, and inhibit atherosclerosis progression by up-regulating ABCA1 expression in apoE^{–/ –} mice [36,37]. Together, these findings provide strong evidence to support ABCA1 as an atheroprotective agent via its ability to remove intracellular cholesterol.

Although its major function is to mediate cholesterol efflux, ABCA1 acts as a multifunctional cell surface protein (Table 1). For example, knockout of ABCA1 leads to increased expression of proinflammatory cytokines and chemokines in mouse bone marrow-derived macrophages challenged with lipopolysaccharide (LPS) [38]. In contrast, overexpression of ABCA1 attenuates interleukin-6 and tumor necrosis factor-α levels in bovine aortic endothelial cells stimulated with LPS [39]. Previous studies from our group showed that treatment of THP-derived macrophages with apoA-I binding protein or apoA-I increases ABCA1 expression and consequently reduces proinflammatory cytokine secretion in response to LPS [40,41]. These observations suggest that ABCA1 exerts an anti-inflammatory action and acts as a critical link between inflammation and lipid metabolism. Additionally, ABCA1 can inhibit retinal ganglion cell apoptosis, promote colorectal cancer cell proliferation, enhance platelet reactivity, and stimulate insulin secretion from pancreatic β-cells [42–45].

Table 1. The biological functions of ABCA1

Experimental models	Functions	References
Macrophages	Promotion of cholesterol efflux	[30–34]
Macrophages and aortic endothelial cells	Inhibition of inflammatory response	[38–41]
Retinal ganglion cells	Prevention of cell apoptosis	[42]
DLD-1 and Caco-2 cells	Stimulation of cell proliferation	[43]
DBA1/lacJ mice and Tangier patients	Enhancement of platelet reactivity	[44]
INS-1 cells	Induction of insulin secretion	[45]

4. Models for ABCA1-mediated cholesterol efflux

ABCA1 plays a central role in promoting intracellular cholesterol efflux. Currently, to the best of our knowledge, nine models have been proposed to reveal the molecular mechanisms of ABCA1-mediated cholesterol efflux (Table 2).

Table 2. The proposed models to explain ABCA1-mediated cholesterol efflux

Models	Cell types	References
The channel trafficking model	HEK293 cells	[17,19,46]
The two-step process model	VSMCs, fibroblasts and HEK293 cells	[47,48]
The mushroom-like protrusion model	J774 mouse macrophages	[63]
The V-ATPase-induced apoA-I unfolding model	BHK cells	[78]
The PIP2-induced apoA-I unfolding model	RAW264.7 cells	[93]
The ABCA1 dimerization model	HeLa cells and BHK cells	[95,96]
The apoA-I-free vesicle model	J774 mouse macrophages and HEK293 cells	[67,100]
The sequential addition of apoA-I model	HEK293 cells	[102]
The retroendocytosis model	RAW264.7 cells and mouse peritoneal macrophages	[80,81]

4.1 The channel trafficking model

A channel is present between TMD1 and TMD2 of ABCA1, which plays a critical role in controlling lipid access [19]. This suggests a channel trafficking model for ABCA1-mediated cholesterol efflux by using human embryonic kidney (HEK) 293 cells (Figure 2). According to this model, the chamber is initially open at the bottom and close at the top. In the presence of lipid loading, accumulated phospholipids in the inner leaflet of plasma membrane are laterally transferred to the chamber by binding to amino acid residues within TMs 1/2/5. In this process, cholesterol is concurrently accessible to the chamber with the aid of phospholipids. ATP is then recruited to both NBDs, which leads to the dimerization of NBDs and consequently closes the chamber [46]. ABCA1 then flops the trapped lipids to the outer leaflet of plasma membrane. The hydrolysis of ATP at both NBDs forms an ADP-bound intermediate that changes the conformation of TMD1 and TMD1 to open the chamber at the top. Cholesterol and phospholipids are egressed from the chamber to the elongated hydrophobic tunnel formed by both ECDs. The conformation of ECDs is also altered by ATP hydrolysis, allowing them to interact with apoA-I [17]. Once bound to ABCA1, apoA-I takes up lipids from the elongated hydrophobic tunnel to assemble nHDL, which is then released from the cell surface. Following dissociation of ADP from NBDs, the chamber is restored to the initial open status for lipid uptake. Given that this is a structure-based model, more direct evidence needs to confirm its rationality.

Figure 2. Schematic illustration of the channel trafficking model.

Accumulated phospholipids along with FC in the inner leaflet of plasma membrane are laterally delivered to the chamber between both TMDs of ABCA1. ATP binding promotes NBD dimerization and closes the chamber, leading to lipid translocation to the outer leaflet. Subsequently, ATP is hydrolyzed to open the chamber at the top, which is followed by lipid export to the elongated hydrophobic tunnel. The hydrolysis of ATP also promotes apoA-I binding to ABCA1 for nHDL biogenesis. The chamber returns to the initial structure to absorb membrane lipids again after ADP dissociates from the NBDs.

4.2 The two-step process model

Fielding et al. and Wang et al. proposed a two-step mechanism for ABCA1-mediated nHDL biogenesis in VSMCs, fibroblasts and HEK293 cells (Figure 3) [47,48]. First, ABCA1 mediates the translocation of phospholipids from the inner leaflet to outer leaflet of plasma membrane. ATP is recruited to both NBDs of ABCA1. The hydrolysis of ATP induces a conformational change of ABCA1, allowing apoA-I to combine with ABCA1 [49]. Phospholipids are loaded onto apoA-I to form a complex, which is a much better acceptor for cholesterol than apoA-I itself. Second, the phospholipid/apoA-I complexes enter the caveolae, 50 ~ 100 nm cellular membrane invaginations enriched in lipids. FC in the caveolae is transported to these complexes in an ABCA1-independent autocrine or paracrine fashion for nHDL production. It is noteworthy that when cells are pretreated with cyclodextrin to deplete intracellular cholesterol, phospholipid efflux and apoA-I binding are not affected while cholesterol removal is disappeared [48]. This suggests that both steps are independent of each other. However, a later study showed that phospholipid efflux is tightly coupled with cholesterol efflux in RAW264.7 cells [50]. Thus, additional work is needed to determine what causes the conflicting findings.

Figure 3. Schematic illustration of two-step process model.

ABCA1 first promotes the translocation of phospholipid molecules to apoA-I. The phospholipid/apoA-I complex then picks up FC in the caveola in an ABCA1-independent manner, leading to nHDL formation.

Caveolin-1, one of the major protein components of caveolae, can bind specifically to cholesterol [51]. Accumulating evidence shows a positive correlation between the amount of caveolin-1 and intracellular cholesterol efflux [52–54]. Overexpression of caveolin-1 in HepG2 cells markedly promotes cholesterol export to acceptors, human plasma and apoA-I [55], whereas knockdown of caveolin-1 by small interfering RNA (siRNA) leads to a significant decrease in cholesterol efflux from RAW264.7 cells [56]. In addition, caveolin-1 co-localizes with ABCA1 in the caveolae, which is required for ABCA1-mediated cholesterol efflux in rat aortic endothelial cells [57,58]. Given caveolin-1 as an important contributor to ABCA1-mediated cholesterol efflux, it is likely that caveolin-1 is responsible for the transfer of cholesterol from the caveolae to the phospholipids/apoA-I complex.

4.3 The mushroom-like protrusion model

Chemical cross-linking studies have demonstrated that the amphipathic α-helix of apoA-I can bind directly to the ECDs of ABCA1 when the latter is in its active conformation [59,60]. The interaction between apoA-I and ABCA1 increases the stability of ABCA1 at plasma membrane by protecting it from proteolytic degradation [61]. Further analysis showed that apoA-I as a monomeric form interacts with ABCA1, and only about 10% of apoA-I associated with the plasma membrane is bound directly to ABCA1 [62]. Thus, ABCA1 activity creates two apoA-I binding sites on the cell surface. One is low-capacity binding site for apoA-I/ABCA1 interaction. The other is high-capacity binding site for apoA-I/lipid interaction. Based on the apoA-I/ABCA1 reaction scheme, Vedhachalam et al. presented a model for nHDL biosynthesis in three steps in J774 mouse macrophages (Figure 4) [63].

First, a small part of apoA-I is bound to cell surface ABCA1, known as low-capacity binding site. The direct interaction between the two molecules leads to a significant increase in ABCA1 translocase activity, which transfers FC and phospholipids from the inner leaflet to the outer leaflet of plasma membrane. This occurs in a chamber created by both TMDs of ABCA1 [46]. Lipid accumulation in the outer membrane leaflet leads to the lateral compression of phospholipid molecules in this region. Accordingly, decreased lipid contents cause the lateral expansion of phospholipid molecules in the inner leaflet. The resultant asymmetric phospholipid packing across the bilayer leads to membrane strain. These protrusions are regarded as a critical site for apoA-I binding and nHDL biosynthesis.

Figure 4. Schematic illustration of mushroom-like protrusion model.

Upon binding to a small amount of apoA-I, ABCA1 translocates FC and phospholipids from the inner leaflet to the outer leaflet of plasma membrane. This process leads to the lateral compression of phospholipid molecules in the outer leaflet and their expansion in the inner leaflet. Due to the unequal molecular packing density across the membrane, the mushroom-like protrusions are formed to release surface tension. A large amount of apoA-I molecules then bind to these protrusion structures. After accepting enough lipids, apoA-I dissociates from plasma membrane, leading to nHDL generation and membrane structure restoration.

Second, apoA-I interacts with the mushroom-like protrusions. Because the phospholipid packing density in the outer leaflet is significantly higher than that in the inner leaflet, the mushroom-like protrusions are formed in the direction of extracellular space to relieve membrane strain. This idea is supported by a recent study showing that the vast majority of apoA-I is bound to the structures protruding from plasma membrane in cholesterol-loaded fibroblasts expressing ABCA1, whereas these protrusions are absent in fibroblasts isolated from a subject with Tangier disease [64]. The mushroom-like protrusions, which are rich in FC and phospholipids but devoid of ABCA1, provide high-capacity binding sites for apoA-I [65]. The hydrophobic C-terminal α-helix in apoA-I molecules then inserts into the phospholipid bilayer for lipid uptake [66]. However, the detailed mechanisms underlying apoA-I interaction with the mushroom-like protrusions are still poor understood.

Third, lipids within mushroom-like protrusions are loaded onto apoA-I. Once apoA-I is bound to the mushroom-like protrusions, it receives a large amount of FC and phospholipids from these protrusions and is then spontaneously released into extracellular space. Concomitantly, plasma membrane restores normal structure. This process leads to the generation of three types of nHDL containing two, three, or four apoA-I molecules per particle [67]. Moreover, these particles have distinct cholesterol amounts and phospholipid components [67]. The heterogeneity of nHDL particles may result from the mushroom-like protrusions created by ABCA1 in distinct membrane environments.

4.4 The vacuolar ATPase (V-ATPase)-induced apoA-I unfolding model

As the acceptor of ABCA1-mediated cholesterol efflux, apoA-I exists in lipid-free, poor, and bound states. The former has a stronger capacity to accept cholesterol compared with the other two forms. During nHDL biogenesis, lipid-free apoA-I undergoes a variety of conformational changes, such as increased helicity and dimeric formation [68,69]. Recently, Wang et al. showed that ABCA1 can mediate the N-terminal unfolding of lipid-free apoA-I on the cell surface, which acts as an intermediate state and is essential for its lipidation [70]. Thus, the ability to unfold apoA-I N-terminus is a third activity of ABCA1 besides apoA-I binding and plasma membrane remodeling.

V-ATPase, a member of the rotary ATPase protein family, is composed of a peripheral cytoplasmic domain (V₁, 8 subunits) and an integral membrane domain (V0, 6 subunits). The V₁ domain is responsible for the hydrolysis of ATP. The V_o domain then utilizes the released energy to translocate protons across the membrane, leading to the establishment of acidic environment in intracellular compartments or extracellular space [71,72]. V-ATPase is expressed in a variety of cell types, including macrophages, peptidergic cells, and melanoma cells [73–75]. It has been reported that under acidic conditions, lipid-free apoA-I undergoes a conformational change to expose its hydrophobic α-helical structure, resulting in increased formation of reconstituted HDL from pure phosphatidylcholine liposomes [76]. Another study showed that incubation of macrophage-derived foam cells with HDL at acidic pH markedly promotes cholesterol efflux due to HDL remodeling and lipid-poor apoA-I release [77]. Given its acidic ability, it is likely that V-ATPase plays an important role in promoting ABCA1-dependent cholesterol efflux. As expected, overexpression of ABCA1 in baby hamster kidney (BHK) cells dramatically increases the cell surface levels of V-ATPase without altering its total levels [78]. This finding suggests that ABCA1 promotes V-ATPase translocation to the plasma membrane instead of its transcription. As a consequence, V-ATPase acidifies not only plasma membrane to increase lipid fluidity but also apoA-I to unfold its N-terminus. The partially unfolded apoA-I dimerizes via the domain swapping around repeat 5 and then spontaneously inserts into the membrane [79]. After being loaded with sufficient lipids, the apoA-I dimers are released from the cells to produce nascent HDL particles. This model reveals that recruitment of V-ATPase to the cell surface is required for the unfolding of apoA-I at its N-terminal domain by ABCA1, thereby providing a novel mechanistic insight into HDL biogenesis (Figure 5). Enhancement of cell surface V-ATPase activity and apoA-I acidification could be valuable approaches to increase cholesterol efflux capability.

Figure 5. Schematic illustration of V-ATPase-induced apoA-I unfolding model.

ABCA1 recruits V-ATPase to the cell surface. V-ATPase then acidifies apoA-I and plasma membrane to unfold its N-terminus and enhance lipid fluidity, respectively. After partial unfolding, apoA-I forms the dimers that can spontaneously interact with membrane lipids, leading to the generation and release of nHDL. The dashed line indicates that V-ATPase may also contribute to the acidification and unfolding of apoA-I in the endosomes, followed by retroendocytosis and nHDL secretion.

In addition to the plasma membrane, V-ATPase is located in the endosomes and normally pumps protons into the endosomal compartments to decrease their pH. It has been proposed that ABCA1 can promote cholesterol efflux to apoA-I via the retroendocytosis pathway [80,81]. Given that ABCA1, apoA-I, and V-ATPase are co-localized in endosomes, V-ATPase-induced acidification of the endosomal lumen may also contribute to apoA-I lipidation and subsequent nHDL secretion (Figure 5). Future studies are needed to confirm this possibility.

4.5 The phosphatidylinositol 4,5-bisphosphate (PIP2)-induced apoA-I unfolding model

HDL is made up of 100 ~ 200 lipid molecules surrounded by two or more apoA-I molecules. The major phospholipid components within HDL are phosphatidylserine (PS), PIP2, phosphatidylethanolamine, and phosphatidylcholine (PC), which originate from lipid rafts in plasma membrane [82]. Phospholipids are unequally distributed across the two leaflets of plasma membrane with an exclusive enrichment of PS in the inner leaflet [83]. Like PS, PIP2 resides predominantly on the inner leaflet where it participates in a variety of biological processes, including endocytosis, exocytosis, protein transport, and endosomal recycling [84–87]. PIP2 biogenesis is tightly modulated by myristoylated alanine-rich C-kinase substrate and small GTPases, such as Rab10 and Rab27a [87–89].

Phospholipid flip-flop between two leaflets of a membrane bilayer is a bi-directional process that is essential for the maintenance of membrane lipid homeostasis. This transport process is known to rely on two classes of enzymes. The flippases mediate the inward translocation of phospholipids, while the floppases have an opposite effect [83]. It is well established that ABCA1 promotes PS translocation to the outer leaflet, which is blocked by the mutations in its first large ECD [90,91]. Although the PS floppase activity of ABCA1 does not lead to direct interaction between apoA-I and the cell surface, it is required for efficient cholesterol efflux to apoA-I [92].

Cellular membrane contains various types of phospholipids. It is possible that ABCA1 translocates other phospholipids to the cell surface besides PS. Indeed, Gulshan et al. showed that the ECD2 of ABCA1 has PIP2 floppase activity, which is necessary for apoA-I binding and lipidation [93]. Interestingly, these two floppase activities of ABCA1 are independent of each other [93]. Thus, these authors proposed a novel model for ABCA1-dependent cholesterol efflux that includes PS and PIP2 in RAW264.7 cells (Figure 6). In this model, ABCA1 flops PS to the outer leaflet, facilitating the remodeling of plasma membrane to increase its susceptibility to apoA-I. Meanwhile, it delivers PIP2 to the cell surface where PIP2 binds to apoA-I. The direct interaction between PIP2 and apoA-I promotes the unfolding of apoA-I N-terminus. The partially unfolded apoA-I forms a dimer that inserts into the membrane to assemble nHDL particles, which are released from the cells. This model provides another important mechanism for ABCA1-induced apoA-I unfolding beyond V-ATPase. Stimulating PIP2 translocation to the cell surface may be a promising strategy to promote cholesterol removal from cells and restore cellular cholesterol homeostasis.

Figure 6. Schematic illustration of PIP2-induced apoA-I unfolding model.

ABCA1 flops PS and PIP2 to the outer leaflet where PIP2 interacts with apoA-I. This leads to apoA-I unfolding at its N-terminal domain and dimerization. The apoA-I dimer then inserts into plasma membrane to form nHDL.

4.6 The ABCA1 dimerization model

The majority of ABCA1 on the plasma membrane exists as an oligomeric complex, which constitutes the minimum functional unit required for apoA-I lipidation [94]. Moreover, the formation of ABCA1 oligomerization is not dependent on apoA-I binding [94]. On the basis of these findings, two later studies proposed a working model to reveal the mechanism underlying nHDL generation by using HeLa cells and BHK cells (Figure 7) [95,96]. In this model, ABCA1 monomers constantly transport FC and phospholipids from plasma membrane to their ECDs due to ATP hydrolysis. Following sequestration of sufficient lipids, these monomers undergo conformational changes to dimerize. The lipidated ABCA1 dimers interact with the membrane-skeletal actin filaments and other stable structures in plasma membrane, and become ready for apoA-I access. Lipid-free apoA-I combines with the ECDs of the ABCA1 dimers [97,98]. The lipids reserved by ABCA1 are then delivered to apoA-I. Lipid loading induces a conformational change of apoA-I, leading to its dissociation from ABCA1 and nHDL production [95]. Upon release of the reserved lipids, the ABCA1 dimers transform into monomers, entering next cycle to receive lipids. This model suggests that ABCA1 dimerization is of critical importance to cholesterol export from cells. Promoting the conversion of ABCA1 monomers to dimers may be another effective strategy for inhibiting intracellular lipid accumulation.

Figure 7. Schematic illustration of ABCA1 dimerization model.

The ABCA1 monomers translocate FC and phospholipids from plasma membrane to their ECDs in an ATP-dependent manner. When enough lipid molecules are sequestered by the ECDs, these ABCA1 monomers are converted to the dimers. Lipid-free apoA-I then binds to the ECDs of dimerized ABCA1. Lipids are loaded onto apoA-I, leading to nHDL generation. The ABCA1 dimers dissociate into monomers to transport lipids again.

4.7 The apoA-I-free vesicle model

Two apoA-I monomers form a half-circle dimer as revealed by the crystal structure, which is necessary for lipid binding [99]. Interestingly, dimerized apoA-I can acquire lipids from not only cellular membrane but also membrane-derived vesicles shed by ABCA1 during nHDL biogenesis in J774 mouse macrophages and HEK293 cells (Figure 8) [67,100]. In this process, ABCA1 first translocates FC and phospholipids to the outer leaflet independent of apoA-I. Higher phospholipid levels in this region create the mushroom-like protrusions in close proximity to ABCA1 for alleviating surface tension [63]. These protrusions are released from plasma membrane as apoA-I-free vesicles, representing apoA-I-independent lipid efflux. Subsequently, apoA-I docks to plasma membrane through its hydrophobic C-terminus and binds to the ECDs of ABCA1 via its N-terminal domain. The interaction between both molecules leads to the unfolding of apoA-I N-terminus, allowing it to form a dimer. A large amount of FC and phospholipid molecules transported by ABCA1 are loaded onto dimerized apoA-I, leading to the production and release of nHDL. It is worth noting that nHDL can continue to gain lipids from apoA-I-free vesicles to form larger particles. This model is different from the mushroom-like protrusion model proposed by Vedhachalam et al. [63], because it shows that upon creating a high-curvature structure, the interaction between apoA-I and ABCA1 is essential for nHDL generation.

Figure 8. Schematic illustration of apoA-I-free vesicle model.

In the absence of apoA-I, FC and phospholipid molecules are transferred by ABCA1 to the outer leaflet leading to the formation of mushroom-like protrusions, which are released into the extracellular space as apoA-I-free vesicles. Upon binding to ABCA1, apoA-I unfolds at its N-terminal domain and then dimerize. The apoA-I dimers receive FC and phospholipids to form small nHDL. Lipids, including GM1, within apoA-I-free vesicles are subsequently loaded onto nHDL to generate larger particles.

The formation of apoA-I-free vesicles is an important step in this model. It is well known that the lipid components of nHDL are similar to those of lipid rafts [82]. Lipid analysis showed that monosialotetrahexosylganglioside (GM1), a lipid raft marker, is present in both apoA-I-free vesicles and nHDL [100]. However, ABCA1 cannot transfer GM1 to apoA-I, excluding direct acquisition of GM1 from plasma membrane. A plausible explanation for this phenomenon is that GM1 is delivered to nHDL from apoA-I-free vesicles via a lipid exchange manner. Of note, flotillin, another lipid raft marker, does not appear in apoA-I-free vesicles [100]. Thus, it remains largely unknown whether these vesicles are derived from raft-like regions of cellular membrane.

4.8 The sequential addition of apoA-I model

HDL is a heterogeneous population of particles with a range of sizes and molecular compositions [101]. Recently, Sorci-Thomas et al. found that nHDL particles are produced as three relatively homogenous populations containing one, two, or three apoA-I molecules [82]. Given that there is no precursor–product relationship between the smallest and largest nHDL particles [102], the authors presented a model involving the sequential addition of apoA-I during nHDL assembly by using HEK293 cells (Figure 9). If only one free-lipid apoA-I binds to ABCA1 on the cell surface, the <6 nm diameter nHDL is generated and released as poor-lipid apoA-I carrying four lipid molecules. When a second apoA-I is available, ABCA1 can transfer more trapped lipids to apoA-I, leading to the formation of the 7.5 nm diameter nHDL particle containing 24 lipid molecules. In the presence of a third apoA-I, apoA-I acquires maximal amounts of lipids to produce the 10 ~ 12 nm diameter nHDL with about 240 lipid molecules. Although this model provides a reasonable explanation for different-sized nHDL formation, future research is required to determine what causes the sequential addition of apoA-I to ABCA1.

Figure 9. Schematic illustration of sequential apoA-I addition model.

Each nHDL particle is derived from a separate step in the process of lipidation. Binding of one free-lipid apoA-I molecule to ABCA1 at plasma membrane leads to the production of poor-lipid apoA-I (<6 nm). If two free-lipid apoA-I molecules interact with ABCA1, 7.5 nm nHDL is formed. However, 10 ~ 12 nm nHDL is generated when three free-lipid apoA-I molecules bind to ABCA1.

It is well accepted that apoA-I receives cholesterol transported by ABCA1 to generate nHDL, while ABCG1 and SR-BI facilitate subsequent cholesterol efflux to these particles to form mature HDL [25]. Epidemiologic studies have demonstrated that plasma HDL-C levels are negatively correlated with the risk of CVD [103,104]. Recently, Du et al. showed that small dense HDL has a stronger capacity to receive cholesterol transported by ABCG1 than large HDL [105]. Thus, HDL-raising treatments should aim to increase the concentrations of nHDL species with one or two apoA-I molecules.

4.9 The retroendocytosis model

In addition to plasma membrane, the endosomes are regarded as an important reservoir of cellular cholesterol [106]. ABCA1 resides not only on the cell surface, but also in the endosomal compartments [107]. Consistently, apoA-I lipidation occurs at both sites. It has been proposed that apoA-I can be internalized to the endosomes for nHDL assembly in a pathway called retroendocytosis in RAW264.7 cells and mouse peritoneal macrophages (Figure 10) [80,81], which was first suggested by Schmitz et al. in 1985 [108]. First, apoA-I binds to ABCA1 on the cell surface to form a complex, which then enters clathrin-coated pits. Second, the complex is endocytosed to early endosomes in a Rab5-dependent manner [109]. ABCA1 located in the early endosomal membrane translocates lipids to the endosomal lumen for apoA-I lipidation. ApoA-I is further lipidated when early endosomes become larger late endosomes, leading to the biogenesis of nHDL particles. Forth, these particles are transferred to cycling endosomes, which then return to the cell surface with the help of Rab4 [109]. After fusion with plasma membrane, nHDL is secreted into the extracellular space, and ABCA1 recycles to the cell surface. This model suggests that stimulating apoA-I internalization provides an alternative approach to enhance ABCA1-mediated cholesterol efflux.

Figure 10. Schematic illustration of the retroendocytosis model.

ApoA-I interacts with cell surface ABCA1 to form a complex that is endocytosed to endosomal compartments via a Rab5-dependent pathway. ABCA1 translocates lipids to apoA-I for nHDL production. Rab4 promotes internalized ABCA1 and synthesized nHDL back to the cell surface. The nHDL particles are then released into the extracellular space, and ABCA1 is located in plasma membrane again.

Calpain is known to cleave ABCA1 through interaction with its PEST sequence [18]. Deletion of the PEST sequence (ABCA1-dPEST) significantly increases cell surface ABCA1 levels but decreases its localization to late endosomes. In cells stably expressing ABCA1-dPEST, overall cholesterol efflux is dramatically increased despite a reduction in cholesterol efflux from late endosomes [110]. This suggests that increased apoA-I lipidation on the cell surface can fully compensate for the loss of retroendocytosis-associated cholesterol efflux. Another study showed that most apoA-I is lipidated on the cell surface of macrophages, and only a small portion of apoA-I is endocytosed and resecreted as nHDL particles [111]. Consistent with this observation, Denis et al. reported that internalized apoA-I contributes to approximately 1.4% of total nHDL generation, and pharmacologic inhibition of its endocytosis with sucrose or cytochalasin D does no reduce intracellular cholesterol efflux [112]. They also found that about 80% of internalized apoA-I is transported to the lysosomes for degradation and only the remaining part can produce nHDL [112]. Thus, the retroendocytosis pathway may play a minor role in ABCA1-mediated cholesterol efflux, and nHDL is predominantly produced from the cell surface.

The mechanisms for nHDL biogenesis in cholesterol-loaded cells differ from those in cholesterol-depleted cells [113]. When apoA-I internalization is blocked, cholesterol efflux is significantly decreased in cholesterol-loaded cells but increased in cells without cholesterol accumulation [109]. In agreement with this, the amount of apoA-I endocytosed by cholesterol-normal cells is not sufficient to produce nHDL particles [111]. Thus, the retroendocytosis pathway may contribute to nHDL formation only in cells with excessive cholesterol.

ABCA1 recycles rapidly between plasma membrane and the endosomal compartments, which has a significant effect on endosomal cholesterol removal. Of internalized ABCA1, only 30% is recycled back to plasma membrane, implying the presence of a degradative pathway [114]. It has been proposed that ADP-ribosylation factor 6 (Arf6), a small GTPase, directs endocytosed ABCA1 into the lysosomes for degradation, leading to reduced efflux of intracellular cholesterol [115]. Niemann-Pick type C (NPC) disease, an autosomal recessive condition, is characterized by excessive cholesterol accumulation in late endosomes [116]. In fibroblasts isolated from NPC patients, overexpression of small GTPase Rab8 markedly inhibits late endosomal cholesterol deposition by translocating cholesterol to plasma membrane [117]. Importantly, knockdown of Rab8 with siRNA attenuates cell surface ABCA1 expression and suppresses endosomal cholesterol efflux to apoA-I in primary human macrophages, suggesting Rab8 as a significant contributor to the retroendocytosis pathway [118]. However, how Rab8 affects ABCA1 localization to the cell surface remains unknown. Palmitoylation, a reversible post-translational modification, plays an important role in promoting protein localization to specific membrane domains [119]. Blockade of ABCA1 palmitoylation leads to a significant reduction in cell surface ABCA1 and cholesterol efflux, while overexpression of the palmitoyl transferase DHHC8 stimulates ABCA1-mediated cholesterol efflux [120]. Collectively, these observations suggest that targeting Arf6, Rab8, or ABCA1 palmitoylation could be attractive strategies for promoting endosomal cholesterol efflux to apoA-I.

5. Conclusion and future directions

ABCA1 plays a central role in nHDL biogenesis by promoting the transport of FC and phospholipids to apoA-I. Decreased cholesterol efflux mediated by ABCA1 is closely associated with a variety of diseases, especially CVD. To date, nine models have been proposed to account for the mechanisms by which ABCA1 exports cholesterol from cells. It remains largely unknown, however, which model represents the dominant pathway for ABCA1-mediated cholesterol efflux. Thus, there is a great need for additional research to precisely define the relative contributions of each of these pathways to nHDL production. The two-step process model has demonstrated that ABCA1-dependent phospholipid efflux precedes FC efflux to apoA-I, while other models show a simultaneous translocation of two substrates. This paradoxical finding highlights the need for further research in this area. Both V-ATPase and PIP2 have been shown to unfold apoA-I at its N-terminal domain [78,93], while whether other factors can exert such an effect remains to be clarified. Given the lack of ABCA1 in the mushroom-like protrusions, additional work is required to determine how lipids in these membrane microdomains are transferred to apoA-I. In addition, the following issues need to be addressed. What are the biological roles of Walker C signature of ABCA1? How are apoA-I-free vesicles released from plasma membrane into extracellular environment? Does V-ATPase participate in the retroendocytosis pathway? Are there any other components involving the endocytosis of the ABCA1/apoA-I complex besides Rab5? In summary, answers to these questions will provide insightful knowledge about the mechanisms of ABCA1-dependent cholesterol efflux and promote the development of novel HDL-based therapies.

Acknowledgments

We thank Dr. Xiao-Hua Yu for helpful discussions and valuable suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by grant from the National Natural Science Foundation of China (82103782).