Genetic variants of ABCA1 (ATP-binding cassette A1), an ATP-driven transporter that pumps cholesterol out of cells, recently joined the ranks of potential risk factors for late-onset Alzheimer’s disease (see Wollmer et al., 2003 and Katzov et al., 2004). Why these variants may predispose carriers to AD is uncertain, but the role of the transporter in the periphery is to mobilize cholesterol out of cells and onto lipid-poor apolipoproteins, and it may work similarly in the central nervous system (CNS). If so, ABCA1 could provide a link between two major AD risk factors, cholesterol and apolipoprotein E (ApoE), the major lipoprotein of the CNS (see ARF related news story on the link among cholesterol, ApoE and AD). Ironically, just as two papers from independent labs report that ABCA1 keeps ApoE levels high and saturated with cholesterol, a third paper casts doubt on the link between the transporter and Alzheimer’s.

The first two papers are currently in press in the Journal of Biological Chemistry and are already available online. In the first, Dave Holtzman and colleagues, at Washington University, St. Louis, and the Carnegie Mellon University, Pittsburgh, used ABCA1 knockout mice to test the role of the transporter in the CNS. When first author Suzanne Wahrle and colleagues measured ApoE in these animals, they found that levels in the cortex and cerebrospinal fluid (CSF) were 80 and 98 percent lower than in normal mice, while animals missing only one copy of the gene had intermediate levels of ApoE (13 and 46 percent, for cortex and CSF, respectively).

To determine why levels may be so ablated, the authors looked in the CSF where the majority of ApoE exists as lipoprotein particles around 10-17 nanometers in diameter. When Wahrle fractionated CSF from ABCA1 knockout mice, she found that some particles were much smaller than normal, about 7 nanometers wide. Given the role of ABCA1 as a cholesterol transporter, this suggests that the particles may be poorly bound with the lipid. To investigate this, Wahrle examined lipoproteins secreted from cultured astrocytes, cells that are the major source of ApoE in the brain. The authors found that about 75 percent of ApoE secreted from ABCA1-negative astrocytes ends up in the smaller lipoproteins, and these particles were indeed cholesterol poor (0.69 mg cholesterol/mg ApoE, compared to 2.3 mg cholesterol/mg ApoE in particles from normal astrocytes). The authors conclude that “ABCA1 plays a major role in maintaining normal ApoE levels in vivo,” and suggest that “modulation of ABCA1 function and levels may be a novel therapeutic target for AD.”

Also reporting in the Journal of Biochemistry, Cheryl Wellington and collaborators from the University of British Columbia, Vancouver, and the Clinical Research Institute of Montreal, came to very similar conclusions. First author Veronica Hirsch-Reinshagen and colleagues also found that ApoE is depleted in the brains of ABCA1-negative mice, being 65 percent lower overall than normal, and 76 and 79 percent lower in the hippocampus and striatum, respectively.

Hirsch-Reinshagen also looked at secretion of cholesterol by cultured astrocytes and microglia, adding exogenous ApoA1 or ApoE isoforms to the culture medium to provide an apolipoprotein “sink.” The ABCA1-negative astrocytes secreted poorly in comparison to their wild-type counterparts, releasing less than five percent of their total cholesterol in comparison to 7-10 percent released from normal cells. The microglia fared slightly better, secreting about 30-40 percent less than normal cells. ApoE3 and ApoE4 provided more “pulling power” than the other lipoproteins because when these were added to the medium, more cholesterol was secreted by the cells. However, this difference was only significant in the case of astrocytes.

As for secretion of ApoE, Hirsch-Reinshagen’s data were again in agreement with the work from Holtzman’s lab. The authors found that ApoE secretion was reduced in cells lacking the cholesterol transporter. It was down by 30 percent in astrocytes and by about 90 percent in microglia.

Taken together, these papers provide a clear link between ABCA1 and ApoE metabolism in the brain. However, as Hirsch-Reinshagen and colleagues write, “the mechanisms by which ABCA1 affect ApoE metabolism in glial cells are not yet understood.” It is interesting, for example, that neither group found any alteration in ApoJ levels in the CNS of the knockout mice, suggesting that the transporter may selectively impinge on ApoE metabolism.

As for ABCA1 variants as risk factors for AD, Andrew Grupe and colleagues from Celera Diagnostic, Alameda, California, and elsewhere, report the results of a case control study in the August 19 Neuroscience Letters (currently available online).

First author Yonghong Li and colleagues genotyped ABCA1 polymorphisms in DNA samples from a total of 2,146 individuals, 980 diagnosed with AD and 1,166 controls. They identified nine single nucleotide polymorphisms (SNPs) in the ABCA1 gene, three of which were identical to those previously predicted to confer risk for AD (see Katzov et al., 2004). Li, however, found that none of the SNPs showed any significant association with late-onset Alzheimer’s, even when they conducted pairwise linkage disequilibrium analysis. The discrepancy may be due to the larger sample size analyzed in the newest study.—Tom Fagan


  1. After ApoE, which has an unusually large effect size as a
    late-onset AD gene, the remaining AD genes would be expected to have
    more modest to moderate effect sizes. Thus, one really needs to
    routinely use at least a thousand or so uniformly ascertained
    subjects in a case-control study to have a chance of replicating a
    given AD candidate gene. In this case, the collaboration between Li
    and colleagues and Celera tested 796 individuals (which is pretty
    sizeable) and could not replicate the original findings. In addition,
    at the recent AD meeting in Philadelphia, we could not replicate the
    association with ABCA1 using family-based association on the large
    NIMH AD sample. So, the possibilities are 1) the original result was
    a false positive which often plagues case control studies, 2) the
    original association was real but was due to linkage disequilibrium
    with a nearby gene and thus not readily able to be confirmed in all
    populations studied, or 3) the association was real, but the ABCA1
    has a very small effect size in AD, and is, thus, difficult to
    replicate across various populations. I should note that in our
    presentation in Philadelphia of negative findings for ABCA1, which
    sits in the chromosome 9 AD linkage peak region, in contrast, we
    presented evidence for strong association with the gene encoding
    ubiquilin 1 within that same peak in two independent family samples.

  2. ABCA1 has been shown to be critical outside the brain for effluxing phospholipid and cholesterol from cells onto HDL. In the periphery, ApoAI is the main apolipoprotein in HDL. The absence of ABCA1 function results in Tangier's disease in which plasma HDL is very poorly lipidated and is rapidly metabolized resulting in very low plasma HDL levels. The two new papers by Wahrle et al. and Hirsch-Reinshagen et al. show, using ABCA1 knockout mice, that ABCA1 is also critical for effluxing phospholipid and cholesterol from glial cells onto ApoE-containing HDL in the brain. Since ApoE is the most abundant apolipoprotein produced in the brain, this results in the production of very cholesterol- and phospholipid-poor CNS HDL. There are also very low levels of ApoE in the CNS of these mice, probably because the poorly lipidated ApoE is metabolized more rapidly. These results have important implications for any effect that ApoE may have in the normal brain, but perhaps more importantly, in the physiological setting of CNS diseases such as Alzheimer's disease. As one example, it is clear from animal studies that the amount of ApoE regulates the onset, amount, conformation, and toxicity of the amyloid-β peptide. These studies suggest that by altering ABCA1 function, the amount and lipidation state of ApoE will change in the brain. This is likely to impact the onset of amyloid-β-related pathology.

    In regard to the genetics paper from the groups of Grupe and Goate, there was no association with certain SNPs in ABCA1 and LOAD.
    If the main effect that ABCA1 might have on AD is related to altering ApoE levels and lipidation (or altering Aβ levels as suggested by some other recent papers), unless the SNPs actually alter the function of ABCA1 enough to alter ApoE or Aβ levels in the CNS, one might not expect to see them alter risk for AD. The SNPs in this paper will need to be examined to determine if they actually alter ABCA1 function or not.

  3. Cholesterol
    and Alzheimer's

    I would like to point Alzforum readers to see the key
    note by Cheryl Wellington
    , senior author of one of the discussed articles
    entitled "Deficiency of ABCA1 impairs apolipoprotein E metabolism in brain".
    It is freely available at Neurobiology of Lipids noteworthy articles' collection.
    Also, please see NoL
    collections index
    on different aspects of the role of fats in brain
    function and Alzheimer's disease.

    I would like also to point readers to our 2001 FASEB Journal
    article "Essential
    role for cholesterol in synaptic function and neuronal degeneration
    and our last year's ARF
    hypothesis submission
    setting the pathogenic primacy of cholesterol
    metabolism dysfunction in Alzheimer's, and explaining why Abeta and TAU changes
    are secondary neurodegeneration features.

  4. The field of complex disease genetics is in a rapid developmental mode. However, we still know frighteningly little about how gene sequence variation can affect gene function, and most genetic association studies are not designed with this in mind. Exploratory studies (a category under which the vast majority of association studies must be said to fall) merely highlight genomic regions for further analysis. Engaging in debate around genetic association data can be futile if the effect sizes we are dealing with on an epidemiological level are small. This is also true if the number of pathogenic alleles in a genomic region, their impact upon a gene, and their potential interactions are not yet known. Efforts to find genes that influence disease might be better seen as a community undertaking where the accumulating bulk of data from many independent groups can eventually lead to a strong case for (or against) a particular gene. Very few groups have available clinical materials which can capture small genetic effects and demonstrate association conclusively (if one chooses to ever regard high probability as conclusive). The ABCA1 story is in its infancy and ultimately may turn out not to be relevant to AD at all, but a number of additional hypotheses are being tested in our laboratory (at present a major focus is upon CSF Aβ42 levels). The recent publications describing the relationship between ABCA1 and ApoE are extremely encouraging from a biological viewpoint, but the ultimate genetic question is whether different "versions" of the ABCA1 gene exist in human populations that differentially affect AD risk or severity.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Lipoproteins and Amyloid-β—A Fat Connection

Paper Citations

  1. . ABCA1 modulates CSF cholesterol levels and influences the age at onset of Alzheimer's disease. Neurobiol Aging. 2003 May-Jun;24(3):421-6. PubMed.
  2. . Genetic variants of ABCA1 modify Alzheimer disease risk and quantitative traits related to beta-amyloid metabolism. Hum Mutat. 2004 Apr;23(4):358-67. PubMed.

Further Reading

Primary Papers

  1. . Deficiency of ABCA1 impairs apolipoprotein E metabolism in brain. J Biol Chem. 2004 Sep 24;279(39):41197-207. PubMed.
  2. . Association of ABCA1 with late-onset Alzheimer's disease is not observed in a case-control study. Neurosci Lett. 2004 Aug 19;366(3):268-71. PubMed.
  3. . ABCA1 is required for normal central nervous system ApoE levels and for lipidation of astrocyte-secreted apoE. J Biol Chem. 2004 Sep 24;279(39):40987-93. PubMed.