The Neuroprotective Effects of Phytoestrogens on Amyloid b Protein-induced Toxicity Are Mediated by Abrogating the Activation of Caspase Cascade in Rat Cortical Neurons*

Amyloid b protein (A b ) elicits a toxic effect on neurons in vitro and in vivo . In present study we attempt to elucidate the mechanism by which A b confers its neurotoxicity. The neuroprotective effects of phytoestrogens on A b -mediated toxicity were also investigated. Cortical neurons treated with 5 m M A b -(25–35) for 40 h decreased the cell viability by 45.5 6 4.6% concomitant with the appearance of apoptotic morphology. 50 m M kaempferol and apigenin decreased the A b -induced cell death by 81.5 6 9.4% and 49.2 6 9.9%, respectively. A b increased the activity of caspase 3 by 10.6-fold and to a lesser extent for caspase 2, 8, and 9. The A b -induced activation of caspase 3 and release of cytochrome c showed a biphasic pattern. Apigenin abrogated A b -induced cytochrome c release, and the activation of caspase cascade. Kaempferol showed a similar effect but to a less extent. Kaempferol was also capable of eliminating A b -induced accumulation of reactive oxygen species. These two events accounted for the remarkable effect of kaempferol on neuroprotection. Quercetin and probucol did not affect the A b -mediated neurotoxicity. However

Extensive studies have shown that A␤-induced neurotoxicity in multiple cell types may be mediated by several different mechanisms. The neurotoxic effect may be attributable to the disturbance in calcium homeostasis (12)(13)(14) and consequently inducing the accumulation of reactive oxygen species (ROS) (13,(15)(16)(17)(18). ROS provokes membrane damage compromising membrane integrity and increasing the permeability of ions, including calcium. The increase of calcium influx leads to generating more ROS, thereby initiating the positive feedback loop. Cultured neurons treated with A␤ or transgenic mice expressing A␤ renders neurons vulnerable to apoptosis, indicating that caspase activation plays a role in A␤-induced neurotoxicity. Several caspases involved in apoptosis have been described to be activated by A␤ (6, 19 -22). However, the mechanisms by which A␤ activates caspase cascade remain unclear. Furthermore, the activation of cyclin-dependent kinase has also been indicated to be a mediator of neurotoxicity induced by A␤ (23).
Several agents have been shown to be neuroprotective in in vitro system by targeting to specific pathway responsible for A␤-induced toxicity. These agents include antioxidants or free radical scavengers (12, 15, 17, 18, 24 -26), calcium ion channel blockers (13,26); growth factors (27), inhibitor of cyclin-dependent kinase (23), and caspase inhibitors (15,19,21). Recent evidence shows that estrogen deficiency in postmenopausal women is one of the most significant risk factors for onset of AD (28,29). Thus, estrogens have become a research focus. It has been shown that estrogen protects neurons against a number of toxic insults, including A␤ (30 -34). The neuroprotective effects of estrogen are suggested to be independent of their classic nuclear estrogen receptors (32,34).
In the present report, we demonstrate that apigenin blocked the release of cytochrome c and activation of caspase cascade induced by A␤. Kaempferol only inhibited the activation of caspase cascade, and the effect was less potent than that of apigenin. However, kaempferol was more effective than apigenin in counteract the deleterious effect of A␤ on cortical neurons. Kaempferol exhibited antioxidative activity and decreased the ROS accumulation induced by A␤, whereas apigenin lacked antioxidative activity and showed a marginal effect on ROS level. Furthermore, quercetin or probucol facilitated the neuroprotective effect of apigenin on A␤-mediated toxicity. Therefore, these results indicated that blockade of activation of caspase cascade conferred the neuroprotective effects of phytoestrogens on A␤-mediated neurotoxicity. The inhibition of caspase cascade in combination with antioxidative activity will further eliminate A␤-mediated neurotoxicity.

Methods
Cell Culture-Primary cultures of neonatal cortical neurons were prepared from the cerebral cortex of Harlan Sprague-Dawley rat pups at postnatal day 1 (40,41). Briefly, each pup was decapitated and the cortex was digested in 0.5 mg/ml papain at 37°C for 15 min. The tissue was dissociated in Hypernate A medium (containing B27 supplement) by aspirating trituration. Cells were plated (5 ϫ 10 4 cells/cm 2 ) onto poly-D-lysine-coated dishes and maintained in Neurobasal medium containing B27 supplement (40), 10 units/ml penicillin, 10 g/ml streptomycin, and 0.5 g/ml glutamine (5% CO 2 /9% O 2 ) for 3 days. Cells were then exposed to cytosine-␤-D-arabinofuranoside (5 M) for 1 day to inhibit proliferation of non-neuronal cells. The cells were used for the experiment on the fifth day.
Measurement of Cell Viability-The reduction of MTT, cleavage of FDA, and trypan blue exclusion were used to evaluate the cell viability. Cells were incubated with minimum essential medium containing 0.5 mg/ml MTT for 1 h. The medium was aspirated, and the formazan particle was dissolved with lysis buffer (10% sodium dodecyl sulfate, 3.3 mM HCl, 50% dimethylformamide). A 600 nm was measured by using enzyme-linked immunosorbent assay reader (42). Cells were loaded with 15 M FDA for 5 min at 25°C, and then 1 ml of 1% deoxycholate was added to lyse the cells. The fluorescent intensity of the lysate was determined by using a spectrofluorometer with excitation and emission wavelength of 490 nm and 514 nm, respectively (26,43). Cell viability was also assessed by using trypan blue exclusion as described previously (42).
Measurement of ROS-Intracellular reactive oxygen species were measured by CM-H 2 DCFDA assay (26). In brief, cells were loaded with 50 M CM-H 2 DCFDA for 30 min, and then 1 ml of 1% deoxycholate was added to lyse the cells. The fluorescent intensity of the lysate was determined by using a spectrofluorometer with excitation and emission   (10) 45.0 Ϯ 5.0 wavelength of 488 nm and 525 nm, respectively. Measurement of Cytochrome c Release-Treated cells were collected in harvest buffer (50 mM Hepes, pH 7.5, 1 mM EDTA, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, 5 g/ml aprotinin, and 10 g/ml leupeptin) and suspended by passage through a 200-l pipette tip. The cell suspensions were centrifuged at 10,000 ϫ g for 10 min, and supernatants were diluted with the same volume of immunoprecipitation buffer (harvest buffer containing 2% Nonidet P-40). Immunoprecipitation of cytochrome c was performed by using anti-cytochrome c antibody. The immunoprecipitates were subjected to SDS-polyacrylamide gel electrophoresis. Immunoreactive proteins were detected by enhanced chemiluminescence detection reagents.
Measurement of Cellular Activity of Caspases-Treated cells were harvested in cell lysis buffer (50 mM Hepes, pH 7.4, 1 mM dithiothreitol, 0.1 mM EDTA, 0.1% Chaps, and 0.1% Triton X-100). The detailed experiments were performed according to the manufacturer's protocol.
Other Methods-The antioxidant activity of phytoestrogens were determined by trolox equilibrium antioxidant capacity (TEAC) method as described by Pellegrini et al. (44). The activity of lactate dehydrogenase (LDH) was determined by a method described previously (45). Results are expressed as means Ϯ S.D. and were analyzed by ANOVA with post hoc multiple comparison with a Bonferroni test.

Phytoestrogens Protected Neurons from A␤-induced
Neurotoxicity-Treatment of rat cortical neurons with the toxic fragment of fibril A␤ (f-A␤-(25-35)) for 40 h decreased cell viability in a concentration-dependent manner as determined by MTT reduction or trypan blue exclusion ( Fig. 2A). 10 M A␤ decreased the MTT reduction and trypan blue exclusion by 58.2 Ϯ 3.9% and 63.3 Ϯ 11.5%, respectively. However, treatment with A␤ elicited a marginal effect on the ability to cleave fluorescein diacetate and had no effect on the release of LDH. The effect of A␤ on cell viability of cortical neurons was time-dependent as measured by MTT reduction (Fig. 2B). The appearance of irregularly shaped cell bodies and discontinuous neurite was concomitant with the decrease of cell viability (Fig. 2C). The percentage of cells with injured morphology was elevated as the concentration of A␤ was increased.
The effects of a series of phytoestrogens on A␤-induced neurotoxicity were investigated. Cells were incubated with various concentrations of pytoestrogens for 2 h then exposed to 5 M A␤ for 40 h. Cell viability was verified by MTT reduction analysis. Results showed that kaempferol and apigenin reduced the A␤induced neurotoxicity in a concentration-dependent manner (Fig. 3, A and B). Kaempferol at 50 M decreased the percentage of cell death from 45.5 Ϯ 4.6% to 7.9 Ϯ 3.6%. Apigenin was less potent to attenuate A␤-induced neurotoxicity (Fig. 3B). For the morphology, 50 M kaempferol diminished the extent of cells with injured morphology induced by A␤ (Fig. 3C). The protective effect of kaempferol on A␤-mediated neurotoxicity was further confirmed by the measurement of trypan blue exclusion (data not shown). Luteolin, quercetin, flavanones, isoflavones, and ␤-estradiol did not show any significant effects on A␤-induced neurotoxicity (Table I).
A␤ Induced the Activation of Caspase Cascade-Inhibitors of caspases were employed to investigate whether apoptosis was involved in A␤-mediated toxicity. Ac-DEVD-CHO and Ac-LEHD-CHO are the cell-permeable inhibitors for caspases 3 and 9, respectively. These inhibitors block the activity of caspases but do not interfere with its activation. Both inhibitors reduced A␤-induced cell death in a concentration-dependent manner (Fig. 4, A and B). In contrary, Ac-IETD-CHO, the cell-permeable inhibitor of caspase 8, did not decrease A␤induced cell death (data not shown). Morphological study also showed that Ac-DEVD-CHO and Ac-LEHD-CHO eliminated the cells with injured morphology induced by A␤ (Fig. 4C).
Analysis of the activity of caspases were performed to further determine whether the activation of caspase cascade was involved in the A␤-mediated neurotoxicity. Treatment of neurons with A␤ induced activation of caspase 2, 3, 8, and 9 in a time-dependent manner as measured by substrate cleavage (Fig. 5). Exposure of neurons to A␤ for 24 h, the specific activity of caspase 2, 3, 8, and 9 were increased by 6.4-, 11.6-, 6.0-, and 4.7-fold of control, respectively. The activity of caspase 3 was enhanced to a larger extent by A␤ in comparison with caspase 2, 8, and 9. A␤ exhibited a biphasic effect on the activation of caspase 3. There was a transit activation of caspase 3 at 2 h and followed by a sustained activation from 8 to 24 h. The similar result for caspase 9 was also obtained. However, the first wave of activation of caspase 9 at 2 h was not statistically significant. Caspases 2 and 8 did not show early phase of activation. The significant activation of caspases 2 and 8 occurred from 12 to 24 h.
The Effects of Caspase Inhibitors and Phytoestrogens on Caspase Cascade-The inhibitor of caspases 3, 9, and 8 decreased the activity of caspase 3 and 9 at 2 h (Table II). Both inhibitors of caspase 3 and 9 inhibited the activity of caspase 2, 3, and 9 at 24 h (Table III). The inhibitor of caspase 3 diminished the activity of caspase 2, 3, and 9 by 72.8 Ϯ 8.7, 90.8 Ϯ 4.4, and 60.9 Ϯ 15.0%, respectively, and the inhibitor of caspase 9 decreased the activity of caspases to a lesser extent. However, both inhibitors did not show significant effect on the activity of caspase 8. The inhibitor of caspase 8 decreased the activity of caspase 8 and 9 by 17.0 and Ϯ 4.0% and 43.4 Ϯ 4.0%, respectively.
Apigenin and kaempferol did not show significant effects on the activation of caspase cascade at 2 h (Table II)   by about 55 to 60% (Table III). Kaempferol also diminished the activity of these caspases but to a lesser extent. Quercetin only showed inhibitory effect on the activity of caspase 9 at 24 h (Table III). Apigenin Inhibited A␤-induced Release of Cytochrome c-Kaempferol and apigenin attenuated A␤-induced activation of caspases 9 and 3. The release of cytochrome c from mitochondria, the upstream signaling component of these two caspases, was therefore investigated (Fig. 6). The release of cytochrome c induced by A␤ exerted a biphasic pattern similar to the activation of caspase 3. The first phase occurred at 2 h (Fig. 6A) and followed by the second phase from 12 to 24 h (data not shown). Apigenin significantly inhibited A␤-induced cytochrome c release at 2 and 12 h by 34.1 and 55.7%, respectively. However, kaempferol did not affect A␤-induced cytochrome c release either at 2 or 12 h.
Antioxidative Activity Potentiated the Neuroprotection of Apigenin-The effect of A␤ on intracellular level of ROS was examined by using 5-(and-6)-chloromethyl-2Ј,7Ј-dichlorodihydrofluorescein diacetate (CM-H 2 DCFDA). A␤ induced ROS accumulation significantly from 8 to 24 h (Fig. 7A). Treatment with 5 M A␤ for 8 and 16 h elevated the level of ROS to 129 Ϯ 5% and 158 Ϯ 10% of control, respectively. A␤ increased the level of ROS in a concentration-dependent manner (Fig. 7B). Kaempferol, luteolin, and quercetin reduced the level of ROS by 30 -50% in control cells and by 50 -60% in A␤-treated cells (Fig. 7C). Apigenin decreased the ROS level by 25% in A␤treated cells but did not affect that in control cells. The antioxidant capacity of phytoestrogens was also determined by TEAC method (Table IV). The results showed that apigenin had lower antioxidant capacity than quercetin and luteolin. Kaempferol also showed higher antioxidant capacity than apigenin. Thus, the sequence of antioxidant capacity for these phytoestrogens did not correlate well to their neuroprotective ability. The results also implied that high antioxidative activity of flavonoid per se was not able to protect neuron against A␤-induced neurotoxicity.
Apigenin was much more able than kaempferol to block caspase activation (Table III). Apigenin, however, exhibited the neuroprotective effect to a lesser extent (Fig. 3). We speculated, therefore, that the antioxidative activity and ability for ROS reducing may modulate the neuroprotective effect of kaempferol. To address the hypothesis that antioxidative activity may enhance the neuroprotective effect of apigenin, the effect of antioxidants on neuroprotection of apigenin was evaluated (Fig. 8). Quercetin did not reduce A␤-mediated cell death from 1 to 50 M. Cotreatment with quercetin and apigenin enhanced the neuroprotective effect of apigenin (Fig. 8A). Probucol, an antioxidant, was more potent at potentiating the neuroprotection effect of apigenin (Fig. 8B). The results demonstrated that, although antioxidant activity of quercetin and probucol per se did not show neuroprotective effect, they did modulate the neuroprotective effect of apigenin. DISCUSSION In the present report, we demonstrate that kaempferol and apigenin prevented death of cultured neurons exposed to fibril A␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35). The action modes of these phytoestrogens were neither due to an activation of the nuclear estrogen receptor nor solely based on an antioxidative mechanism. The antiapoptotic signaling activity conferred neuroprotective effect of kaempferol and apigenin. Antioxidative activity of flavonoids or other antioxidants did not exhibit direct effects on neuroprotection. However, antioxidative activity facilitated the neuroprotective effect of apigenin.
Kaempferol and apigenin protected neurons from A␤-mediated toxicity, whereas quercetin and luteolin failed to protect neurons. This result suggests that the substitution of hydroxyl group at C-3Ј position severely impairs the neuroprotective ability of kaempferol. The deficiency of neuroprotective effects of narigenin and 2,3-dihydroluteolin also implicates that the double bond between the C-2 and C-3 positions is required for apigenin and kaempferol to exert their neuroprotective effects. 4-Hyroxytamoxifen, an estrogen receptor antagonist, did not abrogate the protective effects, suggesting that the mode of action is independent of nuclear estrogen receptor (data not shown). That 17␤-Estradiol did not protect cortical neurons against A␤-induced toxicity further confirms the hypothesis that the effect of phytoestrogen segregates from the action of estrogen receptor. However, some studies have shown that estrogens counteract the toxic effect of A␤ on PC12 phenochromocytoma cells and human neuroblastoma cells (32,33,45,46). A possible explanation of these discrepancies could be A␤ exerting toxic effect on cell line and primary culture of cortical neurons via distinct mechanisms. Phytoestrogens reducing the loss of MTT reduction by A␤ has also been described in PC12 phenochromocytoma cells. The report (45), however, concludes that phytoestrogens affect the plasma membrane sensitivity to formazan crystal rather than protect cells against A␤-induced cell death (45). Furthermore, quercetin shows protective effect to PC12 cells but not to primary culture of cortical neurons. These results suggest that cell lines and the primary culture of cortical neurons are protected by pytoestrogens through distinct mechanism.
Both caspase 3 inhibitor and caspase 9 inhibitor prevented neuronal death induced by A␤, suggesting that A␤-exerted neurotoxicity is mediated by an apoptotic pathway. Evidence supporting this speculation includes the facts that A␤ elevated the activity of caspase 3, 9, 8, and 2 and promoted the release of cytochrome c from mitochondria. The results also showed that caspase 3 displayed a biphasic activation by A␤ treatment. This activation consisted of a transient activation at 2 h and a

TABLE III
The inhibition of caspase activation at 24 h by either phytoestrogen or caspase inhibitors Cortical neurons were incubated with phytoestrogens (50 M) or caspase inhibitor (10 M) for 2 h and exposed to 5 M A␤ for 24 h. Cell viability was assessed by MTT reduction analysis. The data are means Ϯ S.D. of four independent experiments and expressed relative to cell treated with A␤ alone. Significant differences between cells treated with A␤ and A␤ plus phytoestrogen or caspase inhibitor are indicated in footnotes. sustained activation after 8 h. For caspase 9, the first wave activation was not supported by the statistics. Nevertheless, the elevation of caspase 9 activity at 2 h may be biologically relevant. The release of cytochrome c also showed a similar pattern as the activation of caspase 3 and 9, although A␤ activated caspase 2 and 8 after 16 h and to a lesser extent. These data suggest that the apoptosis signaling induced by A␤ is mediated primarily by activation of caspase 3 and 9 and cytochrome c release. The activation of caspase 2 and 8 and the production of ROS may be the secondary responses. Caspase 8 inhibitor was found unable to prevent neuronal death induced by A␤, which further confirms that caspase 8 may not play the major role in A␤-induced neuronal death.

Reagents
Although the A␤-mediated neurotoxicity becomes the focus of intense interest. The underlining mechanisms are still controversial. The study of Giovanni et al. shows that cortical neurons from caspase 3 knockout mice are resistant to A␤-mediated cell death, suggesting that caspase 3 is the major component mediating A␤-induced apoptosis (22). However, the studies of Troy et al. (21) show down-regulation of caspase 3 does not block A␤-(1-42)induced cell death. They also show that sympathetic neurons from caspase 2 null mice are resistant to A␤-(1-42)-mediated cell death, implicating an important role of caspase 2 in A␤-(1-42)induced apoptosis (21). Moreover, cell death is blocked by the down-regulation of caspase 2 in hippocampal neurons, sympathetic neurons, or neuronal PC12 cells with antisense oligonucleotides. Beside caspase 3 and 2, other caspases are also thought to be involved in A␤-induced apoptosis. Nakagawa et al. (20) shows that caspase-12-deficient cortical neurons are not susceptible to the apoptosis induced by A␤-(1-40) (20). Another study, however, shows that the apoptotic pathway activated by A␤ requires both caspase 8 and Fas-associated death domain (FADD) (47). Our results demonstrate that A␤ elicited the activation of caspase 2, 3, 8, and 9 in cortical neurons. The release of cytochrome c and activation of caspase 9 and caspase 3 may be the major pathway mediating the A␤-induced apoptosis of neurons. Kaempferol and apigenin abrogated the activation of all four caspases (Table III). Apigenin was more potent than kaempferol in blocking the release of cytochrome c and activation of caspase cascade induced by A␤. However, kaempferol was more effective than apigenin to protect neurons from A␤-induced cell death. On the basis of these data, we speculate that there may be other factors involved in the protective effect of kaempferol. ROS scavenging activity of kaempferol may be the most possible candidate to promote its neuroprotective effect.
Many reports have demonstrated or proposed that ROS is responsible for A␤-induced neurotoxicity (13,(15)(16)(17)(18). Behl et al. (17) shows that A␤ increases the intracellular level of H 2 O 2 and   (48) further provide evidence that 4-hydroxynonenal, an aldehydic product of membrane lipid peroxidation, is a key mediator of neuronal apoptosis induced by A␤. Therefore, it is possible that scavenging of ROS may also contribute to the neuroprotective activity of kaempferol and apigenin. The level of ROS in A␤-treated neurons was elevated after 8-h incubation, implying that ROS production may be another mediator for A␤induced cell death. The level of ROS in A␤-treated neurons were reduced by both neuroprotective and nonprotective flavonoids. Furthermore, nonprotective flavonoids, quercetin and luteolin, were more potent in reducing the level of ROS and had higher antioxidative activity. These results indicate that the antioxidative activity of flavonoids per se do not confer the neuroprotective effect on A␤-mediated toxicity. Nevertheless, quercetin and probucol, which possess antioxidative activity, did show a significant facilitating effect on the neuroprotection of apigenin. These data provide the convincing explanation for the inconsistent results between inhibition of caspase cascade and the neuroprotective effect of kaempferol and apigenin. Taken together, the results presented here provide a plausible mechanism by which A␤ provokes death of cortical neurons. In this model (see Fig. 9), A␤-mediated apoptosis consists of the first and second waves of caspases activation. A␤ primarily induces the release of cytochrome c from mitochondria and subsequent activation of caspase 9 and 3 at 2 h (Figs. 5 and 6). The inhibitor of caspase 8 blocked the activation of caspase 9 and 3 at 2 h (Table II), indicating caspase 8 may be involved in the first wave of caspase activation. Thereafter, caspase 3 evokes a second wave of cytochrome c release and activation of caspase cascade from 12 to 24 h. In the second wave of caspase activation, there is a positive feedback between caspase 3 and cytochrome c release, thereby establishing a signaling cascade and amplifying the signal (49). The inhibitor of caspase 3 abrogated the activation of caspase 2 at 24 h (Table III), indicating that caspase 2 may be the downstream factor of caspase 3 and complicated in the second wave of caspase activation. The sustained activation of caspase 8 occurred after 12 h, indicating that caspase 8 may also participate in the second wave of caspase activation. Furthermore, The damaged mitochondria in the second wave of caspase activation may also cause the accumulation of ROS, or alternatively, the generation of ROS after 8 h may also be involved in the release of cytochrome c and the second wave activation of caspase cascade. Both activation of caspase cascade and elevation of ROS may account for the A␤-mediated toxicity. Both apigenin and kaempferol, but not quercetin, abrogated the second wave of cytochrome c release or activation of caspase 2, 3, 8, and 9 (Table  III, Fig. 6). Despite apigenin abrogating the initial wave of cytochrome c release, both flavonoids did not affect the initial wave activation of caspase 3 and 9. These results suggest that apigenin and kaempferol protect neurons against A␤-induced toxicity by diminishing the second wave activation of the caspase cascade. Our results clearly demonstrate that kaempferol and apigenin exhibited differentially neuroprotective effects on A␤-induced toxicity by abrogating the release of cytochrome c and activation of the caspase cascade. Antioxidative activity of flavonoids or other antioxidants facilitated the caspase-dependent neuroprotective effect of phytoestrogen, thereby conferring a significant neuroprotective ability against A␤-mediated toxicity.
Our results demonstrate that inhibition of caspase and scavenging of ROS act cooperatively to save neurons from A␤-mediated toxicity. Therefore, inhibition of caspase cascade and decrease in the level of ROS are proposed as neuroprotective strategies in AD. Base on these results, the development of neuroprotective agents such as a compound that combines potent antioxidant and caspase inhibitory properties may prevent the incidence of AD or retard the progression of AD. Thus, our findings point toward new approach to drug discovery for clinical therapies of AD.