Address correspondence and reprint requests to David H. Small, Department of Pathology, University of Melbourne, Victoria 3010, Australia. E-mail: firstname.lastname@example.org
Accumulation of beta amyloid (Aβ) in the brain is central to the pathogenesis of Alzheimer's disease. Aβ can bind to membrane lipids and this binding may have detrimental effects on cell function. In this study, surface plasmon resonance technology was used to study Aβ binding to membranes. Aβ peptides bound to synthetic lipid mixtures and to an intact plasma membrane preparation isolated from vascular smooth muscle cells. Aβ peptides were also toxic to vascular smooth muscle cells. There was a good correlation between the toxic effect of Aβ peptides and their membrane binding. ‘Ageing’ the Aβ peptides by incubation for 5 days increased the proportion of oligomeric species, and also increased toxicity and the amount of binding to lipids. The toxicities of various Aβ analogs correlated with their lipid binding. Significantly, binding was influenced by the concentration of cholesterol in the lipid mixture. Reduction of cholesterol in vascular smooth muscle cells not only reduced the binding of Aβ to purified plasma membrane preparations but also reduced Aβ toxicity. The results support the view that Aβ toxicity is a direct consequence of binding to lipids in the membrane. Reduction of membrane cholesterol using cholesterol-lowering drugs may be of therapeutic benefit because it reduces Aβ-membrane binding.
Alzheimer's disease (AD) is a progressive neurodegenerative disorder characterized by the accumulation of amyloid deposits in the form of amyloid plaques and cerebral amyloid angiopathy (CAA; Price et al. 1991). The accumulation of CAA causes the loss of cerebrovascular smooth muscle cells (SMCs) and weakening of the small and mid-sized vessels in the cerebral cortex and leptomeninges (Vinters 1987).
The major component of the amyloid is a 4-kDa polypeptide known as β-amyloid protein (Aβ; Glenner and Wong 1984;Masters et al. 1985), which is derived from a much larger β-amyloid protein precursor (APP; Kang et al. 1987). The major form of Aβ that is produced in the brain contains 40 amino acid residues. However, minor forms containing 42 or 43 residues are also formed. Production of these minor forms is closely linked to AD pathogenesis (Scheuner et al. 1996). Aβ is toxic and accumulation of Aβ in the neuropil contributes to degenerative changes such as tangle formation and gliosis (Small et al. 2001).
One approach to AD therapy is to inhibit production of Aβ in the brain. Proteolytic cleavage of APP by β- and γ-secretase generates the full-length Aβ, which is then released from cells (Nunan and Small 2000). Therefore, inhibitors of either β- or γ-secretase may be of therapeutic value. Alternatively, a number of studies have shown that cholesterol can influence Aβ release (Simons et al. 1998; Frears et al. 1999; Fassbender et al. 2001; Friedhoff et al. 2001; Hartmann 2001). Therefore, inhibitors of cholesterol biosynthesis (statins) may also be of therapeutic value. One advantage of statins is that their toxicities are relatively low and their mode of action much better understood than many other compounds currently being examined as AD therapeutics.
In the present study, we have examined the binding of Aβ peptides to both synthetic lipid bilayers and to plasma membrane-enriched preparations derived from vascular SMCs using surface plasmon resonance. We show that the extent of binding to membranes correlates very well with the extent of toxicity. Importantly, we show that Aβ binding to synthetic lipid membranes and intact SMC membranes is influenced by the amount of cholesterol in the membrane and that reduction of cholesterol with a cholesterol biosynthesis inhibitor reduces Aβ toxicity. The results strongly support the view that cholesterol-lowering drugs have the effect of reducing Aβ toxicity by reducing Aβ-membrane binding.
Materials and methods
Aβ peptides were synthesised using manual solid-phase Boc (N-tert-butocarbonyl) amino acid synthesis. Peptide purification was achieved using an acetonitrile/water (0.01% trifluoroacetic acid) gradient on a reversed-phase preparative Zorbax high-performance liquid chromatography (HPLC) column heated to 60°C. The purity (≥ 95%) and identity of the peptide was analysed by analytical HPLC, electrospray mass spectrometry and amino acid analysis.
Dulbecco's modified Eagle's medium (DMEM) and penicillin/streptomycin were purchased from Gibco Life Technologies (Mulgrave, Victoria, Australia) and fetal bovine serum (heat-inactivated) was purchased from Commonwealth Serum Laboratories (Parkville, Victoria, Australia). N-octyl-α-d-glucopyranoside, dimyristoyl-l-α-phosphatidylcholine (DMPC), dimyristoyl-l-α-phosphatidyl-dl-glycerol (DMPG), dimyristoyl-l-α-phosphatidylethanolamine (DMPE), dimyristoyl-l-α-phosphatidylserine (DMPS) and d-cholesterol were purchased from Sigma (St Louis, MO, USA). Lovastatin was purchased from Calbiochem (Sydney, NSW, Australia) and activated to its open-ring form as previously described (Jakobisiak et al. 1991).
Solubilization and ageing of Aβ peptides
Aβ1–42, Aβ1–40, Aβ1–28, Aβ17–42 and Aβ29–42 were dissolved in dimethyl sulphoxide (DMSO) at a concentration of 2 mm. The peptide solutions were then sonicated (42 kHz) for 5 min and centrifuged at 6000 g for 1 min at room temperature using a Hermle Z160M bench microfuge (Hermle, Wehingen, Germany). Solubilized peptides were immediately snap-frozen and stored at − 80°C.
Prior to use, peptides were thawed on ice for 5 min and vortex mixed for 15 s. The peptides were then diluted into DMEM for cell culture experiments or 0.02 m sodium phosphate buffer, pH 6.8 for biosensor experiments to give a final concentration of 10 µm. To ‘age’ peptides, a process which increases the proportion of fibrillar oligomeric species (Jarrett and Lansbury 1993), peptides were incubated at 37°C in a humidified atmosphere of 5% CO2 for 5 days at a concentration of 100 µm.
Congo red assay of amyloid fibrils
The concentration of amyloid fibrils was measured using the assay of Klunk et al. (1999). Aβ peptides were mixed with Congo red (CR) in 0.02 m sodium phosphate buffer, pH 6.8. The final concentration of peptide and CR was 10 µm. Solutions of CR alone were also prepared in 0.02 m sodium phosphate buffer, pH 6.8. The mixture was vortexed briefly and then incubated at room temperature for 15 min. The absorbances at 403 and 541 nm were measured using a Bio-Rad SmartSpec 3000 spectrophotometer (Bio-Rad, Hercules, CA, USA). Background absorbance values of buffer alone were subtracted from the values obtained from each sample. The concentration of Aβ fibrils in each preparation were then determined using the formula:
All preparations were prepared in triplicate and the assay was conducted independently three times with similar results in each experiment.
Vascular SMC culture
Vascular SMCs from aortae of Sprague–Dawley rats were provided by Dr G. Dusting (Howard Florey Institute, Melbourne, Australia). The cells were cultured in DMEM supplemented with 10% fetal bovine serum and penicillin/streptomycin at 37°C in a humidified atmosphere of 5% CO2. Vascular SMCs were plated at a density of 104 cells/well in a 96-well plate or at a density of 106 cells/75 cm2 cell culture flask (Nunc, Roskilde, Denmark) in 20 mL culture medium and grown to 80% confluence, after which the medium was removed and replaced with fresh, serum-free medium lacking or containing 10 µm lovastatin. The cells were then incubated for 72 h. The cells were then either used for the preparation of membranes, or incubated with Aβ peptides (10 µm) for cytotoxicity assay. In the latter case, Aβ peptides were added to the culture medium and the cells incubated for a further 24 h. In control incubations, only vehicle (lacking peptide) was added.
MTS assay of cytotoxicity
Cytotoxicity was determined using the CellTiter 96 AQueous One Solution Cell Proliferation Assay kit (Promega Corporation, Madison, WI, USA; Cory et al. 1991). The [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2(4-sulfophenyl)-2H-tetrazolium (MTS) reagent solution was added at a concentration of 10% (by volume) to the culture medium. The cells were then incubated for a further 2 h at 37°C, and the absorbance of the samples read at a wavelength of 560 nm using a Wallac Victor 1420 plate reader (Wallac, Turku, Finland).
Plasma membrane preparation
A crude plasma membrane preparation was prepared from vascular SMCs by differential centrifugation (Hubbard et al. 1983). Cells were scraped off 10 × 75-cm2 flasks using a cell scraper, and centrifuged in DMEM at 1600 g in a Beckman Coulter Allegra 21R centrifuge (Beckman, Gladsville, NSW, Australia) at 4°C for 3 min. The pellet was then washed with phosphate-buffered saline (PBS), added to 10 mL of STM buffer (0.25 m sucrose/5 mm Tris–HCl, pH 7.4/1.0 mm MgCl2), and homogenized on ice using 10 up and down strokes in a 40-mL Dounce-type glass homogenizer with a loose fitting pestle. The homogenate was centrifuged at 220 g for 5 min. The supernatant fraction was saved and the pellet rehomogenized in 5 mL of STM buffer. The suspension was again centrifuged, and the first and second supernatant fractions combined, then centrifuged at 100 000 g for 2 h (Beckman L8-M Ultracentrifuge, 70Ti rotor, no brake) and the resulting crude plasma membrane fraction resuspended in 1.0 mL of 0.02 m sodium phosphate buffer, pH 7.4. Total membrane cholesterol was determined using the Amplex Red cholesterol assay kit (Molecular Probes, Eugene, OR, USA). The protein content of the membrane preparations was determined using the bicinchoninic acid (BCA) assay using bovine serum albumin as standard.
Preparation of synthetic model membranes
Small 100 nm unilamellar vesicles (SUV), containing DMPC, DMPG, DMPS and DMPE, and cholesterol, were prepared in 0.02 m phosphate buffer (pH 6.8) by sonication and extraction. Briefly, 1.5 mg of total lipid was dissolved in 1.5 mL of CHCl3: MeOH (3 : 1, v/v). Aliquots (408 µL) were removed and evaporated under a stream of nitrogen, and the lipids further dried in vacuo overnight. The lipids were then resuspended in 600 µL of 0.02 m sodium phosphate buffer, pH 6.8. The resulting lipid dispersion was then sonicated in a bath-type sonicator until clear and then extruded through 100-nm pore diameter polycarbonate filters 17 times using Liposofast apparatus (Avestin, Ottawa, Canada) to obtain 100 nm SUV. The mixed lipid vesicles contained 80% (w/w), 60% (w/w), 40% (w/w), 30% (w/w) or 0% (w/w) cholesterol. The remaining lipid comprised a mixture of DMPC : DMPE : DMPS : DMPG in a ratio of 75 : 20 : 2.5 : 2.5 (body weight).
Binding experiments involved surface plasmon resonance technology and were carried out with a BIAcore X analytical system (Biacore, Uppsala, Sweden) using an L1 sensor chip (Biacore), containing alkyl groups immobilized to a dextran matrix. The running buffer used for all experiments was 0.02 m sodium phosphate buffer, pH 6.8 (phosphate buffer). In experiments aimed at determining the effect of ionic strength on binding, 0.15 m, 0.3 m or 0.5 m NaCl in phosphate buffer was injected at the time of removal of the peptide over a period of 5 min. The washing solution was 40 mmN-octyl β-d-glucopyranoside. The regeneration solution was 10 mm sodium hydroxide. All solutions were freshly prepared, degassed and filtered through a 0.22-µm filter. The operating temperature was routinely 25°C unless otherwise stated.
The alkyl surface of the L1 chip was cleaned by an injection of 25 µL of 40 mmN-octyl β-d-glucopyranoside at a flow rate of 5 µL/min. SUV (100 µL) or vascular SMC membranes (100 µL containing 0.33 mg protein) were then immediately applied to the chip surface at a flow rate of 5 µL/min. To remove any multilamellar structures from the synthetic lipid surface, 30 µL of 10 mm sodium hydroxide was injected at a flow rate of 50 µL/min, resulting in a stable baseline corresponding to the successful formation of an immobilized layer of SUVs.
Peptide solutions were prepared at concentrations ranging from 0.5 to 10 µm. The solutions were injected over the lipid surface at a flow rate of 5 µL/min for 20 min. The peptide solution was then replaced by phosphate buffer and the peptide–membrane complex allowed to dissociate. The removal of the bound peptide and regeneration of the L1 chip surface (without removal of the synthetic lipid or vascular SMC membrane layer) was achieved by an injection of sodium hydroxide (30 µL, 10 mm) at a flow rate of 50 µL/min. The difference between the RU value obtained just before addition of the peptide and the value obtained 20 s after removal of the peptide was used as an index of the amount of peptide bound to the membrane.
Succinylation of Aβ1–40
Aβ1–40 (2 mg) was dissolved in 10 mL of 0.2 m sodium borate buffer, pH 8.5, and then solid succinic anhydride (1.25 mg) was added slowly in 0.25-mg portions over a period of 15 min to give a 50-fold molar excess. The pH of the reaction was maintained at 7.0 with 1 m NaOH. The reaction was allowed to proceed for an additional 30 min. The modified peptide was then purified on a C8 SepPak cartidge (Waters, Milford, MA, USA). The succinylated peptide was eluted from the column with acetonitrile and dried in vacuo. Confirmation that all of the peptide had been succinylated was obtained by matrix-assisted laser desorption-ionization time of flight (MALDI–TOF) mass spectrometry. No underivatized peptide was detected.
Aβ toxicity experiments
To measure Aβ toxicity on vascular SMCs, cultures of cells were treated with Aβ peptides and analogues (10 µm) and the amount of toxicity determined using the MTS assay (Cory et al. 1991). Aβ1–40, Aβ1–42, Aβ29–42 and Aβ17–42 were all found to cause significant toxicity (Fig. 1). Aβ1–28 was less toxic, suggesting that the toxicity of Aβ may be associated with the C-terminal region of the peptide. Ageing the peptides by incubation for 5 days caused a significant increase in toxicity. Previous studies (e.g. Pike et al. 1991) have shown that the aggregation of the Aβ into fibrils may be important for the generation of toxic species. Our own results support this conclusion. Incubation of Aβ peptides for 5 days significantly increased the proportion of fibrillar species as determined by a CR binding assay (Fig. 2). In general, the amount of Aβ toxicity correlated approximately with the proportion of fibrillar species. For example, there was a significant increase in cytotoxicity after ageing Aβ1–42 (p < 0.05; Student's t-test). In addition, aged Aβ1–28, which formed fewer fibrillar species, was significantly less toxic to vascular SMC cultures.
Binding of Aβ peptides to lipid membranes
As previous studies have shown that Aβ can interact directly with lipid membranes, we examined the possibility that the toxic effects of Aβ were due to a direct interaction with the vascular SMC membrane. To study the binding of Aβ to lipids, biosensor technology was employed. Initially, a biosensor chip was coated with a synthetic lipid mixture containing 60% cholesterol, 30% DMPC, 8% DMPE, 1% DMPS and 1% DMPG, and sensorgrams obtained for Aβ1–42 (Fig. 3a) and Aβ1–40 (Fig. 3b). Maximum binding was approached approximately 20 min after application of the peptide. A significant bulk effect was observed, as the signal rose rapidly at the start of the injection period and fell rapidly at the end of injection period. The difference between the RU value obtained just before addition of the peptide and the value obtained 20 s after removal of the peptide was taken as an indication of peptide bound to the membrane.
A comparison of Aβ peptides and analogues for their abilities to bind to the synthetic lipid mixture showed that there was good agreement between the extent of lipid binding (Fig. 4) and the amount of toxicity (Fig. 1) caused by each peptide. For example, the more toxic C-terminal Aβ fragments (Aβ29–42 and Aβ17–42) bound more strongly than the less toxic N-terminal fragment (Aβ1–28).
Effect of ionic strength on binding
The effect of changing the ionic strength on the dissociation of bound Aβ1–42 or Aβ1–40 from the membrane was also examined. Previous experiments had shown that it was not possible to change the ionic strength of the solution in which the peptides were dissolved without changing the state of aggregation of the peptides (data not shown). Therefore, Aβ1–42 and Aβ1–40 were applied to the lipid membrane in low ionic strength buffer (0 m NaCl) and then after maximum binding was achieved (20 min), the dissociation of the peptide from the membrane was followed by washing the membrane with various concentrations of NaCl. The percentage decrease in Aβ bound after a 5-min dissociation period was determined.
Similar results were obtained for both Aβ1–40 and Aβ1–42 (Fig. 5). The rate of dissociation from the membrane increased with increasing concentrations of NaCl. In the presence of 0 m NaCl, < 5% of the Aβ was released from the membrane after 5 min. However, in the presence of 0.5 m NaCl, approximately 75–80% of the peptide was dissociated after 5 min.
As increasing the ionic strength decreased membrane binding, this suggested that Aβ1–40 and Aβ1–42 bound through electrostatic interactions, probably to the negatively charged phospholipid head groups. Therefore, we examined whether positively charged groups on Aβ1–40 contribute to membrane binding. Primary amino groups on Aβ1–40 were succinylated by reaction with succinic anhydride and then the binding of the succinyl-Aβ1–40 to the lipid membrane was measured. Succinylation of Aβ1–40 almost completely abolished membrane binding (Fig. 6a). In addition, succinylated Aβ1–40 was markedly less toxic in the VSMC assay (Fig. 6b). Once again, this experiment confirmed the close association of membrane binding with toxicity.
Effect of cholesterol on binding
The ratio of cholesterol to phospholipid in the synthetic lipid mixture was directly related to the extent of high-affinity lipid binding (Fig. 7). When a pure phospholipid mixture was used, very little binding was observed. However, at higher concentrations between 30 and 80% cholesterol, the amount of binding increased.
As cholesterol is known to alter the fluidity of lipid membranes in a temperature-dependent manner (Sugahara et al. 2001), we examined the temperature dependence of the effect of cholesterol on Aβ binding. Temperature was found to cause a marked change in Aβ peptide binding, but only at low cholesterol concentrations (Fig. 8). When cholesterol comprised 60% of the total lipid, little effect of temperature was seen, with the amount of binding ranging between 6000 and 7000 RU for both Aβ1–40 and Aβ1–42. In contrast, in the absence of cholesterol, the binding of both Aβ1–40 and Aβ1–42 increased about sixfold when the temperature was increased from 25°C to 37°C.
Role of cholesterol in vascular SMC toxicity
To determine whether the cholesterol content of the vascular SMC membrane influences the binding of Aβ, we prepared a plasma membrane-enriched fraction from vascular SMCs and applied this fraction to the biosensor chip. When applied at a concentration of 10 µm, both Aβ1–40 and Aβ1–42 bound to the membrane fraction (Fig. 9). The total amount of binding was only about 10% of that observed for a synthetic lipid mixture containing 60% cholesterol and 40% phospholipid (Fig. 7).
The effect of a cholesterol biosynthesis inhibitor (lovastatin) on Aβ binding to the membrane was examined. Cells were pre-treated with lovastatin for 72 h and then a plasma membrane-enriched fraction prepared. The protein composition of the membrane fractions was closely similar in the control (3.29 ± 0.15 mg/mL) and lovastatin-treated groups (3.24 ± 0.03 mg/mL), indicating that lovastatin treatment did not greatly alter the amount of total membrane protein. Treatment of the cells with lovastatin was found to strongly decrease binding of both Aβ1–40 and Aβ1–42 to the membrane fraction (Fig. 9a). After lovastatin treatment, the cholesterol content of the membrane fraction was reduced to approximately 55% of that recovered from untreated cells (Fig. 9b). Aβ1–40 and Aβ1–42 binding to the plasma membrane fraction of lovastatin-treated cells was approximately 10–15% of that achieved without lovastatin pretreatment.
To examine whether this decrease in binding might have any consequences for Aβ toxicity, the amount of Aβ toxicity was measured using the MTS assay following lovastatin treatment of cells (Fig. 10). Lovastatin on its own had little effect on the ability of vascular SMCs to reduce MTS. However, cells pre-treated with lovastatin were more resistant to Aβ toxicity, as the Aβ1–40 and Aβ1–42-induced decrease in MTS reduction was approximately 25–40% lower in lovastatin-treated cells than in controls.
This study demonstrates that cholesterol is required for the binding of Aβ to synthetic lipid mixtures and to vascular SMC membranes. Our studies also suggest that lowering plasma membrane cholesterol can decrease Aβ toxicity. Taken together, the results indicate that the binding of Aβ to the lipid component of the plasma membrane is required for Aβ toxicity.
Ageing the peptides by incubation for 5 days was found to increase Aβ binding and to increase the concentration of amyloid fibrils. Unaged Aβ also contained some amyloid fibrils, suggesting that different oligomeric species have different affinities for lipid binding.
There have been conflicting reports on the role of cholesterol in Aβ toxicity. Zhou and Richardson (1996) reported that methyl-β-cyclodextrin-cholesterol protects cells from Aβ-mediated toxicity. However, in that study, cholesterol was added exogenously and the lipid composition of the plasma membrane was not analysed. In contrast, Wang et al. (2001) reported that methyl-β-cyclodextrin on its own attenuated Aβ toxicity by lowering cell cholesterol, and suggested that the effect of methyl-β-cyclodextrin-cholesterol might be more related to a loss of cellular cholesterol, rather than the addition of exogenous cholesterol. Our own studies strongly support this view, as well as providing a biochemical explanation for the decreased toxicity.
There is good evidence that Aβ-membrane binding is involved in Aβ toxicity. Studies by Waschuk et al. (2001) demonstrate that Aβ can bind to phospholipids, and Hertel et al. (1997) have shown that inhibition of electrostatic interactions between Aβ and the negative phospholipids can inhibit Aβ toxicity. In contrast to many peptides, which bind hydrophobically and penetrate deeply into the lipid bilayer (Mozsolits et al. 2001), in our study, bound Aβ was easily removed from lipid membranes with sodium hydroxide, suggesting that electrostatic, rather than hydrophobic, forces are involved. This conclusion was supported by experiments that showed that the rate of dissociation of Aβ peptides from the membrane was increased at higher ionic strength and that succinylation of primary amino groups in Aβ1–40 abolished membrane binding and toxicity.
Ji et al. (2002) have shown that Aβ can bind to cholesterol and that this binding inhibits fibril formation. However, in our study, the binding of Aβ peptides to the membrane was probably not due to a direct interaction between Aβ and cholesterol as cholesterol is hydrophobic and unable to form electrostatic interactions. Instead, cholesterol plays some other role in Aβ binding, perhaps through the control of membrane structure or fluidity. This concept is supported by the finding that the effect of cholesterol on Aβ-membrane binding is temperature-dependent, as temperature is known to regulate the gel–liquid crystal phase transition of lipid membranes (Sugahara et al. 2001).
The mechanism by which Aβ binding to the membrane causes cell toxicity is still unclear. Free radical production and lipid peroxidation have been implicated as well as alterations in ion channel function (reviewed by Small and McLean 1999). If Aβ binds directly to the lipid component of the membrane, then alterations in membrane fluidity may occur. Chochina et al. (2001), using synaptic plasma membranes, have reported that Aβ1–40 can increase neuronal membrane fluidity, although Kremer et al. (2001), using synthetic lipids, found that Aβ can decrease membrane fluidity. Certainly, changes in membrane fluidity could affect the function of a variety of proteins on the cell surface, including ion channels. For example, alterations in fluidity are known to affect the submembrane localization and function of the nicotinic receptor (Baenziger et al. 2000).
The extent of Aβ aggregation was shown to correlate with the vascular SMC toxic response. Although the process of ageing increases the number of amyloid fibrils formed from Aβ (as assessed using a CR binding assay), this is not proof that fibrils are the major toxic form of Aβ. For example, even though aged Aβ1–28 failed to form fibrils, it was still toxic to vascular SMC albeit in a significantly less pronounced manner to the other peptides. Aβ is probably secreted as a monomer and subsequently aggregates into soluble oligomers or fibrils (Podlinsny et al. 1998). It is likely that the levels of soluble oligomeric species of Aβ are also increased by the process of ageing. A study by Lambert et al. (1998) found that small, low molecular weight oligomers of Aβ1–42 are several orders of magnitude more potent neurotoxins than high molecular weight fibrillar species of Aβ1–42. Therefore more work is needed to define the precise nature of the toxic form of Aβ and to ascertain the mechanism of toxicity.
There is some epidemiological evidence that lowering cholesterol may be of benefit in AD. In at least one population-based study, the incidence of AD has been found to be higher in individuals with higher cholesterol levels (Roher et al. 1999). The ε4 allele of the apolipoprotein E gene is known to be a risk factor for AD (Corder et al. 1993; Bales et al. 1997) and demented individuals who are homozygous for the ε4 allele have been reported to have higher plasma cholesterol levels than normal elderly controls (Czech et al. 1994). This idea is also supported by the observation that AD-like pathology is less severe in APP transgenic mice that have been treated with a cholesterol-lowering drug (Refolo et al. 2001). Indeed, two retrospective studies using cholesterol-lowering statins have reported drastic decreases in the risk of developing AD (Jick et al. 2000; Wolozin et al. 2000).
Cholesterol could have more than one role in the pathogenesis of AD. A number of studies have shown that cholesterol is also important for the regulation of Aβ production (Mizuno et al. 1998). High cholesterol uptake can increase Aβ deposition in transgenic mice (Sparks et al. 1994; Refolo et al. 2000). Cholesterol depletion can inhibit the generation of Aβ in hippocampal neurones (Simons et al. 1998) and this effect may be mediated by APP secretases (Kojro et al. 2001). Thus it is possible that inhibition of cholesterol biosynthesis as a therapeutic strategy for AD may have a dual beneficial role, not only by decreasing Aβ production in the brain, but also by decreasing the toxic consequences of Aβ accumulation. As cholesterol biosynthesis inhibitors have few major toxic side-effects, this approach coupled with other therapeutic strategies such as the use of secretase inhibitors (Nunan and Small 2000) or Aβ immunization (Schenk et al. 1999) could become the treatment of choice.