Curcumin has potent anti-amyloidogenic effects for Alzheimer's β-amyloid fibrils in vitro

Authors

  • Kenjiro Ono,

    1. Department of Neurology and Neurobiology of Aging, Kanazawa University Graduate School of Medical Science, Kanazawa, Japan
    Search for more papers by this author
  • Kazuhiro Hasegawa,

    1. Department of Pathology, Fukui Medical University, Fukui, Japan
    2. CREST of Japan Science and Technology Corporation, Kawaguchi, Japan
    Search for more papers by this author
  • Hironobu Naiki,

    1. Department of Pathology, Fukui Medical University, Fukui, Japan
    2. CREST of Japan Science and Technology Corporation, Kawaguchi, Japan
    Search for more papers by this author
  • Masahito Yamada

    Corresponding author
    1. Department of Neurology and Neurobiology of Aging, Kanazawa University Graduate School of Medical Science, Kanazawa, Japan
    • Department of Neurology and Neurobiology of Aging, Kanazawa University Graduate School of Medical Science, Kanazawa 920-8640, Japan
    Search for more papers by this author

Abstract

Inhibition of the accumulation of amyloid β-peptide (Aβ) and the formation of β-amyloid fibrils (fAβ) from Aβ, as well as the destabilization of preformed fAβ in the central nervous system, would be attractive therapeutic targets for the treatment of Alzheimer's disease (AD). We reported previously that nordihydroguaiaretic acid (NDGA) and wine-related polyphenols inhibit fAβ formation from Aβ(1–40) and Aβ(1–42) and destabilize preformed fAβ(1–40) and fAβ(1–42) dose-dependently in vitro. Using fluorescence spectroscopic analysis with thioflavin T and electron microscopic studies, we examined the effects of curcumin (Cur) and rosmarinic acid (RA) on the formation, extension, and destabilization of fAβ(1–40) and fAβ(1–42) at pH 7.5 at 37°C in vitro. We next compared the anti-amyloidogenic activities of Cur and RA with NDGA. Cur and RA dose-dependently inhibited fAβ formation from Aβ(1–40) and Aβ(1–42), as well as their extension. In addition, they dose-dependently destabilized preformed fAβs. The overall activities of Cur, RA, and NDGA were similar. The effective concentrations (EC50) of Cur, RA, and NDGA for the formation, extension, and destabilization of fAβs were in the order of 0.1–1 μM. Although the mechanism by which Cur and RA inhibit fAβ formation from Aβ and destabilize preformed fAβ in vitro remains unclear, they could be a key molecule for the development of therapeutics for AD. © 2004 Wiley-Liss, Inc.

Alzheimer's disease (AD) is pathologically characterized by the abundance of extracellular deposition of amyloid β-peptide (Aβ) as amyloid plaques and vascular amyloid, and the intraneuronal accumulation of neurofibrillary tangles (Selkoe, 2001). AD therapy is currently not essential. There are few drugs available, and most compounds for AD are acetylcholine esterase inhibitors, which aim at stabilizing acetylcholine levels in the synaptic cleft to maintain neurotransmission (Cacabelos et al., 2000). Many researchers favor other therapeutic approaches that target formation, deposition, and clearance of Aβ from nervous tissue. Vaccination (Schenk et al., 1999; Janus et al., 2000; Morgan et al., 2000) and secretase inhibitors (Skovronsky et al., 2000) have been reported as experimental therapies and for clinical trials.

Based on a nucleation-dependent polymerization model to explain the mechanism of Alzheimer's β-amyloid fibril (fAβ) formation in vitro (Jarrett and Lansbury, 1993; Lomakin et al., 1997; Naiki et al., 1997; Naiki and Gejyo, 1999), we reported previously that nordihydroguaiaretic acid (NDGA) and rifampicin (RIF) dose-dependently inhibit fAβ formation from Aβ and fAβ extension in vitro (Naiki et al., 1998). In addition, we reported that NDGA also destabilizes fAβ(1–40) and fAβ(1–42) in a concentration-dependent manner within a few hours at pH 7.5 at 37°C, based on fluorescence spectroscopic analysis with thioflavin T (ThT) and electron microscopic studies (Ono et al., 2002b). Very recently, we also showed that wine-related polyphenols, such as myricetin, dose-dependently inhibit formation and extension of fAβ and destabilize preformed fAβs in vitro (Ono et al., 2003).

Many studies have demonstrated that oxidative damage is closely associated with the hallmark pathologies of AD and may play a critical role in the development of AD (Pratico and Delanty, 2000; Rottkamp et al., 2000; Smith et al., 2000; Varadarajan et al., 2000). Oxidative stress in AD may result from aging, energy deficiency, inflammation, or excessive production of Aβ (Grundman and Delaney, 2002). Many antioxidant compounds, such as vitamin E (DL-α-tocopherol) (Behl et al., 1992; Zhou et al., 1996; Subramaniam et al., 1998; Pereira et al., 1999; Yatin et al., 1999), vitamin A (Jama et al., 1996; Perrig et al., 1997), quercetin (Roth et al., 1999), and nicotine (Kihara et al., 1997) have been suggested to reduce oxidative stress associated with AD. One phenolic antioxidant alternative is curcumin (Cur), a major component of the yellow curry spice turmeric. This spice is used in the traditional diet and as an herbal medicine in India (Kelloff et al., 1996, 2000). The frequency of AD in India is roughly one-quarter of that in the US (e.g., 0.7 vs. 3.1% in patients between 70 and 79 yr; Ganguli et al., 2000). Cur is much stronger than vitamin E as a free radical scavenger (Zhao et al., 1989), protects the brain from lipid peroxidation (Martin-Aragon et al., 1997), and scavenges nitric oxide (NO)-based radicals (Sreejayan and Rao, 1997). Oral administration of Cur has been shown to be centrally neuroprotective (Kaul and Krishnakantha, 1997; Rajakrishnan et al., 1999). Recently, Lim et al. (2001) reported that Cur reduced oxidative damage and amyloid pathology in an Alzheimer transgenic amyloid precursor protein with Swedish mutant (APPSw) mouse model (Tg2576). Although Lim et al. (2001) suggested that Cur blocks AD pathogenesis at multiple sites in the inflammation cascade, the direct effects of Cur on the formation and destabilization of fAβ remain unclear.

Using fluorescence spectroscopy with ThT and electron microscopy, we examined the effects of Cur and its analog, rosmarinic acid (RA), on the formation and extension of fAβ(1–40) and fAβ(1–42), as well as their activity to destabilize fAβs at pH 7.5 at 37°C in vitro.

MATERIALS AND METHODS

Preparation of Aβ and fAβ Solutions

Aβ(1–40) (trifluoroacetate salt, 520130; Peptide Institute, Osaka, Japan) and Aβ(1–42) (trifluoroacetate salt, 520625; Peptide Institute) were dissolved by brief vortexing in 0.02% ammonia solution at a concentration of 500 μM (2.2 mg/mL) and 250 μM, respectively, in a 4°C room and then stored at −80°C before assaying (fresh Aβ[1–40] and Aβ[1–42] solutions). fAβ(1–40) and fAβ(1–42) were formed from the fresh Aβ(1–40) and Aβ(1–42) solutions, respectively, sonicated, and then stored at 4°C as described elsewhere (Hasegawa et al., 1999).

Fresh, nonaggregated fAβ(1–40) and fAβ(1–42) was obtained by extending sonicated fAβ(1–40) or fAβ(1–42) with fresh Aβ(1–40) or Aβ(1–42) solutions, respectively, just before the destabilization reaction (Ono et al., 2002a, b). The reaction mixture volume was 600 μL and contained 10 μg/mL (2.3 μM) fAβ(1–40) or fAβ(1–42), 50 μM Aβ(1–40) or Aβ(1–42), 50 mM phosphate buffer, pH 7.5, and 100 mM NaCl. Measurement of ThT fluorescence showed that the extension reaction proceeded to equilibrium after incubation at 37°C for 3–6 hr under nonagitated conditions. In the following experiment, the concentration of fAβ(1–40) and fAβ(1–42) in the final reaction mixture was regarded as 50 μM.

Fluorescence Spectroscopy, Electron Microscopy, and Polarized Light Microscopy

A fluorescence spectroscopic study was carried out as described elsewhere (Naiki and Nakakuki, 1996) on a Hitachi F-2500 fluorescence spectrophotometer. Optimum fluorescence measurements of fAβ(1–40) and fAβ(1–42) were obtained at the excitation and emission wavelengths of 445 and 490 nm, respectively, with the reaction mixture containing 5 μM ThT (Wako Pure Chemical Industries, Osaka, Japan) and 50 mM of glycine-NaOH buffer, pH 8.5. An electron microscopic study and polarized light microscopic study of the reaction mixtures were carried out as described elsewhere (Hasegawa et al., 1999).

Polymerization Assay

Polymerization of Aβ with or without fAβ added as seeds was assayed as described elsewhere (Naiki et al., 1998). Briefly, the reaction mixture contained 50 μM Aβ(1–40), or 25 or 50 μM Aβ(1–42), 0 or 10 μg/mL fAβ(1–40), or fAβ(1–42), 0–50 μM Cur, RA, or NDGA, 1% dimethyl sulfoxide (DMSO), 50 mM phosphate buffer, pH 7.5, and 100 mM NaCl. Cur, RA, and NDGA (Sigma, St. Louis, MO) were first dissolved in DMSO at concentrations of 1 μM, 10 μM, 100 μM, 1 mM, and 5 mM, then added to the reaction mixture to make final concentrations of 0.01, 0.1, 1, 10, and 50 μM, respectively.

Aliquots (30 μL) of the mixture were put into oil-free PCR tubes (size, 0.5 mL, code number 9046; Takara Shuzo, Otsu, Japan) and the tubes were then put into a DNA thermal cycler (PJ480; Perkin Elmer Cetus, Emeryville, CA). Starting at 4°C, the plate temperature was elevated at maximal speed to 37°C. Incubation times ranged between 0–8 days (as indicated in each figure), and the reaction was stopped by placing the tubes on ice. The tubes were not agitated during the reaction. Aliquots (5 μL) from each tube in triplicate were subjected to fluorescence spectroscopy and the mean of the three measurements determined. In the ThT solution, the concentration of Cur, RA, and NDGA examined in this study was diluted up to 1/200 of that in the reaction mixture. We confirmed that these compounds did not quench ThT fluorescence at the diluted concentration (data not shown).

Measurement of Fibril Destabilizing Activity

Destabilization of fAβ was assayed as described elsewhere (Ono et al., 2002b). Briefly, the reaction mixture contained 25 μM fresh fAβ(1–40) or fAβ(1–42), 0–50 μM Cur, RA, or NDGA, 1% DMSO, 50 mM phosphate buffer, pH 7.5, 100 mM NaCl, and 1% (wt/vol) polyvinyl alcohol (Wako) to avoid fAβ aggregation and adsorption of fAβ onto the inner wall of the tube during the reaction. Cur, RA, and NDGA were dissolved in DMSO at concentrations of 1 μM, 10 μM, 100 μM, 1 mM, and 5 mM, and then added to the reaction mixture to make final concentrations 0.01, 0.1, 1, 10, and 50 μM, respectively.

After being mixed by pipetting, triplicate 5-μL aliquots of the reaction mixture were subjected to fluorescence spectroscopy and 30-μL aliquots were put into PCR tubes. Reaction tubes were then transferred into a DNA thermal cycler. Starting at 4°C, the plate temperature was elevated at maximal speed to 37°C. Incubation times ranged from 0–6 hr (as indicated in each figure), and the reaction was stopped by placing the tubes on ice. The reaction tubes were not agitated during the reaction. Aliquots (5 μL) from each tube in triplicate were subjected to fluorescence spectroscopy and the mean of the three measurements was determined.

Other Analytical Procedures

Protein concentrations of the reaction mixture supernatants after centrifugation were determined by the method of Bradford (1976) with a protein assay kit (Bio-Rad, Hercules, CA). The Aβ(1–40) solution quantified by amino acid analysis was used as the standard. Linear least squares fit was used for statistical analysis. The effective concentrations (EC50) were defined as the concentrations of NDGA, Cur, or RA required to inhibit the formation or extension of fAβs to 50% of the control value, or the concentrations to destabilize fAβs to 50% of the control value. EC50 were calculated by the sigmoidal curve fitting of the data using Igor Pro v.4 (WaveMetrics, Lake Oswego, OR).

RESULTS

Effects of Cur and RA on Kinetics of fAβ Formation

As shown in Figure 1A–D, when fresh Aβ(1–40) or Aβ(1–42) was incubated at 37°C, the fluorescence of ThT followed a characteristic sigmoidal curve. This curve is consistent with the nucleation-dependent polymerization model (Jarrett and Lansbury, 1993; Naiki et al., 1997). fAβ(1–40) and fAβ(1–42) stained with Congo red showed typical orange-green birefringence under polarized light (data not shown). The final equilibrium level decreased after incubation of Aβ(1–40) or Aβ(1–42) with 10 and 50 μM Cur or RA (Fig. 1A–D).

Figure 1.

Effects of Cur (A,B) and RA (C,D) on the kinetics of formation of fAβ(1–40) (A,C) and fAβ(1–42) (B,D) from fresh Aβ(1–40) and Aβ(1–42), respectively. Reaction mixtures containing 50 μM Aβ(1–40) (A,C) or 25 μM Aβ(1–42) (B,D), 50 mM phosphate buffer, pH 7.5, 100 mM NaCl, and 0 (filled circles), 10 (open circles), or 50 μM (open squares) of Cur (A,B) or RA (C,D) were incubated at 37°C for indicated times. Each figure representative of three independent experiments. E: Dose-dependent inhibition of fAβ(1–40) formation from fresh Aβ(1–40). Reaction mixtures containing 50 μM Aβ(1–40), 50 mM phosphate buffer, pH 7.5, 100 mM NaCl, and 0, 0.01, 0.1, 1, 10, and 50 μM NDGA (filled circles), Cur (open circles), or RA (filled squares) were incubated at 37°C for 7 days. Points represent means of three independent experiments. At all points, standard errors were within symbols. The average without compounds was regarded as 100%.

As shown in Figure 2A–D, when fresh Aβ(1–40) or Aβ(1–42) was incubated with fAβ(1–40) or fAβ(1–42), respectively, at 37°C, the fluorescence increased hyperbolically without a lag phase and proceeded to equilibrium much more rapidly than without seeds (compare Fig. 1 and 2). This curve is consistent with a first-order kinetic model (Naiki and Nakakuki, 1996). When Aβ(1–40) and fAβ(1–40) were incubated with 10 or 50 μM Cur or RA, the final equilibrium level decreased (Fig. 2A,C). A similar effect of Cur and RA was observed for the extension of fAβ(1–42) (Fig. 2B,D). At a constant fAβ(1–40) concentration, a perfect linearity was observed between Aβ(1–40) concentration and the initial rate of fAβ(1–40) extension in both the presence and absence of Cur (Fig. 2E). This linearity is again consistent with a first-order kinetic model and indicates that at each Aβ(1–40) concentration, the net rate of fAβ(1–40) extension is the sum of the rates of polymerization and depolymerization (Naiki and Nakakuki, 1996, Hasegawa et al., 2002). In the presence of 10 μM Cur, the slope of the straight line decreased to about 1/2. The interpretation of this figure, implicating the mechanism of the anti-amyloidogenic effect of Cur, will be discussed later.

Figure 2.

Effects of Cur (A,B) and RA (C,D) on kinetics of extension of fAβ(1–40) (A,C) and fAβ(1–42) (B,D). Reaction mixtures containing 10 μg/mL (2.3 μM) sonicated fAβ(1–40) (A,C) or fAβ(1–42) (B,D), 50 μM Aβ(1–40) (A,C) or Aβ(1–42) (B,D), 50 mM phosphate buffer, pH 7.5, 100 mM NaCl, and 0 (filled circles), 10 (open circles), or 50 μM (open squares) of Cur (A,B), or RA (C,D), were incubated at 37°C for the indicated times. Each figure representative of three independent experiments. E: Effect of Aβ(1–40) concentration on the initial rate of fAβ(1–40) extension in the presence (open circles) and absence (filled circles) of Cur. Reaction mixtures containing 10 μg/mL (2.3 μM) sonicated fAβ(1–40), 50 mM phosphate buffer, pH 7.5, 100 mM NaCl, 0 (filled circles) or 10 μM (open circles) Cur, and 0, 10, 20, 30, 40, and 50 μM Aβ(1–40), were incubated at 37°C for 1 hr. Points represent means of three independent experiments. At all points, standard errors were within symbols. Linear least-square fit was carried out for each straight line (R2 = 1.000 and 0.991 for filled circles and open circles, respectively). F: Dose-dependent inhibition of fAβ(1–40) extension. The reaction mixtures containing 10 μg/mL (2.3 μM) sonicated fAβ(1–40), 50 μM Aβ(1–40), 50 mM phosphate buffer, pH 7.5, 100 mM NaCl, and 0, 0.01, 0.1, 1, 10, and 50 μM NDGA (filled circles), Cur (open circles), or RA (filled squares) were incubated at 37°C for 1 hr. Points represent means of three independent experiments. At all points, standard errors were within symbols. The average without compounds was regarded as 100%.

After incubation of fresh Aβ(1–40) with fAβ(1–40) at 37°C, clear fibril extension was observed by electron microscopy (Fig. 3B); however, 50 μM Cur completely inhibited extension of sonicated fAβ(1–40) (Fig. 3A,C). Cur inhibited the extension of fAβ(1–42) (data not shown). Similarly, RA also inhibited the extension of fAβ(1–40) and fAβ(1–42) (data not shown).

Figure 3.

Electron micrographs of extended fAβ(1–40). Reaction mixtures containing 10 μg/mL (2.3 μM) fAβ(1–40), 50 μM Aβ(1–40), 50 mM phosphate buffer, pH 7.5, 100 mM NaCl, and 0 (B) or 50 μM Cur (A,C), were incubated at 37°C for 0 (A), or 6 hr (B,C). Scale bars = 250 nm.

Fibril Destabilizing Assay

As shown in Figure 4A–D, the fluorescence of ThT was almost unchanged during the incubation of fresh fAβ(1–40) or fAβ(1–42) at 37°C without additional molecules. On the other hand, the ThT fluorescence decreased immediately after addition of Cur and RA to the reaction mixture. After incubation of 25 μM fresh fAβ(1–40) with 50 μM Cur for 1 hr, many short, sheared fibrils were observed (Fig. 5B). At 4 hr, the number of fibrils was reduced markedly, and small amorphous aggregates were occasionally observed (Fig. 5C). Similar morphology was observed after incubation of 25 μM fresh fAβ(1–42) with 50 μM Cur (data not shown). RA also destabilized preformed fAβ(1–40) and fAβ(1–42) similarly (data not shown).

Figure 4.

Effects of Cur (A,B) and RA (C,D) on kinetics of destabilization of fAβ(1–40) (A,C) and fAβ(1–42) (B,D). Reaction mixtures containing 25 μM fAβ(1–40) (A,C) or fAβ(1–42) (B,D), 50 mM phosphate buffer, pH 7.5, 100 mM NaCl, and 0 (filled circles), 10 (open circles), or 50 μM (open squares) of Cur (A,B), or RA (C,D), were incubated at 37°C for the indicated times. Each figure representative of three independent experiments. E: Dose-dependent destabilization of fAβ(1–40). Reaction mixtures containing 25 μM fAβ(1–40), 50 mM phosphate buffer, pH 7.5, 100 mM NaCl, and 0, 0.01, 0.1, 1, 10, and 50 μM NDGA (filled circles), Cur (open circles), or RA (filled squares) were incubated at 37°C for 4 hr. Points represent means of three independent experiments. At all points, standard errors were within symbols. The average without compounds was regarded as 100%.

Figure 5.

Electron micrographs of destabilized fAβ(1–40). The reaction mixture containing 25 μM fAβ(1–40), 50 mM phosphate buffer, pH 7.5, 100 mM NaCl, and 50 μM Cur was incubated at 37°C for 0 (A), 1 (B) or 4 hr (C). Scale bars = of 250 nm.

After incubation with 50 μM Cur or RA for 4 hr, fAβ(1–40) and fAβ(1–42) were not stained with Congo red as much as fresh fAβ(1–40) and fAβ(1–42) were (data not shown). They all showed orange-green birefringence under polarized light (data not shown), however, indicating that a significant amount of intact fAβ(1–40) and fAβ(1–42) remained in the mixture after the reaction. When the protein concentration of the supernatant after centrifugation (at 16,000 × g for 2 hr at 4°C) was measured by the Bradford assay, no proteins were detected in the supernatant in any case (data not shown). This implied that although Cur and RA could destabilize fAβ(1–40) and fAβ(1–42) to visible aggregates (Fig. 5C), they could not depolymerize fAβ(1–40) and fAβ(1–42) to monomers or oligomers of Aβ(1–40) and Aβ(1–42).

Comparison of NDGA, Cur, and RA Activity

NDGA, Cur and RA dose-dependently inhibited fAβ formation and extension and destabilized preformed fAβs. We calculated EC50, the concentrations of NDGA, Cur, or RA to inhibit the formation or extension of fAβs to 50% of the control value or to destabilize fAβs to 50% of the control value, by the sigmoidal curve fitting of the data as shown in Figures 1E, 2F, and 4E (Table I). In all molecules examined, EC50 values required to inhibit fAβ formation or extension were similar to those required to destabilize fAβs. All data presented in Table I suggest that NDGA, Cur, and RA have similar anti-amyloidogenic activity.

Table I. EC50 of NDGA, Cur, and RA for Formation, Extension, and Destabilization of fAβ(1–40) and fAβ(1–42)*
CompoundsFormation (μM)aExtension (μM)bDestabilization (μM)c
fAβ(1–40)fAβ(1–42)fAβ(1–40)fAβ(1–42)fAβ(1–40)fAβ(1–42)
  • *

    EC50 defined as the concentrations of nordihydroguaiaretic acid (NDGA), curcumin (Cur), or rosmarinic acid (RA) to inhibit the formation or extension of fAβs to 50% of the control value or the concentrations to destabilize fAβs to 50% of the control value. EC50 was calculated by sigmoidal curve fitting of the data.

  • a

    Reaction mixtures containing 50 μM Aβ(1–40) or 25 μM Aβ(1–42), 50 mM phosphate buffer, pH 7.5, 100 mM NaCl, and 0, 0.01, 0.1, 1, 10, and 50 μM NDGA, Cur, or RA were incubated at 37°C for 7 days and 24 hr, respectively.

  • b

    Reaction mixtures containing 10 μg/mL (2.3 μM) sonicated fAβ(1–40) or fAβ(1–42), 50 μM Aβ(1–40) or Aβ(1–42), 50 mM phosphate buffer, pH 7.5, 100 mM NaCl, and 0, 0.01, 0.1, 1, 10, and 50 μM NDGA, Cur, or RA were incubated at 37°C for 1 hr.

  • c

    Reaction mixtures containing 25 μM fAβ(1–40) or fAβ(1–42), 50 mM phosphate buffer, pH 7.5, 100 mM NaCl, and 0, 0.01, 0.1, 1, 10, and 50 μM NDGA, Cur, or RA were incubated at 37°C for 4 hr.

NDGA0.160.740.140.091.000.67
Cur0.190.630.190.520.420.32
RA0.291.100.260.810.830.60

DISCUSSION

Recently, our systematic in vitro study indicated that the overall activity of the anti-amyloidogenic molecules may be in the order of: NDGA = wine-related polyphenols (myricetin, morin, quercetin) >> RIF = > poly(vinylsulfonic acid, sodium salt) = 1,3-propanedisulfonic acid, disodium salt > iAβ5 > nicotine (Ono et al., 2002a, b, 2003). The Indian spice Cur and its analog RA dose-dependently inhibit fAβ formation from fresh Aβ, and destabilize preformed fAβ in vitro. NDGA, Cur, and RA examined in this study had similar anti-amyloidogenic activity (Table I). All of these molecules have two 3,4-dihydroxyphenyl rings (NDGA, RA) or 3,4-methoxyhydroxyphenyl rings (Cur) symmetrically bound by a short carbohydrate chain (Fig. 6). This compact and symmetric structure might be suitable for specifically binding to free Aβ and subsequently inhibiting polymerization of Aβ into fAβ. Alternatively, this structure might be suitable for specific binding to fAβ and subsequent destabilization of the β-sheet-rich conformation of Aβ molecules in fAβ. Tomiyama et al (1994, 1996) suggested that RIF binds to Aβ by hydrophobic interactions between its lipophilic ansa chain and the hydrophobic region of Aβ, thus blocking associations between Aβ molecules leading to fAβ formation. The anti-amyloidogenic activity of tetracyclines (TCs), small-molecule anionic sulfonates or sulfates, melatonin, iAβ5, and nicotine may also be related to the propensity to bind to the specific sites of Aβ (Kisilevsky et al., 1995; Soto et al., 1996; Pappolla et al., 1998; Forloni et al., 2001; Zeng et al., 2001).

Figure 6.

Structure of NDGA, Cur, and RA.

Cur, RA, and NDGA did not extend the length of the lag phase in the formation fAβs from Aβs (Fig. 1; Naiki et al., 1998). In addition, they did not extend the time to proceed to equilibrium in the extension reaction (Fig. 2; Naiki et al., 1998). These results are in sharp contrast to those of apolipoprotein E (apoE), in which apoE extends in a dose-dependent manner the length of lag phase and the time to proceed to equilibrium (Naiki et al., 1998). Although apoE was suggested to inhibit the formation of fAβs in vitro by making a complex with Aβs, thus eliminating free Aβs from the reaction mixture (Naiki et al., 1997, 1998), Cur, RA, and NDGA may inhibit the formation of fAβs by different mechanisms. As shown in Figure 2E, the extension of fAβ(1–40) followed a first-order kinetic model even in the presence of Cur. The net rate of fAβ(1–40) extension is the sum of the rates of polymerization and depolymerization (Naiki and Nakakuki, 1996, Hasegawa et al., 2002). One possible explanation for the finding in Figure 2E thus may be that Cur could bind to the ends of extending fAβ(1–40) and increase the rate of depolymerization by destabilizing the conformation of Aβ(1–40) that has just been incorporated into the fibril ends. Alternatively, Cur would bind to Aβ(1–40) and consequently decrease the rate of polymerization. Further studies are essential to clarify the mechanisms by which Cur and RA inhibit fAβ formation in vitro.

Cur is a potent antioxidant and an effective antiinflammatory compound (Zhao et al., 1989; Sreejayan and Rao, 1997; Xu et al., 1998). Part of its nonsteroidal antiinflammatory drug-like activity is based on the inhibition of nuclear factor κB (NFκB)-mediated transcription of inflammatory cytokines (Xu et al., 1998), inducible nitric oxide synthase (iNOS) (Chan et al., 1998), and cyclooxygenase 2 (Cox-2) (Zhang et al., 1999). Because of antitumor activity, relative safety, and its long history of use, Cur is currently being developed for clinical use as a cancer chemopreventive agent in India (Kelloff et al., 1996, 2000). RA is an ester of caffeic acid and 3,4-dihydroxyphenyllactic acid (Petersen and Simmonds, 2003). It is commonly found in species of the Boraginaceae and the subfamily Nepetoideae of the Lamiaceae (Petersen and Simmonds, 2003), and has several interesting biological activities, e.g., antioxidant, antiinflammatory, antimutagen, antibacterial, and antiviral (Parnham and Kesselring, 1985). The latter activity is used in the therapy of Herpes simplex virus infections with RA-containing extracts of Melissa officinalis. It is thought that the antiinflammatory properties are based on the inhibition of lipoxygenases and cyclooxygenases, as well as the interference with the complement cascade (Parnham and Kesselring, 1985). RA is eliminated rapidly from the blood circulation after intravenous administration and shows very low toxicity in mice (Parnham and Kesselring, 1985). NDGA is a pure compound isolated from the creosote bush, Larrea tridentata (Luo et al., 1998). It is a potent oxygen radical scavenger and a lipoxygenase inhibitor (Goodman et al., 1994), and has been capable of lowering plasma glucose concentration in mouse models of non-insulin-dependent diabetes mellitus (Luo et al., 1998). Our study and several other reports suggest that Cur, RA, and NDGA could be key molecules for development of therapeutics for AD. First, Kim et al. (2001) reported that Cur protects PC12 and human umbilical vein endothelial cells from Aβ insult by strong antioxidant properties. Goodman et al. (1994) also reported that NDGA can interrupt the cytotoxicity of Aβ to cultured rat hippocampal neurons by suppressing Aβ-induced accumulation of reactive oxygen species and intracellular free Ca2+. Second, Lim et al. (2001) reported that Cur suppresses indices of inflammation and oxidative damage in the brain of Tg2576 APPSw transgenic mouse and that low, nontoxic doses of Cur decrease levels of insoluble and soluble Aβ and plaque burden in many affected brain regions. They speculated that mechanisms are based mainly on inflammation-related targets, such as inhibition of NFκB-induced iNOS, Cox-2, and inflammatory cytokine production. Finally, we showed that Cur, RA, and NDGA dose-dependently inhibit fAβ formation from fresh Aβ and destabilize preformed fAβ in vitro. In addition, cell culture experiments with human embryonic kidney (HEK) 293 cells indicated that fAβ destabilized by NDGA might be less toxic than are intact fAβ (Ono et al., 2002b). It thus may be reasonable to speculate that Cur, RA, and NDGA could prevent the development of AD not only through scavenging reactive oxygen species but also through directly inhibiting fAβ deposition in the brain. Recently, Caughey et al. (2003) showed that Cur potently inhibits protease-resistant prion protein accumulation in scrapie agent-infected neuroblastoma cells. Although the exact mechanism of anti-amyloidogenic activity of Cur, RA, and NDGA is unclear, these structurally similar compounds could be key molecules for the development of therapeutics for AD and other human conformational diseases.

Acknowledgements

This work was supported in part by the Ministry of Education, Culture, Sports, Science and Technology, Japan (Grant-in-Aid for Scientific Research to M.Y. and a Grant-in-Aid for Scientific Research on Priority Areas [C] -Advanced Brain Science Project to H.N.) and by the Ministry of Health, Labour, and Welfare, Japan (grant to M.Y.). We thank Drs. S. Okino and K. Iwasa (Kanazawa University) for cooperation in the experiments, and Drs. H. Okada and N. Takimoto (Fukui Medical University) for excellent technical assistance.

Ancillary