The FASEB Journal express article 10.1096/fj.03-1490fje. Published online May 20, 2004. Association of ABCA2 expression with determinants of Alzheimer's disease

Zhijian J. Chen,* Bojana Vulevic,* Kristina E. Ile,* Athena Soulika,* Warren Davis, Jr.,* Peter B. Reiner,† Bruce P. Connop,† Parimal Nathwan,† John Q. Trojanowski,‡ and Kenneth D. Tew*

*Department of Pharmacology, Fox Chase Cancer Center, Philadelphia, Pennsylvania 19111; †Active Pass Pharmaceuticals Inc., Vancouver, BC Canada V5Z 4H5; ‡Center for Neurodegenerative Disease Research, Institute on Aging and Dept. of Pathology and Laboratory Medicine, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania 19104- 4283

Corresponding author: K. D. Tew, Department of Pharmacology, MUSC, 173 Ashley Avenue, Charleston, SC 29425. E-mail: [email protected]

ABSTRACT

With the use of a novel method for detecting differential expression, alterations in functional gene clusters related to transport or oxidative stress response and β-amyloid (Aβ) peptide metabolism were identified in a HEK293 cell line engineered to overexpress the human ATP binding cassette transporter ABCA2. These included fatty acid binding , phospholipid binding protein, phospholipid synthesis protein, transporter cofactors, seladin-1, Aβ precursor protein (APP), vimentin, and low-density lipoprotein receptor-related protein. ABCA2 was highly expressed in neuroblastoma cells and colocalized with Aβ and APP. Additionally, increased APP protein levels were detected within ABCA2/APP double-transfected cells, and increased Aβ was detected in the media of ABCA2-transfected cells relative to controls. The transporter was abundant in the temporal and frontal regions of both normal and Alzheimer’s disease (AD) brain but was detected at lower concentrations in the parietal, occipital, and cerebellar regions. The ABCA2 transfected cell line expressed resistance to a free radical initiator, confirming involvement in protection against reactive oxygen species and suggesting a further possible link to AD.

Key words: β-amyloid • Aβ precursor protein • AD

he 49 of the human ABC transporter family are characterized by the presence of an ATP binding cassette unit, which is composed of Walker A and B motifs and an ATP Tbinding cassette signature between them (updates: http://nutrigen.4t.com/humanabc.htm). The ABC family has been divided into seven subfamilies, ABCA to ABCG. Although substrate specificities and functions have yet to be ascribed for the majority of these transporters, there is a growing literature linking mutations with human disease states. For example, ABCA4 has been associated with Stargardt disease (1) and mutations in ABCA1 have been linked to Tangier disease and familial high-density cholesterol deficiency syndrome (2).

An initial indication of a function for human ABCA2 was gained from the observation of a gene amplification in a region of 9q34 in an estramustine-resistant human ovarian carcinoma cell line (3). Manipulation of the expression of ABCA2 by transfection or antisense treatments caused a corresponding increase or decrease in resistance to estramustine (3, 4).

Page 1 of 21 (page number not for citation purposes) ABCA2 expression has also been associated with myelination and rat nervous system development (5). The expression of ABCA2 in macrophages has been found to respond to cholesterol loading (6). Since ABCA2 is localized in the endolysosomal compartment of the cell (4), it seemed plausible that it may serve to sequester multiple substrates into these vesicles and influence how material is processed within this compartment.

Our initial goal was to identify those gene products that may be altered in the cellular response to the forced expression of high levels of ABCA2. To this end, standard DNA microarray technology is somewhat limited by low sensitivity for genes with small quantitative expression changes (7). Amplified differential gene expression (ADGE) microarray, developed in this laboratory, magnifies the ratios of differential gene expression (8, 9) and improves detection sensitivity and fidelity, while reducing the requirement for biological material and maintaining high throughput. This method has been successfully applied to a drug-resistant cell line resulting in characterization of genes that are overexpressed in the drug-resistant phenotype (9). In the present study, ADGE microarray was used to profile the gene expression of an ABCA2- transfected cell line. The data indicate that concomitant adaptations in expression of transport and oxidative stress response-related gene clusters occur. Oxidative stress, indicated by the generation of reactive oxygen species leading to protein oxidation and lipid peroxidation, has been found to be elevated in Alzheimer's disease (AD) brain tissue (10, 11). Cells exposed to a segment of the amyloid β-peptide (Aβ, 25-35) develop detectable levels of free radicals in minutes, and these can initiate lipid peroxidation (12). In addition, Aβ1-42 causes protein oxidation and death of neuronal cells (11). An indication from our present analysis is that increased expression of ABCA2 may be causally linked with altered expression of genes associated with the pathogenesis of AD.

METHODS

Tissues and cell lines

The A16 cell line was derived from the HEK293 cell line by transfection with a 7304 bp Hind III/EcoR1 ABCA2 fragment subcloned into the pcDNA 3.1 vector (Invitrogen, Carlsbad, CA). The C1 cell line is the parental HEK293 transfected with the pcDNA3.1 vector (4). Both cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 50 µg/ml streptomycin, 50 units/ml penicillin, 2 mM glutamine, 10% (v/v) fetal bovine serum, and 0.9 mg/ml G418.

Neuroblastoma cells

Human BE(2)-M17 neuroblastoma cells were obtained from the American Type Culture Collection (ATCC, Rockville, MD) and maintained in 1:1 modified Eagle’s medium/Ham’s F12 supplemented with 10% fetal bovine serum and 100 mg/ml streptomycin/penicillin at 37°C, 5% CO2.

ABCA2 and APP double-transfected cells

293 EBNA (Invitrogen, Carlsbad, CA) cells stably transfected with wild-type amyloid precursor protein-695 (WT6 cells) were cultured in DMEM supplemented with sodium pyruvate (1 mM) and 10% fetal calf serum. Cells were plated at a density of 105 cells per well in 35 mm2 culture dishes (Falcon™) 18 h before transient transfection. Cultures were then transfected for 48 h with the use of a Fugene-6 transfection procedure (Boehringer Mannheim, Laval, QC) that entailed the use of 2 µg DNA at a DNA:Fugene-6 ratio of 1:3. Control cells were transfected with a Page 2 of 21 (page number not for citation purposes) construct referred to as pCEPβGal, which is a vector backbone composed of pCEP4 (Invitrogen) into which the LacZ sequence was inserted. An ABC Cntl cell line was developed by transfecting WT6 cells with an ABC transporter of approximately the same size as ABCA2 cloned by Active Pass Pharmaceuticals and subcloned in the pCEP4 vector. The ABCA2 cell line was developed by transfecting WT6 cells with ABCA2 cDNA in the pCEP4 vector.

ADGE microarray

Both A16 and C1 cells were harvested at ~80% confluence and used for RNA isolation with RNeasy Maxi Kit (Qiagen, Valencia, CA). The total RNA was used for ADGE microarray (9). Briefly, 10 µg of total RNA were reverse-transcribed into single-stranded cDNA with oligo(dT)12-18. Then the double-stranded cDNA for C1 and A16 was generated with the cDNA Synthesis System (Life Technologies, Rockville, MD). After phenol extraction and ethanol precipitation, the cDNA was resuspended in 25 µl of ddH2O. Both C1 and A16 cDNA were digested with 3 µl (30 units) of the restriction enzyme TaqI in a final volume of 30 µl. The TaqI fragments of C1 cDNA were ligated with the CT adaptor at 16°C overnight with 3 µl (9 units) of T4 ligase (Promega, Madison, WI), while the TaqI fragments of A16 cDNA were ligated with the TT adaptor. Fifteen microliters each of the adapterized C1 cDNA and A16 cDNA were mixed with 30 µl of 2x HB buffer in a final volume of 60 µl, denatured at 95°C for 5 min, and annealed at 68°C for 20 h. The reassociated DNA was used for generating the C1 and A16 probe DNA. To generate C1 probe, a PCR reaction was set up with 1 µl of the reassociated DNA, 5 µl of 10x Clontech PCR buffer, 1 µl of 10mM dNTPS, 2 µl of 10 µM CT primer, 1 µl of Clontech cDNA polymerase, and 34 µl of ddH2O. In the PCR reaction for the A16 probe DNA, TT primer was used instead of CT primer. The reaction cycling conditions were 72°C for 5 min (for filling in the adaptor ends), 94°C for 1 min; then 25 cycles of 94°C for 30 s, 66°C for 30 s, 72°C for 60 s; and then 72°C for a 5 min extension. Three PCR reactions were set up for each probe. The PCR reactions were purified with Qiagen PCR purification kit and then incorporated with aminoallyle-dUTP (aa-dUTP) by using the following PCR reaction: 42 µl of the C1 elute, 5 µl of 10x Clontech PCR buffer, 1 µl of 10mM aa-dUTP/dNTPS (6 mM aa-dUTP, 4 mM dTTP, 10 mM of dGTP, dCTP, and dATP each), 1 µl of 10 µM CT primer, and 1 µl of Clontech cDNA polymerase. For the A16 elute, TT primer was used instead of CT primer. The reaction cycling conditions were 94°C for 1 min; then three cycles of 94°C for 30 s, 64°C for 30 s, 72°C for 60 s; and then 72°C for 2 min extension. Three reactions of C1 or A16 DNA were combined together, then purified with ethanol precipitation, and resuspended in 7.5 µl coupling buffer. The C1 and A16 probes were coupled with the same volume of Cy3 and Cy5 dyes, respectively, at room temperature. The coupling reaction was ended by adding 50 µl 100 mM sodium acetate, purified with Qiagen PCR purification kit, eluted into 50 µl warmed H2O (45°C), and reduced to a final volume of 7.5 µl with a speed vacuum. The C1 and A16 probe was mixed with 3.75 µl of 20x SSC, 0.75 µl of 10% SDS, 1.5 µl of 1 µg/µl salmon DNA, and 1.5 µl of 50x Denhardt’s solution, denatured at 95°C for 5 min, cooled on ice, and incubated at 42°C for 15 min. The denatured Cy3 (C1) and Cy5 (A16) DNA were mixed and loaded onto a microarray chip. A hybridization chamber was assembled with the microarray chip and submerged in a water bath at 60°C overnight. The microarray chip was washed in wash buffer I (2x SSC, 0.1% SDS) for 5 min, then in wash buffer II (1x SSC) for 5 min, and wash buffer III (0.2x SSC) for 5 min. The slide was dried by centrifuging at 650 rpm for 5 min and scanned with Affymetrix 428 array scanner using the Cy3 and Cy5 channels. Two replicates were performed on the three sets of human microarray chips containing 45,000 genes in total (chips were made at the Fox Chase Cancer Center Microarray Facility).

Page 3 of 21 (page number not for citation purposes) Analysis of microarray data

The spots from the microarray images were quantified with ImaGene4.1 (Biodiscovery, Los Angeles, CA). The Cy3 and Cy5 data were integrated into a dataset and transformed with GeneSight3.0 (Biodiscovery) by using the following sequence: local background correction, removal of flagged spots, logarithm of base 2, ratio calculation, and linear regression normalization. The transformed data were exported into Microsoft Excel. M values [log2(Cy5/Cy3) = log2Cy5- log2Cy3] were calculated for each gene. M represents the ratios in the power of 2, with positive values for up-regulated genes (Cy5/Cy3), negative values for down- regulated genes (Cy3/Cy5), and zero for unchanged genes. Differentially expressed genes were identified, and their functions, where known, were annotated.

RT-PCR

Quantification of seven genes (4 up-regulated, 3 down-regulated), along with the actin gene, was performed to verify the results of ADGE microarray with RT-PCR. Primers for RT-PCR were designed with Software OLIGO 4.0 based on the sequences of the genes corresponding to the identified spots on the chip. The total RNA samples of C1 and A16, used for ADGE microarray, were reverse-transcribed with Superscript II reverse transcriptase. Serial dilutions were made for templates to ensure amplification in the linear range. For genes examined by gel quantification of RT-PCR bands, the cDNA templates of the two samples were normalized to expression of β- actin. For each gene, specific PCR cycle conditions were selected to optimize the levels of differential expression. PCR results within the linear range were selected, and bands were quantified using the Kodak 1-D image analysis software program (Scientific Imaging Systems, Eastman Kodak Company). For genes examined by real time PCR, cDNA and primers were prepared the same way. Reactions were performed on the Cepheid Smart Cycler using the Qiagen Sybr Green kit.

Quantification of expression of the selected genes was calculated as described by Su et al. (13). Briefly, the CT value, or the cycle at which significant levels of PCR product are detected, is determined for each gene and the calibrator (18S). The relative CT for each set of primers was -∆∆CT determined using the Eq. 2 , where ∆CT is determined by subtracting the 18S rRNA CT value from the gene CT value. ∆∆CT was calculated by subtracting the average calibrator (C1) ∆CT from the tester (A16) ∆CT. Three trials for each PCR were performed, and values were averaged.

Cytotoxicity assays

The cytotoxicity of AAPH (2,2′-azobis-(2-amidinopropane)) was determined using a colorimetric assay. Briefly, C1 and A16 cells were plated at 2000 cells per well in a 96-well plate and allowed to grow for 3-4 h. Serial dilutions of AAPH were added, and cells were cultured at 37°C with 5% CO2 for times indicated in the figure legends. At the end of the incubation, cells were fixed by adding TCA for a final concentration of 10% and were incubated at 4°C for 1 h. TCA was removed, and the plate was air dried. Fixed cells were treated with 0.4% sulforhodamine B (SRB)/1% acetic acid in each well and incubated at room temperature for 30 min. Excess SRB was removed by washing with 1% acetic acid, and plates were air dried. Finally, 10 mM unbuffered Tris base were added to each well, and the plate was read at 560 nm on a spectrophotometer. Average readings of each row were calculated and converted to percentage of control.

Page 4 of 21 (page number not for citation purposes) Immunofluorescence and confocal microscopy

Cells were prepared for immunofluorescence by standard methods. Fixation was accomplished by -20°C methanol or by 4% paraformaldehyde with 0.1% Tween 20. Primary antibodies used were a polyclonal rabbit anti-ABCA2 and monoclonal antibodies against APP (Zymed Laboratories, Inc., San Francisco, CA) and Aβ (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) at dilutions of 1:100. Secondary antibodies used were species specific for the primary antibodies and were conjugated with Rhodamine Red X (ABCA2) and Cy2 (APP or Aβ; Jackson ImmunoResearch Laboratories, Inc., West Grove, PA). A Mowiol antifade in glycerol was used in an aqueous mounting medium.

Immunofluorescently stained cells were imaged using a Bio-Rad MRC-600 laser scanning confocal microscope (LSCM). With the use of the two photomultiplier (PMT) setting on the LSCM, dual acquisition of Cy2 and Rhodamine Red X was accomplished. Optical sections were acquired with 60x and 100x objectives using a 0.5 µM step size. The volumes were rendered using Voxel View (Vital Images, Plymouth, MN).

ABCA2 immunoblot quantitation

To determine relative levels of ABCA2 in white and gray matter of the five lobes, immunoblots were performed on brain samples from these regions. Frozen brain samples were used from patients with no brain pathology, AD, or schizophrenia confirmed by postmortem examination as described by Schmidt et al. (14). White and gray matter were separated manually, placed in high salt buffer (0.75 M NaCl, 0.02 M NaF in TBS) with proteolysis inhibitors, and homogenized. Insoluble material was removed by centrifugation at 100,000 g, and the supernatant was separated by SDS-PAGE. SDS gels were transferred to nitrocellulose membranes and probed with a polyclonal anti-ABCA2 antibody, and appropriate secondary antibody and bands were detected using the ECL system (Amersham Biosciences). The membranes were also probed with an anti-actin primary and appropriate secondary antibody for quantitation/standardization. Images of the blots were scanned and densitometry was measured using the Kodak 1-D image analysis software program (Scientific Imaging Systems, Eastman Kodak Company). Lanes and bands were found by the program and adjusted if necessary.

APP detection

After the 48 h transient transfection with ABC Cntl and ABCA2, the double-transfected cells were washed once with warm PBS (37°C) and then exposed to DMEM supplemented with sodium pyruvate (1 mM) for 4 h. The cells were then washed once with PBS; harvested in 100 µl of ice-cold lysis buffer containing 20 mM MOPS (pH 7.2), 5 mM EDTA, 0.01% Nonidet P-40, 75 mM β-glycerol phosphate, and a cocktail of protease inhibitors (Boehringer Mannheim); and sonicated on ice for 8 s using a probe sonicator. Cellular APP levels were quantitated by 10% Tris-Glycine SDS-PAGE Western blot analysis using an anti-APP N-terminal antibody (22C11, Boehringer Mannheim; 15, 16). Immunoreactive bands were visualized using ECL detection (Amersham, Oakville, ON) and quantified with standard densitometry.

Aβ detection

To examine Aβ levels, the harvested culture media were subjected to trichloroacetic acid precipitation. The pellet was resuspended in Laemmli buffer, and the normalized levels of protein were subjected to electrophoresis on a 16.5% Tris-Tricine sodium dodecyl sulfate gel as described previously (15, 16). Total Aβ was measured using monoclonal antibody 6E10 (Senetek

Page 5 of 21 (page number not for citation purposes) Research, Maryland Heights, MO). Bands were visualized using ECL detection (Amersham) and analyzed by standard densitometric techniques.

RESULTS

Transport function and oxidative stress response gene cluster identification

With the use of a screening strategy that permitted the analysis of 40,000 genes on three sets of chips, 152 genes were detected with greater than threefold changes (M>1.5) in replicate samples of control and ABCA2-transfected cells (Table 1). Among them, 100 genes were up-regulated and 52 genes were down-regulated. Among 68 annotated genes, 22 were associated with transport functions and 6 with oxidative stress response and/or pathogenesis of AD (Table 2). Expression changes of these genes were consistent among replicates, and differential expression for a selection of the genes was quantifiably confirmed by RT-PCR (Table 3).

As shown in Table 2, the 22 genes from the transport-related group can have roles in transport, membrane composition, substrate binding, and metabolism. For example, syntaxin 11 is a member of the syntaxin family, with a potential role in regulating vesicle docking and fusion (17). Syntaxin 1A has been shown to interact with the GABA transporter GAT1 in rat brain (18), and syntaxin 6 may be involved with glucose transport through the transporter Glut4 (19). Therefore, it is plausible that syntaxin 11 may act as a cofactor for ABCA2. The OCSYN protein has been shown to interact with syntaxin (20). CACNA2D2 is a voltage-dependent calcium channel. DOC2, with double C2-like domains, interacts with Ca2+, and phospholipids and may have a role in Ca2+-dependent intracellular vesicle trafficking in cells (21). Stomatin is a major lipid-raft component of platelet α granules. Lipid rafts have cholesterol and sphingolipid-rich domains that are also involved in intracellular trafficking (22).

Several substrate binding genes were identified. Chromogranin A has a role in steering secretory into a regulated pathway. Expression of the gene is regulated through MAP kinase pathways and is influenced by secretin levels (23). Copine I is a Ca2+-dependent phospholipid binding protein. Expression of fatty acid binding protein 7 was increased while fatty acid binding protein 5 was decreased. Autotaxin is involved in the synthesis of the bioactive phospholipid, lysophosphatidic acid. SULT1A3 and SULT1A2 catalyze sulfate conjugation in the metabolism of drugs and hormones, and angiopoietin-like 3 has a role in regulating lipid metabolism (24), suggesting that phospholipid may be a candidate substrate for ABCA2.

Six genes linked with oxidative stress response and pathogenesis of AD were identified (Table 2). For example, seladin-1 is an oxido-reductase that has been shown to confer resistance to oxidative stress and its expression has been shown to be down-regulated in the affected areas of AD (25, 26). Vimentin is an oxidation sensitive protein, and is known to be up-regulated in cells exposed to Aβ (27). Slc23a1 is a sodium-dependent vitamin C transporter. Both APP and low- density lipoprotein receptor-related protein have been correlated with the pathogenesis of AD (28, 29). Calcineurin is a calcium- and calmodulin-dependent protein phosphatase and has also been linked with tau hyperphosphorylation in AD (30).

ABCA2 association with oxidative stress response and AD

Although the observed changes in transport-related gene expression after ABCA2 transfection were not unexpected, the apparent connection between ABCA2 and oxidative stress/AD was interesting enough to merit further study. Initially, cytotoxicity assays showed that ABCA2- transfected cells were more resistant than mock transfected to oxidative stress induced by AAPH

Page 6 of 21 (page number not for citation purposes) (Fig. 1). This chemical generates free radicals that cause lipid, protein, and nucleic acid oxidation and peroxidation. We have previously shown that ABCA2 transcript levels are high in brain tissue (4). Autopsy samples of human brain tissue were analyzed by immunoblot of SDS-PAGE separated protein, and immunoreactive bands were scanned and quantified using actin as loading control. Densitometric quantitation showed that ABCA2 levels were highest in temporal and frontal regions of the AD brain, with lower, but significant levels, in parietal, occipital, and cerebellar regions (Table 4); however, similar ABCA2 protein expression patterns were also observed in schizophrenia samples. These data suggest that although no significant quantitative differences were apparent, qualitative expression of the transporter in specific brain regions does occur. Further, the AD pathology is variably distributed within affected regions (e.g., frontal and temporal lobes of AD brains), which is readily evident by microscopy. Although immunohistochemistry did not reveal any concentrated expression within these regions (data not shown), the immunoblots do reflect average changes in an entire brain homogenate that may obscure potentially significant focal changes in protein concentrations in the AD brain.

ABCA2 colocalization and coexpression with APP and Aβ

Confocal microscopy images showed that ABCA2 colocalized with both Aβ and APP in BE(2)- M17 neuroblastoma cells (Fig. 2). The protein levels of APP increased in a manner consistent with the expression pattern of ABCA2 in WT6 cells (Fig. 3). The differences from the control vector and ABC Cntl were significant with Tukey's post hoc test at P < 0.05. The difference between ABC Cntl and the control vector did not reach statistical significance. Introduction of ABCA2 in WT6 cells also raised the level of Aβ significantly (P<0.05) when compared with the control vector and ABC Cntl (Fig. 4). However, introduction of ABC Cntl did not change the Aβ levels significantly. Therefore, the changes in expression levels of APP and Aβ were associated specifically with ABCA2.

DISCUSSION

In the present study, we used a sensitive technique for measuring differential gene expression and determined those cellular adaptations that are a consequence of overexpression of the ABCA2 transporter in HEK293 cells. Gene expression measurements by this laboratory and others have documented that ABCA2 is highly expressed in the brain. The levels produced in the HEK 293 overexpressing cell line are comparable to the endogenous levels measured in several neuroblastoma cell lines that we have studied (4, 31, 32). We believe that the genes identified in this screen represent physiological changes in response to elevation of ABCA2 expression. With the use of a threefold threshold in the initial screen, 100 genes were found to have increased expression. Of these, two identifiable cluster patterns emerged. In particular, the altered expression of transporter related gene products shown in Table 2 would support the hypothesis that an imbalance caused by the enhanced levels of ABCA2 may cause compensatory adaptations in the expression of these related gene products. In the transfected cells, the expression level of the calcium channel gene CACNA2D2 and its potential cofactor DOC2 was decreased, while a possible cofactor of ABCA2, syntaxin 11, was increased. Three genes coding for phospholipid or fatty acid binding proteins and two for phospholipid/lipid metabolism were also identified. Such expression data point to the association of ABCA2 with lipid and/or steroid transport, also providing an indirect link with AD. Although these adaptations may be linked with particular functional roles, equally intriguing was the cluster of genes more directly related to oxidative stress and AD, particularly APP and Aβ.

We have previously reported that ABCA2 expression is elevated in response to chronic exposure to the estrogen-based anticancer drug estramustine. This agent is a complex of nor-nitrogen Page 7 of 21 (page number not for citation purposes) mustard and estradiol and crystallographic studies showed that it maintains the three-dimensional properties of the steroid ring system (33). Although its anti-tumor properties have been attributed to an unusual anti-microtubule activity (34, 35), in vivo, the drug does produce estrogenic effects (36). Transient transfection of ABCA2 resulted in a 2- to 3-fold increase in resistance to both estramustine and β-estradiol (37), and previous results with antisense treatment produced a sensitization to estramustine (3). These results suggest that there is a connection between ABCA2 function and steroid transport.

ABCA2 has been localized to the endolysosomal subcellular compartment (4). Here we present data showing a colocalization of the transporter with APP and Aβ in discrete intracellular vesicles. These same vesicles also stain positively for endolysosome markers LAMP1 and LAMP2 (data not shown). The topological orientation of ABCA2 in the membrane of these vesicles has not yet been determined. It is presumed that the transporter will convey material from the cytosol into the endolysosome compartment, in effect sequestering cytotoxic compounds into an "inert" compartment. Subsequent processing within the vesicle could be expected to detoxify these cytotoxic compounds and enhance cell survival. For example, with the oxidizing agent AAPH, a broad range of oxidized and peroxidized by-products might be expected. In particular, we detected lipid peroxides at high levels in the treated cells (data not shown). ABCA2 sequestration of these products into the endolysosome compartment could serve to protect the cell by limiting damage to critical membrane components. The Aβ peptide has marked cytotoxicity in cultured cells and induces free radical formation, protein oxidation, and lipid peroxidation (11). Colocalization of this peptide with ABCA2 in these vesicles would also support a functional link between the two. It is also pertinent to note that APP undergoes axonal transport, and APP may be cleaved after transport in axons to generate Aβ, a process that could be accomplished in the endolysosomal compartment. It is also significant to note that one pathological hallmark of AD is the formation of senile plaques that contain the peptide Aβ, the cleavage product of APP. Our results demonstrate the expression of both AD biomarkers was specifically associated with the increased expression of ABCA2, further strengthening the connection between ABCA2 and the pathogenesis of AD.

ABCA2 shares a high degree of homology with ABCA1 (46%), although the latter is located in the plasma membrane (38). The involvement of ABCA1 in reverse-cholesterol transport and ABCA1 deficiency in Tangier disease provided impetus to consider that the homologous family member ABCA2 may also have a functional role in the regulation of cholesterol homeostasis. ABCA2 expression is elevated in macrophages treated with enzymatically-degraded low-density lipoprotein (eLDL) and this evidence suggests that ABCA2 may have a function in the trafficking of LDL-derived cholesterol (6). The uptake and trafficking of LDL to the endolysosome for the release of free cholesterol and colocalization of ABCA2 to this compartment suggest a possible role for ABCA2 in the trafficking of LDL-derived cholesterol. Once again, such a connection would not be inconsistent with a link to AD.

Figure 5 provides a scheme for how those genes altered by ABCA2 overexpression might be involved in pathways of oxidative stress response and the pathogenesis of AD. The forced overexpression of ABCA2 could necessitate an adaptive cellular response resulting in a quantitative alteration in the expression of cofactor proteins necessary for ABCA2 function. Overall, our present data confirm the high level of expression of ABCA2 in human brain, particularly in two regions (frontal and temporal). Pathology of these two regions, together with the parietal, has been previously implicated in AD (39, 40).

In summary, in this study a number of independent lines of enquiry point to a possible cause/effect relationship between ABCA2 expression and the etiology of AD: 1) with gene Page 8 of 21 (page number not for citation purposes) expression profiling of ABCA2 transfectants, a number of accompanying changes involve genes linked with the disease, either through transport functions or in pathways dealing with reactive oxygen species. 2) Colocalization in the endolysosomal compartment and coexpression of ABCA2 with both Aβ and APP occurs. 3) Overexpression of ABCA2 causes increased protein levels of APP and Aβ. 4) Toxicity produced through oxidation of lipids or proteins is abrogated by enhanced ABCA2 expression. 5) ABCA2 expression is high in frontal and temporal regions of the brain.

ACKNOWLEDGEMENTS

This work was supported in part by National Institutes of Health Grants CA-06927 and RR- 05539, by National Institutes of Health Grant CA-83778 to K. D. Tew and AG-10124 to J. Q. Trojanowski, and by appropriation from the Commonwealth of Pennsylvania.

REFERENCES

1. Shroyer, N. F., Lewis, R. A., Allikmets, R., Singh, N., Dean, M., Leppert, M., and Lupski, J. R. (1999) The rod photoreceptor ATP-binding cassette transporter gene, ABCR, and retinal disease: from monogenic to multifactorial. Vision Res. 39, 2537–2544

2. Brooks-Wilson, A., Marcil, M., Clee, S. M., Zhang, L. H., Roomp, K., van Dam, M., Yu, L., Brewer, C., Collins, J. A., Molhuizen, H. O., et al. (1999) Mutations in ABC1 in Tangier disease and familial high-density lipoprotein deficiency. Nat. Genet. 22, 336–345

3. Laing, N. M., Belinsky, M. G., Kruh, G. D., Bell, D. W., Boyd, J. T., Barone, L., Testa, J. R., and Tew, K. D. (1998) Amplification of the ATP-binding cassette 2 transporter gene is functionally linked with enhanced efflux of estramustine in ovarian carcinoma cells. Cancer Res. 58, 1332–1337

4. Vulevic, B., Chen, Z., Boyd, J. T., Davis, W., Jr., Walsh, E. S., Belinsky, M. G., and Tew, K. D. (2001) Cloning and characterization of human adenosine 5′-triphosphate-binding cassette, sub-family A, transporter 2 (ABCA2). Cancer Res. 61, 3339–3347

5. Zhou, C. J., Inagaki, N., Pleasure, S. J., Zhao, L. X., Kikuyama, S., and Shioda, S. (2002) ATP-binding cassette transporter ABCA2 (ABC2) expression in the developing spinal cord and PNS during myelination. J. Comp. Neurol. 451, 334–345

6. Kaminski, W. E., Piehler, A., Pullmann, K., Porsch-Ozcurumez, M., Duong, C., Bared, G. M., Buchler, C., and Schmitz, G. (2001) Complete coding sequence, promoter region, and genomic structure of the human ABCA2 gene and evidence for sterol-dependent regulation in macrophages. Biochem. Biophys. Res. Commun. 281, 249–258

7. Yue, H., Eastman, P. S., Wang, B. B., Minor, J., Doctolero, M. H., Nuttall, R. L., Stack, R., Becker, J. W., Montgomery, J. R., Vainer, M., et al. (2001) An evaluation of the performance of cDNA microarrays for detecting changes in global mRNA expression. Nucleic Acids Res. 29, E41–41

8. Chen, Z. J., Shen, H., and Tew, K. D. (2001) Gene expression profiling using a novel method: amplified differential gene expression (ADGE). Nucleic Acids Res. 29, E46

Page 9 of 21 (page number not for citation purposes) 9. Chen, Z. J., Gate, L., Davis, W., Ile, K. E., and Tew, K. D. (2003) Sensitivity and fidelity of DNA microarray improved with integration of amplified differential gene expression (ADGE). BMC Genomics 4, 28

10. Smith, C. D., Carney, J. M., Starke-Reed, P. E., Oliver, C. N., Stadtman, E. R., Floyd, R. A., and Markesbery, W. R. (1991) Excess brain protein oxidation and enzyme dysfunction in normal aging and in Alzheimer disease. Proc. Natl. Acad. Sci. USA 88, 10540–10543

11. Butterfield, D. A. (2002) Amyloid beta-peptide (1-42)-induced oxidative stress and neurotoxicity: implications for neurodegeneration in Alzheimer's disease brain. A review. Free Radic. Res. 36, 1307–1313

12. Butterfield, D. A., Hensley, K., Harris, M., Mattson, M., and Carney, J. (1994) beta- Amyloid peptide free radical fragments initiate synaptosomal lipoperoxidation in a sequence-specific fashion: implications to Alzheimer's disease. Biochem. Biophys. Res. Commun. 200, 710–715

13. Su, Y. R., Linton, M. F., and Fazio, S. (2002) Rapid quantification of murine ABC mRNAs by real time reverse transcriptase-polymerase chain reaction. J. Lipid Res. 43, 2180–2187

14. Schmidt, M. L., Schuck, T., Sheridan, S., Kung, M. P., Kung, H., Zhuang, Z. P., Bergeron, C., Lamarche, J. S., Skovronsky, D., Giasson, B. I., et al. (2001) The fluorescent Congo red derivative, (trans, trans)-1-bromo-2,5-bis-(3-hydroxycarbonyl-4-hydroxy)styrylbenzene (BSB), labels diverse beta-pleated sheet structures in postmortem human neurodegenerative disease brains. Am. J. Pathol. 159, 937–943

15. Mills, J., Laurent Charest, D., Lam, F., Beyreuther, K., Ida, N., Pelech, S. L., Reiner, P. B. (1997) Regulation of amyloid precursor protein catabolism involves the mitogen-activated protein kinase signal transduction pathway. J. Neurosci. 17, 9415-9422

16. Connop, B. P., Thies, R. L., Beyreuther, K., Ida, N., and Reiner, P. B. (1999) Novel effects of FCCP [carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone] on amyloid precursor protein processing. J. Neurochem. 72, 1457–1465

17. Hu, J. Y., Meng, X., and Schacher, S. (2003) Redistribution of syntaxin mRNA in neuronal cell bodies regulates protein expression and transport during synapse formation and long- term synaptic plasticity. J. Neurosci. 23, 1804–1815

18. Wang, D. D. S., Whitworth, T. L., and Quick, M. W. (2003) Syntaxin 1A inhibits GABA flux, efflux, and exchange mediated by the rat brain GABA transporter GAT1. Mol. Pharmacol. 64, 905-913

19. Perera, H. K., Clarke, M., Morris, N. J., Hong, W., Chamberlain, L. H., and Gould, G. W. (2003) Syntaxin 6 regulates Glut4 trafficking in 3T3-L1 adipocytes. Mol. Biol. Cell. 14, 2946-2958

20. Safieddine, S., Ly, C. D., Wang, Y. X., Wang, C. Y., Kachar, B., Petralia, R. S., and Wenthold, R. J. (2002) Ocsyn, a novel syntaxin-interacting protein enriched in the subapical region of inner hair cells. Mol. Cell. Neurosci. 20, 343–353

Page 10 of 21 (page number not for citation purposes) 21. Sakaguchi, G., Orita, S., Maeda, M., Igarashi, H., and Takai, Y. (1995) Molecular cloning of an isoform of Doc2 having two C2-like domains. Biochem. Biophys. Res. Commun. 217, 1053–1061

22. Mairhofer, M., Steiner, M., Mosgoeller, W., Prohaska, R., and Salzer, U. (2002) Stomatin is a major lipid-raft component of platelet alpha granules. Blood 100, 897–904

23. Mahapatra, N. R., Mahata, M., O'Connor, D. T., and Mahata, S. K. (2003) Secretin activation of chromogranin A gene transcription. Identification of the signaling pathways in cis and in trans. J. Biol. Chem. 278, 19986–19994

24. Dubin, R. L., Hall, C. M., Pileri, C. L., Kudlacek, P. E., Li, X. Y., Yee, J. A., Johnson, M. L., and Anderson, R. J. (2001) Thermostable (SULT1A1) and thermolabile (SULT1A3) phenol sulfotransferases in human osteosarcoma and osteoblast cells. Bone 28, 617–624

25. Greeve, I., Hermans-Borgmeyer, I., Brellinger, C., Kasper, D., Gomez-Isla, T., Behl, C., Levkau, B., and Nitsch, R. M. (2000) The human DIMINUTO/DWARF1 homolog seladin-1 confers resistance to Alzheimer's disease-associated neurodegeneration and oxidative stress. J. Neurosci. 20, 7345–7352

26. Iivonen, S., Hiltunen, M., Alafuzoff, I., Mannermaa, A., Kerokoski, P., Puolivali, J., Salminen, A., Helisalmi, S., and Soininen, H. (2002) Seladin-1 transcription is linked to neuronal degeneration in Alzheimer's disease. Neuroscience 113, 301–310

27. Choi, J., Malakowsky, C. A., Talent, J. M., Conrad, C. C., Carroll, C. A., Weintraub, S. T., and Gracy, R. W. (2003) Anti-apoptotic proteins are oxidized by Abeta25-35 in Alzheimer's fibroblasts. Biochim. Biophys. Acta 1637, 135–141

28. Gylys, K. H., Fein, J. A., Tan, A. M., and Cole, G. M. (2003) Apolipoprotein E enhances uptake of soluble but not aggregated amyloid-beta protein into synaptic terminals. J. Neurochem. 84, 1442–1451

29. Ma, S. L., Ng, H. K., Baum, L., Pang, J. C., Chiu, H. F., Woo, J., Tang, N. L., and Lam, L. C. (2002) Low-density lipoprotein receptor-related protein 8 (apolipoprotein E receptor 2) gene polymorphisms in Alzheimer's disease. Neurosci. Lett. 332, 216–218

30. Frey, N., Richardson, J. A., and Olson, E. N. (2000) Calsarcins, a novel family of sarcomeric calcineurin-binding proteins. Proc. Natl. Acad. Sci. USA 97, 14632–14637

31. Luciani, M. F., Denizot, F., Savary, S., Mattei, M. G., and Chimini, G. (1994) Cloning of two novel ABC transporters mapping on human . Genomics 21, 150–159

32. Zhao, L. X., Zhou, C. J., Tanaka, A., Nakata, M., Hirabayashi, T., Amachi, T., Shioda, S., Ueda, K., and Inagaki, N. (2000) Cloning, characterization and tissue distribution of the rat ATP-binding cassette (ABC) transporter ABC2/ABCA2. Biochem. J. 350, 865–872

33. Punzi, J. S., Duax, W. L., Strong, P., Griffin, J. F., Flocco, M. M., Zacharias, D. E., Carrell, H. L., Tew, K. D., and Glusker, J. P. (1992) Molecular conformation of estramustine and two analogues. Mol. Pharmacol. 41, 569–576

Page 11 of 21 (page number not for citation purposes) 34. Stearns, M. E., Jenkins, D. P., and Tew, K. D. (1985) Dansylated estramustine, a fluorescent probe for studies of estramustine uptake and identification of intracellular targets. Proc. Natl. Acad. Sci. USA 82, 8483–8487

35. Laing, N., Dahllof, B., Hartley-Asp, B., Ranganathan, S., and Tew, K. D. (1997) Interaction of estramustine with tubulin isotypes. Biochemistry 36, 871–878

36. Hudes, G. R., Greenberg, R., Krigel, R. L., Fox, S., Scher, R., Litwin, S., Watts, P., Speicher, L., Tew, K., and Comis, R. (1992) Phase II study of estramustine and vinblastine, two microtubule inhibitors, in hormone-refractory prostate cancer. J. Clin. Oncol. 10, 1754– 1761 37. Ile, K. E., Davis, W. Jr., and Tew, K. D. (2004) Identification of novel first exon of human ABCA2 transporter gene encodes a unique N-terminus. BBA, 1678, 22-32

38. Dean, M., and Allikmets, R. (2001) Complete characterization of the human ABC gene family. J. Bioenerg. Biomembr. 33, 475–479

39. Petrella, J. R., Coleman, R. E., and Doraiswamy, P. M. (2003) Neuroimaging and early diagnosis of Alzheimer disease: a look to the future. Radiology 226, 315–336

40. Jagust, W. J., Eberling, J. L., Reed, B. R., Mathis, C. A., and Budinger, T. F. (1997) Clinical studies of cerebral blood flow in Alzheimer's disease. Ann. N. Y. Acad. Sci. 826, 254–262

Received January 13, 2004; accepted April 7, 2004.

Page 12 of 21 (page number not for citation purposes) Table 1

Differentially expressed genes between C1 and A16

Genes of Total Genes of Genes of Genes Genes differential Slide genes known unkown up- down- expression on slide function function regualated regulated (>3-fold) I 15,000 40 14 26 39 1 II 10,000 36 25 11 14 22 III 15,000 76 45 31 47 29 Total 40,000 152 84 68 100 52

Page 13 of 21 (page number not for citation purposes) Table 2

Functional groups of differentially expressed genes

Gene ID–gene name M, M, replicate 1 replicate 2 Transporter related R33852–syntaxin 11 2.4 1.73 T61269–hypothetical protein FLJ20366, OCSYN -1.62 -3.06 AA598868–coatomer protein complex, subunit β 2.88 1.65 H79351–homologue of yeast Golgi Yif1p 2.10 3.83 AA461071–solute carrier family 23 member 1 2.39 1.95 N53512–calcium channel, α 2/delta 2, CACNA2D2 -2.18 -1.64 AI478508–double C2-like domains, β -2.94 -2.01 R39936–stomatin (EBP72)-like 1 -2.12 -1.69 AA504785–nucleoporin 153kD 2.25 2.78

Membrane components AA862371–transmembrane protein 2 1.63 1.52 W46985–matrix metalloproteinase 24 2.10 2.62

Substrate binding AA976699–chromogranin A (parathyroid secretory protein 1) 2.89 1.70 AA485922–copine I 1.82 2.16 W72051–fatty acid binding protein 7, brain 2.01 1.66 N47717–fatty acid binding protein 5 (psoriasis-associated) -1.50 -2.01

Substrate metabolism R38717–ectonucleotide pyrophosphatase/phosphodiesterase 2 2.06 1.69 (autotaxin) AA398458–sulfotransferase family, cytosolic, 1A, member 3 2.23 2.36 AI018607–sulfotransferase family, cytosolic, 1A, member 2 2.20 2.12 AI335086–angiopoietin-like 3 -2.21 -1.67 AA778196–UDP-Gal:betaGlcNAc β 1,4- galactosyltransferase 2.51 2.34 AI360366–prolylcarboxypeptidase (angiotensinase C) 1.54 1.52 AA025142–ferrochelatase (protoporphyria) -2.15 -2.97

Oxidative stress response and pathogenesis of Alzheimer’s disease AI057612–seladin-1 1.96 1.53 W42849–amyloid β (A4) precursor protein 1.55 2.48 AA486321–vimentin 1.78 1.65 AI360371–LDL receptor-related protein 3 1.63 1.91 AA461071–Slc23a1 (vitamin C transporter) 2.39 1.95 AA065090–calsarcin-1 1.91 1.82

Page 14 of 21 (page number not for citation purposes) Table 3

Comparison of detected expression levels with ADGE microarray and quantified RT-PCR

Gene PCR detected change ADGE average Actina -1.01 1.1 Copinea 1.54 1.99 A-βa 1.35 2 Seladina 1.58 1.75 DOC2a -1.37 -2.48 Stomatina -2.12 -1.91 Vimentinb 2.14 1.715 Ferrochelataseb -1.23 -2.56 aPCR change measured by band quantitation from RT-PCR at cycles in linear range of amplification; bPCR change measured by real time PCR.

Page 15 of 21 (page number not for citation purposes) Table 4

Comparative expression of ABCA2 in brain regions

Brain region Normal Alzheimer Schizophrenic

Frontal white 3.1 ± 1.9 1.44 ± 0.9 6.6 ± 6.2 gray 0.47 ± 0.2 2.55 ± 3.1 1.66 ± 0.6

Temporal white 10.13 ± 1.21 14.95 ± 29.4 17.2 ± 5.0 gray 0.23 ± 0.08 0.36 ± 0.2 0.24 ± 0.07

Parietal white 0.35 ± 0.12 0.33 ± 0.1 0.19 ± 0.16 gray 0.21 ± 0.09 0.28 ± 0.1 0.19 ± 0.09

Occipital white 0.36 ± 0.07 0.35 ± 0.2 0.3 ± 0.12 gray 0.42 ± 0.21 0.4 ± 0.11 0.31 ± 0.11

Cerebellum white 0.33 ± 0.1 0.31 ± 0.19 0.56 ± 0.25 gray 0.73 ± 0.18 0.63 ± 0.17 0.52 ± 0.15

Page 16 of 21 (page number not for citation purposes) Fig. 1

Figure 1. Survival curves of C1 and A16 ABCA2 transfected cells after treatment with AAPH for 60 h. Cell survival is represented by the percentage of control. Standard deviations are represented with vertical bars.

Page 17 of 21 (page number not for citation purposes) Fig. 2

Figure 2. Immunofluorescent images of neuroblastoma cells stained for ABCA2, β-amyloid, and APP. Merged images (right hand panels) demonstrate the colocalization of the transporter with both Aβ and APP.

Page 18 of 21 (page number not for citation purposes) Fig. 3

Figure 3. Cellular APP levels in cells transiently transfected with vector, ABC Cntl, and ABCA2. The top panel is Western blotting, and the bottom panel is quantified densitometry relative to control. Data are expressed as means ± SE with n = 5 and statistical significance determined by ANOVA with Tukey’s post hoc test at *P < 0.05.

Page 19 of 21 (page number not for citation purposes) Fig. 4

Figure 4. Aβ levels in media of cells transiently transfected with the control vector, ABC Cntl and ABCA2. The top panel is the Western blotting and the bottom graph represents the corresponding densitometric values. Data are means ± SE with n > 3 and statistical significance determined by ANOVA with Tukey’s post hoc test at *P < 0.05.

Page 20 of 21 (page number not for citation purposes) Fig. 5

Figure 5. Differentially expressed genes associated with oxidative stress and pathogenesis of AD. This diagram is constructed from published references, cited in the text. The arrows indicate association between genes and between a gene and a pathway. The underlined genes were differentially expressed in the ABCA2 transfected cell line and can be found in Table 2.

Page 21 of 21 (page number not for citation purposes)