Apoptotic Neuronal Cell Death Induced by the Non-fibrillar Amyloid-β Peptide Proceeds through an Early Reactive Oxygen Species-dependent Cytoskeleton Perturbation*

In the present study, we have determined the nature and the kinetics of the cellular events triggered by the exposure of cells to non-fibrillar amyloid-β peptide (Aβ). When cortical neurons were treated with low concentrations of soluble Aβ (1–40), an early reactive oxygen species (ROS)-dependent cytoskeleton disruption precedes caspase activation. Indeed, caspase activation and neuronal cell death were prevented by the microtubule-stabilizing drug taxol. A perturbation of the microtubule network was noticeable after being exposed to Aβ for 1 h, as revealed by electron microscopy and immunocytochemistry. Microtubule disruption and neuronal cell death induced by Aβ were inhibited in the presence of antioxidant molecules, such as probucol. These data highlight the critical role of ROS production in Aβ-mediated cytoskeleton disruption and neuronal cell death. Finally, using FRAP (fluorescence recoveryafter photo bleaching) analysis, we observed a time-dependent biphasic modification of plasma membrane fluidity, as early as microtubule disorganization. Interestingly, molecules that inhibited neurotubule perturbation and cell death did not affect the membrane destabilizing properties of Aβ, suggesting that the lipid phase of the plasma membrane might represent the earliest target for Aβ. Altogether our results convey the idea that upon interaction with the plasma membrane, the non-fibrillar Aβ induces a rapid ROS-dependent disorganization of the cytoskeleton, which results in apoptosis.

A common feature of Alzheimer's disease (AD), 1 the most common form of dementia, is the accumulation and the aggregation of the amyloid-␤ peptide (A␤), a 39-to 43-amino acid peptide derived from the proteolytic cleavage of the amyloid precursor protein (1,2). Although A␤ represents a key factor in AD (3), the nature of the toxic form of A␤ early involved in AD pathology remains unclear. Whether it is the fibrillar or the non-fibrillar peptides that are the more deleterious remains a controversial issue (4). The amyloid cascade hypothesis causally links AD clinico-pathological process and neuronal cell death to the aggregation and deposition of A␤ (5)(6)(7). However, this hypothesis has been challenged by recent evidences indicating that the non-fibrillar A␤ also plays a major role in AD (8,9). A recent elegant study has demonstrated that the fibrils from AD brain are composed of amyloid peptide moieties arranged at right angles to the backbone of the amyloid P protein wrapped in glycosaminoglycans (10). Thus, the fibrils are not simply made of chains of self-aggregated A␤ and do not comprise long chains of multimeric A␤, similar to those used to evaluate the neurotoxicity of the fibrillar A␤ in vitro and in vivo. Moreover, the synaptic loss in AD brain has been correlated with the soluble pool of A␤ peptides rather than the fibrillar one, implying that the non-fibrillar A␤ may be a crucial pathological factor in AD (11)(12)(13). Several studies, based on the use of transgenic mice, have demonstrated that neurodegeneration and specific spatial learning deficits might occur without amyloid plaque formation (14 -17).
These results emphasize the necessity to clarify the initial response of neurons to the non-fibrillar A␤ and to identify the cellular targets involved in non-fibrillar A␤-induced neurotoxicity. Tailing with these observations, our studies and others rely on the hypothesis of a close association between neuronal loss and a proapoptotic effect of soluble forms of A␤ (18 -21). Indeed, it has been established that the amphiphilic non-aggregated A␤ may intercalate into the plasma membrane of neurons, directly altering membrane activities and inducing neuronal cell death (18,(22)(23)(24)(25). Accumulative evidences have laid emphasis on the critical role of an oxidative stress in AD and in the neurotoxicity induced by A␤ (26,27). However, most of the molecular mechanisms involved in the neuronal cell death induced by non-fibrillar forms of A␤ are yet to be characterized.
The aim of the present paper was to identify the primary targets of the non-fibrillar A␤ and the chronology of the early cellular events involved in apoptotic neuronal cell death upon A␤ exposure. Microtubules fulfill a plethora of cellular functions, including axonal and dendritic growth and stability (28,29). We have proved recently that the non-fibrillar A␤  induces apoptosis to rat cortical neurons (18,22).
In the present study, we have investigated whether the microtubule network could be an early cellular target for the non-fibrillar A␤. We have demonstrated the following se-quence: low concentrations of soluble A␤ , or of its shorter (29 -40) C-terminal domain, induce apoptotic neuronal cell death by perturbing the fluidity of the plasma membrane, leading to a disruption of the neurotubule network depending on the induction of an early oxidative stress. , A␤ (29 -40), the caspase substrate, and inhibitor peptides were purchased from Bachem, and DCFH-DA was from Molecular Probes. Unless otherwise indicated, materials used for cell culture were obtained from Invitrogen. The drug stabilizing cytoskeleton, taxol, and all other chemicals were of high purity grade from Sigma.

Materials-A␤
Peptide Solubilization-To overcome problems of amyloid peptide solubility at high concentrations, fresh peptide stock solutions were prepared at 5 mg ϫ ml Ϫ1 in hexafluoro-2-propanol (Sigma) as described previously (30). For the incubation of the peptides with the neurons, aliquots of peptide stock solution were quickly dried under nitrogen and directly solubilized at the experimental concentrations into the culture medium. Peptide solutions were then applied onto the cells. Under those conditions, all the amyloid peptides remained soluble for the determination of their neurotoxic properties (18).
Neuronal Viability-Experiments were performed on six to seven DIV neurons. Cell viability was first determined by morphological observation and cell counting after 5 min of trypan blue staining (0.4%; Sigma) to evaluate membrane integrity, and the metabolic activity was assessed by the MTT reduction assay (18,22). Moreover, the release of lactate dehydrogenase into the culture medium was assessed using a cytotoxicity detection kit (Roche Molecular Biochemicals) according to the recommendations of the manufacturer.
Monitoring of Apoptosis-Cell nuclei were visualized using 4,6-diamidino-2-phenylindole (DAPI; Sigma). The cells, grown on a glass coverslips, were washed in PBS, incubated at room temperature for 10 min with DAPI (0.1 g ϫ ml Ϫ1 ), washed with PBS, and examined under a microscope equipped for epifluorescence. To evaluate the percentage of apoptotic cells, five independent fields of microscope were counted (around 100 cells) in three separate experiments, with two determinations each. Under control conditions, neuronal cells exhibited 12-15% of apoptotic cells at nine DIV. For experiments in the presence of caspase inhibitors, cells were incubated 2 h with 50 or 100 M inhibitors before and throughout A␤ peptide exposure. Alternatively, DNA fragmentation was monitored by enzyme-linked immunosorbent assay for the detection of oligonucleosomes using a kit form purchased from Roche Molecular Biochemicals. Briefly, cortical neurons were plated in 24-well dishes (around 200.000 cells per well) and treated at seven DIV for 24 h with A␤. After having been washed, the cells were lysed directly on wells, and oligonucleosomes were determined according to the recommendations of the manufacturer.
Measurement of Caspase-like Proteolytic Activities-The caspase activities were measured by means of the cleavage of the substrates DEVD-pNA, YVAD-pNa, LEHD-AMC, and IEPD-AMC (Bachem). Briefly, at the indicated time points following peptide treatments, the cells were rinsed three times with ice-cold PBS and incubated for 20 min on ice in a buffer of 25 mM Hepes, pH 7.5, 1% (v/v) Triton X-100, 5 mM EDTA, 1 mM EGTA, 5 mM MgCl 2 , 5 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml each of pepstatin and leupeptin, and 5 g/ml aprotinin. The lysate was centrifuged for 15 min at 12,000 rpm and assayed for protein by Bradford (Bio-Rad). 50 g of proteins were incubated for 2 h with 100 M caspase substrates initially dissolved in Me 2 SO. The cleavage of the caspase substrates was monitored by absorbance measurements at 405 nm for DEVD-pNA and YVAD-pNa and by fluorescence emission at 460 nm after exciting LEHD-AMC and IEPD-AMC at 360 nm, using a Fluostar reader plate (BMG-Labtechnologies).
DCFH-DA Assay-The measurement of cell oxidation is based on the oxidation of the non-fluorescent compounds, DCFH-DA, to a fluorescent derivative, DCF, in a peroxidase-mediated reaction. Increases in fluorescence emission reflect enhanced cellular oxidative stress. Briefly, treated cortical neurons were loaded with 100 M DCFH-DA for 45 min. Before analysis, cells were washed three times in PBS, and DCF fluorescence was recorded directly on culture dishes by a Fluostar reader plate (BMG-Labtechnologies), using 488-nm excitation and 510-nm emission filters.
Electron Microscopy-Cortical neurons were fixed for 2 h at 4°C in 2.5% glutaraldehyde and 0.5% tannic acid in 0.1 M cacodylate buffer, pH 7.4. Then the neurons were postfixed for 2 h at 4°C in 2% osmic acid in phosphate buffer. After having been dehydrated in a graded alcohol series, the samples were embedded in Epon resin (Poly/Bed 812; Polysciences, Warrington, PA), and ultra thin sections (70-nm) were obtained using a Reichert Ultracut. Thin sections were counterstained with uranyl acetate, and lead citrate and examined with a Jeol 100CX microscope.
Immunofluorescence-For immunofluorescence studies, the neurons were cultured on glass coverslips that had been coated overnight with 15 g/ml poly L-ornithine. Following the treatments, the neurons were fixed in PBS containing 4% paraformaldehyde for 30 min at room temperature. The cells were permeabilized with 0.1% Triton X-100 made up in PBS containing 3% bovine serum albumin for 30 min and then incubated with a monoclonal anti-␤-tubulin antibody (1:500) (Chemicon) for 1 h under constant agitation. After several washes in PBS, the cells were incubated for 1 h with a fluorescein isothiocyanateconjugated donkey anti-mouse IgG (1: 250) (Santa Cruz Biotechnology), washed with PBS, labeled with DAPI as described above, and mounted in Fluoprep (BioMérieux). The microtubules were visualized with a Nikon microscope using a PlanFluor X40/1.3 objective. For semiquantitative analysis of microtubule organization, at least five microscope fields/condition were imaged using a Nikon DXM1200 digital camera, and microtubule organization in 100 -120 cells/field was classified as normal, mildly disrupted, or severely disrupted.
Cell Labeling for Photo Bleaching-The cells were incubated for shorter times with A␤ (1-40) in Hanks' balanced salt solution at room temperature and loaded with 4 M NBD-SM (1 mg/ml stock solution in chloroform). In the case of longer treatments (up to 24 h), the neurons were incubated with A␤ (1-40) in normal culture conditions, then rinsed with Hanks' balanced salt solution and stained with the fluorescent lipid, and analyzed 10 min thereafter.
Measurement of the Lateral Fluidity of Cell Membrane-The measurement of the lateral diffusion of the molecules in the membrane by means of FRAP (fluorescence recovery after photo bleaching) has already been described (32). A fluorescent probe, here a labeled lipid, was incorporated into the cell membranes. When a small defined area of the labeled membrane was photobleached by a high powerful laser pulse, the intensity of fluorescence was reduced immediately. As the fluorescent probe present in the vicinity underwent a constant diffusion motion, it diffused into the bleached area, gradually increasing the fluorescence intensity after bleaching. Thus, the kinetics of the fluorescent recovery depends on the diffusion rate of the probe, measured as the diffusion coefficient, D 20,w . Analyses were performed at 37°C with a Zeiss LSM 510 confocal laser scanning microscope. We used the 488-nm line of a 25-milliwatt argon laser with a Zeiss C-Apochromat, ϫ63, numerical aperture 1.2, oil-immersion objective. The pinhole diameter was set to 1 airy unit, which correspond to a 1.2-m depth of field, to reduce the contribution of cytoplasm fluorescence as much as possible. The area of bleaching was defined as a circle of 3.0-m diameter, centered on an apical body cell membrane. The area was photobleached at full laser power (100% power, 100% transmission) for 230 ms. The extent of the bleaching typically reached 50 to 80%. Before bleaching, five images were monitored to define the initial fluorescence. The postbleached images were scanned at 0.6% transmission with a delay of 100 ms during the last 60 s, resulting in a total acquisition of 80 points for 90 s, the time required to complete the maximum recovery of fluorescence. No photobleaching was observed during recovery. The sets of scans in which the fraction mobile of fluorophore was less than 65% or more than 110% and bleaching less than 35% were discarded. Cells presenting debris, heterogeneous labeling, or movements during scanning were not used for FRAP measurements. D 20,w parameters were calculated according to the method described by Kubitscheck et al. (33).
Statistical Analysis-STAT VIEW computer software was used for the statistical analysis. Most of the data were from three separate experiments with three to four determinations each. Values were expressed as means ϩ S.E. Differences between control and treated groups were analyzed using Student's t test. Multiple pairwise comparisons among the groups of data were performed using ANOVA followed by a Scheffe's post hoc test. Statistical differences were determined at p Ͻ 0.05.

Early Perturbations of the Neurotubule Organization upon
A␤ Treatment-To investigate the kinetics of microtubule network disorganization, cortical neurons were exposed to 5 M soluble A␤ (1-40) for short incubation times at the end of which no morphological feature of apoptosis (e.g. membrane bleeding, cell shrinkage, and chromatin condensation) was detected (see Figs. 1 and 2). After a 3-h A␤ (1-40) exposure, we observe a dramatic perturbation of the neurotubule network in most of the neurites of the treated neurons (Fig. 1c), as compared with the control cells in which neurotubules elongated within the neurites normally, in a parallel organization (Fig. 1a). Similar results were obtained using the shorter A␤ (29 -40) which displays also membrane perturbing properties (data not shown). These observations made through electron microscopy were confirmed using immunocytochemistry. In the control cells, a dense and constant microtubule network radiates from the cell bodies to the periphery (Fig. 2, A and B). By contrast, in the cortical neurons treated with 5 M non-fibrillar A␤ (1-40) for 1 and 3 h (Fig. 2, C and D, respectively), we observed a peripheral fragmentation and loss of microtubules without any fragmentation or condensation of nuclear DNA (Fig. 2C). Despite this severe microtubule disruption, the treated cells maintained their spreading shape, implying that the non-fibrillar A␤ (1-40) did not affect neurofilaments. Higher peptide concentrations or longer incubation times resulted in a more extensive and rapid loss of the microtubule network (not shown).
Taxol Prevents Cytoskeleton Disruption and Neurotoxicity Induced by the Soluble A␤ Peptide-To determine the kinetics of onset of the early microtubule perturbations induced by the non-fibrillar A␤  and of the apoptotic neuronal cell death, we investigated the effects of taxol, a microtubule-stabilizing drug, on A␤-induced cytoskeleton disruption and neurotoxicity. Cortical neurons incubated with 100 nM taxol only displayed a typical microtubule organization (Fig. 1, b and bЈ) as described previously (34). Interestingly, the non-fibrillar A␤ (1-40) was unable to disrupt the neuronal microtubule network in the cells preincubated for 2 h with 100 nM taxol before A␤ (1-40) addition (Fig. 1D). These results have been confirmed using immunocytochemistry and with taxol being added to the cortical neurons at the same time as A␤ (1-40) (not shown). We next investigated the effects of taxol on non-fibrillar A␤-induced neuronal cell death. As described previously (18), the treatment of cortical neurons with 5 M A␤ (1-40) resulted in a time-dependent decrease in cell viability monitored by the MTT assay (Fig. 3A). Upon a 6-h exposure to A␤ (1-40), the MTT reduction level decreased significantly to 18.3% (p Ͻ 0.05) compared with the control level. Interestingly, whereas taxol alone displayed no effect on MTT even after 48 h of treatment, its presence protected the neurons against A␤ (1-40)-induced neurotoxicity (Fig. 3A). After a 48-h exposure to A␤ in the presence of 100 nM taxol, the MTT reduction level remained at 82% of control, whereas the protective effects of taxol diminished after prolonged incubations. Moreover, the presence of taxol in the culture medium inhibited the release of lactate dehydrogenase after a 48-h A␤ (1-40) treatment (Fig. 3B). This suggests that the stabilization of the microtubule organization by taxol prevents A␤-induced neuronal cell death.
In agreement with these morphological and biochemical observations, we demonstrated that taxol inhibited apoptosis induced by low concentrations of non-fibrillar A␤ . The apoptotic nuclei were visualized and quantified after DAPI staining of cultures treated with 5 M A␤  in the absence and presence of 100 nM taxol. Upon A␤ exposure, cortical neurons shared a time-dependent increase in the number of apoptotic nuclei, which was statistically different from control after 24 h of incubation and reached 58.6 ϩ 3.1% (p Ͻ 0.001) after 48 h of incubation (Fig. 4A). The presence of taxol in the culture medium almost completely inhibited the apoptotic cell death induced by A␤ (1-40) after a 24-h incubation, and the effects persisted for up to 48 h (23.4 ϩ 2.5% of apoptotic nuclei) (Fig. 4A).
To further improve our results, we quantified the cytoskeleton perturbation induced by the non-fibrillar A␤ (1-40) using immunocytochemistry with an anti-␤-tubulin antibody. The A␤-induced microtubule disruption was time-dependent (  . Apoptotic nuclei were visualized and quantified after DAPI staining (A and C), and the microtubule perturbations were quantified using immunofluorescence with an antibody against ␤-tubulin (B and C). Data are means (Ϯ S.E.) of three different experiments with four determinations each and are normalized to the effect of vehicle, designated as 100% (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001). Differences among the subgroups for each condition were performed by ANOVA followed by a Scheffe's post hoc test. #, p Ͻ 0.05 between cells treated with A␤ alone and cells treated with A␤ in the presence of taxol. No significant differences were found between taxol-treated and control cells.
bules after a 24-h incubation (Fig. 4B). Interestingly, the presence of taxol completely abolished microtubule perturbation induced by the non-fibrillar A␤ peptide (Fig. 4B). These data strongly emphasize that early cytoskeleton disruption is required in the neuronal cell death induced by non-fibrillar A␤ peptide. Moreover, Fig. 4C showed only few cells exhibiting both disorganized microtubules and apoptotic nuclei. Altogether, these results strongly suggest that microtubule perturbations precede, and might be involved in, a pathway leading to apoptosis upon soluble A␤ exposure.
A␤ Peptide-induced Cytoskeleton Disruption Precedes Caspase Activation-In a previous report, we demonstrated that the caspase 3 inhibitor, DEVD-CHO peptide, reduced neuronal cell death induced by the non-fibrillar A␤ (1-40) significantly (18). Here, we clearly show that the apoptotic cell death induced by non-fibrillar A␤ (1-40) requires the activation of caspases 3 and 9 by directly measuring caspase-like activity in the lysates of the treated cells (Fig. 5). The activity of caspases 3 and 9 increased significantly (p Ͻ 0.05 as compared with the control cells) after a 6-h incubation with 5 M A␤ (1-40) (Fig. 5,  A and D, respectively), whereas the activation of caspases 1 and 8 was not detected (Fig. 5, B and C, respectively). To establish a causative relationship between microtubule disruption and caspase activation, we performed kinetic experiments of microtubule perturbation in the absence and presence of caspase inhibitors (Fig. 6). Fig. 6A demonstrates that both caspase inhibitors markedly reduced apoptosis induced by 5 M nonfibrillar A␤ (1-40). However, unlike taxol, the caspase inhibitors had no effect on the early cytoskeleton perturbations induced by A␤ (1-40), even after a prolonged incubation (Fig.  6B). Moreover, the presence of 100 nM taxol during A␤ exposure inhibited the activation of caspases 3 and 9 significantly (not shown). These results suggest that the microtubule perturbation might occur before caspase activation under treatment with non-fibrillar A␤ (1-40).

A␤-induced Cytoskeleton Disruption Involves an Oxidative
Stress-We showed that cell exposure to 5 M non-fibrillar A␤ (1-40) peptide induced a time-dependent increase ROS formation, as measured by the oxidative stress-sensitive dye DCFH-DA. A 1-to 3-h treatment with A␤ caused a significant increase in ROS production (Fig. 7). During these short incubation times, the neuronal membrane appeared intact as determined by trypan blue exclusion (data not shown). The first clear signs of cell damages, e.g. membrane bleeding and trypan blue staining, appeared only after a 6-h exposure to 5 M non-fibrillar A␤ (1-40) (7% of the control cells were trypan blue-positive versus 16% of the treated neurons, of 200 cells counted for each condition). In addition, we tested the effects of several antioxidant molecules on A␤-induced neurotoxicity (Fig. 8). A 2-h preincubation of the cortical neurons with 10 M probucol, 1 M promethazine, or 5 M propyl gallate prior to non-fibrillar A␤ (1-40) peptide exposure induced a persistent increase in cell survival monitored by the MTT assay (Fig. 8A) and an inhibition of neuronal apoptosis as determined by DAPI staining (Fig. 8B). Finally, we clearly demonstrated, using immunocytochemistry, that antioxidant molecules prevented A␤-induced cytoskeleton disruption (Fig. 9, A, B, and C). The quantification of the number of cells exhibiting disorganized microtubules indicates that after a 6-h incubation, the presence of 10 M probucol during A␤ (1-40) treatment significantly reduced the number of cells exhibiting disordered microtubules, even after a 24-h incubation (18.9 ϩ 5.1%, p Ͻ 0.01, versus 40.5 ϩ 4.1% in the absence of probucol) (Fig. 9D). Moreover, we observed that the effects of probucol on A␤-induced cytoskeleton perturbation was concentration-dependent (not shown).
We reported previously that A␤ (29 -40) exhibited membrane fusion properties (30) and that part of the neurotoxicity of A␤ (1-40) and A␤ (1-42) was triggered by the insertion of their C-terminal ends into the plasma membrane of the neurons (18). Here, we have demonstrated that A␤ (29 -40) in- duced similar kinetics of caspase activation as A␤ (1-40) (Fig.  5). Furthermore, the incubation of the cortical neurons with A␤ (29 -40) resulted in the disorganization of the microtubule network, prevented by the presence of probucol (Fig. 9). These data highlight the critical role of ROS production in A␤-mediated cytoskeleton disruption and subsequent neuronal cell death.
Non-fibrillar A␤ Induces Primary Plasma Membrane Disorder-We next investigated whether the interaction of the nonfibrillar A␤  with the plasma membrane of cortical neurons could directly modify the properties of the membrane in living cells. To that end, cortical neurons were incubated with 1 M A␤ peptide, a concentration at which no significant cell death and ROS production could be observed even after an A␤ treatment of 3 h. After A␤ (1-40) treatment during increasing incubation times, the plasma membrane of the neurons was loaded with 4 M NBD-SM for 5 min, and the plasma membrane fluidity was assessed by measuring the lateral diffusion coefficient (D 20,w ) of the phospholipids using FRAP. The intensity of the fluorescent signal enabled us to discriminate the apical plasma membrane from the inside of the cell using confocal microscopy. All experiments were therefore performed at 37°C before the accumulation of the fluorescent probe in the inner membranes. After a 2-h incubation with 1 M A␤ (1-40), the cortical neurons displayed a slight increase of the lateral diffusion of NBD-SM, i.e. 0.280 ϩ 0.052 m 2 /s (n ϭ 27) and 0.402 ϩ 0.044 m 2 /s (n ϭ 26) (p Ͻ 0.05) in control cells and in cells treated for 2 h with A␤, respectively (Fig. 10). Prolonged incubation times up to 24 h, with the non-fibrillar A␤ resulted in a continuous decrease of the D 20,w (after 24 h of incubation, D 20,w ϭ 0.109 ϩ 0.064, p Ͻ 0.05), as compared with the control cells (Fig. 10). Interestingly, the alteration of the lateral diffu- We further investigated the effects of taxol on A␤ peptideinduced modifications of the membrane fluidity. In separate experiments than those described in Fig. 10 Finally, we demonstrated that the presence of antioxidant molecules did not counteract the disturbing effects of the non-fibrillar A␤ on the membrane (not shown). Altogether, these data suggest that A␤-induced membrane perturbation might represent a primary event of A␤induced neuronal cell death. DISCUSSION In the present paper, we demonstrate for the first time that (a) the cytoskeleton is an early cellular target for the nonfibrillar A␤ and that the disruption of the microtubule network is required for A␤-induced neuronal cell death, (b) the nonfibrillar A␤ induces a rapid oxidative stress that precedes and induces the microtubule disorganization, and (c) the earliest detectable action of soluble A␤ on the neurons is a rapid and biphasic modification of the plasma membrane fluidity preceding latter cellular events leading the cells into an apoptotic pathway. This process involves the activation of caspases, the neo-synthesis of proteins, and DNA fragmentation (18,22). Cell death occurs in a time-and dose-dependent manner via a pathway involving the activation of several classes of caspases, including caspases 3 and 9. However, this pathway did not involve the activation of caspase 8, as demonstrated through the measurement of the caspase 8-like activity and the absence of effect of caspase 8 inhibitors on soluble A␤-induced cell death. Our results contrast with previous studies demonstrating that cell death induced by fibrillar A␤ consisted partly in a rapid activation of caspase 8 (35,36). This corroborates our previous hypothesis (18,22), which was based on the idea that the cellular targets and the molecular mechanisms involved in cell death induced by A␤ strongly depended on peptide conformation.
It has already been suggested that early modifications in the progression of AD pathology might involve the disorganization of the cytoskeleton of neurons (37,38). Here, we show that when exposed to soluble A␤, most of the neurons displayed a disrupt microtubule architecture, even after 3 h of incubation and before the morphological and biochemical alterations typical of apoptotic cell death. Indeed, the perturbations of the microtubules precede caspase activation and nuclear DNA fragmentation and condensation. Moreover, the presence of a microtubule-stabilizing drug, e.g. taxol, inhibited neuronal cell death together with non-fibrillar A␤-induced perturbation of the cytoskeleton. It remains to establish the specificity of A␤induced microtubule perturbations. Using immunoblotting, we observed that the treatment of the cortical neurons with nonfibrillar A␤ did not modify the tubulin pools (data not shown). It could be hypothesized that the microtubule perturbations might occur via post-translational modifications and/or proteolytic cleavage of microtubule-associated proteins (MAP). Modifications of the phosphorylation level of tau and MAP 2 in AD brains have been described (39 -41). In agreement with these observations, it has been demonstrated that the products of lipid peroxidation disrupted the microtubules of cortical neurons directly (42).
Accumulative evidences emphasize the critical role of oxidative stress in AD and in the neurotoxicity of A␤ (26,27). Our results highlight the critical role of ROS production in the cytoskeleton disruption and the subsequent neuronal cell death induced by soluble A␤. The incubation of cortical neurons with low concentrations of soluble A␤ (1-40) results in an early time-dependent increase in the ROS production, preceding both microtubule perturbation and caspase activation. Indeed, the presence of anti-oxidant molecules prevents all of the intracellular events triggered by non-fibrillar A␤.
Finally, our results demonstrate for the first time that soluble A␤ (1-40) peptide induces a rapid modification of the membrane fluidity in living cells. Using FRAP analysis, we show that low concentrations of soluble A␤ (1-40) (1 M) induce a biphasic modification of the fluidity of the neuron plasma membrane as monitored by a global determination of the lateral diffusion coefficient of lipids. Upon short incubation times of cortical neurons with soluble A␤, we observed a reproducible increase in the membrane fluidity. In agreement with previous works (42)(43)(44), longer incubation times result in a significant decrease of the membrane fluidity, which could be attributed to lipid peroxidation associated with A␤-induced oxidative stress described in the present paper.
We have demonstrated recently that the non-fibrillar A␤ (1-40) displays fusogenic properties because of the membrane perturbing activity of its C-terminal domain, e.g. 29 -40 (30, 45). Here, we demonstrate that A␤ (29 -40) induces effects on the membrane fluidity and subsequent cell death similar to those induced by A␤ . These results strongly support the idea that the effects of A␤ on the membrane fluidity are partly mediated by the fusogenic properties of its C-terminal fragment and correlate the physicochemical properties to the neurotoxicity of non-fibrillar A␤.
Tailing our data, several observations suggest that the deleterious effects of A␤ on neuron viability might involve the interaction of A␤ with the lipid phase of the plasma membrane. A recent study by Cotman and co-workers (46) reports that the effects of A␤ on neuronal viability were not mediated by specific interactions with a receptor but more likely by changes in the structure and dynamics of the lipid constituents of the membrane (46). Most studies describe the action of aggregated forms of A␤ on model membranes or synaptosomes, demonstrating that the fibrils of A␤ displayed a high affinity with the membrane and decreased the fluidity of the lipid phase (47,48). Furthermore, aggregated A␤ (1-40) has strong electrostatic interactions with the surface of model membranes that appear to mediate its neurotoxicity (49). However, whereas both soluble and aggregated A␤ interact differently with rat synaptic plasma membranes, both transiently increase their membrane fluidity (50,51). Our results emphasize the importance of the early interactions of non-fibrillar A␤ with lipids in A␤-mediated neuronal cell death.
The direct relationships between the perturbation of the plasma membrane properties and cell death induced by the non-fibrillar A␤ are presently under investigation. The transient increase of the plasma membrane fluidity induced by low concentrations of soluble A␤ may explain the reported effects of A␤ on the permeability of the neurons to extracellular calcium and the increase in KCl-induced neuronal calcium in brain neurons and lymphocytes (52,53). Indeed, the presence of inhibitors of calcium influx reduced markedly neuronal cell death induced by soluble A␤  and (29 -40). 2 The interactions between soluble A␤ and the lipids might modify the function of ion channels or/and directly create ion pore, as described recently (25,54).
Altogether, our data strongly highlight the role of the nonfibrillar forms of A␤ in the progression of AD. Accordingly, recent findings have demonstrated that only in a non-aggregated conformation, A␤ activates the phosphoinositide signaling pathway (55) and mediates vasoconstriction activity both in vitro and in vivo (56,57). Uncovering the neuronal toxicity of non-fibrillar A␤ gives new insights into the development of AD therapy, which might take into account both amyloid deposit prevention and an efficient clearance of non-fibrillar A␤ peptides.