ABCA1, from pathology to membrane function

The annotation of genomic databases led to the identification of 49 human ABC proteins classified into seven distinct subfamilies, ABC-A through ABC-G on the basis of similarity in gene structure, sequence or phylogenesis [36, 98]. In 1994, the identification of ABCA1 and its structural peculiarities led to the definition of a new subclass further named A [63]. The complete genomic sequence of the human and mouse ABCA1 genes, including their promoters and regulatory elements, has been determined in the late 1990s [51, 81, 89]. In terms of molecular architecture, ABCA1 is a full-sized ABC transporter, in that it shows a symmetrical structure with a transmembrane spanning domain (TMD) consisting of six transmembrane segments (TMS) and a nucleotide-binding domain (NBD), repeated in tandem. In the NBDs, all the diagnostic motifs can be found. ABCA1 possesses two extremely large extracellular loops, located between TMS1 and 2 and between TMS7 and 8 (Fig. 1). This is the most original feature of the A class of the ABC transporters shared by all the members [33]. Detailed topological studies have been performed in the case of ABCR (ABCA4) and ABCA1 and demonstrated a clear conservation of the topological arrangement [12]. This also holds true for ABCA7 and for the other members on the basis of sequence alignment and hydrophobicity plot comparisons [9, 44]. These extracellular loops show sequence hypervariability among the ABCA members and may thus participate directly to the functional specification of individual transporters in their appropriate expression compartment.

Open image in new window

A precise description of the molecular function of ABCA1 as a membrane translocator is so far not achieved and most of the knowledge on the substrates that it may handle is actually rather indirect. The bottom line experiment would consist in the purification of the transporter and its functional reconstitution in artificial systems such as proteoliposomes. However, considering that ABCA1 is a polytopic membrane protein of very large size, “en masse” production and purification for reconstitution or crystallographic purposes is extremely demanding. Hence, alternative methods to approach the problem of its quaternary structure may be extremely valuable.

Establishing precisely the cellular localisation of ABCA1 and the most relevant functional sites is critical to understand its physiological function. However, the lack of valuable antibodies has prevented a detailed immunohistological description of the ABCA1 expression pattern in the living animal so far. A number of studies in tissues and cell lines have analysed the expression pattern of ABCA1 transcript in human, mouse or baboon via Northern blot, reverse transcriptase-polymerase chain reaction, dot blot or in situ hybridisation analysis [8, 17, 44, 51, 54, 63, 117]. There is a general consensus toward a wide distribution of ABCA1 mRNA with variation in abundance in specific sites. The highest mRNA expression levels are detected in placenta, pregnant uterus, liver, lung, adrenal glands, heart, small intestine and fetal tissues; the lowest expression is found in pancreas, skeletal muscle, ovary, colon, prostate, mammary glands and bone marrow. The discrepancies concern mainly the level of expression rather than the presence or absence of the transcript.

Wellington et al. [117] have proposed an extensive review of the distribution pattern of mouse ABCA1, taking into account both transcript and protein levels. According to this work, the level of transcript may not accurately predict protein abundance, notably in tissue with intermediate to low abundance of ABCA1 transcript. In conclusion, liver, adrenal glands, testis, pregnant uterus and placenta appear as the major sites of ABCA1 expression. At the cellular level, tissue macrophages as well as macrophage-like cell lines of mouse or human origin are consistently expressing high levels of the transporter. Expression in both liver and macrophage appears of extreme physiological relevance. The expression of ABCA1 on the basolateral membrane of hepatocytes has been shown to indeed contribute significantly to the maintenance of circulating HDL level, in agreement with the prevailing view that the liver is the major site responsible for HDL biogenesis [42, 43]. On the other hand, despite the ubiquitous expression of ABCA1, the accumulation of cholesterol in both human and mouse models of ABCA1 dysfunction occurs principally in macrophages. However, the global amount of lipid efflux from these cells provides only a minimal contribution to circulating HDL levels. The progression of foam cell formation and the atheroma development are consistently significantly influenced by the ABCA1 expression on macrophages but largely independent of the levels of circulating HDL [1, 2, 30].

At the cellular level, ABCA1 is predominantly located on the plasma membrane. It is currently thought that this localisation is necessary and sufficient for its function as a promoter of lipid effluxes. On the other hand, ABCA1–EGFP chimeras also decorate the endo-lysosomal compartment [34, 82, 88] from where they can actively shuttle to the plasma membrane [71, 88]. This may be relevant for the control of cellular lipid handling considering the endocytic recycling of cell-bound apoproteins. Takahashi and Smith [102] reported that uptake and resecretion of ApoA-I indeed occur via endocytic vesicles (retroendocytosis) where ApoA-I and ABCA1–EGFP could co-localise. Thus, in line with these results, it may be envisioned that ABCA1 not only mediates lipid desorption at the cell surface but, upon internalisation, may also pump lipids into the vesicular lumen. Those would associate with the endocytosed apoproteins there and be further released as mature HDL particles upon fusion of the vesicle with the plasma membrane [7] (Fig. 3).

Open image in new window

The mutations in the ABCA1 transporter gene have been identified as the molecular defect responsible for Tangier disease, an autosomic recessive disorder of lipid metabolism [10, 55, 86, 87]. The clinical hallmarks of Tangier disease include the virtual absence of circulating HDL and ApoA-I. These are associated to a whole spectrum of signs, from orange tonsils, peripheral neuropathy, splenomegaly to thrombocytopenia and increased incidence of cardiovascular disease, whose clinical penetrance is however extremely variable [4]. Seminal to these signs is an impairment of cellular effluxes of phospholipids and cholesterol to lipid-poor apolipoproteins, diagnostic of Tangier disease among the other forms of hypoalphalipoproteinemias [4]. The mutations of ABCA1 have also been detected in a nosographically distinct disorder of HDL deficiency, familial HDL deficiency, that is clinically superimposable to Tangier disease but is transmitted as a dominant trait. This discrepancy may find its origin in the recent notion that ABCA1 assembly into oligomeric structures is required for full functional competence. We have indeed recently applied biophysical methods such as intermolecular fluorescence resonance energy transfer [46] and biochemical methods such as electrophoretic analysis in native condition (native polyacrylamide gel electrophoresis) [78] to the study of ABCA1 molecular assembly. These studies have provided evidence that ABCA1 is predominantly expressed as a dimer and that the transition between oligomeric states is part of its enzymatic cycle (Trompier et al., submitted for publication). Thus, the mutations specifically impairing oligomerisation would show a transdominant negative effect and the associated phenotypic abnormalities would be transmitted genetically as dominant traits.

An investigation on the relationship between discrete mutation and phenotype may also be instrumental in understanding what governs the variable penetrance of clinical signs in Tangier pedigrees. As direct genotype/phenotype correlations are impossible due to the rarity of the disease (60 pedigrees reported as for now) [94], alternative approaches such as the generation of appropriate animal models exclusively expressing informative ABCA1 mutants have to be envisioned. Finally, as ABCA1 can be considered one of the most clearly identified therapeutic targets in the prevention of cardiovascular disease [58, 73], several efforts have arisen in the aim of predicting how and whether genetic polymorphism may contribute to differences in phenotypic traits [11, 79].

The generation of knockout mice irrevocably demonstrated that ABCA1 dysfunction is sufficient to generate hypolipidemic profiles [17, 34, 68, 75]. These mice faithfully reproduce the human syndrome; however, a few traits appear limited to the mouse models. First, the engulfment of cells dying by apoptosis is clearly impaired during embryonic development in the mice [34]. This defect, which is devoid of major developmental consequences, has obviously not been investigated in humans. In addition, ABCA1−/− females are infertile due to impaired placental development, a phenotype related to the high expression level of ABCA1 in the pregnant uterus [17, 75]. A similar sign is lacking in human Tangier patients though an increased incidence of spontaneous abortions has been informally recorded.

The availability of animal models carrying the invalidation of ABCA1 gene has prompted a large body of work to be targeted at the definition of the protective role of the transporter in the development of atheroma lesion. Though the results globally confirm the predicted protective effect, a number of inconsistencies can be observed [42]. These are likely to originate from the heterogeneity of genetic backgrounds used in the different experimental setups. Though the alterations in metabolic profile and most phenotypes related to ABCA1 loss of function appear robust and background-insensitive, exquisite differences in the ABCA1 expression indeed exist between the various mouse strains (Trompier, unpublished results) [65]. This together with the well-known genetically determined differences in the cytokine regime and the activation of immunoregulatory pathways may well contribute to blur the experimental results obtained from heterogeneous sources.

The invalidation of ABCA1 gene in mice leads also to clinically relevant phenotypes unrelated to lipid metabolism; these span from aberrant responses to malaria infection to increased development of Alzheimer’s related degenerative lesions. In ABCA1−/− mice, generated in the pure DBA1/J background, the infection with Plasmodium berghei ANKA indeed leads to a robust resistance to the development of fatal cerebral malaria [18]. This can be traced to a reduced activation of macrophages in response to Plasmodium antigens as very low levels of circulating TNF α and plasma microparticles were observed. Modified macrophages polarisation, consequent to the ABCA1 deletion, may hence be envisioned and is indeed supported by the exquisite responses of ABCA1−/− myelomonocytic lineages to inflammatory stimuli and cytokine activation (Broccardo, unpublished results). The development of crosses between ABCA1 deleted mice and Alzheimer’s prone models, in heterogeneous genetic backgrounds, more recently revealed that the absence of the transporter induces an aggravation of fibrillar amyloid deposit [35, 49, 113]. This may be related directly not only to the ABCA1 function as a provider of lipid to nascent HDL-like particles in the brain but also to the fact that cholesterol content/distribution in ABCA1−/− membranes is greatly altered. It is known that the amyloidogenic proteolysis of Alzheimer’s precursor protein is indeed largely determined by its partitioning in/out membrane microdomains [37].

In this respect, it may be worth mentioning that the sorting of ABCA1 in specialised membrane domains is still a controversial issue. In our hands and at odds with previous results [23, 69], the experimental evidence indicates that a substantial fraction of the transporter is located in Lubrol WX and Brij 98 insoluble membrane domains while largely excluded from the classical “core microdomains” defined as Triton X-100 insoluble (Zarubica, unpublished observation). This finding has been quite recently supported by the lipidomic analysis of ABCA1-generated membrane microvesicles, clearly indicating that a conspicuous proportion derives from membrane microdomains [24]. Thus, irrespective of the transporter partitioning in the liquid-ordered or liquid-disordered membrane phases, its ability to facilitate desorption of membrane phospholipids and cholesterol may be expected to have consequences on the lateral arrangement of membrane lipids and on the sorting of proteins [84].

The shedding of lipid from cell membranes is a ubiquitous event. In most cases, it concerns the transport of a specific monomer form of lipid from one leaflet of the membrane to the other or to an external acceptor. Many human diseases have been related to defective desorption of lipids from cell membranes and, in nearly all cases, members of the ABC transporter family appear at the origin of defect [110]. In their vast majority, they belong to the A, D and G classes [3, 36] and dedicated reviews in this issue]. The D class includes half-transporters localised at the membrane of the peroxisome and involved in peroxisome biogenesis and/or in the transport of fatty acids across the membrane of these organelles [105]. Apart from ABCA1, most of the A class members are supposed to transport lipid substrates from the cytosol to either the exofacial leaflet of the membrane or the extracellular space or onto specialised membranes in intracellular organelles [45, 93, 107]. Four out the ABCG hemitransporters have been implicated in the transport of sterols, either in the intestine where ABCG5 and ABCG8 pump back absorbed sterols into the gut lumen [119] or in macrophages and liver cells where ABCG1 and ABCG4 work in concert with ABCA1 to actively transport cholesterol to mature HDL [91].

In all of these cases, the direct evidence addressing the transport itself and the substrate specificity is relatively limited, though the latter one seems to be quite diverse among the different members. How this is controlled and more generally how transport occurs in mechanistic terms are still open questions. Several models have been suggested: from the classical flippase model to the refined variation recently proposed by van Meer et al. [110] (Fig. 4). This latter proposal suggests that the lipids could be translocated without undergoing reorientation of their headgroups, from the inner membrane leaflet to the internal core of the transporter. From there, it can either be released onto an external acceptor or, in its absence, be reoriented, at energetic expenses, in the exofacial membrane leaflet. In other words, for example in our case, ABCA1 could either pump the phospholipid substrates onto apoprotein acceptors docked in its close proximity or flip them into the external leaflet of the membrane. This model has the distinctive advantage to solve some of the incongruence of the more classical flippase model, such as the evident need for acceptors or the energetic demand required for release of lipid substrates into aqueous medium. In addition, this bimodal and non-exclusive model can easily accommodate molecular partners acting as regulators of extrusion vs membrane flipping.

Open image in new window