Curcumin Inhibits Formation of Amyloid β Oligomers and Fibrils, Binds Plaques, and Reduces Amyloid in Vivo

Alzheimer's disease (AD) involves amyloid β (Aβ) accumulation, oxidative damage, and inflammation, and risk is reduced with increased antioxidant and anti-inflammatory consumption. The phenolic yellow curry pigment curcumin has potent anti-inflammatory and antioxidant activities and can suppress oxidative damage, inflammation, cognitive deficits, and amyloid accumulation. Since the molecular structure of curcumin suggested potential Aβ binding, we investigated whether its efficacy in AD models could be explained by effects on Aβ aggregation. Under aggregating conditions in vitro, curcumin inhibited aggregation (IC50 = 0.8 μm) as well as disaggregated fibrillar Aβ40 (IC50 = 1 μm), indicating favorable stoichiometry for inhibition. Curcumin was a better Aβ40 aggregation inhibitor than ibuprofen and naproxen, and prevented Aβ42 oligomer formation and toxicity between 0.1 and 1.0 μm. Under EM, curcumin decreased dose dependently Aβ fibril formation beginning with 0.125 μm. The effects of curcumin did not depend on Aβ sequence but on fibril-related conformation. AD and Tg2576 mice brain sections incubated with curcumin revealed preferential labeling of amyloid plaques. In vivo studies showed that curcumin injected peripherally into aged Tg mice crossed the blood-brain barrier and bound plaques. When fed to aged Tg2576 mice with advanced amyloid accumulation, curcumin labeled plaques and reduced amyloid levels and plaque burden. Hence, curcumin directly binds small β-amyloid species to block aggregation and fibril formation in vitro and in vivo. These data suggest that low dose curcumin effectively disaggregates Aβ as well as prevents fibril and oligomer formation, supporting the rationale for curcumin use in clinical trials preventing or treating AD. Alzheimer's disease (AD) involves amyloid β (Aβ) accumulation, oxidative damage, and inflammation, and risk is reduced with increased antioxidant and anti-inflammatory consumption. The phenolic yellow curry pigment curcumin has potent anti-inflammatory and antioxidant activities and can suppress oxidative damage, inflammation, cognitive deficits, and amyloid accumulation. Since the molecular structure of curcumin suggested potential Aβ binding, we investigated whether its efficacy in AD models could be explained by effects on Aβ aggregation. Under aggregating conditions in vitro, curcumin inhibited aggregation (IC50 = 0.8 μm) as well as disaggregated fibrillar Aβ40 (IC50 = 1 μm), indicating favorable stoichiometry for inhibition. Curcumin was a better Aβ40 aggregation inhibitor than ibuprofen and naproxen, and prevented Aβ42 oligomer formation and toxicity between 0.1 and 1.0 μm. Under EM, curcumin decreased dose dependently Aβ fibril formation beginning with 0.125 μm. The effects of curcumin did not depend on Aβ sequence but on fibril-related conformation. AD and Tg2576 mice brain sections incubated with curcumin revealed preferential labeling of amyloid plaques. In vivo studies showed that curcumin injected peripherally into aged Tg mice crossed the blood-brain barrier and bound plaques. When fed to aged Tg2576 mice with advanced amyloid accumulation, curcumin labeled plaques and reduced amyloid levels and plaque burden. Hence, curcumin directly binds small β-amyloid species to block aggregation and fibril formation in vitro and in vivo. These data suggest that low dose curcumin effectively disaggregates Aβ as well as prevents fibril and oligomer formation, supporting the rationale for curcumin use in clinical trials preventing or treating AD. The 4-kDa (40–42-amino acid) amyloid-β peptide (Aβ) 1The abbreviations used are: Aβ, amyloid β; APP, amyloid precursor protein; AD, Alzheimer's disease; CR, Congo Red; dH2O, distilled H2O; BSA, bovine serum albumin; PBS, phosphate-buffered saline; ELISA, enzyme-linked immunosorbent assay; Tricine, N-[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]glycine; ANOVA, analysis of variance; EM, electron microscopy; LDH, lactate dehydrogenase; NSAID, nonsteroidal anti-inflammatory drug; HFIP, hexafluoroisopropanol. is derived from the amyloid precursor protein (APP) through sequential proteolysis by the aspartyl protease β-secretase and presenilin-dependent γ-secretase cleavage (1Selkoe D.J. Science. 1997; 275: 630-631Crossref PubMed Scopus (848) Google Scholar). Mutations at the cleavage sites in APP or in presenilin that increase production or aggregation of Aβ provide a compelling argument for a central role for Aβ aggregation in the pathogenesis of Alzheimer's disease (AD). The progressive accumulation of Aβ aggregates is widely believed to be fundamental to the initial development of neurodegenerative pathology and to trigger a cascade of events such as neurotoxicity, oxidative damage, and inflammation that contribute to the progression of AD (2Pike C.J. Walencewicz A.J. Glabe C.G. Cotman C.W. Brain Res. 1991; 563: 311-314Crossref PubMed Scopus (825) Google Scholar, 3Klein W.L. Krafft G.A. Finch C.E. Trends Neurosci. 2001; 24: 219-224Abstract Full Text Full Text PDF PubMed Scopus (916) Google Scholar, 4Cummings J.L. Vinters H.V. Cole G.M. Khachaturian Z.S. Neurology. 1998; 51 (S65-S67): S2-S17Crossref PubMed Google Scholar, 5Harris M.E. Hensley K. Butterfield D.A. Leedle R.A. Carney J.M. Exp. Neurol. 1995; 131: 1-10Crossref PubMed Scopus (328) Google Scholar). Therefore, many therapeutic efforts are targeted at reducing Aβ production, including inhibiting secretase, increasing Aβ clearance with amyloid vaccines, or blocking Aβ aggregation (with antibodies, peptides, or small organic molecules that selectively bind and inhibit Aβ aggregate and fibril formation). Aβ fibrillization involves formation of dimers and small oligomers followed by growth into protofibrils and fibrils via a complex multistep-nucleated polymerization. Polymerizing Aβ fibrils and intermediates can be stained by amyloidophilic dyes such as Congo Red (CR) (6Puchtler H. Sweat F. Levine M. J. Histochem. Cytochem. 1962; 10: 355-364Crossref Google Scholar). Congo Red can inhibit the toxic and inflammatory activity of polymerizing Aβ (7Lorenzo A. Yankner B.A. Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 12243-12247Crossref PubMed Scopus (1298) Google Scholar, 8Abe K. Kato M. Saito H. Neurosci. Res. 1997; 29: 129-134Crossref PubMed Scopus (15) Google Scholar) and prevent the natural oligomer formation found at low Aβ concentrations (9Podlisny M.B. Walsh D.M. Amarante P. Ostaszewski B.L. Stimson E.R. Maggio J.E. Teplow D.B. Selkoe D.J. Biochemistry. 1998; 37: 3602-3611Crossref PubMed Scopus (190) Google Scholar). However, CR is toxic and negatively charged and therefore poorly brain-penetrant (10Klunk W.E. Debnath M.L. Pettegrew J.W. Neurobiol. Aging. 1994; 15: 691-698Crossref PubMed Scopus (166) Google Scholar). Curcumin (diferulomethane) is a low molecular weight molecule with potent antioxidant and anti-inflammatory activities that has a favorable toxicity profile and is under development as a potential cancer chemotherapeutic agent (11Kelloff G.J. Crowell J.A. Hawk E.T. Steele V.E. Lubet R.A. Boone C.W. Covey J.M. Doody L.A. Omenn G.S. Greenwald P. Hong W.K. Parkinson D.R. Bagheri D. Baxter G.T. Blunden M. Doeltz M.K. Eisenhauer K.M. Johnson K. Knapp G.G. Longfellow D.G. Malone W.F. Nayfield S.G. Seifried H.E. Swall L.M. Sigman C.C. J. Cell. Biochem. Suppl. 1996; 26: 72-85PubMed Google Scholar). Our previous results demonstrated that chronic dietary curcumin lowered Aβ deposition in 16-month-old APPsw transgenic mice (Tg2576) (12Lim G.P. Chu T. Yang F. Beech W. Frautschy S.A. Cole G.M. J. Neurosci. 2001; 21: 8370-8377Crossref PubMed Google Scholar). However, it remains unresolved whether curcumin was reducing plaques in vivo in part by effects on aggregation. As shown in Fig. 1, curcumin has a structure similar to Congo Red but with the charge replaced by polar groups like the brain-permeable compound chrysamine G (10Klunk W.E. Debnath M.L. Pettegrew J.W. Neurobiol. Aging. 1994; 15: 691-698Crossref PubMed Scopus (166) Google Scholar, 13Klunk W.E. Jacob R.F. Mason R.P. Anal. Biochem. 1999; 266: 66-76Crossref PubMed Scopus (262) Google Scholar). It is also similar to RS-0406, a novel compound selected from a screen of 113,000 compounds as a potent inhibitor of Aβ oligomer formation (14Nakagami Y. Nishimura S. Murasugi T. Kaneko I. Meguro M. Marumoto S.K.H. Koyama K. Oda T. Br. J. Pharamacol. 2002; 137: 676-682Crossref PubMed Scopus (88) Google Scholar). We hypothesized that, like these polar Aβ binding compounds, curcumin might be able to cross the blood-brain barrier and bind to amyloid and related aggregates. Therefore, in these studies, we used an in vitro model of Aβ fibrillization to show that curcumin can bind amyloid to inhibit Aβ aggregation as well as fibril and oligomer formation with dosing at achievable levels. We also demonstrate that curcumin can label plaques in vitro and in vivo, block toxicity of oligomers in vitro, and significantly reduce amyloid levels in aged Tg2576 mice (22 months old) fed a curcumin diet beginning at 17 months after established amyloid deposition. Aβ40 and Aβ42 peptides were purchased from the laboratory of Dr. Charles Glabe (University of California, Irvine, CA) and American Peptide (Sunnyvale, CA). Aβ fragment peptides 1–13, 1–28, 25–35, 34–40, and 37–42 were purchased from Bio-Synthesis (Lewisville, TX), and Aβ 12–28 was purchased from American Peptide Company. Aβ 8–17, 14–24, 17–24, and 17–28 were purchased from Quality Control Biochemicals (Hopkinton, MA). Curcumin, naproxen, and ibuprofen were purchased from Cayman Chemicals (Ann Arbor, MI). Stocks of curcumin, naproxen, and ibuprofen (5 mm) were dissolved in 100% ethanol and stored at -80 °C. A 5 mm stock of Congo Red was prepared in dH2O immediately before use. The 6E10 antibody (Aβ 1–17) was purchased from Signet Labs, and the A11 antibody was generously provided by C. Glabe (University of California, Irvine). All other reagents were from Sigma. Just before use, curcumin was diluted from 2.5 mm to 50 nm in 0.1 m TBS (pH 7.4) containing 3% BSA with 0.5% Tween 20. Thioflavin S (1%) was freshly prepared in dH2O, stirred for 0.5 h, and filtered. Human AD hippocampus was fixed in buffered formalin and snap-frozen. Tg2576 mice (22 months old) were perfused with Hepes buffer and protease inhibitors (15Lim G.P. Yang F. Chu T. Chen P. Beech W. Teter B. Tran T. Ubeda O. Hsiao Ashe K. Frautschy S.A. Cole G.M. J. Neurosci. 2000; 20: 5709-5714Crossref PubMed Google Scholar), and brains were removed, fixed in buffered formalin, and snap-frozen. Both brains were cryosectioned at 12 μm and stored at -70 °C. Sections were warmed to room temperature for 10 min and dipped in 75% ethanol, treated with 0.3% Triton X-100 and 0.1 m TBS (pH 7.4) containing 3% BSA with 0.5% Tween 20 for 10 min each. Different concentrations of curcumin were applied to sections for 1 h at 37 °C in a humidified chamber. Sections were washed in TBS three times, rinsed once in dH2O, and coverslipped with fluorescent mounting medium. Adjacent sections were stained with 1% thioflavin S for 10 min at room temperature and dipped for 1 min in 75% ethanol, 95% ethanol, 100% ethanol, 95% ethanol, and 75% ethanol before rinsing in dH2O. All sections were examined and photographed with a fluorescence microscope using fluorescein isothiocyanate optics. Groups of APPsw Tg2576 transgenic mice raised on Purina 5015 breeder chow were placed on 500 ppm curcumin (Sabinsa, Piscataway, NJ) or control safflower oil-based test diets (TD#02347 and TD#02346, respectively; Harlan Teklad) and aged until 22 months. A curcumin stock solution was diluted 1:100 into sterile PBS. Three 22-month-old Tg2576 mice (50 g body weight) were anesthetized with 25 mg/ml nembutal (intraperitoneally). One mouse on the chronic 500 ppm curcumin diet was injected with 200 μl of curcumin/PBS into the right carotid artery over a 5-min period. Because blood volume for a 50-g mouse was estimated to be about 4 ml (80 ml/kg × 0.05 kg), we expected blood levels of curcumin to reach ∼2 μm. The other two mice were injected with PBS only as a control, and one of these mice had also received the 500-ppm curcumin diet. After 1 h, the mice were perfused as described previously (15Lim G.P. Yang F. Chu T. Chen P. Beech W. Teter B. Tran T. Ubeda O. Hsiao Ashe K. Frautschy S.A. Cole G.M. J. Neurosci. 2000; 20: 5709-5714Crossref PubMed Google Scholar). Brains were removed and snap-frozen in 2-methylbutane chilled by liquid nitrogen. Freshly cut cryosections (12 μm) were air-dried for 10 min in the dark and coverslipped with VECTASHIELD mounting medium for fluorescence protection (Vector Laboratories, Burlingame, CA). Sections were examined and photographed with a Nikon Microphot-Fx fluorescence microscope using a fluorescein isothiocyanate filter set. Image Analysis of Plaque Pathology—Plaque burden was assessed using a characterized polyclonal antibody against Aβ 1–13 (DAE) (15Lim G.P. Yang F. Chu T. Chen P. Beech W. Teter B. Tran T. Ubeda O. Hsiao Ashe K. Frautschy S.A. Cole G.M. J. Neurosci. 2000; 20: 5709-5714Crossref PubMed Google Scholar). Coronal sections were made through anterior (bregma; -1.00 to -1.46 mm), middle (bregma; -1.58 to -2.30 mm), and posterior hippocampus (bregma; -2.46 to -3.16 mm) of control (n = 6) and curcumin-treated mice (n = 8). Immunolabeling was examined in various cortical and hippocampal areas of animals, and image analysis was performed as described previously (12Lim G.P. Chu T. Yang F. Beech W. Frautschy S.A. Cole G.M. J. Neurosci. 2001; 21: 8370-8377Crossref PubMed Google Scholar, 15Lim G.P. Yang F. Chu T. Chen P. Beech W. Teter B. Tran T. Ubeda O. Hsiao Ashe K. Frautschy S.A. Cole G.M. J. Neurosci. 2000; 20: 5709-5714Crossref PubMed Google Scholar). Amyloid Extracts and Assays—Amyloid levels were evaluated in the cortex of control (n = 6) and curcumin-fed (n = 4) Tg2576 mice. Samples were processed in TBS and lysis buffer (detergent) as described previously (16Calon F. Lim G.P. Yang F. Morihara T. Teter B. Ubeda O. Rostaing P. Triller A. Salem Jr., N. Ashe K.H. Frautschy S.A. Cole G.M. Neuron. 2004; 43: 633-645Abstract Full Text Full Text PDF PubMed Scopus (626) Google Scholar). Detergent-insoluble pellets from cortex were sonicated in 8 volumes of 5 m guanidine, 50 mm Tris-HCl and solubilized by agitation at room temperature for 3–4 h. Guanidine-soluble extracts were diluted 1:10,000 with TBS containing 5% BSA and 1× protease inhibitor mixture (Calbiochem) and assayed for total Aβ by ELISA as described previously (17Howlett D.R. Perry A.E. Godfrey F. Swatton J.E. Jennings K.H. Spitzfaden C. Wadsworth H. Wood S.J. Markwell R.E. Biochem. J. 1999; 340: 283-289Crossref PubMed Scopus (190) Google Scholar). Inhibition of Aggregation—Aβ40 was dissolved at 1 mg/ml in dH2O, and curcumin stock was diluted in 0.1 m TBS with 0.02% Tween 20. Aβ40 (100 μg/ml) was mixed a 1:1 volume ratio with curcumin in Eppendorf tubes and incubated for 6 days at 37 °C. Final concentrations of Aβ40 were 50 μg/ml (11.6 μm), and curcumin concentrations were 0, 0.125, 0.25, 0.5, 1, 2, 4, and 8 μm. Disaggregation of Preformed Fibrils—Aβ40 (100 μg/ml) was incubated for 3 days at 37 °C to generate fibrils. Preformed fibrils were mixed with the same concentrations of curcumin as above for an additional 3 days at 37 °C. At the end of both incubations, an aliquot (3 μl) was taken from each tube for electron microscopy (EM) analysis. The remaining Aβ solution was assayed for aggregates using the 6E10/6E10 sandwich ELISA. 6E10/6E10 Aggregation ELISA—To determine the relative amount of aggregated Aβ, the monoclonal antibody 6E10 was used for both detection and capture as described by Howlett et al. (17Howlett D.R. Perry A.E. Godfrey F. Swatton J.E. Jennings K.H. Spitzfaden C. Wadsworth H. Wood S.J. Markwell R.E. Biochem. J. 1999; 340: 283-289Crossref PubMed Scopus (190) Google Scholar). Briefly, 96-well plates were coated with 3 μg/ml 6E10 in 55 mm sodium bicarbonate buffer (pH 9) at 4 °C overnight before wells were blocked for 2 h at room temperature with TBS-Tween 20 (TBS-T; 0.1 m TBS, pH 7.4, 0.02% Tween 20) with 1% BSA. Samples were transferred to the wells, and plates were placed on an orbital shaker (30 rpm) for 2 h at room temperature. After washing four times with TBS-T, biotinylated 6E10 antibody diluted 1:1200 in TBS-T containing 1% BSA was applied to the plate for 1 h at room temperature with shaking. After washing, alkaline phosphatase streptavidin (Vector Laboratories, Burlingame, CA) diluted at 1:4000 in TBS-T with 1% BSA was applied to the plates, followed by the addition of the fluorescent substrate, Attophos (Promega). Fluorescence was measured using a CytoFluor II fluorescence plate reader (excitation 450 nm and emission 580 nm; Applied Biosystems, Foster City, CA). Aβ 1–40 was dissolved at 1 mg/ml in dH2O. A 50 μg/ml solution was made with 0, 0.5, 2, or 8 μm curcumin. Naproxen and ibuprofen were tested at 0, 8, 16, and 32 μm. Samples were sealed in Eppendorf tubes and incubated for 6 days at 37 °C. Aβ aggregates were assayed with the 6E10/6E10 ELISA as above. Separate drug-only controls were used to assess possible drug fluorescence interference in the assay. EM was used to observe inhibition of Aβ fibril formation in disaggregation and aggregation inhibition experiments (see Fig. 5). To determine whether filaments had formed, 3 μl of Aβ peptide solution was applied to 150-mesh copper grids coated with Formvar/carbon film (EM Sciences, Fort Washington, PA) for 30 s. Excess solution on the other side of grids was absorbed with filter paper, grids were stained with one drop of 0.5% filtered uranyl acetate for 20 s, and staining solution was absorbed with filter paper again. After air drying for 4 h or overnight, grids were examined with an electron microscope (TECNAI 10/12; Phillips) at 80 kV. Aβ40 and Aβ42 were solubilized in HFIP, dried overnight at room temperature and speed-vacuumed for 10 min (18Chromy B.A. Nowak R.J. Lambert M.P. Viola K.L. Chang L. Velasco P.T. Jones B.W. Fernandez S.J. Lacor P.N. Horowitz P. Finch C.E. Krafft G.A. Klein W.L. Biochemistry. 2003; 42: 12749-12760Crossref PubMed Scopus (479) Google Scholar). Aliquots were stored at -80 °C until they were needed, when they were resolubilized in 20 μl of Me2SO and dissolved in dH2O (1 mg/ml). Additional samples of Aβ40, Aβ42, and shorter Aβ peptides were dissolved in dH2O, whereas Aβ 17–24 was dissolved in Me2SO. Peptides were aggregated in TBS (0.05 m Tris-HCl, pH 7.4, 75 mm NaCl, 0.025% NaN3) at 500 μg/ml. Two hundred μl of each peptide solution was sealed in Eppendorf tubes and incubated without shaking at 37 °C for 10 days. After 10 days, an aliquot was removed for EM analysis of structure before solutions from each tube were mixed with curcumin at a final concentration of 1 μm. Samples were incubated at 37 °C for 1 h and then spun at 16,000 × g for 10 min. To assess curcumin staining of synthetic peptide aggregates, pellets were mixed with 10 μl of dH2O, smeared to slides, and air-dried for 1 h. Supernatant, spun down after mixing Aβ peptides with 1 μm curcumin, were applied to cryosections of Tg2576 mouse brain (22 months) at 37 °C for 1 h. Slides were coverslipped with anti-quenching mounting medium before examination under a fluorescent microscope (Table I).Table ICurcumin binds to peptide sequences that form fibrils or protofibrils Shown are a schematic diagram and table of curcumin binding experiments. Aβ was aggregated for 10 days at 37 °C before an aliquot was removed for EM analysis. Curcumin was incubated with samples before they were centrifuged. Pellets were smeared on glass slides to check for Aβ binding while supernatants were applied to Tg2576 mouse brain sections for plaque labeling. Weak staining was defined as + or ++, and strong staining was +++ or ++++. Fibrils/protofibrils were determined by estimating thickness and length of fibers. Open table in a new tab Immunoblot of Curcumin- and Congo Red-treated Aβ42 Oligomers—An aliquot of Aβ42 (0.045 mg) was dissolved in 20 μlofMe2SO and diluted in Ham's F-12 media without phenol red (BIOSOURCE, Camarillo, CA). Aβ42 (5 μm) was incubated with curcumin (0, 0.25, 1, 4, 16, and 64 μm) or Congo Red (16 and 64 μm) in a 37 °C water bath for 4 h. After the incubation, the samples (72 μl) were spun at 14,000 × g (4 °C for 10 min), and the supernatant (65 μl) was mixed with an equal part of Tricine sample buffer without reducing agents (Bio-Rad). The unaggregated Aβ42 control was not incubated at 37 °C, and after dilution in F-12 medium, it was mixed with sample buffer (no centrifuging) and stored at -20 °C before it was electrophoresed. Samples (30 μl, no boiling) were electrophoresed at 100 V on a 10–20% Tris-Tricine SDS gel, transferred at 100 V for 1 h, and blocked overnight (4 °C) with 10% nonfat milk in PBS plus 0.1% gelatin. The blots were probed with 6E10 (1:2000; 3 h at room temperature), followed by goat anti-mouse horseradish peroxidase (1:10,000; 1 h at room temperature), and developed with ECL. Dot Blot Assay—Aβ40 oligomer was prepared from HFIP-solubilized Aβ (4 mg/ml, 10–20 min in HFIP at 25 °C). An 80-μl Aβ aliquot was diluted 1:10 with 800 μl of sterile H2O. The final Aβ concentration was 400 μg/ml or 88 μm. Curcumin was added to Aβ solutions to give final 0, 2, and 16 μm curcumin in 0.01% methanol. After the pH was adjusted (pH 3), the samples were incubated for 2.5 h at 42 °C followed by a 48-h incubation at room temperature with stirring. Samples (500 ng of oligomer) were applied to nitrocellulose membrane in a Bio-Dot apparatus (Bio-Rad). The membrane was blocked with 10% nonfat milk in TBS-T at room temperature for 1 h, washed with TBS-T, and probed with anti-oligomer A11 antibody (19Kayed R. Head E. Thompson J.L. McIntire T.M. Milton S.C. Cotman C.W. Glabe C.G. Science. 2003; 300: 486-489Crossref PubMed Scopus (3452) Google Scholar) solution (1:10,000) or 6E10 (1: 10,000) in 3% BSA-TBS-T overnight at 4 °C. After washing, it was probed with anti-rabbit horseradish peroxidase- or anti-mouse horseradish peroxidase-conjugated antibody (Pierce) solution (1:12,000) for 1 h at room temperature. The blot was developed with SuperSignal (Pierce) for 2–5 min. Dots were scanned and analyzed with a model GS-700 densitometer using Molecular Analyst software (Bio-Rad). Toxicity Assays—Confirmation of Aβ oligomer toxicity was performed in human APPSwe neuroblastoma (N2a) cells, which were cultured in 50% Dulbecco's modified Eagle's medium, 50% Opti-MEM (Invitrogen), 5% fetal bovine serum, 200 μg/ml Glutamax. Cells were plated at equal densities (8000 cells/well) with 1.5% bovine calf serum and maintained at 37 °C in an atmosphere of 5% CO2. The medium was removed before treatment and replaced with medium containing 70% Dulbecco's modified Eagle's medium, 30% Opti-MEM with 0.1% BSA. An LDH assay was performed on media using the LDH assay toxicity kit (Promega, Madison, WI). To determine whether curcumin could block this oligomer toxicity, SH-SY5Y human neuroblastoma cells were grown and maintained as described previously (20Yang F. Sun X. Beech W. Teter B. Wu S. Sigel J. Vinters H.V. Frautschy S.A. Cole G.M. Am. J. Pathol. 1998; 152: 379-389PubMed Google Scholar). Cells were plated at 10,000 cells/well in 96-well plates and were differentiated in low serum Dulbecco's modified Eagle's medium with N2 supplement and 1 × 10-5m all-trans-retinoic acid for 7 days. The medium was removed and replaced with fresh maintenance medium containing 0.1% BSA. Aβ42 oligomer (100 nm) was added to cells for 48 h at 37 °C with or without 0.1, 1, 2.5, and 5 μm curcumin. After treatment, LDH assay was performed as mentioned above. Cell viability was assayed using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide reduction assay (21Mosmann T. J. Immunol. Methods. 1983; 65: 55-63Crossref PubMed Scopus (46517) Google Scholar), where absorbance was measured at 550 nm. Each data point was determined in triplicate, and the S.D. value did not exceed 5%. Because curcumin is a fluorochrome, we first compared plaque-associated curcumin fluorescence with that of thioflavin S, using sections from APPsw (Tg2576) mouse brain (Fig. 2, A, B, and E) and AD hippocampus (Fig. 2, C and D). Like thioflavin S (Fig. 2, A and C), curcumin brightly labeled amyloid plaques (1 μm; Fig. 2, B–E). In AD brain, curcumin labeled plaques with a yellow fluorescence (Fig. 2D) similar to thioflavin S in an adjacent section (Fig. 2C). Thioflavin S labeled tangles very strongly (Fig. 2C), whereas curcumin labeled tangles only weakly or not at all (Fig. 2D). Curcumin fluorescence was initially yellow-orange, but with prolonged (>2-min) light exposure, the fluorescence gradually shifted to yellow/green fluorescence. This phenomenon is illustrated in Fig. 2E, where the right side of the field was illuminated for 3 min before shifting the field back to make the photoexposure. Color change depended on the intensity of plaque staining and the fluorescent exposure time, and was stable for several weeks at 4 °C. Lipofuscin age pigments in this aged mouse brain (small arrows) are orange/red under these thioflavin optics and, unlike curcumin labeling, did not show a fluorescent shift with prolonged illumination. Optimal staining concentrations for curcumin were 0.5–2 μm. Unlike thioflavin S and Congo Red, curcumin is highly hydrophobic and should readily enter the brain to bind to plaques in vivo. In order to evaluate this idea, we injected curcumin (50 μm in 200 μl) or vehicle (PBS) into the carotid artery of aged Tg2576 mice. Mice were sacrificed 1 h later, and unfixed cryosections were immediately prepared from snap-frozen brains. Untreated mice injected with PBS only showed no plaque staining (Fig. 3A), but a mouse on a chronic curcumin diet (500 ppm) showed discernable but weak plaque staining when injected with PBS alone (Fig. 3B). Plaques in the curcumin-injected mouse were brightly labeled (Fig. 3, C and D). These data suggested that curcumin can cross the blood-brain barrier and bind to plaques in transgenic mice after oral feeding or peripheral injection. Having established that curcumin can stain in vitro and in vivo amyloid, we sought to test whether it can interfere with Aβ aggregation and its disaggregation. Because curcumin fluoresces over the same wavelengths as thioflavin S, we could not use thioflavin-based assays to quantify amyloid fibrils in curcumin-treated samples, and thioflavin-based assays require high concentrations of Aβ. Therefore, we employed a sandwich ELISA for aggregated Aβ using the same N-terminal Aβ antibody (6E10) for both capture and detection. Aβ40 (final concentration 50 μg/ml) was incubated under fibril-forming conditions with 0–8 μm curcumin. As shown in Fig. 4A, Aβ aggregation was significantly inhibited with increasing doses of curcumin, with an approximate IC50 of 0.81 μm (gray circles; p < 0.001). When Aβ was preaggregated for 3 days before incubation with curcumin, increasing doses of curcumin were capable of inducing the disaggregation of preaggregated Aβ40, with an IC50 of 1 μm (filled circles; p < 0.005; Fig. 4B). In control experiments, curcumin over this range did not interfere with the Aβ aggregate ELISA and did not interfere with an ELISA with 6E10 as the capture and anti-Aβ 34–40 as the detection (not shown), implying that curcumin did not simply block 6E10 antibody binding. We also sought to test whether curcumin was better than other NSAIDs at inhibiting Aβ aggregation. Aβ40 (final concentration 50 μg/ml) was incubated with curcumin (0–8 μm), naproxen (0–32 μm), or ibuprofen (0–32 μm) for 6 days at 37 °C. The presence of Aβ aggregates was determined with the 6E10/6E10 ELISA. In this assay, curcumin again showed dose-dependent inhibition of Aβ aggregate formation over the low dose range (Fig. 4C), in contrast to naproxen and ibuprofen (Fig. 4, D and E). These data indicate that although high concentrations of naproxen and ibuprofen can inhibit Aβ aggregation (20Yang F. Sun X. Beech W. Teter B. Wu S. Sigel J. Vinters H.V. Frautschy S.A. Cole G.M. Am. J. Pathol. 1998; 152: 379-389PubMed Google Scholar), curcumin is a better aggregation inhibitor. EM Analysis of Inhibition of Fibril Formation—To determine whether curcumin could inhibit the formation of fibrils, EM was used to examine samples of Aβ40 incubated with and without curcumin as described under "Experimental Procedures." Aβ40 (50 μg/ml) incubated for 6 days at 37 °C formed extensive, but less mature fibrils (Fig. 5A) than fibrils formed with a higher concentration of Aβ40 (100 μg/ml; Fig. 5D). These fibrils were extensive and more uniform in thickness than those at the lower concentration, with less branching, bumps, or nodules. In order to test whether low physiologically relevant concentrations of curcumin might inhibit fibril formation, we tested curcumin's impact on Aβ assembly at the lower Aβ40 (50 μg/ml) concentration. If a low dose of curcumin (0.125 μm) was included in the initial incubation, fibril formation appeared reduced (Fig. 5B). A higher dose (2 μm) resulted in similar but more potent inhibition of fibril formation (Fig. 5C). Droplets of curcumin/Tween accumulated with curcumin dose as fibril formation was inhibited. In order to determine whether curcumin could also be inhibitory when added after initial aggregation events, we incubated 100 μg/ml Aβ40 for 3 days at 37 °C and then incubated another 3 days at 37 °C without (Fig. 5D) or with 0.125 μm (Fig. 5E) and 2 μm curcumin (Fig. 5F). The results show that curcumin limits fibril formation even when added midway through the incubation, consistent with an impact on fibril maturation or dissolution. Curcumin Prevents Formation of Oligomers—Soluble oligomers of Aβ or "ADDLs" are a neurotoxic species implicated in AD pathogenesis (22Walsh D.M. Klyubin I. Fadeeva J.V. Cullen W.