Recent reports suggest that some commonly used nonsteroidal anti-inflammatory drugs (NSAIDs) unexpectedly shift the cleavage products of amyloid precursor protein (APP) to shorter, less fibrillogenic forms, although the underlying mechanism remains unknown. We now demonstrate, using a fluorescence resonance energy transfer method, that Abeta(42)-lowering NSAIDs specifically affect the proximity between APP and presenilin 1 and alter presenilin 1 conformation both in vitro and in vivo, suggesting a novel allosteric mechanism of action.

The expression, function, and regulation of the cholesterol efflux molecule, ABCA1, has been extensively examined in peripheral tissues but only poorly studied in the brain. Brain cholesterol metabolism is of interest because several lines of evidence suggest that elevated cholesterol increases the risk of Alzheimer's disease. We found a largely neuronal expression of ABCA1 in normal rat brain by in situ hybridization. ABCA1 message was dramatically up-regulated in neurons and glia in areas of damage by hippocampal AMPA lesion after 3-7 days. Immunoblot analysis demonstrated ABCA1 protein in cultured neuronal and glial cells, and expression was induced by ligands of the nuclear hormone receptors of the retinoid X receptor and liver X receptor family. ABCA1 was induced by treatment with retinoic acid and several oxysterols, including 22(R)-hydroxycholesterol and 24-hydroxycholesterol. Expression of an ABCA1-green fluorescent protein construct in neuroblastoma cells demonstrated fluorescence in perinuclear compartments and on the plasma membrane. Because the A␤ peptide is important in Alzheimer's disease pathogenesis, we examined whether ABCA1 induction altered A␤ levels. Treatment of neuroblastoma cells with retinoic acid and 22(R)-hydroxycholesterol caused significant increases in secreted A␤40 (29%) and A␤42 (65%). Treatment with a nonsteroidal liver X receptor ligand, TO-901317, similarly increased levels of secreted A␤40 (25%) and A␤42 (126%). The increase in secreted A␤ levels was reduced by RNAi blocking of ABCA1 expression. These data suggest that the cholesterol efflux molecule ABCA1 may also be involved in the secretion of the membrane-associated molecule, A␤.ABCA1 encodes an ATP-binding cassette protein that promotes efflux of cholesterol and phospholipids from intracellular compartments to high density lipoprotein, lipid-deficient apoAI, 1 and other apolipoproteins (1). Disruption of ABCA1 byTangier disease mutations in humans or by engineered knockout in mice is associated with a loss of cellular cholesterol efflux, an ablation of circulating high density lipoprotein, and, interestingly, peripheral neuropathies (2-5). Like many genes involved in cholesterol homeostasis, ABCA1 is regulated by the liver X receptors (LXR) (6 -8), nuclear receptors activated by oxysterols. ABCA1 is also regulated by peroxisome proliferatoractivated receptor ␦, which can be activated by fatty acid metabolites (9). Both of these classes of receptors form heterodimers with retinoid X receptors (RXR), which bind retinoic acid; as heterodimers, they alter gene transcription. These systems for ABCA1 induction help decrease cellular cholesterol after cholesterol loading (10). The importance of cerebral cholesterol metabolism in Alzheimer's disease (AD) risk and pathogenesis is supported by genetic, cell culture, mouse model, and epidemiologic data. ApoE in the central nervous system is implicated in supplying appropriate membrane lipid for development, nerve growth, and responses to injury and repair in the central nervous syste...

sorLA is a recently identified neuronal receptor for amyloid precursor protein (APP) that is known to interact with APP and affect its intracellular transport and processing. Decreased levels of sorLA in the brain of Alzheimer's disease (AD) patients and elevated levels of amyloid-␤ peptide (A␤) in sorLA-deficient mice point to the importance of the receptor in this neurodegenerative disorder. We analyzed APP cleavage in an APP-shedding assay and found that both sorLA and, surprisingly, a sorLA tail construct inhibited APP cleavage in a ␤-site APP-cleaving enzyme (BACE)-dependent manner. In line with this finding, sorLA and the sorLA tail significantly reduced secreted A␤ levels when BACE was overexpressed, suggesting that sorLA influences ␤-cleavage. To understand the effect of sorLA on APP cleavage by BACE, we analyzed whether sorLA interacts with APP and/or BACE. Because both full-length sorLA and sorLA C-terminal tail constructs were functionally relevant for APP processing, we analyzed sorLA-APP for a potential cytoplasmatic interaction domain. sorLA and C99 coimmunoprecipitated, pointing toward the existence of a new cytoplasmatic interaction site between sorLA and APP. Moreover, sorLA and BACE also coimmunoprecipitate. Thus, sorLA interacts both with BACE and APP and might therefore directly affect BACE-APP complex formation. To test whether sorLA impacts BACE-APP interactions, we used a fluorescence resonance energy transfer assay to evaluate BACE-APP interactions in cells. We discovered that sorLA significantly reduced BACE-APP interactions in Golgi. We postulate that sorLA acts as a trafficking receptor that prevents BACE-APP interactions and hence BACE cleavage of APP.

A common polymorphism in the cystatin C gene is associated with increased risk of developing Alzheimer's disease (AD). To explore possible neuropathological consequences of this genetic association, we examined expression of cystatin C in brains from 22 AD and 11 control patients by immunohistochemistry. In the temporal cortex of all AD brains, there was strong cystatin C immunostaining of neurons and activated glia, whereas staining was absent or minimal in 7 of the 11 control brains. Neuronal staining of cystatin C in AD brains was primarily limited to pyramidal neurons in cortical layers III and V, which are the neurons most susceptible to cell death in AD. The increase in cystatin C staining in AD was independent of cystatin C genotype. Immunostaining of cystatin C within neurons showed a punctate distribution, which co-localized with the endosomal/lysosomal proteinase, cathepsin B. A primarily glial source for cystatin C was suggested by parallel studies using in situ hybridization of mouse brain. In human AD brain, there was little co-localization of cystatin C with parenchymal Abeta deposits, although a small fraction of cerebral blood vessels and neurofibrillary tangles were cystatin C-positive. The regional distribution of cystatin C neuronal immunostaining also duplicated the pattern of neuronal susceptibility in AD brains: the strongest staining was found in the entorhinal cortex, in the hippocampus, and in the temporal cortex; fewer pyramidal neurons were stained in frontal, parietal, and occipital lobes. These neuropathological observations reinforce the association between cystatin C and AD, and support a model of cystatin C involvement in the process of neuronal death in AD.

CST3 is a susceptibility gene for late-onset AD, especially in patients aged 75 years and older. To our knowledge, CST3 B is the first autosomal recessive risk allele in late-onset AD.