Inhibitors of Catalase-Amyloid Interactions Protect Cells from β-Amyloid-Induced Oxidative Stress and Toxicity*

Compelling evidence shows a strong correlation between accumulation of neurotoxic β-amyloid (Aβ) peptides and oxidative stress in the brains of patients afflicted with Alzheimer disease (AD). One hypothesis for this correlation involves the direct and harmful interaction of aggregated Aβ peptides with enzymes responsible for maintaining normal, cellular levels of reactive oxygen species (ROS). Identification of specific, destructive interactions of Aβ peptides with cellular anti-oxidant enzymes would represent an important step toward understanding the pathogenicity of Aβ peptides in AD. This report demonstrates that exposure of human neuroblastoma cells to cytotoxic preparations of aggregated Aβ peptides results in significant intracellular co-localization of Aβ with catalase, an anti-oxidant enzyme responsible for catalyzing the degradation of the ROS intermediate hydrogen peroxide (H2O2). These catalase-Aβ interactions deactivate catalase, resulting in increased cellular levels of H2O2. Furthermore, small molecule inhibitors of catalase-amyloid interactions protect the hydrogen peroxide-degrading activity of catalase in Aβ-rich environments, leading to reduction of the co-localization of catalase and Aβ in cells, inhibition of Aβ-induced increases in cellular levels of H2O2, and reduction of the toxicity of Aβ peptides. These studies, thus, provide evidence for the important role of intracellular catalase-amyloid interactions in Aβ-induced oxidative stress and propose a novel molecular strategy to inhibit such harmful interactions in AD.

The major ROS in cells are superoxide and the more reactive hydrogen peroxide (H 2 O 2 )-derived hydroxyl radical (24,25). Both superoxide and H 2 O 2 are primarily produced in the mitochondria (26 -28). A␤ peptides have been shown to accumulate in the mitochondria (29,30), and, therefore, could exert their detrimental effects through interaction with mitochondrial proteins (23,(31)(32)(33)(34). Superoxide is produced by several enzymecatalyzed reactions in the mitochondria (25). Behl et al. (11) however, showed that a broad range of inhibitors of several of these enzymes had no effect on 〈␤ toxicity in clonal and primary neuronal cell cultures. Furthermore, Zhang et al. (35) reported that superoxide levels in cells were not substantially elevated upon exposure to 〈␤. These findings suggest that superoxide is not a dominant contributor to 〈␤ toxicity.
On the other hand, 〈␤-induced cellular increase in H 2 O 2 or its metabolites is strongly correlated with 〈␤ toxicity (11,13,36). H 2 O 2 can be generated by several mitochondrial enzymes including monoamine oxidases, superoxide dismutase, and xanthine oxidase (25). Behl et al. (11) showed that inhibitors of monoamine oxidases and xanthine oxidase had no effect on 〈␤-induced H 2 O 2 accumulation or 〈␤ toxicity. Gsell et al. (37) also found that the activity of superoxide dismutase was unaltered in the brains of AD patients. Additionally, Rensink et al. (38) reported that the Dutch mutation of 〈␤ peptides (HCHWA-D 〈␤) did not bind directly to superoxide dismutase and Kaminsky et al. (39) reported only a relatively small effect of 〈␤ on superoxide dismutase activity and H 2 O 2 production upon chronic exposure of rat brains to 〈␤. These results suggest that 〈␤ peptides do not significantly affect production of H 2 O 2 in cells. Consequently, these findings imply that 〈␤-induced oxidative stress may arise from reduced degradation of H 2 O 2 in 〈␤-challenged cells.
Degradation of H 2 O 2 in cells is primarily achieved by the enzymes catalase and glutathione peroxidase (GPx), both inside and outside of the mitochondria (25). Sagara et al. (40) found that cells resistant to 〈␤ toxicity had elevated levels of catalase and GPx. The activity of both enzymes was reduced in rat brains exposed to 〈␤. In particular, Pappolla et al. and Lovell et al. found that catalase was associated with senile plaques in human brain sections from patients with AD (6,41). Several groups have shown that transfection of cells with catalase (40) or addition of catalase to the extracellular environment of cultured cells protected cells from 〈␤ toxicity (11,38). This cytoprotective effect was attributed to a catalase-induced reduction of the H 2 O 2 concentration outside and inside the cells, because H 2 O 2 readily diffuses across cell membranes (25,43). These observations are consistent with the hypothesis that H 2 O 2 , or its metabolites, but not superoxide, play a dominant role in 〈␤-induced oxidative stress. Moreover, Milton et al. (21) and our previous work (55) revealed that 〈␤ binds directly to catalase in cell free assays, whereas 〈␤ does not bind to GPx (supplemental Fig. S12). The binding of 〈␤ to catalase leads to deactivation of the H 2 O 2 -degrading activity of catalase in solution (21). Evidence for direct catalase-amyloid interactions and their resulting detrimental effects within living cells, however, has not yet been reported. Testing the hypothesis that intracellular catalase-〈␤ interactions play a significant role in 〈␤-induced increases in cellular H 2 O 2 may, therefore, provide a critical and missing mechanistic link between 〈␤ accumulation and oxidative stress in AD.
Here, we investigated the effect of catalase-amyloid interactions on A␤-induced increases in H 2 O 2 and on the resulting toxicity in live cells. In these studies, we exposed human neuroblastoma cells to a mixture of aggregated A␤(1-42) peptides (containing soluble oligomers and protofibrils) to mimic the heterogeneous nature of aggregated A␤ species found in the brains of AD patients (44 -46). Because the identity of the most toxic aggregated form of A␤ remains highly debated (47)(48)(49)(50)(51)(52)(53), we chose to use this mixture of different species of aggregated A␤ in these studies in order to increase the likelihood that as many toxic species as possible were included in all of our experiments. To assess whether catalase-amyloid interactions lead to 〈␤-induced increases in H 2 O 2 in cells, we designed and synthesized molecular probes with the following six characteristics: 1) capability of generating protein-resistant surface coatings on aggregated 〈␤ peptides (54 -56) (to inhibit catalase-amyloid interactions in cells), 2) lack of toxicity, 3) cell permeability, 4) capability of accessing the same subcellular compartments of cells as 〈␤, 5) intrinsic fluorescence properties (to visualize the intracellular location of the molecules), and 6) chemical stability in oxidizing environments. We achieved these characteristics with two oligo(ethylene glycol) derivatives of 6-methylbenzothiazole aniline (BTA-EG 4 and BTA-EG 6 ) (55); BTA analogs have wellknown, high affinity binding properties to both A␤ oligomers and fibrils (57,58). These molecules made it possible to challenge neuroblastoma cells with A␤ and to evaluate the effects of coating aggregated A␤ peptides with a protein-resistant surface on the viability of cells.

EXPERIMENTAL PROCEDURES
Materials-A detailed list of materials used in this research can be found in the supplemental information.
Preparation of Aggregated A␤  Peptides-Aggregated A␤  was prepared by incubation in deionized water (100 M) at 37°C for 72 h and characterized as described in the supporting information. For cellular experiments using aggregated A␤, a stock solution of aggregated A␤ prepared in sterile water was diluted with an equivalent volume of 2ϫ complete culture medium to create the desired concentration of aggregated A␤ dissolved in 1ϫ complete medium. For cellular experiments using aggregated A␤ in the presence of the BTA-EG x molecules, aggregated 〈␤ (in sterile water) was preincubated for 12 h with various concentrations of BTA-EG 4 or BTA-EG 6 , followed by addition of an equal volume of 2ϫ medium to create the desired solution of aggregated A␤ and BTA-EG x dissolved in 1ϫ medium.
MTT Cell Viability Assay-SH-SY5Y cells were plated at a density of 50,000 cells/well in 100 l of complete medium consisting of a 1:1 mixture of Eagle's Minimum Essential Medium (EMEM) and Ham's F12, supplemented with 10% fetal bovine serum (FBS). Cells were incubated overnight before treatment with 100 l of various sample solutions. Cells were exposed to solutions containing a final concentration of 25 M A␤ and various concentrations of the BTA-EG x molecules (0 -40 M) for 24 h at 37°C. The MTT reagent (20 l of the solution from the commercial kit) was added, and the cells were incubated for 3 additional hours. The cells were subsequently solubilized with detergent reagent (100 l of the solution from the commercial kit) and incubated at room temperature overnight. The cell viability was determined by measuring the absorbance at 570 nm using a Spectramax 190 microplate reader (Molecular Devices). All results were expressed as percent reduction of MTT relative to untreated controls (defined as 100% viability), and the average absorbance value for each treatment was blanked with the absorbance reading of wells containing only medium, MTT reagent, and detergent reagent.
Measurement of Hydrogen Peroxide Release in Cells-Cells were plated in 50 l of Dulbecco's modified Eagle's medium (DMEM) without phenol red, supplemented with 10% FBS and 4 mM L-glutamine. Cells were incubated overnight before dosing with sample solutions. The solutions of A␤ aggregates in the presence or in the absence of the BTA-EG x molecules were incubated with the cells for 24 h. Cellular H 2 O 2 release was determined by adding 20 l/well of a solution containing 250 M Amplex red reagent and 0.5 units/ml horseradish peroxidase (HRP) dissolved in complete medium. The absorbance at 560 nm was subsequently measured 30 min after addition of the Amplex red/HRP solution (59). For control experiments, cells were incubated for 24 h with 40 M BTA-EG x , 20 mM 3AT, or 20 mM 3AT with 40 M BTA-EG x , followed by measurement of H 2 O 2 released. All absorbance measurements for each treatment were measured in triplicate and the average absorbance values of control wells containing medium alone were subtracted.
Immunofluorescence Analysis of the Cellular Co-localization of A␤ and Catalase-SH-SY5Y cells were cultured on poly-L-lysine coated glass cover slips and incubated overnight in a 1:1 mixture of EMEM and Ham's F12 supplemented with 10% FBS. The growth medium was removed and solutions of aggregated fluorescently labeled A␤ (5 M dissolved in medium), or fluorescently labeled A␤ (5 M) that had been preincubated with BTA-EG x (40 M), were added to the different glass cover slips. The cells were incubated for 12 h. To visualize co-localization of aggregated A␤ with catalase, the cells were fixed with 4% paraformaldehyde in PBS (pH 7.4) and permeabilized with 0.25% Triton-X in PBS, followed by incubation with a rabbit anti-catalase polyclonal antibody on a shaker at 4°C overnight. To detect the primary antibody, a fluorescently labeled goat anti-rabbit antibody (diluted 1:1000) was added to the glass slides and incubated in the dark, on a shaker at room temperature for 2 h. Coverslips were mounted using Hydromount mounting medium (product no. HS-106) from National Diagnostics (Atlanta, GA) according to the manufacturers guidelines and allowed to dry for 48 h in the dark at 4°C before imaging with a Delta Vision Deconvolution Microscope System (Applied Precision, Issaquah, WA) equipped with a Nikon TE-200 inverted light microscope, with infinity corrected lenses, and with a mercury arc lamp as the illumination source. The fluorescence of the HiLyte Fluor 488-labeled A␤ (Abs/Em: 503/528 nm) was detected using an Ex/bp: 490/20 nm excitation filter and an Em/ bp: 528/38 nm emission filter. The fluorescence of the Alexa Fluor 647 goat anti-rabbit antibody (Abs/Em: 650/668) was monitored using an Ex/bp: 640/20 nm excitation filter and an Em/bp: 685/40 nm emission filter. The co-localization was visualized and Pearson's correlation coefficient of the entire three-dimensional volume of the cell was determined using softWoRx image analysis software (Applied Precision, Issaquah, WA). The images shown in Fig. 3 are fluorescence micrographs of representative z-slices within cells.
Co-immunoprecipitation of A␤ with Catalase-SH-SY5Y cells were grown to confluence on 10-cm tissue culturetreated dishes. The medium was removed, and solutions of aggregated A␤(1-42) (25 M, in medium), A␤ (25 M) that had been preincubated with BTA-EG x (40 M) or fresh medium were added to the different dishes and incubated overnight. The solutions were removed, and the cells were carefully washed three times with ice-cold PBS. The cells were incubated with 1 ml of ice-cold lysis buffer (from Cell Signaling Technology, Inc, Danvers, MA, Product No: 9803) containing a protease inhibitor mixture (Sigma-Aldrich, Product No: P2714) and 1 mM phenylmethylsulfonyl fluoride and placed on a shaker for 30 min at 4°C. The cell lysate was collected, centrifuged, and the supernatant of the sample was precleared with 50 l of protein G-agarose beads (Roche Diagnostics Co., Indianapolis, IN, Product No: 11719416001) to remove proteins that bind nonspecifically to protein G. The precleared cell lysate was subjected to immunoprecipitation with a goat polyclonal to catalase antibody (Abcam Inc., Cambridge, MA, Product No: ab50434) and protein G-agarose beads were used to extract the anti-catalase antibody and catalase from the lysate. The samples were subjected to 4 -12% SDS-polyacrylamide gel electrophoresis. Catalase (from human erythrocytes) and synthetic aggregated A␤  were also separated on the same gel as the immunoprecipitated proteins for comparison purposes. Western blot analyses using rabbit polyclonal anti-catalase antibody (Abcam Inc., Product No: ab1877) and 6E10 monoclonal anti-A␤(1-16) antibody (Covance) were performed. Chemiluminescence was used to detect the proteins.
Assessment of the Cellular Internalization of A␤ and BTA-EG x Molecules by Fluorescence Deconvolution Microscopy-SH-SY5Y cells were plated in DMEM without phenol red (supplemented with 10% FBS and 4 mM L-glutamine) on 35 mm glass bottom dishes (MatTek Co., Ashland, MA) and in-cubated overnight. The growth medium was removed, and solutions of pre-aggregated fluorescently labeled A␤ (5 M dissolved in medium) with or without BTA-EG x (40 M dissolved in medium) were added to the cells and allowed to incubate for 12 h before imaging. The cells were washed with DMEM (without phenol red or FBS) immediately prior to imaging with a Delta Vision Deconvolution Microscope System (Applied Precision, Issaquah, WA) equipped with a Nikon TE-200 inverted light microscope, with infinity corrected lenses, and with a mercury arc lamp as the illumination source. The co-localization was determined using softWoRx image analysis software (Applied Precision, Issaquah, WA). The intrinsic fluorescence of the BTA-EG x molecules (Abs/Em: 360/ 440 nm (60) Flow Cytometry-SH-SY5Y cells were grown to confluence on 35 mm tissue culture-treated dishes. The medium was removed, and solutions of aggregated fluorescently labeled A␤ (5 M dissolved in medium), fluorescently labeled A␤ (5 M) that had been preincubated with BTA-EG x (40 M), or fresh medium only were added to different dishes, and incubated overnight. The solutions were removed, and the cells were carefully washed with PBS. The cells were dissociated from the tissue culture dishes with solutions of 0.25% trypsin-EDTA, collected, and counted. The dead cells in each sample were stained using Live/Dead fixable far red stain (Invitrogen Co., Product No: L10120) according to the manufacturer's protocol, and all cell samples were fixed with 4% paraformaldehyde in PBS (pH 7.4). The flow cytometry data were acquired using a FACSCanto flow cytometer (Becton Dickinson, San Jose, CA) equipped with an argon 488 nm laser and a 530/30 band-pass filter to analyze the fluorescently labeled A␤ as well as a helium-neon 633 nm laser and a 660/20 bandpass filter to analyze the far red dye. FACSDiva software (Becton Dickinson) was used to control the setup, acquisition, and analysis of flow cytometry data from the FACSCanto flow cytometer. The samples were gated to only include single cell populations, excluding larger cell aggregates and cellular debris. This population contained more than 95% viable cells (Ͼ20,000 cells/sample). The live cells were analyzed for the presence of fluorescently labeled A␤, and the cellular fluorescence from this cell population was presented as histograms.  (61,62). The average absorbance values for each concentration of BTA-EG x were subtracted with the absorbance value of blanks consisting of 25 M A␤, the corresponding concentration of the BTA-EG x molecules, 25 nM catalase, the colorimetric reagents, and 50 mM sodium phosphate buffer (instead of a solution of H 2 O 2 ). The results were expressed relative to absorbance values measured when catalase was not exposed to A␤ (defined as 100% catalase activity), and the absorbance measured when the assay was run in the absence of catalase (defined as 0% catalase activity).

BTA-EG x Molecules Reduce the Toxicity of A␤ Peptides-
We hypothesized that small molecules (e.g. BTA-EG x , Fig. 1A) with the capability of forming protein-resistant surface coatings on aggregated A␤ peptides (55) could protect cells from A␤ toxicity. To test this hypothesis, we determined the viability (63, 64) of human SH-SY5Y neuroblastoma cells after exposure to aggregated A␤ peptides in the presence or in the absence of these small molecules. For these experiments, we mimicked the heterogeneous population of aggregated A␤ species in AD brains (44 -46) by using a preparation of A␤ peptides that contained ϳ15% small oligomers (MW ϳ15 kDa corresponding to trimers), ϳ25% medium-sized oligomers (MW 20 -65 kDa corresponding to 5-15 mers), and ϳ60% soluble protofibrils (MW Ͼ 150 kDa corresponding to Ͼ 30 mers) (see supplemental Fig. S1). Fig. 1B shows that exposure of cells to 25 M concentrations of this A␤ prepara-tion reduced their viability to ϳ27% compared with control cells that were cultured in the absence of A␤.
Before exploring the effect of the BTA-EG x molecules on reducing the cytotoxicity of A␤, we tested the tolerance of cells to these BTA-EG x molecules by themselves. These experiments revealed that the BTA-EG x molecules were not toxic at concentrations up to 40 M 4 ( Fig. 1B and  supplemental Fig. S2). Remarkably, the BTA-EG x molecules were, however, capable of reducing the toxicity of aggregated A␤, causing an increase in cell viability to ϳ72% (Fig. 1C). 5 Because we used a mixture of aggregated A␤ species in these experiments, the results shown in Fig. 1C suggest that the BTA-EG x molecules could simultaneously protect cells from several potentially toxic forms of aggregated A␤ peptides that are present in this mixture (47)(48)(49)(50)(51)(52)(53). In all of the work shown here, we describe results using both of the structurally related BTA-EG 4 and BTA-EG 6 molecules because testing two molecules provides 2-fold evidence that this class of molecules can protect cells from A␤-related injury. 4 Further examination showed that these molecules are not toxic to this cell line at concentrations below 60 M (supplemental Fig. S2). 5 Interestingly, when SH-SY5Y cells are preincubated with the BTA-EG x molecules for 24 h prior to challenging cells with untreated aggregated A␤ peptides (i.e., aggregated A␤ that was not preincubated with BTA-EG x ), we did not observe significant inhibition of A␤-induced toxicity in the cells. Because the effect of the BTA-EG x molecules on protection of A␤induced toxicity in cells depended on whether they were preincubated with the aggregated A␤ prior to exposure of the peptides to the cells, we hypothesize that distribution of the BTA-EG x molecules in cells may be different depending on whether they are introduced to cells alone (i.e., in the absence of A␤) or whether the BTA-EG x molecules are preincubated and introduced to cells together with A␤ peptides. Preincubation with A␤ may, therefore, improve internalization of the BTA-EG x molecules and facilitate, enhance, or stabilize intracellular interactions between BTA-EG x and A␤.

BTA-EG x Molecules Inhibit A␤-induced Increases in Cellular H 2 O 2 Levels-To assess whether the BTA-EG x molecules can minimize A␤-induced increases in cellular H 2 O 2 levels,
we measured the amount of H 2 O 2 released by SH-SY5Y cells in the presence of aggregated A␤ peptides. Fig. 2A demonstrates that cells incubated with 25 M A␤ for 1 day released 3-fold more H 2 O 2 than control cells that were not exposed to A␤. Moreover, Fig. 2B shows that concentrations of BTA-EG x above 10 M significantly inhibited A␤-induced elevations in H 2 O 2 levels in these cells. The presence of the BTA-EG x molecules together with A␤ resulted in low H 2 O 2 levels that were indistinguishable from control cells. Additional control experiments showed that 40 M solutions of the BTA-EG x molecules by themselves (i.e. in the absence of A␤) did not affect cellular H 2 O 2 levels compared with cells that were not exposed to these small molecules ( Fig. 2A).
To evaluate if the BTA-EG x molecules exerted their action by either specifically inhibiting A␤-induced increases in cellular H 2 O 2 levels or by reacting chemically with H 2 O 2 or other ROS, we inhibited catalase with a known inhibitor (3-amino-1,2,4-triazole, 3AT (25, 65)). We expected this treatment to increase cellular levels of H 2 O 2 . Fig. 2A shows that a 20 mM solution of 3AT caused a 3-fold increase in H 2 O 2 released by cells compared with untreated control cells. This increase in H 2 O 2 levels was not affected by incubating cells with 3AT in the presence of 40 M BTA-EG x ( Fig. 2A), demonstrating that the anti-oxidant properties of the BTA-EG x molecules are specific to A␤-mediated oxidative stress. Mass spectrometry analysis provided additional evidence that the BTA-EG x molecules did not react directly with H 2 O 2 in solution (see supplemental information). The BTA-EG x molecules, therefore, exerted their activity by a mechanism that is different from a chemical anti-oxidant mechanism.
BTA-EG x Molecules Readily Internalize in Cells and Reduce the Intracellular Co-localization of A␤ and Catalase-To test the hypothesis that intracellular protein-amyloid interactions contribute to A␤-induced increases in H 2 O 2 levels and to A␤ toxicity in live cells, we exposed SH-SY5Y cells to a solution containing N-terminal, fluorescently labeled 〈␤ peptides. This experiment made it possible to determine the location of these peptides inside of cells. Fig. 3, A-D DECEMBER 10, 2010 • VOLUME 285 • NUMBER 50

JOURNAL OF BIOLOGICAL CHEMISTRY 38937
locations such as the lysosomes (supplemental Fig. S3) and mitochondria (supplemental Fig. S4). Because catalase is the primary enzyme responsible for degradation of H 2 O 2 in cells (25,71), we investigated whether the interaction of A␤ peptides with intracellular catalase (72) could contribute to the observed A␤-induced increase in H 2 O 2 levels. Fig. 3, A-D and supplemental Fig. S5 show that a significant fraction of the aggregated A␤(1-42) peptides co-localized with catalase in the peroxisomes and in the cytosol; this observation has not been reported previously in live cells (30, 33, 66 -70). Fig. 3 also shows that BTA-EG 4 (Fig. 3H) and BTA-EG 6 (Fig. 3L) significantly reduced the intracellular co-localization of A␤ and catalase. Although Fig. 3 shows only representative z-slice fluorescence micrographs of these cells, we quantified the degree of co-localization of A␤ and catalase within the entire three-dimensional volume of cells with the Pearson's correlation coefficient (73). Before addition of the BTA-EG x molecules, this coefficient was 0.46, whereas it was 0.10 or 0.29 when cells were introduced to A␤ together with BTA-EG 4 or BTA-EG 6 , respectively. The molecular association between A␤(1-42) peptides and catalase was further supported by exposing SH-SY5Y cells to aggregated A␤(1-42) peptides and immunoprecipitating catalase with an anti-catalase antibody. Western blot analysis revealed a protein band of ϳ4.5 kDa in the lysates (corresponding to the MW of A␤ monomers), which was immunoreactive to an anti-A␤ antibody and coimmunoprecipitated with catalase (one subunit ϳ60 kDa) (Fig. 4). This ϳ4.5 kDa band was not observed in the Western blots from control cells that were not treated with A␤. Additionally, immunoprecipitation of catalase in the presence of the BTA-EG x molecules reduced the co-immunoprecipitation of A␤ peptides with the catalase from the cell lysate (Fig. 4). Interestingly, the fluorescence microscopy experiments and the co-immunoprecipitation experiments revealed a consistent trend for the slightly better performance of BTA-EG 4 for reducing the interaction of A␤ with catalase in cells compared with the BTA-EG 6 . We attribute this apparent enhanced performance of BTA-EG 4 to its increased lipophilicity compared with BTA-EG 6 (55), which may facilitate internalization in cells.
As a control for the fluorescence microscopy co-localization experiment, we also investigated the possible intracellular inter-action of aggregated A␤ peptides with superoxide dismutase (SOD). Supplemental Fig. S6 shows results from fluorescence confocal imaging experiments revealing a reduced degree of overlap of aggregated A␤ and SOD throughout the entire volume of the cell (Pearson's coefficient ϭ 0.27) relative to the overlap of A␤ and catalase (Pearson's coefficient ϭ 0.46) under the same conditions. Additionally, the presence of the BTA-EG x  molecules did not affect the degree of overlap of A␤ and SOD (Pearson's coefficient ϭ 0.30 for both BTA-EG x molecules) compared with cells that were treated with A␤ alone (supplemental Fig. S6). Finally, we exposed SH-SY5Y cells to aggregated A␤  peptides and immunoprecipitated SOD with an anti-SOD antibody. These experiments reveal that SOD does not appear to co-immunoprecipitate with A␤ (supplemental Fig. S7), which is in contrast to the observed co-immunoprecipitation of A␤ with catalase under the same conditions (Fig. 4). These results support that A␤ and SOD do not likely interact significantly within cells. Fig. 3 and supplemental Fig. S5 clearly illustrates that the BTA-EG x molecules are capable of reducing the co-localization of catalase and A␤ in cells. To test if this effect could be due to the interaction of the BTA-EG x molecules with A␤, we determined whether the BTA-EG x molecules could also interact with A␤ in the cells. Fig. 5, A-F shows that the intrinsically fluorescent (60) BTA-EG x molecules distributed throughout the cells and can occupy overlapping subcellular regions of the cells that are also occupied by 〈␤. Additionally, Fig. 5G shows results from flow cytometry experiments demonstrating that the BTA-EG x molecules (especially BTA-EG 4 ) do not significantly affect the uptake of the aggregated fluorescently labeled A␤ peptides in the cells. These results are consistent with the possibility that the BTA-EG x molecules were capable of entering cells and could, hence, interact with intracellular pools of A␤ peptides. We hypothesize that this interaction between the BTA-EG x molecules and A␤ protected cells from harmful protein-amyloid interactions by forming protein-resistant molecular surface coatings on A␤ (55). Such a mechanism would explain the observed reduction in intracellular co-localization of catalase and A␤ in the presence of the BTA-EG x molecules. To test this hypothesis, we examined whether the BTA-EG x molecules protected the H 2 O 2 -degrading activity of catalase in A␤-rich environments.
BTA-EG x Molecules Protect the H 2 O 2 -degrading Activity of Catalase in the Presence of A␤-We reported previously that BTA-EG 4 and BTA-EG 6 could inhibit catalase-A␤ binding interactions in solution by forming protein-resistant surface coatings on aggregated A␤(1-42) peptides (55). In order to determine whether these molecules could also protect the enzymatic activity of catalase in an A␤-rich environment, we developed an assay for quantifying catalase activity (see supplemental Fig. S8). We designed this assay to be compatible with the presence of the BTA-EG x molecules and with the presence of aggregated A␤ peptides in solution (see supplemental Fig. S9). Fig. 6A shows that micromolar concentrations of A␤ suppressed catalase activity substantially, in agreement with previous reports (21). Importantly, Fig. 6, B and C shows that the presence of BTA-EG x molecules preserved 80 -90% of catalase activity despite the presence of A␤. The concentration of BTA-EG x molecule that resulted in halfmaximal preservation of catalase activity was ϳ8 nM for BTA-EG 4 and ϳ40 nM for BTA-EG 6 . 6 These results are consistent with the hypothesis that the BTA-EG x molecules protect cells from 〈␤-induced increases in H 2 O 2 levels and from 〈␤ toxicity through a mechanism involving the formation of proteinresistant molecular coatings on aggregated A␤. 7 FIGURE 6. Inhibition of catalase activity by A␤ and preservation of catalase activity by BTA-EG x molecules in 〈␤-rich solutions. A, catalase activity in the presence of increasing concentrations of aggregated A␤. The results are expressed relative to activity of catalase in the absence of A␤. B and C, catalase activity in the presence of presence of 25 M aggregated A␤ and increasing concentrations of BTA-EG 4 (B) or BTA-EG 6 (C). Catalase activity was defined as 100% when no A␤ was present and as 0% when no catalase was present in the sample. Data are expressed as mean values Ϯ S.D., n ϭ 4 for each concentration.
To consider another possible mechanism by which the BTA-EG x molecules could affect A␤-catalase interactions, for instance, by affecting the aggregation state of A␤, we performed dynamic light scattering experiments of A␤ preparations (74 -76). We carried out these experiments in the presence and in the absence of the BTA-EG x molecules and found that the aggregation states of A␤ were not altered significantly due to the presence of the BTA-EG x molecules (see supplemental Fig. S10). In contrast, incubation of aggregated A␤ peptides with vanillin, a molecule known to inhibit A␤ aggregation (77), resulted in a population of A␤ monomers.

DISCUSSION
A␤ has been implicated as a possible causative agent for oxidative stress in AD (9,78). Proposed mechanisms for A␤induced oxidative damage in cells include: 1) A␤ peptides interact directly with anti-oxidant enzymes resulting in accu-mulation of ROS (20 -23); 2) A␤ peptides interact directly with cellular membranes resulting in uncontrolled calcium influx as a consequence of forming transient, ion channel-like pores (79 -87); the resulting disruption of calcium homeostasis can lead to mitrochondrial dysfunction, followed by increased H 2 O 2 production and generation of high levels of ROS (86,88); and 3) A␤ peptides interact directly with intracellular transition metal ions such as Fe 2ϩ (89 -92), resulting in the catalytic reduction of O 2 to H 2 O 2 in cells (17,18,89).
Here, we developed a small molecule probe to investigate, in detail, the first mechanism, i.e. the specific contribution of protein-amyloid interactions toward A␤-induced oxidative stress. The results presented here support that deactivation of catalase by intracellular catalase-amyloid interactions can lead to A␤-induced increases in H 2 O 2 levels and to subsequent toxicity in cells. We base this conclusion on the following three key observations. First, A␤ internalized in cells and co- FIGURE 7. Schematic diagram of the proposed deactivation of catalase as a contributor of A␤-induced oxidative stress. Cellular internalization of A␤ peptides via endocytotic and/or non-endocytotic pathways result in accumulation of A␤ within cells. The internalized A␤ can interact with proteins such as catalase present in the cytosol, peroxisomes, and other cellular organelles. The diagram of the expanded region illustrates the deactivation of the H 2 O 2degrading activity of catalase in the cytosol or peroxisomes (pathway 1). The BTA-EG x molecules, however, protect the activity of catalase by forming protein-resistant surface coatings on aggregated A␤ and inhibiting harmful intracellular catalase-amyloid interactions (pathway 2). localized with catalase. The results from fluorescence imaging experiments ( Fig. 3 and supplemental Figs. S3-S6) demonstrate that exogenous aggregated A␤(1-42) is capable of entering the cells through an endocytotic lysosomal pathway (67,70,93,94) or a non-endocytotic pathway (95). After the initial uptake of A␤, the A␤ peptides can accumulate in the cytosol, peroxisomes, mitochondria, and other intracellular compartments. Fig. 7 summarizes the results presented here in a graphical illustration that describes the proposed deactivation of catalase as a contributor of A␤-induced oxidative stress. Because we found a significant fraction of A␤ co-localized with catalase in the peroxisomes and in the cytosol (25,96) (Fig. 3C) and co-immunoprecipitated with catalase ( Fig.  4), it is apparent that A␤ interacts directly with catalase inside of live cells (Fig. 7, pathway 1). Moreover, addition of small molecule inhibitors of catalase-A␤ interactions significantly reduced co-localization of A␤ and catalase inside the cells (Figs. 3, H and L and 7, pathway 2) without affecting significantly the uptake of A␤ in cells (Fig. 5G).
Second, binding interactions between catalase and A␤ deactivate catalase inside of live cells, leading to decreased degradation of H 2 O 2 in solution (Fig. 6A) and increased cellular levels of H 2 O 2 ( Fig. 2A). The results presented here reveal that small molecule inhibitors of catalase-A␤ interactions are capable of protecting the activity of catalase in A␤-rich solutions (Fig. 6, B and C) and of attenuating A␤-induced increases in H 2 O 2 levels in cells (Fig. 2B).
Third, A␤ significantly reduced cell viability (Fig. 1A) at concentrations that led to a 3-fold increase in cellular H 2 O 2 levels ( Fig. 2A). These results provide evidence for a correlation between A␤-induced oxidative stress and A␤ toxicity as proposed by others (11,13,36). Remarkably, the BTA-EG x molecules (by inhibiting catalase-amyloid interactions) significantly reduced the toxicity of A␤ peptides in cells (Fig. 1B) at concentrations that also reduced A␤-induced increases in cellular H 2 O 2 levels (Fig. 2B). This result further supports the important role of catalase-A␤ interactions in A␤-induced elevations of H 2 O 2 and in A␤-mediated toxicity.
Although the results presented here clearly demonstrate that intracellular catalase-amyloid interactions induce elevations in cellular H 2 O 2 levels, an important question is whether these interactions are relevant to AD pathology. The accumulation of catalase in senile plaques (41) suggests that in vivo environments do indeed provide evidence for binding of catalase to A␤. We hypothesize that these in vivo catalase-amyloid interactions lead to deactivation of catalase, resulting in neuronal loss as a consequence of increased levels of H 2 O 2 and its metabolites in the brain. This hypothesis is supported by the strong correlation of deactivation of intracellular catalase by either A␤ (Fig. 6A) or by catalase inhibitor 3AT (97) with the ensuing increased H 2 O 2 levels ( Fig. 2A) as well as with reduced viability (Fig. 1B and supplemental Fig. S11) in cells.
The results presented here provide clear evidence for the important role of catalase-amyloid interactions in A␤-induced oxidative stress in cells, as outlined in Fig. 7. The results, however, do not rule out the possibility that amyloid interactions with proteins other than catalase could also contribute to cellular injury. We previously reported that BTA-EG 4 and BTA-EG 6 are capable of inhibiting several different A␤-binding proteins from associating with aggregated A␤ peptides (55). It is, therefore, likely that the BTA-EG x molecules in cells also attenuate harmful effects of A␤ on proteins other than catalase. From a therapeutic point of view, this is, of course, an attractive characteristic. Furthermore, these BTA-EG x molecules exhibit desirable drug-like characteristics (98,99) because they are: 1) non-toxic and cell permeable; 2) small (MW of BTA-EG 4 ϭ 418 and MW of BTA-EG 6 ϭ 504) and have beneficial topological polar surface areas (42, 100); 3) low in the number of H-bond donors (i.e. less than 5); and 4) low in the number of H-bond acceptors (i.e. less than 10). Most importantly, as revealed in this work, these molecules inhibit catalase-A␤ interactions, reduce the extent of co-localization of A␤ and catalase in cells, protect the activity of catalase in A␤-rich environments, inhibit A␤-induced increases in cellular H 2 O 2 levels, and reduce A␤ toxicity. In conclusion, the formation of protein-resistant surface coatings on aggregated forms of amyloid proteins introduces an attractive and novel strategy for probing the effect of A␤ on cells and potentially for treatment of AD.