The Liver X Receptor Ligand T0901317 Decreases Amyloid (cid:1) Production in Vitro and in a Mouse Model of Alzheimer’s Disease*

Recent studies indicate that oxysterols, which are ligands for the nuclear hormone liver X receptors (LXR), decrease amyloid (cid:1) (A (cid:1) ) secretion in vitro . The effect was attributed primarily to the ATP-binding cassette transporter A1 (ABCA1) transcriptionally up-regulated by li-gand-activated LXRs. We now examined the effect of the synthetic LXR ligand T0901317, which can be used in vivo , on A (cid:1) production in vitro and in APP23 transgenic mice. T0901317 applied to a variety of in vitro models, including immortalized fibroblasts from Tangier patients, and primary embryonic mouse neurons caused a concentration-dependent decrease in A (cid:1) secretion, and this effect was increased by the addition of apolipoprotein A-I. The inhi-bition of A (cid:1) production by T0901317 was cell-type specific, mAb) and anti-Ab 42 (G2–13 mAb) monoclonal antibodies conjugated to horseradish peroxidase (Genetics Company, Schlieren, Switzerland) were used as the detection antibodies. The reaction was developed by TMB Microwell peroxidase substrate system (Kierkegaard and Perry Laboratories, Gaithersburg, PA). The final values of A (cid:2) were based on standard curves using synthetic A (cid:2) 1–40 and A (cid:2) 1–42 peptide standards (Bachem Biosciences, King of Prussia, PA). The amount of A (cid:2) was normalized either to the total protein or to the expression of full-length APP as measured by Western blotting and was expressed as pmol/mg total protein or as a -fold of vehicle-treated control. Statistical Analysis— Results are reported as mean (cid:6) S.E. Statistical significance was determined by a two-tailed Student’s t test (GraphPad Prism version 3.0, GraphPad Software, San Diego, CA).

(AD). 1 Recent data demonstrate that vascular risk factors such as high total plasma cholesterol, specifically in the form of low density lipoprotein cholesterol, influence the progression or the incidence of AD (1)(2)(3)(4)(5)(6)(7). In contrast, plasma levels of high density lipoprotein (HDL) cholesterol are inversely associated with risk of cardiovascular disease (8) and AD (9,10). The ATP-binding cassette transporter A1 (ABCA1) is a major regulator of cholesterol efflux and HDL metabolism. Mutations in the ABCA1 gene cause severe HDL deficiencies, the most prominent of which is Tangier disease, characterized by the accumulation of cholesterol in cells and prevalent atherosclerosis (11)(12)(13)(14).
Genetic data support the hypothesis that ABCA1 may have a link to AD through its role in regulating HDL level. Wollmer et al. (15) found that the R219K variant of ABCA1, associated with increased HDL cholesterol levels and a decreased risk of atherosclerosis (16), delayed the age of onset of late onset Alzheimer's disease. Recently Katzov et al. (17) reported that a set of ABCA1 genetic variants modify the AD risk in Scottish, Swedish, and English populations. However, a third study of American and United Kingdom populations found that ABCA1 variants do not appear to influence the risk of late onset Alzheimer's disease (18). Experiments with any of those genetic variants in AD models systems, particularly transgenic or knock-in animals have not been reported, and so there is no definitive conclusion about the genetic association between ABCA1 and AD.
The transcriptional activation of ABCA1 is controlled by nuclear liver X receptors ␣ and ␤ (LXR␣/␤). LXR␣/␤ are members of the nuclear hormone receptor superfamily and transcriptionally regulate gene expression in response to ligand binding. The receptors are activated by oxidized forms of cholesterol and function as heterodimers with retinoic X receptors (RXR). Activated LXRs regulate genes involved in the uptake, catabolism, and transport of excess cholesterol and oxysterols in the body (19 -21). Thus LXRs act as molecular sensors of cholesterol levels and respond by inducing processes that reduce cholesterol levels.
Recently, by using primary cultures isolated from embryonic rat brain, we have demonstrated that oxysterols increased ABCA1 expression and function in different brain cell types. More importantly, we demonstrated that these ligands alone or in combination with apoA-I caused a substantial reduction in the steady state amounts of APP C-terminal fragments and decreased A␤ production (22). The results from this study were independently confirmed by two other laboratories (23,24). Thus, these effects of natural and synthetic LXR ligands provide important information about the design and implementation of a novel strategy to decrease A␤ amyloid secretion and consequently reduce the amyloid burden in the brain.
Here we demonstrate for the first time that pharmacological activation of LXRs by the synthetic ligand T0901317 (T0) reduces amyloidogenic processing of human APP in young APP23 transgenic mice. The results with immortalized fibroblasts from patients with Tangier disease provide additional support to our hypothesis that ABCA1 has a critical role in A␤ generation and secretion.
Antibodies-Rabbit polyclonal anti-ABCA1 antibody (Ab) was purchased from Novus (Littleton, CO). The 6E10 monoclonal Ab (Signet, Dedham, MA) recognizes the first 17 amino acids of the A␤ peptide. 6E10 Ab was used for Western blotting to detect full-length APP and soluble APP␣ and for immunoprecipitation of total A␤. Rabbit C8 polyclonal antibody (25) was used to detect CTF resulting from ␣or ␤-secretase cleavages. Rabbit 869 antibody (25) was used to detect sAPP␤ by Western blotting. This antibody does not recognize full-length APP but recognizes the neoepitope generated after cleavage by ␤-secretase. Murine-specific apoE and mouse anti-␤-tubulin antibodies were from Santa Cruz (Santa Cruz, CA). Glyceraldehyde-3-phosphate dehydrogenase monoclonal antibody was purchased from Chemicon International (Temecula, CA). Secondary antibodies conjugated to horseradish peroxidase were from Jackson ImmunoResearch (West Grove, PA).
Transgenic Mice-The study fully conformed to the guidelines outlined in the Guide for the Care and Use of Laboratory Animals from the United States Department of Health and Human Services and was approved by the University of Pittsburgh Institutional Animal Care and Use Committee. We used transgenic mice expressing APP751 with the Swedish familial AD double mutation (APPsw) at positions 670/671 (APPK670N, M671L), line APP23 (26). The expression of human APPsw is driven by the murine Thy-1 promoter and is restricted to neurons.
Treatments and Animal Tissue Processing-Eleven-week-old APP23 mice were treated by gastric gavage with 50 mg/kg/day T0901317 dissolved in propylene glycol/Tween 80 (4/1) for 6 days or with vehicle only. At the end of the treatment, the mice were perfused with 100 ml of phosphate-buffered saline, pH 7.4, and the brains were quickly removed and put on dry ice. Brains (150 mg/ml of buffer) were homogenized in Tris/sucrose buffer (250 mM sucrose, 20 mM Tris base, 1 mM EDTA, 1 mM EGTA, pH 7.4) as described previously (27). Protein extracts were prepared by a 1:1 dilution of the initial homogenate with RIPA buffer (10 mM Tris/HCl, pH 7.3, 1 mM MgCl 2 , 0.25% SDS, and 1% Triton X-100) in the presence of protease inhibitors (10 g/ml leupeptin, 10 g/ml aprotinin, and 10 g/ml AEBSF), and Western blots for ABCA1, ApoE, full-length APP, and CTF␣/␤ were made as before (22).
ABCA1 cDNA Cloning and Transfection Experiments-Human liver total RNA (Ambion, Austin TX) was used to amplify ABCA1 cDNA by SuperScript One step reverse transcription-PCR (Invitrogen) according to the protocols supplied by the vendors. Two fragments of ABCA1 cDNA with an overlapping KpnI recognition site at the 3Ј and 5Ј ends, respectively, were used for ligation and cloning into mammalian expression vector. The KpnI site is an internal one, 3789 nucleotides downstream from the transcriptional start codon. We used SacI and NotI recognition sites in pEGFP-N1 (Clontech) and generated the recombinant vector by replacing the sequence of EGFP. All sequencing reactions were performed at the University of Pittsburgh Biotechnology Center.

Replication-deficient Herpes Simplex Virus (HSV)-1 Viral Vectors for
Protein Expression in Primary Neuronal Cultures-We generated recombinant HSV-1 viral vectors for multigene delivery and expression in neurons simplifying the selection process and the monitoring for infection efficiency. Recombination of APPsw cDNA into the HSV-1 vector genome was performed in 7b cells, a Vero cell line capable of providing ICP4 and ICP27 IE gene products in trans. For this, the APPsw cDNA was first cloned into plasmid p41H that contains HSV UL41 flanking sequences for the HCMV IE promoter. Recombination of the APPsw cassette into the vector backbone at the UL41 locus yields plaques that are clear upon 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-gal) staining (lacZ) on the 7b complementing cell line. The recombinants were triple plaque-purified and verified by Southern blot analysis, using ϳ1.2 kb of biotin-labeled hAPP cDNA as a probe. Production and purification of high titer viral stocks was carried out according to our standard protocols (28), aliquotted, and stored at Ϫ80°C. Vector stocks were confirmed free of replication-competent herpes virus by the absence of plaques on infected (GFPϩ) non-complementing Vero cells. Following titering of vector stocks on 7b cells, primary rat hippocampal neurons were infected at a series of multiplicities of infection. Briefly, after washing with plain medium containing no additives, neurons in multiwell dishes were incubated for 90 min in neurobasal medium (Invitrogen) containing viral particles at a concentration corresponding to the predetermined multiplicity of infection at a given cell density/ unit of growth area. Neuronal expression of EGFP driven by the CMV promoter (same as in APPsw cassette) enabled easy and immediate monitoring of the infection efficiency as well as the early appearance of cytopathic effects. Under these conditions more than 75% of the cells were GFP-positive within 8 h post-infection.
Cell Culture-Primary neuronal cultures were made from dissociated cortices and hippocampi of 17-18-day-old C57BL/6J mouse embryos as described previously. Briefly, cortices and hippocampi were dissected and incubated with 1ϫ trypsin/EDTA (Invitrogen) for 10 min at room temperature. The trypsin was inactivated with complete defined neurobasal medium supplemented with B27, GlutaMax II, 10% horse serum, 5% fetal bovine serum, and antibiotics (medium and additives were from Invitrogen). A single cell suspension was obtained after filtering through a 70 m Falcon cell strainer (BD Biosciences). For all experiments neurons were plated at high density (2 ϫ 10 5 /ml, 1 ml/well) on poly-D-lysine (100 g/ml)-coated 12-well Costar plates (Stony Brook, NY). One hour after plating, the medium was changed with complete neurobasal medium as above, without serum. At day in vitro 2 the neuronal cultures were treated with cytosine-␤-D-arabinoside (4 M final concentration) to suppress proliferation of non-neuronal dividing cells.
CHOAPPsw cells (a gift from Dr. Ruth Perez, University of Pittsburgh, PA) were maintained in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and 500 g/ml Geneticin (Invitrogen) as described before (29). H4 (human neuroglioma) cells were maintained in Dulbecco's modified Eagle's medium with 10% fetal bovine serum. The establishment of telomerase-immortalized human skin fibroblasts from two unrelated homozygous patients with Tangier disease and a control patient was described elsewhere (30). Cells were grown to confluence in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum and used between 22 and 28 passages.
Transfection-For transfection cells were plated in 24-well Costar plates a day before transfection, at a density that would reach 80% confluence at the time of transfection. Cells were transfected with 1 g of recombinant vector DNA/well using Lipofectamine Plus in Opti-MEM (Invitrogen). Control cells were transfected with 1 g of vector cDNA/well. LXR/RXR Ligand Treatment-Cells were washed in Dulbecco's modified Eagle's medium containing 1% delipidated calf serum (Sigma) and treated for 24 h with LXR ligands 22(R)-hydroxycholesterol or T0901317 and RXR ligand retinoic acid in the presence or absence of cholesterol acceptor (apoA-I). ApoA-I was used at concentrations of 15-30 g/ml. Control cells received vehicle (ethanol) only. After incubation, the medium was centrifuged for 15 min at 12,000 ϫ g to remove any cells. The cells were washed and lysed in RIPA buffer.
Protein Isolation and Western Blot Analysis-For Western blot analysis cellular extracts were prepared from Chinese hamster ovary, H4 cells, Tangier fibroblasts, and primary neurons. Cells were washed and scraped in phosphate-buffered saline and lysed in RIPA buffer in the presence of protease inhibitors. Cellular extracts were centrifuged to remove debris. For ABCA1, APP, sAPP␣/␤ Western blot analysis, extracts containing 20 -50 g of total protein were mixed with Tris/glycine reducing buffer, denaturing loading buffer, loaded, and electrophoresed on 8% Tris/glycine gels (Invitrogen). For Western blot analyses of CTF and A␤ cellular extracts and conditioned media were loaded on 10 -20% Tricine gels (Invitrogen). Gels were transferred to nitrocellulose membranes, incubated with the respective primary antibodies followed by secondary antibodies conjugated to horseradish peroxidase and processed for visualization by enhanced chemiluminescence ECL Plus TM (Amersham Biosciences). Antibodies with anti-␤-tubulin and glyceraldehyde-3-phosphate dehydrogenase were used as an internal standard. The relative intensities of the bands were quantified by densitometry (Molecular Dynamics).
Immunoprecipitation and ELISA for A␤-Immunoprecipitation and ELISA for A␤ were performed essentially as before (31,32). Briefly, A␤ was immunoprecipitated from conditioned medium and immunoblotted using 6E10 antibody. Soluble A␤ was extracted with 2ϫ RIPA buffer and centrifuged at 135,000 ϫ g. In addition, soluble A␤ was extracted using cold 0.4% diethylamine in 100 mM NaCl, centrifuged at 135,000 ϫ g for 1 h at 4°C and neutralized by adding 0.5 M Tris/HCl, pH 6.8. Diethylamine extracts were also used to determine sAPP␣ and sAPP␤ by Western blot. The soluble A␤ extracts were additionally diluted in EC buffer (20 mM sodium phosphate, 2 mM EDTA, 400 mM NaCl, 0.2% bovine serum albumin, 0.05% CHAPS, 0.4% Block Ace, 0.05% NaN 3 , pH 7.0), and A␤ 1-40 and A␤ 1-42 concentrations were measured by Sandwich ELISA as described before. Briefly, ELISA for A␤ was performed using 6E10 as the capture antibody and anti-Ab 40 (G2-10 mAb) and anti-Ab 42 (G2-13 mAb) monoclonal antibodies conjugated to horseradish peroxidase (Genetics Company, Schlieren, Switzerland) were used as the detection antibodies. The reaction was developed by TMB Microwell peroxidase substrate system (Kierkegaard and Perry Laboratories, Gaithersburg, PA). The final values of A␤ were based on standard curves using synthetic A␤ 1-40 and A␤ 1-42 peptide standards (Bachem Biosciences, King of Prussia, PA). The amount of A␤ was normalized either to the total protein or to the expression of full-length APP as measured by Western blotting and was expressed as pmol/mg total protein or as a -fold of vehicle-treated control.
Statistical Analysis-Results are reported as mean Ϯ S.E. Statistical significance was determined by a two-tailed Student's t test (GraphPad Prism version 3.0, GraphPad Software, San Diego, CA).

The Synthetic LXR Ligand T0901317 Induces ABCA1 Expression and Decreases A␤ Secretion in Non-neuronal Cell
Lines-Previously, we and others have found (22,24) that the LXR ligand 22(R)-hydroxycholesterol increased ABCA1 expression and decreased the secretion of A␤ in non-neuronal and neuronal cells. Because it has been shown that the synthetic LXR ligands are more efficacious and less toxic in vivo than endogenous LXR ligands (33) we compared the effects of 22(R)hydroxycholesterol and the synthetic LXR ligand T0901317 on ABCA1 expression and A␤ secretion. We used Chinese hamster ovary cells (CHOAPPsw) stably transfected with human APP containing a Swedish mutation (APPsw695) that increases the formation of total A␤. T0901317 applied at the same molar concentration as 22(R)-hydroxycholesterol induced a higher level of ABCA1 protein expression (Fig. 1A). Because LXR acts in a heterodimeric complex with RXR, we examined whether a RXR activator, such as 9-cis-retinoic acid would affect the response to T0901317. The addition of retinoic acid to T0901317 resulted in a further 2-fold increase of ABCA1 protein expression (Fig. 1B, top panel) and similar decrease in Ab 40 secretion compared with T0901317 alone (Fig. 1C). These results confirmed that LXR and RXR ligands have an additive effect on ABCA1 expression and APP processing. The effect of T0901317 on A␤ 40 secretion as measured by ELISA was also concentrationdependent (Fig. 1C).
ApoA-I Treatment Increases the Effect of LXR Ligands on A␤ Secretion-To examine the effect of apolipoprotein-mediated cholesterol efflux, we treated CHOAPPsw cells with increasing concentrations of T0901317 with or without the addition of apoA-I ( Fig. 2A). Previously we have shown that treatment with apoA-I alone did not affect A␤ secretion (22). Total A␤ secretion was measured by Western blotting in aliquots of conditioned medium using the 6E10 antibody, which recognizes the first 16 amino acids of A␤ peptide, and was normalized to the expression of intracellular APP. T0901317 at 2 M concentration had no effect on A␤ production regardless of the addition of apoA-I ( Fig. 2A). T0901317 at 5 and 10 M inhibited A␤ secretion, and the addition of apoA-1 significantly increased the inhibitory effect of T0901317 as compared with cells without apoA-I (p Ͻ 0.01 and p Ͻ 0.05, respectively). The effect of apoA-I was confirmed in non-transfected human H4 cells expressing wild type APP treated with 22(R)-hydroxycholesterol. As visible from Fig. 2B, apoA-I and 22(R)-hydroxycholesterol treatment of H4 cells significantly decreased the secretion of total A␤. To examine how 22(R)-hydroxycholesterol and apoA-I affect the production of different A␤ species, we measured A␤ 40 and A␤ 42 by ELISA. We found that A␤ 1-40 was decreased more than 5-fold but detected no change in A␤  .
LXR Ligands Inhibit APP Processing in Primary Neuronal Cultures-To examine the role of LXR/RXR ligands on neuronal processing of APPsw, we used primary neurons from cortices and hippocampi of embryonic mouse brains. The neurons, plated at high density (5 ϫ 10 5 /ml) in 12-well plates were infected at day in vitro 4 with herpes simplex replicationdeficient vectors (HSVAPPsw). The expression of human APPsw was determined by Western blotting using the 6E10 antibody, which recognizes only human APP (Fig. 3A). On the next day the infected neurons were treated for 24 h with 22(R)-hydroxycholesterol and apoA-I, and Ab 40 secretion was determined by ELISA. 22(R)-Hydroxycholesterol when combined with apoA-I decreased A␤ 40 secretion more than 5 times (Fig.  3B). T0901317 applied alone had a slightly smaller effect on A␤ 40 secretion (Fig. 3C) even without the addition of apoA-I. The effect of T0901317 on A␤ 42 secretion was less prominent thus confirming the data obtained with H4 cells.

ABCA1 Overexpression in Transfection Experiments
Decreases A␤ Secretion-To determine whether the induction of ABCA1 might be responsible for a decreased secretion of A␤, we transiently transfected CHOAPPsw cells with ABCA1 and incubated cells with or without apoA-I. The transfection of ABCA1 resulted in expression levels of ABCA1 protein comparable with those induced by LXR/RXR activators (Fig. 4A, top  panel) without affecting cellular APP levels (middle panel). Fig.  4B shows that overexpression of ABCA1 caused a statistically significant decrease in A␤ 40 levels, and the addition of apoA-I simultaneously to ABCA1 transfection caused a more than 2-fold decrease as compared with the mock-transfected cells.
The overexpression of ABCA1 alone did not cause a statistically significant decrease in secretion of A␤ 42 , although the addition of apoA-I resulted in a more than 4-fold decrease in A␤ 42 secretion. Therefore, the overexpression of ABCA1 decreased A␤ secretion in a similar fashion as treatment with LXR/RXR ligands suggesting that the effect of the ligands could be mediated through ABCA1. Mutations in ABCA1 Increase A␤ Secretion-Next we examined whether a functional ABCA1 protein was required for the effect of LXR/RXR ligands on A␤ secretion. For these experiments we used permanent cell lines derived from Tangier patients. The TT1, TT2, and CT cell lines are telomerase immortalized skin fibroblasts from two Tangier patients and a control subject (30). TT1 was derived from T1 patient homozygous for an asparagine to serine amino acid substitution in exon 19 (AAT/AGT, amino acid position 935 of the primary translation product) resulting in the expression of full-length, albeit nonfunctional, protein (Fig. 5B, middle panel). The TT2 cell line was established from the T2 patient characterized by a homozygous 1-bp deletion in exon 1. This deletion introduces a stop codon at position 575, resulting in the omission of the majority of the ABCA1 protein sequence and expression of a truncated protein (Fig. 5B, bottom panel). First, we examined the basal level of A␤ in medium from Tangier and control fibroblasts conditioned for 48 h without treatment. Fig. 5A demonstrates that the Tangier cell lines (TT1 and TT2) produced significantly more A␤ 40 as compared with normal fibro-blasts (CT). In addition there was a statistically significant difference between Tangier cells, with TT2 secreting more A␤ 40 than TT1 cell line. Because the expression of endogenous APP in fibroblasts was very low we were not able to determine the secretion of A␤ 42 .
To examine the effect of T0901317 on A␤ 40 and A␤ 42 secretion, Tangier fibroblasts were infected with HSVAPPsw and were treated with T0901317 and retinoic acid without apoA-I. This treatment increased the expression of ABCA1 protein in CT and TT1 cells by 4-and 6-fold, respectively (Fig. 5B, top and middle panels). As expected there was not detectable ABCA1 protein in TT2 (Fig. 5B, bottom panel). The secretion of A␤ 40 and A␤ 42 was decreased only in the control fibroblasts and was not changed in TT2 cells (Fig. 5, C and D). The decrease in control fibroblasts was modest but statistically significant. Surprisingly, ligand treatment in TT1 cells induced a statistically significant increase in the secretion of amyloid beta (A␤ 40 and A␤ 42 ). The effect of T0901317 on control and Tangier cells was confirmed for A␤ 40 in non-infected cells expressing the endogenous APPwt (data not shown). Therefore, the existence of full-length functional ABCA1 is required to mediate the inhibitory effect of LXR/RXR ligands on A␤ secretion.
In Vivo LXR Ligand Treatment Increases Brain ABCA1 Expression and Influences Amyloidogenic Processing of Human APP in APP23 Mice-To examine the role of T0901317 on ABCA1 expression in vivo, we treated 11-week-old APP23 mice orally by gastric gavage for 6 days with 50 mg/kg/day T0901317. Control mice received vehicle only. The treatment of APP23 mice resulted in a substantial increase in the expression of ABCA1 (more than 3-fold, p Ͻ 0.001), whereas the expression of APP was unchanged (Fig. 6, A and B).
To examine the effect of T0901317 on APP processing, we determined the amount of sAPP␣ and sAPP␤ by Western blotting. sAPP␣ and sAPP␤ were determined in diethylamine extracts, because this technique separates soluble A␤ and sAPP from membrane-associated full-length APP and CTF resulting from ␣and ␤-secretase cleavages (CTF␣ and CTF␤). Fig. 6B  (graph panel) shows that the ratio of sAPP␤-to ␣-secretase cleavage product was significantly decreased after T0901317 treatment. This was in agreement with our previous in vitro data demonstrating that LXR ligands increase ␣-secretase cleavage and sAPP␣ secretion (22). T0901317 did not induce a significant change in the level of CTFs (Fig. 6C), which was in contrast with the in vitro effect of LXR/RXR ligands (22). Because LXR/RXR ligands are known to regulate the expression of apoE, we also examined the level of apoE protein in brain homogenates after T0901317 treatment. As visible from Fig.  6D the ligand treatment did not cause significant change in the apoE protein level.
Finally we examined the effect of T0901317 on A␤ production. Because at this age A␤ in the brain of APP23 mice is soluble in SDS, A␤ was extracted from the initial brain homogenate by SDS containing RIPA buffer or diethylamine, and the results from the two determinations compared. Fig. 7, A  and B show the levels of RIPA-extracted A␤ 40 and A␤ 42 in APP23 mice were significantly decreased. Similar results were obtained by ELISA determinations of diethylamine-extracted soluble A␤ 40 and A␤ 42 (not shown), thus confirming that amy-FIG. 6. T0901317 increases the expression of ABCA1 in the brain and affects APP processing in APP23 mice. T0901317 (T0) (50 mg/kg/day) was applied for 6 days to 11-week-old APP23 mice. A, the expression of ABCA1 after T0901317 treatment was examined by Western blotting. Quantification of ABCA1 protein level, normalized to the level of ␤-tubulin expression is presented below. B, the levels of sAPP␣ and sAPP␤ were determined by Western blotting, and bands quantified and presented as the ratio of sAPP␤/sAPP␣. C, Western blotting of CTF␣ and CTF␤. Quantification of CTF␤ and CTF␣, normalized for APP expression is presented below. D, Western blotting for apoE after T0901317 treatment. The expression level of apoE is normalized to the expression level of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Data are mean Ϯ S.E., n ϭ 10 mice/group. Veh, vehicle. loidogenic processing of human APP in APP23 mice was decreased after T0901317 treatment. DISCUSSION Epidemiological data suggest that vascular risk factors such as high blood pressure and high total plasma cholesterol (1, 2) increase the risk not only for vascular dementia but also for late onset Alzheimer's disease (4,7). In addition, low HDL cholesterol and serum apoA-I concentrations are highly correlated with the severity of AD (5,9,10). In mice overexpressing human APPsw, diet-induced hypercholesterolemia correlated with brain A␤ deposits (34). The amyloid burden correlated inversely with plasma levels of HDL (35). ABCA1 is a major regulator of cholesterol efflux and HDL metabolism, and its transcriptional activation is controlled by nuclear liver X receptors ␣ and ␤. A detailed examination of LXR double knockout mice has shown that LXRs have an important function not only in lipid homeostasis in the brain, but that the loss of these receptors results in neurodegenerative disease (36).
Recently, we and others showed (22)(23)(24) that ligands for LXR␣/␤ increased ABCA1 expression in neuronal cells and reduced A␤ secretion. In the present study, we examined the effect of the synthetic LXR ligand T0901317 in vitro and in vivo in a mouse model for AD. To examine the effect of T0901317 on APP processing in vitro, we used non-neuronal and neuronal cell lines and primary neuronal culture. In all these models T0901317 increased ABCA1 expression and decreased A␤ secretion, confirming our previous results with hydroxysterols. Furthermore, apoA-I led to an additional decrease in secreted A␤, suggesting that lipid efflux and related decrease in intracellular cholesterol concentration are important factors in APP processing. Several important observations should be noted, however. Although the increase in ABCA1 expression caused by LXR/RXR was comparable in primary neurons and nonneuronal cells, there were cell-type-specific differences in the magnitude of T0901317-induced effect on A␤ secretion. Ligandmediated suppression of A␤ secretion was robust in primary neurons and neuronal cell lines but much less pronounced in fibroblasts suggesting that additional tissue specific factors are needed to mediate the LXRs/ABCA1 effect on A␤ secretion.
The experiments with Tangier fibroblasts, which lack a functional ABCA1 protein, clearly demonstrated that ABCA1 affects APP processing. The fact that A␤ secretion was lower in control fibroblasts than in Tangier cells demonstrates that the decrease in A␤ is related to an intrinsic cellular activity of ABCA1. Furthermore, the effect of LXR/RXR ligands required a functional ABCA1 (absent in both Tangier cell lines) to affect A␤ secretion. We were surprised, however, to find that T0901317 increased A␤ secretion in TT1 cell line, which expresses a full-length but non-functional ABCA1 protein. It is possible that increased expression of a non-functional ABCA1 has a dominant negative effect on another cellular component that limits the secretion of A␤. If so, ABCA1 may function as part of a multicomponent complex that inhibits the secretion of A␤. The amino acid substitution in TT1 (asparagine to serine) affects the conserved Walker A motif of the N-terminal ATP binding fold. This domain may affect ATP binding and hydrolysis but may also be critical for functional interactions of ABCA1 with intracellular effector molecules and components involved in trafficking and secretion. For example, Rac1, which is a key regulator in the secretion of the non-amyloidogenic sAPP␣ (37), was shown to accumulate in primary fibroblasts of the T1 pedigree (38). It is conceivable that a mutant ABCA1 may affect such protein and that a point mutation may result in a more severe phenotype than complete ABCA1 deficiency. In this context, it is interesting to note that one of the two homozygous Tangier disease patients of the T1 pedigree developed signs of severe dementia and amyloid depositions at age 60. 2 However, none of the three homozygous Tangier patients of the T2 family have signs of dementia. By contrast, the asparagine to serine mutation of T1 is apparently not associated with premature atherosclerosis (39), whereas all three homozygous patients of the T2 pedigree suffer from severe premature atherosclerosis (30). Altogether these biochemical and clinical data suggest (a) that ABCA1 acts in concert with an unknown component in secretion of A␤, (b) that the genotype of the ABCA1 mutations may determine the clinical phenotype observed in these patients, and (c) that the influence of ABCA1 on AD and athrogenesis is mediated by at least partly different mechanisms.
Because the ratio of secreted sAPP␣ over sAPP␤ was increased in the brain of T0901317-treated mice and A␤ 1-40 and A␤ 1-42 was decreased, our results suggest that the synthetic LXR ligand increases the non-amyloidogenic processing of APP. Previously, we and others found (22, 24) a significant decrease in a steady state level of CTF␣/␤ after in vitro treatment with LXR/RXR that was not confirmed after in vivo treatment. The reason for this discrepancy is not clear, but similar results have been described before (40). The complexity 2 Clinical and autopsy data of the patient will appear elsewhere.  40 and A␤ 42 in APP23 mice. T0 (50 mg/kg) was applied for 6 days to 11-week-old APP23 mice and A␤ 40 (A) and A␤ 42 (B) were measured by ELISA in RIPA brain extractions as described under "Materials and Methods." Results are mean Ϯ S.E. and compared by two-tailed Student t test. n ϭ 10 mice/group; *, p Ͻ 0.05; Veh, vehicle. of the changes induced by T0901317 treatment may also result in an alteration of the complete degradation of CTFs in the treated animals.
Previous studies have shown that LXR/RXR ligands increased the expression of apoE in macrophages and adipocytes (41,42). However the role of LXR on apoE expression in the brain is controversial. Whitney et al. (43) failed to observe an effect of T0901317 on apoE expression in the brain, whereas another group has shown a dramatic increase in the expression of apoE after T0901317 treatment in human astrocytoma cells but a modest effect on mouse apoE expression in vitro and in vivo (42). In the present study we also did not find a significant change in apoE protein expression after T0901317 treatment in vivo, in agreement with Whitney et al. (43). The reasons for the discrepancy between human and mouse brain cells are unknown, but it is possible that different regulatory elements of apoE expression existing in humans are absent in mice.
A␤ normally is present in cerebrospinal fluid and in plasma, associated with apolipoproteins and lipoproteins (44 -49), and its clearance from the brain is mediated by low density lipoprotein receptors (50). A lack of ABCA1 in humans and mice causes abnormal lipidation and increased catabolism of HDL, resulting in virtual absence of plasma apoA-I and HDL. Experiments with neuronal cell lines have shown that apoA-I binds A␤ and decreases its aggregation and toxicity (51). Therefore, the absence of HDL and apoA-I may complicate the clearance of A␤ from the brain. Recently two independent groups demonstrated that ABCA1Ϫ/Ϫ mice have dramatic decrease in apoE levels in the whole brain and cerebrospinal fluid (52,53). ApoE is a major brain apolipoprotein and the ApoE4 genotype is a risk factor for AD and cerebral amyloid angiopathy. ApoE is considered an A␤ chaperone that influences A␤ clearance and fibrillogenesis. The role of apoE as a pathological chaperone is demonstrated in apoEϪ/Ϫ mice that accumulate much less fibrillar amyloid than wild type apoE mice (54,55). However, because in ABCA1Ϫ/Ϫ mice the two major brain apolipoproteins apoE and apoA-I are virtually missing, it is difficult to predict how the lack of ABCA1 will affect the AD phenotype.
The critical role of LXR and ABCA1 in central nervous system and APP metabolism is reinforced by the data presented here, thus suggesting that ABCA1 may be a novel therapeutic target for AD. Cellular and in vivo studies should provide a critical proof of concept for the regulation of APP processing by LXR through the transcriptional control of ABCA1, apoE, and possibly other genes in support of the search for potent and specific LXR ligands with properties allowing therapeutic application.