Identification of naturally occurring spirostenols preventing β-amyloid-induced neurotoxicity

Steroids 69 (2004) 1–16

Identification of naturally occurring spirostenols preventing ␤-amyloid-induced neurotoxicity Laurent Lecanu a,b , Wenguo Yao a,b , Gary L. Teper a,b , Zhi-Xing Yao a,b , Janet Greeson c , Vassilios Papadopoulos a,b,d,e,∗ a

Department of Cell Biology, Division of Hormone Research, Georgetown University School of Medicine, Washington, DC 20057, USA b Samaritan Research Laboratories, Georgetown University School of Medicine, Washington, DC 20057, USA c Samaritan Pharmaceuticals, Las Vegas, NV 89109, USA d Department of Pharmacology, Georgetown University School of Medicine, Washington, DC 20057, USA e Department of Neurosciences, Georgetown University School of Medicine, Washington, DC 20057, USA Received 19 December 2002; received in revised form 3 July 2003; accepted 4 September 2003

Abstract 22R-Hydroxycholesterol is an intermediate in the steroid biosynthesis pathway shown to exhibit a neuroprotective property against ␤-amyloid (1–42) (A␤) toxicity in rat PCl2 and human NT2N neuronal cells by binding and inactivating A␤. In search of potent

22R-hydroxycholesterol derivatives, we assessed the ability of a series of naturally occurring entities containing the 22R-hydroxycholesterol structure to protect PC12 cells against A␤-induced neurotoxicity, determined by measuring changes in membrane potential, mitochondrial diaphorase activity, ATP levels and trypan blue uptake. 22R-Hydroxycholesterol derivatives sharing a common spirost-5-en-3-ol or a furost-5-en-3-ol structure were tested. Although some of these compounds were neuroprotective against 0.1 ␮M A␤, only three protected against the 1–10 ␮M A␤-induced toxicity and, in contrast to 22R-hydroxycholesterol, all were devoid of steroidogenic activity. These entities shared a common structural feature, a long chain ester in position 3 and common stereochemistry. The neuroprotective property of these compounds was coupled to their ability to displace radiolabeled 22R-hydroxycholesterol from A␤, suggesting that the A␤-22R-hydroxycholesterol physicochemical interaction contributes to their beneficial effect. In addition, a 22R-hydroxycholesterol derivative inhibited the formation of neurotoxic amyloid-derived diffusible ligands. Computational docking simulations of 22R-hydroxycholesterol and its derivatives on A␤ identified two binding sites. Chemical entities, as 22R-hydroxycholesterol, seem to bind preferentially only to one site. In contrast, the presence of the ester chain seems to confer the ability to bind to both sites on A␤, leading to neuroprotection against high concentrations of A␤. In conclusion, these results suggest that spirost-5-en-3-ol naturally occurring derivatives of 22R-hydroxycholesterol might offer a new approach for Alzheimer’s disease therapy. © 2003 Elsevier Inc. All rights reserved. Keywords: Neurodegeneration; Neuroprotection; Alzheimer’s disease; Steroids; Cholesterol; Spirostenol; Amyloid-derived diffusible ligand (ADDL)

1. Introduction Excessive accumulation in the brain of the ␤-amyloid peptide (A␤), due either to overproduction and/or decreased clearance and the formation of senile plaques, and neurofibrillary tangles due to hyperphosphorylation of the Tau protein, are the hallmarks of Alzheimer disease (AD) pathology [1]. These biochemical modifications have been shown to be associated with a slowly evolving cognitive deficit and memory impairment [2–4]. Recent advances in the understanding of the neuropathology of the AD have permitted the identification of potential ∗

Corresponding author. Tel.: +1-202-687-8991; fax: +1-202-687-7855. E-mail address: [email protected] (V. Papadopoulos).

0039-128X/$ – see front matter © 2003 Elsevier Inc. All rights reserved. doi:10.1016/j.steroids.2003.09.007

targets and the development of new therapeutic approaches. As circulating sexual steroid concentrations have been shown to decrease over time in life and to be lower in aging and Alzheimer’s disease patients [5], one such promising strategy is the utilization of steroids. Among the various steroids examined in vitro for their putative neuroprotective properties against A␤, estrogens showed the most promising results [6–10]. In addition, ovariectomy was shown to impair spatial memory while estradiol replacement therapy allowed the recovery of memory functions in rhesus monkeys [11]. In agreement with these findings, disruption of the estrogen receptor beta gene induced learning disabilities in mice [12]. In addition to the classical estrogens, phytoestrogens, plant-derived compounds displaying estrogenic activity, were also shown to exert protective activity against A␤

2

L. Lecanu et al. / Steroids 69 (2004) 1–16

toxicity in neurons in vitro [13] and improve memory in rats in vivo [14,15]. Moreover, the estrogen precursors DHEA and DHEAS were shown to improve memory in young and ageing mice [16,17]. In humans, endogenous estradiol and testosterone levels have been shown to correlate to cognitive performances in older women [18] and high dose estradiol therapy was shown to improve cognition in AD women [19,20]. Despite these findings, the use of estrogen as a therapeutic means in AD has been controversial because of reports of absence of clear long-term effects [21–26] and the potential side effects of long-term estrogen use [27,28]. We recently reported that the steroid 22R-hydroxycholesterol, an intermediate in the cytochrome P450 C27 side chain cleavage enzymatic pathway of pregnenolone formation from cholesterol [29], is present at lower levels in AD hippocampus and frontal cortex tissue specimens compared to age-matched controls [30]. 22R-hydroxycholesterol was then found to protect, in a dose-dependent manner, against A␤-induced rat sympathetic nerve pheochromocytoma (PC12) and differentiated human NT2N neuronal cell death. The effect of 22R-hydroxycholesterol was stereospecific because its enantiomer 22S-hydroxycholesterol failed to protect the neurons from A␤-induced cell death. Moreover the protective effect of this steroid was specific for A␤-induced cell death because it did not protect against glutamate-induced neurotoxicity. In search of the mechanism of action of 22R-hydroxycholesterol, we found that it binds to A␤ and that the formed 22R-hydroxycholesterol/A␤ complex is not toxic to rodent and human neurons [30]. Because 22R-hydroxycholesterol is a precursor of pregnenolone, and thus of the final steroid products made, it would be rapidly metabolized in steroid forming tissues by the cytochrome P450 C27 side chain cleavage (P450scc); thus, we used 22R-hydroxycholesterol as the lead structure and searched for naturally occurring compounds containing this structure and exhibiting neuroprotective properties against A␤ toxicity. We report herein the identification of biologically active 22R-hydroxycholesterol derivatives containing a common spirost-5-en-3-ol structure that might offer a new approach for AD therapy.

Russia). Cells culture supplies were purchased form GIBCO (Grand Island, NY) and cell culture plasticware was from Corning (Corning, NY) and Packard BioSciences Co. (Meriden, CT). 2.2. In silico screening for 22R-hydroxycholesterol derivatives The Interbioscreen Database of naturally occurring entities was screened for compounds containing the 22R-hydroxycholesterol structure using the ISIS software (Information Systems Inc., San Leandro, CA). The structure of the selected and tested 22R-hydroxycholesterol (SP222) and derivatives (SP223–238) are shown in Fig. 1 and the denomination, chemical name and origin for each of these compounds is shown in Table 1. 2.3. Cell culture and treatments PC12 cells (rat pheochromocytoma neurons) from ATCC (Manassas, VA) were cultured at 37 ◦ C and 5% CO2 in RPMI 1640 medium devoid of glutamine and supplemented with 10% fetal bovine serum and 5% horse serum [31]. Cells were seeded in 96-well plates (8 × 104 cells per well). After an overnight period of incubation, increasing concentrations of freshly solubilized A␤ (0.1, 1 and 10 ␮M) were added to the cells in the presence or absence of the indicated concentrations of the SP compounds to be tested. After 72-h incubation time, various parameters markers of cell viability were determined. Mouse MA-10 tumor Leydig cells were maintained at 37 ◦ C in DMEM/Ham’s F12 (Biofluids, Rockville, MD) medium supplemented with 5% heat-inactivated fetal calf serum and 2.5% horse serum in 5% CO2 . Cells were plated on 96-well plates at the density of 2.5×104 cells per well for overnight. The cells were stimulated with the indicated concentrations of the SP compounds in 0.2 ml per well serumfree medium for 2 h. Culture media were collected and tested for progesterone production by radioimmunoassay. 2.4. MTT cytotoxicity assay

2. Experimental 2.1. Materials A␤1–42 peptide was purchased from American Peptide Co. (Sunnyvale, CA) and stored at −80 ◦ C. Just before use, A␤1–42 was reconstituted with fresh distilled water to reach a concentration of 500 ␮M and then added directly to the cells at the desired concentration. 22R-Hydroxycholesterol (SP222) was purchased from Sigma (St. Louis, MO). [22-3 H]R-hydroxycholesterol (specific activity 20 Ci/mmol) was synthesized by American Radiolabeled Chemical (St. Louis, MO). The 22R-hydroxycholesterol derivatives (SP223–238) were purchased from Interbioscreen (Moscow,

The cellular toxicity of A␤ was assessed using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay (Trevigen, Gaithersburg, MD). Briefly, 10 ␮l of the MTT solution were added to the cells cultured in 100 ␮l medium. After an incubation period of 4 h, 100 ␮l of detergent were added and cells were incubated overnight at 37 ◦ C. Formazan blue formation was quantified at 600 and 690 nm using the Victor quantitative detection spectrophotometer (EGG-Wallac, Gaithersburg, MD) and the results expressed as (DO600 − DO690 ). Although the MTT assay has been widely used to assess cytotoxicity in neuronal cells treated with A␤, it has been suggested that the results obtain in the presence of various steroids might reflect the A␤-dependent vesicle recycling leading to increased MTT

O

N

O

O

H H

SP223 O

O

H H

SP222 O

O

O

O

O

N

N

SP225 O

SP224

O

O

O

O O S

O

O

N

O N

N

O

O

O

O

O

S O

O

O

L. Lecanu et al. / Steroids 69 (2004) 1–16

SP227

SP226

SP229

SP228

O

O

O

O

O

O

O O O H

O

O

O

O

O O

SP232

O

SP233

O H

H

SP230

O

O

O

SP231

O N

O

O

O

O N

O

O

O

O O

O O

O

O

O

O O SP234 O

O

SP235 O

SP237

SP236 O

O

SP238

O O

O

Fig. 1. Chemical structure of 22R-hydroxycholesterol (SP222) and naturally occurring derivatives. The chemical name, denomination and origin of the compounds shown are given in Table 1. 3

4

L. Lecanu et al. / Steroids 69 (2004) 1–16

Table 1 Chemical name, denomination and origin of naturally occurring compounds containing the 22R-hydroxycholesterol (SP222) lead structure Denomination

Chemical name

Origin

SP222 SP223 SP224 SP225 SP226 SP227 SP228 SP229

22R-Hydroxycholesterol (20␰)-26-Acetylamino-(22R)-hydroxyfurost-5-en-3␰-yl acetate (20␣)-25␰-Methyl-(22R,26)-azacyclofurost-5-en-3␰-ol (20␰)-26-Acetylamino-(22R)-methoxyfurost-5-en-3␣-yl acetate (20␰)-25␰-Methyl-N-acetyl-(22R,26)-azacyclofurost-5-en-3␰-ol (22S,25␰)-(20␣)-Spirost-5-en-(2␣,3␰)-diol (20␰)-26-Acetylamino-(22␰)-ethoxyfurost-5-en-3␰-yl acetate (20␣)-25␰-Methyl-N-paratoluenesulfonyl-(22S,26)-azacyclofurost-5-en-3␰-yl paratoluenesulfonate (22R,25␰)-(20␣)-(14␣,20␣)-Spirost-5-en-(3␤,12␤)-diol (22R,25S)-(20␰)-Spirost-5-en-3␰-ol (22R,25␰)-(20␣)-Spirost-5-en-3␤-yl benzoate (22S,25S)-(20S)-Spirost-5-en-3␤-yl hexanoate (22R,25␰)-(20␣)-Spirost-5-en-(1␰,3␰)-diol (22R,25S)-(20␣)-Spirost-5-en-3␤-ol (22R,25S)-(20␣)-Spirost-5-en-3␤-yl succinate 26-Diacetylamino-(22␰)-acetoxy-(16␰)-acetoxy-cholest-5-en-yl acetate (20␣)-25S-Methyl-N-acetyl-(22S,26)-azacyclofurost-5-en-3␤-yl propanoate

Mammalian Gynura sp. (asteraceae) Solanum asperum (solanaceae) Gynura sp. (asteraceae) Solanum asperum (solanaceae) Gynura japonica (asteraceae) Gynura sp. (asteraceae) Solanum aviculare (solanaceae)

SP230 SP231 SP232 SP233 SP234 SP235 SP236 SP237 SP238

G. japonica (asteraceae) G. japonica (asteraceae) Gynura sp. (asteraceae) Gynura sp. (asteraceae) G. japonica (asteraceae) G. japonica (asteraceae) Gynura sp. (asteraceae) Achlya heterosexualis (saprolegniaceae) Solanum asperum (solanaceae)

formazan exocytosis and loss [6]. For that reason, additional cytotoxicity and cell viability assays were used.

on a TopCount NXTTM counter (Packard BioSciences Co.) following the recommendations of the manufacturer.

2.5. Trypan blue cell viability measurement

2.8. Radioimmunoassay

Cell viability was measured using the trypan blue exclusion method as we previously described [31]. In brief, cells were treated for 72 h with SP compounds in the presence or absence of increasing concentrations of A␤. At the end of the incubation, cells were washed three times with PBS and incubated for 15 min with 0.1% trypan blue stain solution at room temperature. After washing three times with PBS, 0.1N NaOH was added to the cells and trypan blue was quantified using the Victor quantitative detection spectrophotometer at 450 nm.

Progesterone production by MA-10 cells was measured by radioimmunoassay using anti-progesterone antisera (ICN, Costa Mesa, CA), following the conditions recommended by the manufacturer. The progesterone production was normalized to the amount of protein in each well. Radioimmunoassay data was analyzed using the MultiCalc software (EG&G Wallac, Gaithersburg, MD).

2.6. Measurement of membrane potential

Purified A␤ (50 ␮M) and 3 H-22R-hydroxycholesterol (SP222) were incubated either alone or in the presence of 100 ␮M of unlabeled 22R-hydroxycholesterol (SP222) or the various 22R-hydroxycholesterol (SP222) derivatives in 20 ␮l volume for 8 or 24 h at 37 ◦ C. At the end of the incubation time, samples were separated by 1.5% agarose (Type I-B) gel electrophoresis under native conditions and transferred to nitrocellulose membrane (Schleicher&Schuell, Keene, NH) in 10× SSC buffer. The membrane was exposed to tritium-sensitive screen and analyzed by phosphoimaging using the Cyclone Storage phosphor system (Packard BioScience). Image-densitometric analysis was performed using the OptiQuant software (Packard BioScience). This method allows for the separation, visualization and identification of A␤ complexes, which have incorporated radiolabeled cholesterol [32] and 22R-hydroxycholesterol [30] or 22R-hydroxycholesterol derivatives under native conditions. Low molecular weight unincorporated 22R- hydroxycholes-

Cells viability was also assessed using the luminescencebased kit CytoLiteTM (Packard BioScience Co.) according to the recommendations of the manufacturer. Briefly, cells were cultured and treated in 96-well plates and after 72-h incubation time, 25 ␮l of Activator solution was added to the cells followed by 150 ␮l of Amplifier solution. Luminescence was measured on a TopCount NXTTM counter (Packard BioSciences Co.) following a 5 min precount delay. 2.7. Determination of cellular ATP levels Cellular ATP concentrations were measured using the ATPLite-MTM luminescence assay (Packard BioSciences Co.). For this assay, cells were cultured on black 96-well ViewPlateTM and the ATP concentrations were measured

2.9. 22R-hydroxycholesterol-protein binding blot assay (CPBBA)

L. Lecanu et al. / Steroids 69 (2004) 1–16

terol and derivatives are separated and eliminated during electrophoresis. 2.10. Western blot analysis of Aβ polymerization and amyloid-derived diffusible ligand (ADDL) formation Increasing concentration of A␤ (0.1, 1 and 10 ␮M) were incubated in PC12 culture medium for 24 and 72 h at 37 ◦ C under 5% CO2 with or without increasing concentrations of SP233 (1, 10 and 100 ␮M). At the end of the incubation time, samples were separated by 4–20% Tris-Glycine gel electrophoresis (Invitrogen, Carlsbad, CA) under native conditions at 125 V for 2 h and electrotrasferred to nitrocellulose membrane (HybondTM ECLTM , Amersham Pharmacia Biotech, Piscataway, NJ) at 130 A for 30 min. Non-specific adsorption of the antibodies was blocked by incubating the nitrocellulose in 5% milk. The blots were treated for immunodetection of A␤ species using a polyclonal antibody to A␤ that recognizes a 30 amino acid peptide of the A␤ protein (Zymed Laboratories, San Francisco, CA). Membranes were incubated in primary antibody for 1 h at room temperature at a dilution of 1:2000. Then, membranes were incubated in the secondary antibody at a dilution of 1:1000 for 1.5 h at room temperature. The blots were visualized using the ECLTM Western Blotting Analysis System (Amersham Biosciences). Image-densitometric analysis was performed using the OptiQuant software (Packard BioScience, Meriden, CT). This method allows for the separation, visualization and identification of Aß complexes, polymers and ADDLs. 2.11. Peptide modeling and docking simulations Computer docking of 22R-hydroxycholesterol (SP222) and 16 of its derivatives with A␤1–42 was accomplished using an A␤ structure initialized by the solution structure of A␤1–40 Met(O) (MMDB Id: 7993 PDB Id:1BA) resulting from data generated by CD and NMR spectroscopy [33]. The Met(O) SME 35 residue was replaced by Met retaining the adjacent backbone dihedral angles and the I41 and A42 residues appended. The energy of the structure was then minimized using the Alchemy 2000 program (Tripos, St. Louis, MO). The 22R-hydroxycholesterol derivative structures were also generated using Alchemy 2000. Molecular docking was accomplished using Monte Carlo simulated annealing as previously described [34] implemented in modified versions of Autogrid/Autodock [35]. For each of the compounds/A␤ pairs, approximately 108 conformations were evaluated to obtain the selected one of minimum energy. Three sessions consisting of 100 runs, each starting at a random initial relative location and orientation of the ligand with respect to the target were executed. Each run was comprised of 100 annealing cycles using about 2 × 104 improvement steps. The average computation time for each ligand/target pair was about 21/2 h using a 1.7 GHz, 1 GB RAM PC.

5

2.12. Statistical analysis Statistical analysis was performed by one-way analysis of variance (ANOVA) and unpaired Student’s t-test using the INSTAT 3.00 (GraphPad, San Diego, CA).

3. Results Three days exposure of PC12 cells to increasing concentrations of A␤ resulted in dose-dependent cell death (Fig. 2), reaching a maximum of 50% of the cells, in agreement with our previous data [30,31]. To stay close to the concentrations of A␤ present in AD brain, 0.1–10 ␮M concentrations of A␤ were used. The compounds tested for their neuroprotective properties were examined at 30 and 50 ␮M concentrations (Figs. 3–6, Table 2). Table 2 shows the effect of the lead compound 22R-hydroxycholesterol (SP222) and the compounds containing the 22R-hydroxycholesterol structure (SP223–238) on 0.1, 1.0 and 10.0 ␮M A␤-induced neurotoxicity determined using the MTT assay, a measurement of the NADPH diaphorase activity. The effect is expressed as a percentage of inhibition of the NADPH diaphorase activity. The 100% inhibition level, taken as a control, corresponds to the decrease of the blue formazan formation induced by A␤ administered alone. SP222 protects PC12 cells against A␤ 0.1 ␮M but provides a limited neuroprotection against A␤ given at 1 and 10 ␮M. It should be noted that a big variability was observed for the effect of 22R-hydroxycholesterol (SP222) on high concentrations of A␤, depending on the passage of the cells used. SP228, SP229, SP233, SP235, SP236, SP237 and SP238 displayed neuroprotective activity against A␤ 0.1 ␮M, but only SP233 and SP235 exerted a significantly more robust effect than 22R-hydroxycholesterol (SP222) (Table 2). SP233, SP236 and SP238 maintained

Fig. 2. Effect of increasing concentrations of A␤1–42 on rat PC12 neuronal cell viability. PC12 cells were treated for 72 h with the indicated concentrations of A␤. Levels of cell viability were measured using the MTT assay as described in Section 2. Results shown are means ± S.D. (n = 6–12). ∗∗ P < 0.01; ∗∗∗ P < 0.001.

6

L. Lecanu et al. / Steroids 69 (2004) 1–16

Fig. 3. Effect of increasing concentrations of 22R-hydroxycholesterol (SP222) and derivatives on rat PC12 neuronal cell membrane potential in the absence or presence of increasing concentration of A␤1–42 . (A) Effect of increasing concentrations of A␤ on PC12 membrane potential. PC12 cells were treated for 72 h with 0.1 ␮M (B), 1.0 ␮M (C) and 10.0 ␮M (D) A␤ in the absence or presence of increasing concentrations of the indicated compounds. Membrane potential was assessed using the luminescence-based kit CytoLiteTM as described in Section 2. The chemical name, denomination and origin of the compounds shown are given in Table 1. Results shown are means ± S.D. (n = 6–12). ∗ P < 0.05; ∗∗ P < 0.01; ∗∗∗ P < 0.001.

their neuroprotective properties against 1 ␮M A␤-induced toxicity, but only SP233 (50 ␮M) and SP238 (30 ␮M) kept this property in the presence of 10 ␮M A␤ (Table 2). Table 3 shows the intrinsic effect of the SP222–SP238 compounds, applied without A␤, on the basal NADPH diaphorase activity measured using the MTT assay. The MTT assay revealed that among the SP compounds exerting a protective activity against A␤, SP229 and SP238 display a neurotoxic activity. Results obtained with the MTT assay were confirmed using the membrane potential-assessing Cytolite assay for 22R-hydroxycholesterol (SP222), SP233, SP235, SP236 and SP238. Fig. 3A shows that A␤ exposure induces a dose-related decrease of the membrane potential-assessing luminescence. Although 22R-hydroxycholesterol (SP222) protected against 0.1 ␮M A␤ (Fig. 3B), it failed to do so against the two highest concentrations of A␤ (Fig. 3C and D). The various SP compounds used displayed a significantly better neuroprotective effect compared to 22R-hydroxycholesterol (SP222) as shown by the increase

in measured luminescence. The neuroprotective effect of SP233 and SP238 against 10 ␮M A␤ seen using the MTT assay (Table 2) was replicated by the raise of the signal under the same conditions (Fig. 3D). ATP levels, an index of mitochondrial function, were measured in PC12 cells treated with increasing concentrations of A␤ in the presence or absence of the SP222–SP238 compounds (Fig. 4). A␤ decreased in a dose-dependent manner ATP production by PC12 cells; 18, 22 and 25% decrease in ATP levels measured in the presence of 0.1, 1 and 10 ␮M A␤, respectively (P < 0.001 by ANOVA; Fig. 4A). From the compounds tested only SP233 and SP236 were able to reverse the 0.1 and 1 ␮M A␤-induced decrease in ATP levels (Fig. 4B and C). No beneficial effect of the SP compounds on ATP synthesis was seen in the presence of 10 ␮M A␤ (Fig. 4D). Trypan blue uptake by the cells was the fourth test used to assess the impact of the promising SP233 compound on A␤-induced toxicity (Fig. 5A). As expected, 0.1, 1 and

L. Lecanu et al. / Steroids 69 (2004) 1–16

7

Fig. 4. Effect of increasing concentrations of 22R-hydroxycholesterol (SP222) and derivatives on rat PC12 ATP levels in the absence or presence of increasing concentration of A␤1–42 . (A) Effect of increasing concentrations of A␤ on PC12 ATP levels. PC12 cells were treated for 72 h with 0.1 ␮M (B), 1.0 ␮M (C) and 10.0 ␮M (D) A␤ in the absence or presence of increasing concentrations of the indicated compounds. ATP levels were determined as described in Section 2. The chemical name, denomination and origin of the compounds shown are given in Table 1. Results shown are means ± S.D. (n = 6–12). ∗ P < 0.05; ∗∗ P < 0.01; ∗∗∗ P < 0.001.

Fig. 5. Effect of SP233 on A␤1–42 -induced toxicity on rat PC12 neuronal cell. (A) PC12 cells were treated for 72 h with increasing concentrations of A␤1–42 in the presence or absence of either 30 or 50 ␮M SP233. Levels of viability were measured using the trypan blue assay as described in Section 2. (B) PC12 cells were treated for 72 h with increasing concentrations of A␤1–42 in the presence or absence of the indicated concentrations of SP233. Levels of viability were measured using the MTT assay as described in Section 2. A␤1–42 0.1 ␮M (filled circle), 1 ␮M (open circle), 10 ␮M (filled triangle). Results shown are means ± S.D. (n = 6–12).

8

L. Lecanu et al. / Steroids 69 (2004) 1–16

Table 2 PC12 cells were treated for 72 h with 0.1, 1.0, and 10.0 ␮M A␤ in the absence or presence of increasing concentrations of the indicated compounds Compounds (␮M)

A␤ (1–42) ␮M

Control SP222 (30) SP222 (50) SP223 (30) SP223 (50) SP224 (30) SP224 (50) SP225 (30) SP225 (50) SP226 (30) SP226 (50) SP227 (30) SP227 (50) SP228 (30) SP228 (50) SP229 (30) SP229 (50) SP230 (30) SP230 (30)

100 68 98.1 104.8 93.2 130.5 151.4 294.3 335.6 183.4 169.3 109.2 113.7 106.1 91.4 83.5 98.1 101.0 154.1

0.1 ␮M ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

1 ␮M 6.7 9.3∗∗∗ 14.8 2.8 7.5 2.1 2.6 23.9 9.3 12.0 11.9 4.4 12.7 2.8 2.3∗∗ 4.7∗∗,++ 4.5 8.3 7.0

100 96.4 112 109 100.0 108.1 117.6 181.4 198.6 132.0 124.8 103.1 107.6 107.8 97.5 91.6 128 95.6 125.5

10 ␮M ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

8.9 4.5 9.9 1.9 3 1.2 0.8 7.3 7.8 6.6 8.2 3.2 8.3 0.3 3.1 4.6∗ 0.4 8.3 2.4

100 109.5 131 111.1 104 105.3 103.7 151 161.4 120.3 124.0 108.8 112.6 107.1 101.2 98.5 121.1 101.7 116.2

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

5.1 4.5 5.7 1.4 3 3.3 3.9 6.9 3.8 4.0 7.7 2.0 3.8 0.7 0.5 15.9 5.4 3.0 3.3

Compounds (␮M)

A␤ (1–42) ␮M

SP231 SP231 SP232 SP232 SP233 SP233 SP234 SP234 SP235 SP235 SP236 SP236 SP237 SP237 SP238 SP238

101.9 97.1 165.9 162.9 75.6 31.6 214.0 371.6 76.0 48.3 82.5 71.8 83.6 369.2 82.9 179.7

(30) (50) (30) (50) (30) (50) (30) (50) (30) (50) (30) (50) (30) (50) (30) (50)

0.1 ␮M

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

1 ␮M

11.2 2.7 10.1 8.0 1.2∗∗∗ 4.4 ∗∗∗,++ 3.2 42.5 2.2∗∗∗ 8.3∗∗∗,++ 1.9 ∗∗∗,++ 3.1∗∗∗,++ 12.8∗,++ 44.1 2.6∗∗∗,++ 3.5

97.9 91.6 108.1 106.6 89.0 68.2 127.0 180.6 102.1 103.8 90.0 90.5 100.1 311.5 73.6 155.8

10 ␮M

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

1.6 0.3 4.8 2.7 3.6 ∗∗,++ 3.4+++,∗∗∗ 9.2 14.3 5.7 8.6 1.0∗,+ 1.9∗,++ 4.8 1.0 1.0∗∗∗,++ 2.3

105.9 103.0 98.4 97.7 97.0 85.1 112.8 129.8 103.0 123.8 90.1 106.2 119.5 282.4 72.0 155.3

± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±

3.0 3.5 2.4 1.3 4.6 4.1∗∗∗,+++ 10.3 7.9 6.0 3.7 20.4 5.5 2.9 2.1 1.8∗∗∗,+++ 1.4

Levels of cell viability were measured using the MTT assay as described in Section 2. The chemical name, denomination and origin of the compounds shown are given in Table 1. The 100% inhibition level corresponds to the decrease of the blue formazan formation induced by A␤ administered alone. Results shown are means ± S.D. (n = 6–12). ∗ P < 0.05. ∗∗ P < 0.01. ∗∗∗ P < 0.001 compared to control and + P < 0.05. ++ P < 0.01. +++ P < 0.001 compared to SP222-treated cells.

10 ␮M A␤-induced a dose-dependent (33, 36 and 97%, respectively; P < 0.001 by ANOVA) increase in trypan blue uptake by PC12 cells. SP233 at 30 and 50 ␮M inhibited the A␤-induced cell death (P < 0.001 by ANOVA). Fig. 5B shows that the neuroprotective effect of SP233 is dose-dependent and it is maintained in the presence of all three concentrations of A␤, although its efficacy decreases in presence of high, supra-physiopathological, A␤ concentrations. One of the reasons in identifying 22R-hydroxycholesterol (SP222) derivatives is the need of biologically active (neuroprotective) compounds that cannot be metabolized by P450scc to pregnenolone and then to tissue-specific final steroid products. To assess the metabolism of SP233 by steroidogenic cells we examined their ability to form steroids

in MA-10 mouse tumor Leydig cells, a well-characterized steroidogenic cell model where 22R-hydroxycholesterol is an excellent P450scc substrate and can produce large amounts of steroids [34]. Fig. 6 shows that, in contrast to 22R-hydroxycholesterol (SP222), SP233 could not be metabolized to final steroid products. Considering our previous study on the mechanism underlying the neuroprotective action of 22R-hydroxycholesterol (SP222), where a direct interaction between 22R-hydroxycholesterol (SP222) and A␤ was shown using the CPBBA method, we undertook similar experiments to investigate whether the 22R-hydroxycholesterol (SP222) derivatives bind to A␤. The direct interaction of these compounds to A␤ was shown in displacement studies performed against the radiolabeled 22R-hydroxycholesterol (SP222)/A␤

Table 3 Effect of the SP compounds on the PC12 cells basal NADPH diaphorase activity measured using the MTT assay

30 ␮M 50 ␮M

30 ␮M 50 ␮M

30 ␮M 50 ␮M

SP222

SP223

SP224

SP225

SP226

109.3 ± 4.9 91.7 ± 4.0

93.1 ± 2.1 97.1 ± 8.3

47.0 ± 0.6 32.9 ± 3.0

84.0 ± 1.5 72.7 ± 0.9

38.3 ± 1.5 38.0 ± 1.1

SP228

SP229

SP230

SP231

SP232

SP233

79.8 ± 1.6 69.0 ± 2.6

97.1 ± 1.7 52.3 ± 2.6

68.0 ± 1.3 68.0 ± 0.2

49.0 ± 2.5 50.1 ± 0.8

106.0 ± 0.8 108.0 ± 0.7

SP234

SP235

SP236

SP237

SP238

102.8 ± 4.0 95.8 ± 2.0

93.6 ± 2.1 95.9 ± 1.6

94.6 ± 0.5 86.3 ± 2.1

81.1 ± 3.0 5.37 ± 0.46

62.0 ± 0.4 18.3 ± 2.0

89.9 ± 0.6 97.0 ± 0.2

SP227 88.5 ± 2.0 72.5 ± 8.4

The data revealed that among the SP compounds exerting a protective activity against A␤, SP229 and SP238 display a neurotoxic activity. Results shown are means ± S.D. (n = 6–12).

L. Lecanu et al. / Steroids 69 (2004) 1–16

Progesterone ng/mg protein

750

500

SP222 SP233 SP235 SP236 SP238

250

0

0

10

20

30

Fig. 6. Effect of SP233 on MA-10 Leydig cell steroid formation. MA-10 cells were treated for 2 h in the presence or absence of either 22R-hydroxycholesterol (SP222) or SP233 tested at the indicated concentrations. Progesterone levels were determined by radioimmunoassay. Results shown are means ± S.D. (n = 9).

complex (Fig. 7). Co-incubation of radiolabeled 22Rhydroxycholesterol (SP222) together with A␤ for 24 h at 37 ◦ C demonstrated the presence of a high molecular weight radiolabeled band (Fig. 7) recognized by an antibody specific to A␤ (Yao et al., 2002 and data not shown). The specificity of radiolabeling of A␤ by 22R-hydroxycholesterol (SP222) was demonstrated by competition studies using unlabeled 22R-hydroxycholesterol (SP222) (Fig. 7), where 100 ␮M 22R-hydroxycholesterol (SP222) displaced by 80%

9

radiolabeled SP222 compound bound to A␤. From the SP compounds tested, SP237, SP238, SP226, SP227 and SP233 displaced radiolabeled 22R-hydroxycholesterol (SP222) binding to A␤ by 46, 44, 65, 38 and 35%, respectively (Fig. 7). These data were further confirmed using computational docking simulations with A␤. The docking results showed that A␤1–42 forms a pocket in the 19–36 amino acids area (Fig. 8) where 22R-hydroxycholesterol (SP222) binds, in agreement with our previous data [30]. The docking energy for the various compounds tested placed in order of minimal energy required for binding to A␤ was: (−10.34 kcal/ mol) SP229 < SP232 < SP224 < SP237 < SP222 < SP233 < SP228 < SP223 < SP230 < SP234 < SP225 < SP238 < SP236 < SP226 < SP235 < SP231 < SP227 (−8.35 kcal/ mol). Fig. 9 compares the binding characteristics of 22Rhydroxycholesterol (SP222) with SP233. This is an analysis of 100 docking runs with each of the compounds. The data shows that about 23% of the time SP233 docks with energy of −7.0 to −7.5 kcal/mol, while 22R-hydroxycholesterol (SP222) docks about 25% of the time with only −5.5 to −6.0 kcal/mol. The probability of SP233 having a stronger (more negative) docking energy is significantly greater than that for 22R-hydroxycholesterol (SP222). Almost 100% of the time SP233 binds with less than −6.0 kcal/mol while the equivalent number for 22R-hydroxycholesterol (SP222) is only about −4.0 kcal/mol. Analysis of the distribution of

Fig. 7. Identification of the Aß-SP binding and binding site by CPBBA. Top, A␤1–42 peptide (50 ␮M) was incubated with 0.1 ␮Ci 3 H-22R-hydroxycholesterol (SP222) in the presence or absence of 100 ␮M unlabeled SP compounds in a 20 ␮l volume for 24 h at 37 ◦ C. CPBBA was performed as described in Section 2. Bottom, Image and densitometric analysis of the phosphoimage performed as described in Section 2. Results shown are representative of two independent experiments.

10 L. Lecanu et al. / Steroids 69 (2004) 1–16 Fig. 8. Computational SP compound docking simulations to A␤1–42 . Simulations were performed as indicated in Section 2. The chemical name and denomination of the compounds shown are given in Table 1. The different SP compounds are represented in pink color and the A␤ molecule in gray.

L. Lecanu et al. / Steroids 69 (2004) 1–16 Binding Energy Frequencies

(A) 30 Frequency - %

25 20 SP222

15

SP233

10 5

-4

-5

-6

-7

-8

-9

-1 0

0

Energy, KCal/mol

Binding Energy Probabilities

(B)

Probability

100.00% 80.00% 60.00%

SP222

40.00%

SP233

20.00%

-4

-5

-6

-7

-8

-9

-1 0

.00%

Energy, Kcal/mol

Fig. 9. Computational 22R-hydroxycholesterol (SP222) and SP233 docking simulations to A␤1–42 . The binding energy frequencies (top) and probabilities (bottom) for 22R-hydroxycholesterol (SP222) and SP233 binding to A␤ are shown. Data represent the analysis of 100 docking runs with each of the compounds run as described in Section 2.

the binding energy frequencies indicates a bimodal profile (Fig. 9A) suggesting the presence of two binding sites in A␤. For SP233 peaks might be present at both −7 to −7.5 and −8 to −8.5 kcal/mol, whereas with 22R-hydroxycholesterol (SP222) the peaks seem to be at −5.5 to 6.0 and −4.0 to −4.5 kcal/mol. Fig. 10 shows the immunoblot analysis performed after a 24 h (A) and 72 h (F) incubation period. In our experimental conditions no monomer species were detectable at any time with A␤ at 0.1 ␮M and no trimer and no tetramer (ADDLs) were detectable at 24 and 72 h with A␤ at 0.1 and 1 ␮M. SP233 decreases in a dose-dependent manner the amount of the monomeric species present at 24 h (Fig. 10A and B) and 72 h (Fig. 10F and G) with the 1 and 10 ␮M concentrations of A␤. This dose–effect relationship is also observed against the trimeric and tetrameric ADDLs at 24 h (Fig. 10A, C and D) and 72 h (Fig. 10G, H and I) when the SP233 is co-incubated with 10 ␮M A␤. Conversely, SP233 induces a dose-dependent increase of the polymeric species amount (Fig. 10E and J) suggesting that SP233 binds the A␤ and inhibit the formation of the neurotoxic ADDLs by forming stable heavy complexes with the peptide.

4. Discussion We recently reported that 22R-hydroxycholesterol (SP222), a steroid present in human brain, declines in hip-

11

pocampus and frontal cortex of AD brains compared to age-matched control specimens [30]. This finding prompted the search of a function of this steroid in brain, leading in the finding that 22R-hydroxycholesterol (SP222) protects rat PC12 and human differentiated NT2N cells against A␤ toxicity [30]. 22R-hydroxycholesterol (SP222) is formed as an intermediate in the cholesterol side chain cleavage reaction catalyzed by the P450scc enzyme responsible for pregnenolone formation [36]. In addition, 22Rhydroxycholesterol (SP222) is hydrosoluble, able to cross cell membranes and reach the inner mitochondrial membrane P450scc where it serves as its substrate [36]. Because of these characteristics, one should expect that the in vivo use of this steroid might lead to its rapid metabolism in steroidogenic tissues and increased steroid formation by the gonads, adrenal and brain. Although the fate and bioactivity of 22R-hydroxycholesterol (SP222) in vivo is not known, we considered this steroid as a lead structure for the identification of derivatives with neuroprotective properties but devoid of steroidogenic activity. This led us to assay different naturally occurring plant or mold 22R-hydroxycholesterol (SP222) derivatives and to assess their ability to counteract the toxicity of A␤1–42 on the well-established rat PC12 neuron cell model. The data obtained identifies a natural steroid, 22R-hydroxycholesterol (SP222) derivative, present in Gynura japonica (asteraceae), with the ability to protect neurons against the A␤1–42 peptide, considered to be causal of the onset and/or progression of AD pathology. We initially used the MTT assay, a widely used marker of cell viability and thus cytotoxicity. Using this assay, some of the compounds tested, namely SP229, SP233, SP236 and SP238, exhibited neuroprotective activity even when PC12 cells were exposed to concentrations as high as 1 and 10 ␮M A␤. Interestingly, these compounds were more efficacious compared to the reference 22R-hydroxycholesterol (SP222) molecule but the SP229 and SP238 displayed toxicity when applied on cells without A␤. Because it was recently reported that some steroid hormones block the MTT formazan exocytosis, an event that might lead to the non-specific overestimation of an eventual protective effect [6], the protective effect of the four compounds of interest was further examined using an assay that measures membrane potential (Cytolite). This luminescence-based membrane potential assay confirmed the neuroprotective property observed with the SP233, SP235, SP236 and SP238 of compounds against the A␤-induced changes in membrane potential. The membrane potential appears to be much higher in some of the derivative-treated cells than the control. Whether this result is representative of a hyper-excitability or not remains to be established and the direct measurement of the single cell membrane potential should allow answering this question. As the real meaning of this data is unclear, it is impossible to know at present its consequence in terms of therapeutic potential. However, an in vitro study will not be able to address that issue and according to our data this consideration does not challenge the ability of the SP compounds to reverse

12 L. Lecanu et al. / Steroids 69 (2004) 1–16 Fig. 10. Immunoblot analysis of A␤ polymerization and ADDL formation. Increasing concentrations of A␤ were incubated at 37 ◦ C in cell culture medium for 24 (A) and 72 (F) hours with or without increasing concentrations of SP233 (1, 10, 100 ␮M). Samples were separated by electrophoresis and A␤ monomers (B, G), trimers (C, H), tetramers (D, I) and polymers were identified by immunoblot analysis and quantified as described in Section 2. A␤ polymer and ADDL (the sum of trimers and tetramers) formation in the presence of increasing concentrations of SP233 is shown in panes E (24 h) and J (72 h).

L. Lecanu et al. / Steroids 69 (2004) 1–16

the effect of A␤ on the membrane potential and therefore to counteract its neurotoxicity. Amyloid aggregates have been shown to interact with the cell membrane and modify its fluidity [37,38]. Decrease in plasma membrane fluidity could hamper the function of cell surface receptors and ions channel proteins with deleterious consequences for cell survival. Considering these observations, the results obtained with the SP compounds suggest that these compounds preserve the integrity of the cytoplasmic membrane. A late event in the mechanism of action of A␤ is the direct or indirect disruption of the mitochondrial respiratory chain, leading to a decrease in ATP production that alone could lead to cell death [39–42]. 22R-hydroxycholesterol (SP222), SP235, and SP238 compounds, which were able to rescue the PC12 cells from A␤-induced toxicity, did not block the A␤-induced changes in ATP synthesis. Although such an apparent discrepancy remains to be explained, it is possible that the MTT assay (mitochondrial diaphorase activity) and ATP synthesis do not reflect the status of the same part of the respiratory chain. In contrast, SP233 and SP236 blocked, although in part, the A␤-induced decrease in ATP production. The ability of SP233 to preserve ATP stocks could explain the potent neuroprotective effect of this compound, which was further confirmed by the trypan blue uptake cell viability assay. It should be noted that SP233, at concentrations as low as 10 ␮M, was found to be not only the most efficacious in all assays used but also the most potent, offering neuroprotection in vitro against A␤. The studies presented herein were performed using 0.1, 1.0 and 10 ␮M A␤1–42 . These concentrations are supraphysiopathological since the concentrations of A␤1–42 present in cerebrospinal fluid of AD patients and controls range from 500 to 1000 ng/l (0.1–0.2 nM) [43,44]. Even if we consider that A␤1–42 might be present in AD brain at 10 times higher concentration, the estimated pathophysiological concentrations of A␤1–42 would be in the range of 1–2 nM which is 100–10,000 times less than the concentrations used in our experiments. With these considerations in mind it is obvious that the 75% protection offered by SP233 against 0.1 ␮M A␤ is pharmacologically relevant. As noted above, one of the reasons of identifying bioactive 22R-hydroxycholesterol (SP222) derivatives was the need of compounds that could not be metabolized to final steroid products in steroidogenic cells. Using the wellestablished MA-10 mouse Leydig cell model, we demonstrated that, unlike 22R-hydroxycholesterol (SP222), SP233 failed to induce steroid formation suggesting that it is not metabolized by P450scc. The neuroprotective property of the SP compounds seems to follow a structure/activity relationship (SAR). SP231 and SP235 are stereoisomers of diosgenin (Fig. 1), but only SP235 is protective against A␤-induced neurotoxicity. The stereochemistry of the SP235 is C3R, C10R, C13S, C20S, C22S, C25S, a motif shared by SP233 and SP236 (Figs. 1 and 11). SP compounds exhibiting high neuroprotective activity and being active in the presence of high concen-

13

Fig. 11. Model of the basic spirostenol structure present in the neuroprotective SP compounds. The oxygen are represented in red, hydrogen in white and carbon in dark gray color. The stereochemistry of the common core to SP235, SP236 and SP233 is C3R, C10R, C13S, C20S, C22S, C25S.

trations of A␤ contained an ester, preferably a fatty acid or a fatty acid-like structure, on C3. Indeed, SP235 that possesses an unsubstituted hydroxyl group in C3 offered limited neuroprotection, acting only against 0.1 ␮M A␤. In contrast, SP236 that is the succinic ester at C3 of SP235 is active against higher A␤ concentrations, and SP233, which is a hexanoic ester at C3 of SP235 was the most potent compound. The finding that SP238 was able to protect PC12 cells against A␤-induced toxicity, although it had no effect on maintaining ATP levels, further supports this hypothesis, because its derivative without any side-chain on C3 (SP226) did not offer neuroprotection. The finding that benzoic acid substitution, present on SP232, was not effective in neuroprotection suggested that the presence of an aliphatic chain at this level is more relevant that an aromatic structure. Although these data are indicative of a SAR and highlights the importance of the presence of a fatty acid chain at C3, further modeling and SAR studies need to be performed to optimize the SP233 structure for neuroprotection. Yao et al. [30] recently demonstrated that the neuroprotective effect of the 22R-hydroxycholesterol (SP222) lies in its ability to bind and inactivate A␤1–42 . Based on this observation, we examined the ability of 22R-hydroxycholesterol (SP222) derivatives to offer neuroprotection by acting in a similar manner. Our findings indicated that SP compounds exhibiting neuroprotective properties against A␤-induced cell death displaced radiolabeled 22R-hydroxycholesterol (SP222) bound to the amyloid peptide. However, the intensity of the displacement could not be related to the level of

14

L. Lecanu et al. / Steroids 69 (2004) 1–16

protection exhibited by the different SP compounds. Moreover, at present, we do not know whether this displacement corresponds to a competitive or non-competitive behavior. To further examine the interaction of SP233 with A␤ we examined the fate of A␤ in cell culture media incubated with and without SP233. A␤ aggregates and ADDLs were separated by electrophoresis and identified by immunoblot analyses. Under our experimental conditions, the incubation of 10 ␮M A␤ resulted in the formation of trimers, tetramers and heavy polymers. The trimers and tetramers belong to ADDLs, which are non fibrillar oligomers ranging approximately from 13 to 108 kDa [45], with potent neurotoxic properties at concentration as low as 5–10 nM [46,47]. A recent report described the ADDLs as baring the neurotoxic properties of A␤ [45]. Interestingly, we were not able to detect any ADDLs formation in the presence of 0.1 and 1 ␮M A␤ whereas these concentrations induced a decrease of the cells viability raising the question of the ADDLs as the only neurotoxic amyloid species. Whether the amounts of ADDLs formed were under the detection limit under these experimental conditions or ADDLs do not solely account for A␤ neurotoxicity remains to be established. However, SP233 decreased in a dose-dependent manner the formation of the A␤ trimers and tetramers after 24 and 72 h incubation, an event that may accounts for its neuroprotective effect. Moreover, the SP233 was able to decrease the amount of monomers available for ADDL formation, further suggesting that SP233 binds to both oligomers and monomers. The dose-dependent decrease of the ADDL levels by SP233 was accompanied by a dose-dependent increase of high molecular weight polymer aggregation, confirming previous data obtained in our lab with the 22R-hydroxycholesterol [31] and suggesting that both the 22R-hydroxycholesterol and SP233 inactivate A␤ by binding to the peptide and forming stable non-toxic polymers. Computational docking simulations were used in a first attempt to further characterize the SP-A␤ interaction. The studies revealed that two binding sites might be present on A␤ for the bioactive SP compounds. One binding site appears to be more specific for 22R-hydroxycholesterol (SP222), whereas the second binding site displays higher affinity for compounds such as SP233 and SP236. Although SP226 is shown to bind to this second binding site too, the calculated binding energy for this compound is much lower than the energy displayed by the neuroprotective SP molecules. A subsequent computational docking simulation study indicated that the binding energies of 22Rhydroxycholesterol (SP222) and SP233 follow a bimodal distribution, a finding that strongly supports the presence of two binding sites on A␤. Further calculation of binding energies indicated that 22R-hydroxycholesterol (SP222) has less affinity for the second binding site compared to SP233 and suggests that the presence of the ester chain might be responsible for the ability of SP233 to bind to both sites on A␤. Based on these observations, we hypothesize that

occupancy of the A␤ second binding site might be required for a sustained inactivation of the amyloid peptide. This hypothesis goes along with the results obtained with the 22R-hydroxycholesterol displacement study where we showed that the SP compounds with the highest affinity for the 22R-hydroxycholesterol binding site were not the best neuroprotective agent because they did not bind or they did not bind with a good affinity the SP233 site. We are now in the process of testing this hypothesis in vitro and in silico. Others mechanisms not related to a direct inactivation of A␤ could also contribute to the neuroprotective activity of SP233. A possible modulation of the steroid receptor family cannot be excluded, although little is known about the binding of spirostenols on nuclear receptors. It has been shown that A␤ inhibits the fusion of GLUT3-containing vesicles [42] leading to the disruption of mitochondrial homeostasis and, thus to neuronal death. On the other hand, the glucose absorption is enhanced in normal and streptozotocininduced diabetic mice by spirostenol derivatives extracted from Polygonati rhizome [48]. Taken together, these results suggest that restoration of glucose transport inside the cell might be a protective mechanism in our model activated by the spirostenol SP233. Natural and synthetic derivatives of diosgenin have been also shown to lower cholesterol absorption by the cell and to decrease cholesterol synthesis by inhibiting the key enzyme 3-hydroxy-3-methylglutaryl coenzyme A reductase [49,50]. It is also well known that an increase of cellular cholesterol concentration induces the activation of ␤- and ␥-secretase leading to A␤ production. Moreover, diosgenin derivatives have been shown to modify intracellular cholesterol pools by inhibiting the cholesteryl ester transfer protein [51], an enzyme reported to positively modulate the generation of A␤ [52]. Although it is unlikely that these protective mechanisms take place in our model, because A␤ is added in the culture medium, they could however be part of the in vivo response to SP233. Despite the tremendous efforts undertaken during the past few years to discover novel therapeutic modalities for the cure and/or slowing of the progression of AD, no major clinical advances have been made since the introduction of acetylcholine-esterase inhibitors, which produce modest improvements in selected patients with mild or moderate AD [53,54]. Although many compounds are actually in clinical trials in an attempt to treat AD, for most of those, AD pathology is a target secondary to their primary action. Such drugs include antioxidants, COX-1 and COX-2 inhibitors, statins, and brain vessels vasodilators. Our results indicate that naturally occurring spirostenol compounds may be of interest to protect neuronal cells against A␤. The finding that spirostenols were isolated from brain extracts of cows fed with plants containing such compounds suggests that they cross the blood brain barrier [48,55], a property required for in vivo activity. Although further in vitro and in vivo studies are required to establish the pharmacology of these compounds, the SP233 lead structure may constitute a new approach for the treatment of AD.

L. Lecanu et al. / Steroids 69 (2004) 1–16

Acknowledgements We would like to thank Drs. T. Reynolds and R. Hochberg (Yale University) for help with the steroid nomenclature, Dr. M. Ascoli (University of Iowa) for the MA-10 cells and Dr. M. Culty (Georgetown University) for critically reviewing the manuscript. References [1] Alzheimer A. Uber eine eigenartige Erkrankung der Hirnrinde. Zeitschrift für Psychiatrie 1906;10:146–8. [2] Selkoe DJ. Amyloid protein precursor and the pathogenesis of Alzheimer’s disease. Cell 1989;58:611–2. [3] Almqvist O. Neuropsychological features of early Alzheimer’s disease: preclinical and clinical stages. Acta Neurol Scand 1996;93(Suppl 185):63–71. [4] Small BJ, Herlitz A, Bäckman L. Cognitive development in Alzheimer’s disease: charting the decline process. In: Edelstein B, editor. Comprehensive clinical psychology: clinical geropsychology, vol. 7. Oxford: Elsevier; 1998. p. 231–45. [5] Berr C, Lafont S, Debuire B, Dartigues J-F, Baulieu E-E. Relationships of dehydroepiandrosterone sulfate in the elderly with functional, psychological, and mental status, and short-term mortality: a French community-based study. PNAS 1996;93:13410–5. [6] Liu Y, Schubert D. Steroid hormones block amyloid fibril-induced 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) formazan exocytosis: relationship to neurotoxicity. J Neurochem 1998;19:1639–62. [7] Chae HS, Bach JH, Lee MW, Kim HS, Kim YS, Kim KY, et al. Estrogen attenuates cell death induced by carboxy-terminal fragment of amyloid precursor protein in PC12 through a receptor-dependent pathway. J Neurosci Res 2001;65:403–7. [8] Hosoda T, Nakajima H, Honjo H. Estrogens protect neuronal cells from amyloid beta-induced apoptotic cell death. Neuroreport 2001;12(9):1965–70. [9] Zhang L, Rubinow DR, Xaing G, Li BS, Chang YH, Maric D, et al. Estrogens protects against beta-amyloid-induced neurotoxicity in rat hippocampal neurons by activation of Akt. Neuroreport 2001;12(9):1919–23. [10] Greenfield JP, Leung LW, Cai D, Kaasik K, Gross RS, RodriguezBoulan E, et al. Estrogen lowers Alzheimer ␤-amyloid generation by stimulating trans-golgi network vesicle biogenesis. J Biol Chem 2002;277(14):12136. [11] Lacreuse A, Wilson ME, Herndon JG. Estradiol, but not raloxifene, improves aspects of spatial working memory in aged ovariectomized rhesus monkeys. Neurobiol Aging 2002;23:589–600. [12] Rissman EF, Heck AL, Leonard JE, Shupnik MA, Gustafsson JA. Disruption of receptor beta gene impairs spatial learning in female mice. Proc Natl Acad Sci USA 2002;99(6):3996–4001. [13] Zhao L, Chen Q, Diaz Brinton R. Neuroprotective and neuroefficacy of phytoestrogens in cultured hippocampal neurons. Exp Biol Med 2002;227(7):509–19. [14] Lund TD, West TW, Ly T, Bu LH, Simmons DL, Setchell KD, et al. Visual spatial memory is enhanced in female rats (but inhibited in males) by dietary soy phytoestrogens. BMC Neurosci 2001;2(1):20. [15] Lephart ED, West TW, Weber KS, Rhees RW, Setchell KD, Adlercreutz H, et al. Neurobehavorial effects of dietary soy phytoestrogens. Neurotoxicol Teratol 2002;24(1):5–16. [16] Roberts E, Bologa L, Flood JF, Smith GE. Effects of dehydroepiandrosterone and its sulfate on brain tissue in culture and on memory in mice. Brain Res 1987;406:357–62. [17] Flood JF, Roberts E. Dehydroepiandrosterone sulfate improves memory in aging mice. Brain Res 1988;448:178–81.

15

[18] Wolf OT, Kirschaum C. Endogenous estradiol and testosterone levels are associated with cognitive performance in older women and men. Horm Behav 2002;41(3):259–66. [19] Simpkins JW, Singh M, Bishop J. The potential role for estrogen replacement therapy in the treatment of the cognitive decline and neurodegeneration associated with Alzheimer’s disease. Neurology 1994;15(Suppl 2):S195–7. [20] Asthana S, Baker LD, Craft S, Stanczyk FZ, Veith RC, Raskind MA, et al. High-dose estradiol improves cognition for women with AD. Neurology 2001;57:605–12. [21] Cholerton B, Gleason CE, Baker L, Asthana S. Estrogen and Alzheimer’s disease: the story so far. Drugs Aging 2002;19(6):405– 27. [22] Diaz-Brinton R. Cellular and molecular mechanisms of estrogen regulation of memory function and neuroprotection against Alzheimer’s disease: recent insights and remaining challenges. Learning Memory 2001;8:121–33. [23] Haskell SG, Richardson ED, Horwitz RI. The effect of estrogen replacement therapy on cognitive function in women: a critical review of the literature. J Clin Epidemiol 1997;50(11):1249–64. [24] Kesslak JP. Can estrogenplay a significant role in the prevention of Alzheimer’s disease? J Neural Trans 2002;62:227–39. [25] Marks SJ, Batra RR, Frishman H. Estrogen replacement therapy for cognitive benefits: viable treatment or forgettable “senior moment”? Heart Dis 2002;4:26–32. [26] Van Duijn C. Hormone therapy replacement and Alzheimer’s disease. Maturitas 2001;1:201–5. [27] Shetty K. Hormone replacement therapy. Logically, long term hormone replacement therapy cannot be recommended. Br Med J 2002;235:1243. [28] Voelker R. Questions about hormone replacement therapy. J Am Med Assoc 2002;288:2395–6. [29] Dixon R, Furutachi T, Lieberman S. The isolation of crystalline 22R-hydroxycholesterol and 20-alpha, 22R-hydroxycholesterol from bovine adrenals. Biochem Biophys Res Commun 1970;49:161–5. [30] Yao ZX, Brown RC, Teper G, Greeson J, Papadopoulos V. 22R-Hydroxycholesterol protects neuronal cells from ␤-amyloidinduced cytotoxicity by binding to ␤-amyloid peptide. J Neurochem 2002;83:1110–9. [31] Yao Z, Drieu K, Papadopoulos V. The Gingko biloba extract EGb 761 rescues PC12 neuronal cells from ␤-amyloid-induced cell death by inhibiting the formation of ␤-amyloid-derived diffusible neurotoxic ligands. Brain Res 2001;889:181–90. [32] Yao Z, Papadopoulos V. Function of ␤-amyloid in cholesterol transport: a lead to neurotoxicity. FASEB J 2002;16:1677–9. [33] Watson AA, Fairlie DP, Craik DJ. Solution structure of methionineoxidized amyloid beta-peptide (1-40). Does oxidation affect conformational switching? Biochemistry 1998;37:12700–6. [34] Li H, Yao Z, Degenhardt B, Teper G, Papadopoulos V. Cholesterol binding at the cholesterol recognition/interaction amino acid consensus (CRAC) of the peripheral-type benzodiazepine receptor and inhibition of steroidogenesis by an HIV TAT-CRAC peptide. Proc Natl Acad Sci USA 2001;98:1267–72. [35] Morris GM, Goodsell DS, Halliday RS, Huey R, Hart WE, Belew RK, et al. Distributed automated docking of flexible ligands to proteins: parallel applications of AutoDock 2.4.. J Comput Chem 1998;19:1639–62. [36] Hall PF. Cellular organization for steroidogenesis. Int Rev Cytol 1984;86:53–95. [37] Kremer JJ, Pallitto MM, Sklansky DJ, Murphy RM. Correlation of ␤-amyloid aggregate size and hydrophobicity with decreased bilayer fluidity of model membranes. Biochemistry 2000;39:10309–18. [38] Kremer JJ, Sklansky DJ, Murphy RM. Profile of changes in lipid bilayer structure caused by ␤-amyloid peptide. Biochemistry 2001;40:8563–71. [39] Pereira C, Santos MS, Oliveira C. Involvement of oxidative stress on the impairment of energy metabolism induced by Abeta peptides on

16

[40]

[41]

[42]

[43]

[44]

[45]

[46]

L. Lecanu et al. / Steroids 69 (2004) 1–16 PC12 cells: protection by antioxidants. Neurobiol Dis 1999;6(3):209– 19. Eckert GP, Wood WG, Muller WE. Effects of aging and ␤-amyloid on the properties of brain synaptic and mitochondrial membranes. J Neural Transm 2001;108:1051–64. Kourie JI. Mechanisms of amyloid beta protein-induced modification in ion transport systems: implications for neurodegenerative diseases. Cell Mol Neurobiol 2001;21(3):173–213. Prapong T, Buss J, Hsu WH, Heine P, West Greenlee H, Uemura E. Amyloid beta-peptide decreases neuronal glucose uptake despite causing increase in GLUT3 mRNA transcription and GLUT3 translocation to the plasma membrane. Exp Neurol 2002;174(2): 253–8. Hulstaert F, Blennow K, Ivanoiu A, Schoonderwaldt HC, Riemenschneider M, De Deyn PP, et al. Improved discrimination of AD patients using ␤-amyloid(1–42) and tau levels in CSF. Neurology 1999;52:1555–62. Riemenschneider M, Schmolke M, Lautenschlager N, Guder WG, Vanderstichele H, Vanmechelen E, et al. Cerebrospinal ␤-amyloid(1–42) in early Alzheimer’s disease: association with apolipoprotein E genotype and cognitive decline. Neurosci Lett 2000;284:85–8. Klein WL. A␤ toxicity in Alzheimer’s disease: globular oligomers (ADDLs) as new vaccine and drug targets. Neurochem Int 2002;41:345–52. Lambert MP, Barlow AK, Chromy BA, Edwards C, Freed R, Liosatos M, et al. Diffusible, nonfibrillar ligands derived from A␤1–42 are potent central nervous system neurotoxins. Proc Natl Acad Sci USA 1998;95:6448–53.

[47] Dahlgren KN, Manelli AM, Stine WB, Baker LK, Krafft GA, LaDu MJ. Oligomeric and fibrillar species of amyloid␤ peptides differentially affect neuronal viability. J Biol Chem 2002;277(35):32046–53. [48] Juarez-Oropeza MA, Diaz-Zagoya JC, Rabinowitz JL. In vivo and in vitro studies of hypocholesterolemic effects of diosgenin in rats. Int J Biochem 1987;19(8):679–83. [49] Kato A, Miura T. Hypoglycemic activity of Polygonati rhizoma in normal and diabetic mice. Biol Pharm Bull 1993;16(11):1118–20. [50] Kim HS, Wison WK, Needleman DH, Pinkerton FD, Wilson DK, Quiocho FA, et al. Inhibitors of sterol synthesis. Chemical synthesis, structure, and biological activities of (25R)-3␤, 26-hydroxy-5␣cholest-8(14)-en-15-one, a metabolite of 3␤-hydroxy-5␣-cholest8(14)-en-15-one. J Lipid Res 1989;30(2):247–61. [51] Kim HS, Oh SH, Kim DI, Kim IC, Cho KH, Park YB. Chemical synthesis of 15-ketosterols and their inhibitions of cholesteryl ester transfer protein. Bioorg Med Chem 1995;3(4):367–74. [52] Puglielli L, Konopka G, Pack-Chung E, MacKenzie Ingano LA, Berezovska O, Hyman BT, et al. Acyl-coenzyme A: cholesterol acyltransferase modulates the generation of the amyloid ␤-peptide. Nat Cell Biol 2001;3:905–12. [53] Birks JS, Melzer D. Dopenezil for mild and moderate Alzheimer’s disease. Cochrane Database Syst Rev 2000;2:CD001190. [54] Grutzendler J, Moris JC. Cholinesterase inhibitors for Alzheimer’s disease. Drugs 2001;61:41–52. [55] Tsukatani H, Takauchi K, Yoshida J, Yamata S, Tokumura A, Hamaguchi C. Isolation of diosgenin [(25R)-spirost-5-en-3beta-ol] from a lysate of a new hypotensive phospholipid occurring in bovine brain. J Neurochem 1985;44(20):658–61.