Reduction in cholesterol and sialic acid content protects cells from the toxic effects of beta-amyloid peptides.

beta-Amyloid (Abeta) is the primary protein component of senile plaques associated with Alzheimer's disease and has been implicated in the neurotoxicity associated with the disease. A variety of evidence points to the importance of Abeta-membrane interactions in the mechanism of Abeta neurotoxicity and indicates that cholesterol and gangliosides are particularly important for Abeta aggregation and binding to membranes. We investigated the effects of cholesterol and sialic acid depletion on Abeta-induced GTPase activity in cells, a step implicated in the mechanism of Abeta toxicity, and Abeta-induced cell toxicity. Cholesterol reduction and depletion of membrane-associated sialic acid residues both significantly reduced the Abeta-induced GTPase activity. In addition, cholesterol and membrane-associated sialic acid residue depletion or inhibition of cholesterol and ganglioside synthesis protected PC12 cells from Abeta-induced toxicity. These results indicate the importance of Abeta-membrane interactions in the mechanism of Abeta toxicity. In addition, these results suggest that control of cellular cholesterol and/or ganglioside content may prove useful in the prevention or treatment of Alzheimer's disease.


From the Department of Chemical Engineering, Texas A & M University, College Station, Texas 77843-3122
␤-Amyloid (A␤) is the primary protein component of senile plaques associated with Alzheimer's disease and has been implicated in the neurotoxicity associated with the disease. A variety of evidence points to the importance of A␤-membrane interactions in the mechanism of A␤ neurotoxicity and indicates that cholesterol and gangliosides are particularly important for A␤ aggregation and binding to membranes. We investigated the effects of cholesterol and sialic acid depletion on A␤-induced GTPase activity in cells, a step implicated in the mechanism of A␤ toxicity, and A␤-induced cell toxicity. Cholesterol reduction and depletion of membraneassociated sialic acid residues both significantly reduced the A␤-induced GTPase activity. In addition, cholesterol and membrane-associated sialic acid residue depletion or inhibition of cholesterol and ganglioside synthesis protected PC12 cells from A␤-induced toxicity. These results indicate the importance of A␤membrane interactions in the mechanism of A␤ toxicity. In addition, these results suggest that control of cellular cholesterol and/or ganglioside content may prove useful in the prevention or treatment of Alzheimer's disease.
An important pathological hallmark of Alzheimer's disease (AD) 1 is the formation and progressive deposition of insoluble amyloid fibrils within the cerebral cortex (1). The key constituent of these amyloid deposits has been identified as a 39 -43amino acid long polypeptide, ␤-amyloid peptide (A␤), which is derived primarily from the proteolytic cleavage of a much larger amyloid precursor protein (2). Presenilin 1 and 2 are believed to be involved in the proteolytic processing of the A␤ fragment (3)(4)(5)(6). Evidence for the causative role of A␤ in the pathogenesis of AD partly comes from genetic studies, which linked mutations in the amyloid precursor protein and in the presenilins to inheritable forms of AD (7)(8)(9). Further evidence originates from in vitro toxicity studies with synthetic A␤ peptides, which have shown that A␤, in an aggregated state (fibril, protofibril, low molecular weight oligomer, or diffusible, nonfibrillar ligand), is toxic to neurons in culture (10 -17). Although it seems certain that A␤ plays a role in neurotoxicity associated with AD, the molecular mechanism of A␤ neurotoxicity remains unclear.
Increasing evidence indicates that the neuronal cell membrane is important in the mechanism of A␤ toxicity. Studies (18 -21) have indicated that membrane components such as cholesterol and gangliosides alter the affinity of A␤ for phospholipid membranes. Once associated with the membranes, negatively charged phospholipids, cholesterol, and gangliosides have been shown to increase the ␤-sheet content and/or rate of aggregation of A␤ (19,(21)(22)(23)(24). Both in vivo and in vitro, alterations in soluble cholesterol and/or cholesterol biosynthesis have also been shown to affect the normal processing of amyloid precursor protein (25)(26)(27). In these studies, the inhibition of cholesterol synthesis led to decreased A␤ formation (25,27). In addition, changing membrane properties such as altering the electrostatic potential of the membrane has been shown to affect A␤ neurotoxicity (28). The presence of cholesterol and gangliosides, either in the cell membrane or in culture medium, has similarly been implicated in neurotoxicity (29 -31).
In the present study, we examined the effects of depleting cholesterol and cell surface sialic acid residues associated with gangliosides and membrane-associated glycoproteins on the membrane-associated GTP hydrolysis induced by two A␤ fragments, A␤-(1-40) and A␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35). We had shown previously that G protein activation is an early step in A␤-induced neurotoxicity (32), and we now show that depletion of either membrane component significantly attenuated the A␤-induced increases in GTP hydrolysis observed in PC12 cells. In addition, depletion of either membrane component almost entirely eliminated A␤-induced toxicity in both PC12 and SH-SY5Y cell lines. These results imply that cell membrane composition, especially cholesterol and sialic acid content, plays an important role in A␤ toxicity. This work has significance both in our understanding of the mechanism of A␤ neurotoxicity and in the development of new treatments for AD.
Peptide Preparation-A␤-  and A␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) peptides were prepared analogously to methods which in our hands consistently lead to peptides that are toxic to cultured cells (32). 2 Stock solutions of 10 mg/ml were prepared by dissolving the A␤ peptides in 0.1% (v/v) trifluoroacetic acid in water. After incubating for 1 h at 25°C, the peptide * This work was supported by grants from the National Science Foundation and the Welch Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: AD, Alzheimer's disease; A␤, ␤-amyloid; MTT, 3-(4,5-dimethylthiazol-2-yl) 2,5-diphenyltetrazolium bromide; LDH, lactate dehydrogenase; AMP-PNP, adenosine 5Ј-(␤,␥-imido) triphosphate. stock solutions were diluted to concentrations of 0.5 mg/ml in sterile phosphate-buffered saline (0.01 M NaH 2 PO 4 , 0.15 M NaCl, pH 7.4) with antibiotics. These solutions were then rotated at 25°C for 24 h. The peptides were diluted to final concentrations of 20 M in sterile medium and rotated for an additional 24 h prior to being added to the culture wells or plates for the toxicity and GTP studies, respectively. Under the conditions employed, A␤-(1-40) and A␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) had more than 50% ␤-sheet structure as determined by circular dichroism, bound Congo Red, and formed fibril networks that precipitated from solution, whereas A␤-(1-16) was predominantly random coil and did not bind Congo Red (data not shown). A␤ peptide structure data were similar to those observed by others (34 -37) where peptides were prepared from the same solvents but aged as quiescent solutions for several days, except that in our solutions a more extensive fibril network appeared to form during the 24-h rotation than formed in quiescent solutions over 7 days.
Bovine calcitonin solutions were prepared analogously except that the stock peptide solutions were of 1.5 mg/ml calcitonin in sterile 5 mM CaCl 2 and 1 mM MgCl 2 with antibiotics or in sterile deionized water. The final diluted concentration of calcitonin in all studies was 80 M. Prepared under these conditions, bovine calcitonin in divalent salt solution has a 55% ␤-sheet structure and binds Congo Red, typical of amyloids, whereas bovine calcitonin in deionized water has a 95% ␣-helix structure and does not bind Congo Red at detectable levels (32).
Cell Culture-Human neuroblastoma SH-SY5Y cells (a gift of Dr. Evelyn Tiffany-Castiglioni, College of Veterinary Medicine, Texas A & M University, College Station, TX) were cultured in a humidified 5% (v/v) CO 2 /air environment at 37°C in minimal essential medium supplemented with 10% (v/v) fetal bovine serum, 3 mM L-glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 2.5 g/ml amphotericin B. Likewise, rat pheochromocytoma PC12 cells and C6 glioma cells (ATCC, Manassas, VA) were cultured in RPMI medium supplemented with 10% (v/v) horse serum, 5% (v/v) fetal bovine serum, 3 mM Lglutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 2.5 g/ml amphotericin B in a 5% (v/v) CO 2 /air environment at 37°C. For the GTP studies, cells were plated at densities ranging from 2.5 to 4 million cells in 35-mm tissue culture dishes. For the viability assays, cells were plated at a density of 1 ϫ 10 5 cells/well in 96-well plates. During both the GTP and viability studies, the peptides were added to the cells 24 h after plating.
Membrane Preparation for the GTPase Assay-After incubation with 20 M A␤ peptides, 80 M calcitonin, 200 M epinephrine, or controls for 30 min at 37°C, the PC12 membranes were isolated using the method of Seifert and Schultz (38). Cells were harvested with a cell scraper, collected by centrifugation (1600 ϫ g, 4°C, 20 min), washed with a buffer consisting of 10 mM triethanolamine and 140 mM NaCl, pH 7.4, and disrupted by nitrogen cavitation in a 50 mM KH 2 PO 4 buffer with 100 mM NaCl, 3 mM EDTA, and 15 mM ␤-mercaptoethanol, pH 7.0. The nuclear portion of the cells was removed by a short centrifugation (1000 ϫ g, 4°C, 2 min), and membrane sedimentation was attained with a long centrifugation (15000 ϫ g, 4°C, 60 min). The resulting membrane pellet was suspended in a 10 mM triethanolamine/HCl buffer, pH 7.4, and the total membrane protein content was measured with the bicinchoninic acid assay (39).
GTPase Assay-The GTPase activities of the PC12 membranes were measured as described previously (40 -44). Reaction mixtures of 100 l consisted of 0.4 M [␥-32 P]GTP (0.5 Ci/tube), 0.5 mM MgCl 2 , 0.1 mM EGTA, 0.1 mM ATP, 1 mM AMP-PNP, 5 mM creatine phosphate, 40 g creatine kinase, 1 mM dithiothreitol, and 0.2% (w/v) bovine serum albumin in 50 mM triethanolamine/HCl, pH 7.4. Following a 5-min preincubation period at 25°C, the reaction was initiated by the addition of 5-8 g of membrane protein. After 15 min at 25°C, the reaction was stopped by the addition of 800 l of a 20 mM KH 2 PO 4 buffer (4°C, pH 7.0) containing 5% (w/v) activated charcoal. The released 32 P i was separated from the nucleotide-bound phosphate by centrifugation (15000 ϫ g, 4°C, 20 min), and the 32 P i concentration in the supernatant was determined via scintillation counting (Topcount Microplate Scintillation Counter, Packard Instrument Co.).
Low affinity or nonspecific GTPase activity was measured by adding excess unlabeled GTP (50 M) to the reaction mixture and conducting the reaction as described. Specific high affinity GTPase activity was calculated as the difference between the total GTPase activity in the absence of unlabeled GTP and the low affinity GTPase activity.
MTT Reduction Assay-SH-SY5Y, PC12, and C6 cell response to A␤ was measured using the MTT reduction assay (45). The peptides were incubated with the cells for 24 h, after which time MTT was added to the culture medium to yield a final MTT concentration of 0.5 mg/ml. Cells were incubated with the MTT for 4 h in a CO 2 incubator after which time 100 l of a 5:2:3 N,N-dimethylformamide:SDS:water solution, pH 4.7, was added to dissolve the formed formazan crystals. Then, after 18 h of incubation in a humidified CO 2 incubator, the results were read using an Emax Microplate reader at 585 nm (Molecular Devices, Sunnyvale, CA). Percent MTT reduction was reported relative to control cells unexposed to the peptides.
LDH Release Assay-PC12 cell loss of viability was measured by the release of the cytosolic enzyme, LDH, from cells after 24 h of incubation with A␤. The activity of LDH released was measured using a Sigma kit according to the manufacturer's directions. Results are reported as a percentage of the maximum possible LDH release, which was determined via lysis of control cells by osmotic swelling (in water) or by addition of 1% Triton X-100 to cells.
Cholesterol Depletion and Synthesis Inhibition-The plated PC12 and SH-SY5Y cells were incubated for 3 h in serum-free medium (RPMI and minimal essential medium, respectively), and then they were treated with 5 mM methyl-␤-cyclodextrin in serum-free medium for 8 min at 37°C. The methyl-␤-cyclodextrin-containing medium was then removed from cells and replaced with fresh, serum-containing medium, with either the peptide or epinephrine for the GTPase or toxicity assays. Methyl-␤-cyclodextrin has been demonstrated to specifically remove cellular cholesterol (27, 46 -48). Alternatively, the plated PC12 cells were treated with 0.2 g/ml filipin complex or 1 M compactin in medium and incubated for 48 h at 37°C prior to the peptide addition for the toxicity assays. Filipin has been reported to form complexes with cholesterol (49 -52), and compactin has been demonstrated to inhibit cholesterol production (50,53,54). Control cells were treated identically except for the presence of peptide.
Cell Surface Sialic Acid Depletion and Ganglioside Synthesis Inhibition-Membrane-associated sialic acids from gangliosides and cell surface glycoproteins were removed from cells analogously to established procedures (55). The plated PC12 and SH-SY5Y cells were incubated for 3 h in serum-free medium (RPMI and minimal essential medium, respectively), and then they were treated with 11.7 milliunits of V. cholerae neuraminidase and 3.3 milliunits of A. ureafaciens neuraminidase in serum-free medium for 1 h at 37°C prior to the peptide or epinephrine addition for the GTPase or toxicity assays. Alternatively, PC12 cells were treated with 20 M fumonisin B 1 in medium and incubated for 48 h at 37°C prior to the peptide addition for the toxicity assays. Fumonisin B 1 has been reported to inhibit cellular ganglioside synthesis (56 -59). Control cells were treated identically except for the presence of peptide.
Dose Response-The PC12 cells were plated at a density of 1 ϫ 10 5 cells/well in 96-well plates and were incubated for 3 h in serum-free RPMI medium. To obtain the relationship between extent of cholesterol depletion and susceptibility to A␤ toxicity, cells were treated with 0.05, 1, 5, 10, and 20 mM methyl-␤-cyclodextrin in serum-free medium for 8 min at 37°C. Immediately afterward, peptide in serum-containing medium was added to the cells, followed 24 h later by the MTT reduction assay and LDH release assay. Control cells were treated identically except for the presence of peptide.
To obtain the relationship between extent of cell surface sialic acid removal and susceptibility to A␤ toxicity, the plated PC12 cells were treated with 0.1, 0.5, 1, and 3ϫ neuraminidase serum-free medium for 1 h at 37°C prior to the peptide addition and MTT reduction assay and LDH release assay. The stock solution (3ϫ) contained 35.1 milliunits of V. cholerae neuraminidase and 9.9 milliunits of A. ureafaciens neuraminidase in serum-free medium. Control cells were treated identically except for the presence of peptide.
Determination of Total Cholesterol Level-Total lipids were extracted from cells according to the established procedures (60). The plated PC12 cells (ϳ10 6 cells) were removed from Petri dishes with the aid of a rubber policeman and transferred to a centrifuge tube. The cells were washed three times with phosphate-buffered saline, centrifuging at 800 ϫ g for 5 min, and aspirating the supernatant fluid. 0.5 ml of isopropyl alcohol was then added to the cell pellet, and the sample was sonicated for 5 min. After centrifugation for 15 min at 800 ϫ g, the clear supernatant was decanted, and an aliquot was taken for cholesterol analysis. The total cholesterol level was measured enzymatically using a Sigma cholesterol determination kit with cholesterol standards used for calibration according to the manufacturer's directions. Results are reported as a percentage of cholesterol in cells relative to untreated control cells.
Determination of Total Lipid-bound Sialic Acids-Total lipid extraction of cells was performed as described for the cholesterol determination. The total lipid-bound sialic acid content was determined from the extracted lipid sample colorimetrically and was assumed to be an estimate of total cellular ganglioside content. The sialic acid was measured enzymatically using a sialic acid determination kit with sialic standard solution according to the manufacturer's instructions (Roche Molecular Biochemicals). Results are reported as a percentage of ganglioside in cells relative to untreated control cells.
Statistical Analysis-The significance of results was determined using a Student's t test on n independent measurements, where n is specified in the figure legend. Unless otherwise indicated, significance was taken as p Ͻ 0.05.

GTPase Assays with Cholesterol and Cell Surface Sialic Acid
Depletion-We demonstrated previously (32) that A␤ peptide toxicity was linked to G i /G o protein activation through a receptor-independent mechanism. Since A␤-induced increases in G protein activation appeared to act via a peptide-membrane interaction instead of a peptide-receptor interaction, we wanted to examine the effects of membrane composition, particularly cholesterol and ganglioside content, on the ability of A␤ to increase cell membrane-associated GTP hydrolysis. As illustrated in Fig. 1, depletion of cellular cholesterol and removal of cell surface sialic acids associated with gangliosides and membrane glycoproteins led to a significant decrease in the amyloidogenic A␤-(1-40)-and A␤-(25-35)-induced GTPase activity observed in PC12 membranes (p Ͻ 0.05). Similar results were observed with the amyloidogenic bovine calcitonin prepared from the stock solution in divalent salts. As a control, the GTPase activity of the non-amyloidogenic A␤-(1-16) was also measured, and under all conditions, it did not significantly alter GTPase activity relative to the controls (p Ͼ 0.3). Nonamyloidogenic calcitonin prepared from the stock solution in deionized water was also examined and was found to have no effect on cell GTPase activity with or without removal of membrane cholesterol or sialic acids (data not shown).
Also, to demonstrate that the cholesterol and sialic acid removal procedures did not simply affect GTP hydrolysis in a nonspecific manner, we measured the rate of GTP hydrolysis in membranes from cells treated with 200 M epinephrine under all of the conditions (Fig. 1). The increases in the rate of GTP hydrolysis observed for the epinephrine-treated cells was not significantly altered relative to basal levels after cholesterol or sialic acid removal (p Ͼ 0.1), indicating that GTP hydrolysis was not universally altered by cholesterol and sialic acid depletion.
MTT Reduction Assays with Cholesterol and Cell Surface Sialic Acid Depletion-To explore if peptide-cholesterol and/or peptide-ganglioside interactions were important to A␤ toxicity, we examined whether cholesterol and cell surface sialic acid depletion affected the ability of cells to reduce MTT in the presence of A␤ peptides. Inhibition of MTT reduction is typically correlated with other more direct measures of A␤ toxicity (61,62). As seen in Fig. 2A, reduction of cellular cholesterol with methyl-␤-cyclodextrin and cell surface sialic acids with neuraminidase both significantly attenuated the observed A␤-(1-40)-and A␤-(25-35)-induced inhibition of MTT reduction in PC12 cells (p Ͻ 0.001). MTT reduction decreased on average to 52 ϩ 5% that of control cells after a 24-h exposure to A␤-(1-40) and A␤-(25-35), whereas MTT reduction in the cholesterol-and sialic acid-depleted cells was 89 Ϯ 4 and 94 Ϯ 3%, respectively, of controls after A␤ treatment. Cholesterol and sialic acid depletion also attenuated the inhibition of MTT reduction seen in cells treated with amyloidogenic bovine calcitonin. As a control, MTT reduction by cells treated with the non-amyloidogenic A␤-(1-16) was assessed, and under all conditions it did not significantly alter PC12 cell MTT reduction relative to the controls (p Ͼ 0.3). Analogous results were seen with non-amyloidogenic bovine calcitonin (data not shown).
We investigated the effects of A␤ on MTT reduction and the ability of cholesterol and sialic acid removal to prevent A␤-

. The effect of cholesterol and sialic acid depletion on amyloid peptide-induced increases in GTPase activity of PC12 cells.
The data presented are the percent increases in GTPase activity above basal levels for control cells treated with peptide or epinephrine (solid bars), cells treated with peptide or epinephrine from which cholesterol was removed (open bars), and cells treated with peptide or epinephrine from which gangliosides were removed (diagonally striped bars) relative to untreated cells, cells untreated with peptide from which cholesterol was removed, and cells untreated with peptide from which sialic acids were removed, respectively. The means Ϯ S.D. of 3-6 determinations are presented. PC12 cells were exposed to 20 M A␤ peptides, 80 M bovine calcitonin in divalent cation solution (Cal w/salt), or 200 M epinephrine for 30 min at 37°C in a humidified 5% CO 2 environment prior to the GTPase activity measurements. * and ** indicate that the decreases in the rates of hydrolysis relative to the cells treated with peptide, but without cholesterol or sialic acid removal, were significant at p Ͻ 0.05 and p Ͻ 0.005, respectively. from that of control cells (data not shown). In addition, SH-SY5Y cells were protected from A␤ effects on MTT reduction via removal of cholesterol and cell surface sialic acids analogous to the protective trends observed in PC12 cells (Fig. 2B,  p Ͻ 0.001).
To demonstrate that our results were not simply an effect of the particular drug chosen to reduce cholesterol or sialic acid and ganglioside levels, we investigated the protective effects of compactin, a cholesterol de novo synthesis inhibitor, filipin, a cholesterol-binding agent, and fumonisin B 1 , a ganglioside synthesis inhibitor. As seen in Fig. 3, cellular cholesterol reduction and ganglioside reduction, by all of the means examined, significantly attenuated the effects of A␤-(1-40) and A␤-(25-35) on cell MTT reduction (p Ͻ 0.001). Fig. 4, A and B, we show the relationship between cellular cholesterol content and the susceptibility of cells to the effects of A␤-(1-40) on inhibition of cell MTT reduction and LDH release, two measures typically correlated with loss of cell viability. Protection of cells from A␤ effects on MTT reduction and LDH release is enhanced as more cholesterol is removed from cells until ϳ40% of the cholesterol (relative to controls) remains in the cell. At cholesterol levels lower than 40%, whereas some slight protection from A␤ effects is afforded, the viability of cells untreated with A␤ decreases relative to the controls. Included in the figures are data from cells where methyl-␤-cyclodextrin, filipin, and compactin were used to alter cellular cholesterol levels.

Relationship between Cellular Cholesterol and Sialic Acid Content and Susceptibility to A␤ Toxicity-In
The concentrations of methyl-␤-cyclodextrin needed to reduce cholesterol from control levels to the levels used in our experiments are shown on a second x axis (Fig. 4, A and B). At low concentrations of methyl-␤-cyclodextrin, the ability of treatment of cells with methyl-␤-cyclodextrin to attenuate A␤ effects on MTT reduction and LDH release increases with increasing concentration. However, at concentrations above an optimal concentration (5 mM methyl-␤-cyclodextrin), the viability of control cells as well as cells treated with A␤ decreased with increasing concentration of cholesterol-reducing agent.
In Fig. 4, C and D, we show the relationship between cell surface sialic acid content and the susceptibility of cells to the cates that the increase in MTT reduction upon exposure to A␤ or bovine calcitonin after sialic acid or cholesterol removal relative to the untreated cells exposed to the amyloids was significant at p Ͻ 0.001. effects of A␤-(1-40) on MTT reduction and LDH release. As with cell cholesterol content, there is an optimal cell surface sialic acid content, ϳ35% relative to untreated controls, where maximal protection from A␤ effects is achieved. At lower cell surface sialic acid levels, the viability of control cells begins to fall. At higher cell surface sialic acid levels, the protective effect increases with decreasing cellular sialic acid content. Data included in the figures are from cells where neuraminidases and fumonisin B 1 were used to modify cellular sialic acid and ganglioside content, respectively.
The second x axis on Fig. 4, C and D, shows the relative concentrations of neuraminidases needed to reduce cell surface sialic acid levels from control levels to the levels used in our experiments. Analogous to the results shown in Fig. 4, A and B, at low concentrations of neuraminidases, the ability of treatment of cells to attenuate A␤ effects on MTT reduction and LDH release increases with increasing concentration. However, treatment with concentrations above an optimal concentration (11.7 and 3.3 milliunits of V. cholerae and A. ureafaciens neuraminidases, respectively) led to the reduction in viability of the control cells.
In measuring the relationship between cellular cholesterol and membrane-associated sialic acid or ganglioside content on cell susceptibility to the effects of A␤, we used two types of measurements, MTT reduction (Fig. 4, A and C), which is typically correlated with cell viability, and LDH release (Fig. 4,  B and D), which is associated with loss of membrane integrity and loss of viability, to ensure that response we were measuring was actually correlated to cell to viability. Results from the LDH assay and MTT assay were consistent.
Although cholesterol and/or gangliosides have not been previously linked to any of these biological activities of A␤, there is evidence that both may be important in AD. Gangliosides have been implicated in the deposition of A␤ into senile plaques associated with AD. Several studies have suggested that A␤ binds to GM 1 gangliosides in Alzheimer's disease brains (85,86).
Epidemiological studies have shown that cholesterol plays a role in Alzheimer's disease. An increased risk of Alzheimer's disease has been linked with a natural genetic variant of apolipoprotein E (apoE), a molecule associated with cholesterol metabolism. The ⑀4 allele of apoE, the allele associated with increased risk for late onset AD, has been associated with elevated total serum cholesterol (87,88). Demented patients homozygotic for apoE-4 had the highest total plasma cholesterol levels among a referral population of 40 patients with clinically diagnosed Alzheimer's disease compared with a sample of non-demented elderly controls (89). In addition, in a population-based study of individuals without the apoE-4 allele, the risk of AD has been positively correlated with elevated serum cholesterol levels (90).
Based on this collected evidence, we examined the role of cholesterol and gangliosides in G protein activation and toxicity of A␤ fragments. We chose to examine three A␤ fragments, ␤-(1-40), ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35), and ␤-(1-16). ␤-(1-40) is one of the fragments found in amyloid plaques in vivo and forms amyloid fibrils of comparable structure to ␤- , the A␤ peptide more commonly associated with A␤ toxicity in AD (91)(92)(93). ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) comes from the hydrophobic core of A␤, forms amyloid fibrils readily, although the fibrils are less ordered than fibrils from longer sequences, and has been found frequently (12,94) to be toxic to neurons in culture. ␤-(1-16) comes from the A␤ sequence but does not form amyloid fibrils and has not been observed to cause toxicity in cultured cells. As controls, we performed parallel experiments with bovine calcitonin, prepared such that it formed both amyloid and non-amyloid structures. With calcitonin we would be able to test that results observed were due to differences in peptide structure (capable of forming amyloid fibrils, protofibrils, or other intermediates or not capable of forming amyloid fibrils and associated aggregation intermediates). We have shown previously that aggregated A␤ (25-35 and 1-40 fragments but not 1-16 fragment) was able to increase GTP hydrolysis in membranes derived from PC12 cells, and inhibition of GTPase activity attenuated the toxicity of the A␤ peptides (32). We showed similar results with bovine calcitonin; only when in an amyloidogenic form, bovine calcitonin increased GTP hydrolysis, and inhibition of the GTPase activity attenuated the toxic effects of calcitonin (32). We also observed A␤-induced increases in GTPase hydrolysis with purified G i and G o ␣-subunits reconstituted in liposomes and with cell membrane systems in which the receptors had been proteolytically removed. These results suggested that the G protein activation observed upon incubation with A␤ was receptorindependent and was probably membrane-mediated. Our current studies show that membrane composition affects A␤ and amyloidogenic calcitonin-induced G protein activation in the cells studied. Cholesterol and membrane-associated sialic acid depletion both significantly reduced the membrane GTPase activity of cells exposed to the amyloids. Moreover, the cholesterol and sialic acid removal did not affect receptor-mediated GTPase activity.
Previous toxicity studies (32) have shown that inhibition of GTPase activity in PC12 cells attenuated A␤-induced toxicity. In the current work, we show that depleting the cells of cholesterol and removing cell surface sialic acid residues from gangliosides or membrane-associated glycoproteins, which inhibits A␤-induced GTPase activity, reduces the observed A␤induced inhibition of MTT reduction compared with untreated cells exposed to A␤. Inhibition of cellular cholesterol and ganglioside synthesis had similar protective effects as cholesterol and sialic acid removal. Analogous results were seen using a second amyloid protein, bovine calcitonin.
Other authors (30, 95) have presented conflicting results concerning the effects of cholesterol on A␤-induced neurotoxicity. In these studies, methyl-␤-cyclodextrin-solubilized cholesterol and methyl-␤-cyclodextrin alone were found to attenuate A␤-induced PC12 or neuronal toxicity (30, 95). Our results would indicate that removal of cholesterol from membranes by methyl-␤-cyclodextrin, and not methyl-␤-cyclodextrin, was protective in these experiments. Supporting this conclusion are our results that show that cholesterol synthesis inhibition with compactin or cholesterol depletion with filipin complex also render the cells less susceptible to A␤ effects.
In examining the dosage dependence of the protection of cells from A␤ inhibition of MTT reduction or A␤-induced LDH release via methyl-␤-cyclodextrin treatment, we found that there was an optimal time and concentration of methyl-␤-cyclodextrin used such that susceptibility to A␤ effects were minimized without affecting the viability of cells. Analogously, an optimal neuraminidase concentration and treatment time was found that maximized the protective effect of sialic acid removal without compromising cell viability. These results are not surprising given the essential role of cholesterol and gangliosides in cellular function.
A number of authors (62,95,96) have found that the absence of MTT reduction, which is typically correlated with loss of viability, was sometimes caused by the exocytosis of MTT from the cells, particularly in the presence of A␤. In these studies, however, MTT exocytosis and inhibition of MTT reduction in neuron-like cells similar to the PC12 cells used in our studied was found to be correlated directly to neurotoxicity observed in hippocampal cultures as assessed by direct visualization and trypan blue exclusion (33, 61) and was found to correlate with but overpredict LDH release in neuronal cultures (62). Based on this evidence, we believe our measurements of inhibition of MTT reduction should correlate with cell toxicity.
The ability of a cell to exocytose MTT, whether in the absence or presence of A␤ is probably dependent upon membrane properties. Our treatment of cells with methyl-␤-cyclodextrin or neuraminidases may have altered cell exocytosis of MTT in  (3, diamond) are also indicated and plotted as a function of the percentage cholesterol or sialic acids, respectively. Percentage of MTT reduction is measured relative to cells untreated with peptide, methyl-␤-cyclodextrin, or neuraminidases. Percentage of maximal LDH release is measured relative to LDH released from untreated cells lysed using Triton X-100. The means Ϯ S.D. of 8 determinations are presented. ways that are unpredictable without affecting cell susceptibility to MTT toxicity. To examine this possibility, and to confirm that the results we were reporting reflected actual changes in cell viability and not some artifact of our viability assay, we measured cell susceptibility to A␤ before and after treatment with varying concentrations of methyl-␤-cyclodextrin or neuraminidases using the LDH release assay. The practice of verifying MTT reduction results with trypan blue exclusion or LDH release, both of which measure changes in membrane permeability associated with loss of viability, is well characterized in the literature (33, 61, 62). The mechanisms of action of the LDH and MTT assays are significantly different; therefore, we would not expect the assays to have the same biases. In our hands, the MTT assay and LDH assay gave consistent results. Results from both assays indicated that cellular cholesterol and sialic acid reduction protects cells from A␤ toxicity.
In summary, we have shown that cholesterol and membranebound sialic acids associated with gangliosides and other surface molecules, which both play an important role in A␤ binding to membranes, also affect A␤-induced GTPase activity and cell toxicity. These findings have implications for our understanding of the mechanism of A␤-induced neurotoxicity and point to an important role for A␤-membrane interactions in the mechanism of toxicity.