Modulation of Pro-survival Akt/Protein Kinase B and ERK1/2 Signaling Cascades by Quercetin and Its in Vivo Metabolites Underlie Their Action on Neuronal Viability*

Much recent interest has focused on the potential of flavonoids to interact with intracellular signaling pathways such as with the mitogen-activated protein kinase cascade. We have investigated whether the observed strong neurotoxic potential of quercetin in primary cortical neurons may occur via specific and sensitive interactions within neuronal mitogen-activated protein kinase and Akt/protein kinase B (PKB) signaling cascades, both implicated in neuronal apoptosis. Quercetin induced potent inhibition of both Akt/PKB and ERK phosphorylation, resulting in reduced phosphorylation of BAD and a strong activation of caspase-3. High quercetin concentrations (30 (cid:1) M ) led to sustained loss of Akt phosphorylation and subsequent Akt cleavage by caspase-3, whereas at lower concentrations ( < 10 (cid:1) M ) the inhibition of Akt phosphorylation was transient and eventually returned to basal levels. Lower levels of quercetin also induced strong activation of the pro-survival transcription factor cAMP-responsive element-binding protein, although this did not prevent neuronal damage. O- Methylated quercetin metabolites inhibited Akt/PKB to lesser extent and did not induce such strong activation of caspase-3, which was reflected in the lower amount of damage they inflicted on neurons. In contrast, neither quercetin nor its O -methylated collected. Caspase-3-like protease activity in neuronal lysates was assessed based on the hydrolysis of the peptide substrate acetyl-Asp-Glu-Val-Asp p- nitroanilide by caspase-3, resulting in the release of the p- nitroaniline moiety, which absorbs at 405 nm. Quantification of the absorbance at 405 nm was carried out using a Versa Max UV microplate reader (Molecular Devices). Suitable control experiments were performed us- ing the caspase-3 inhibitor acetyl-Asp-Glu-Val-Asp-al. All of the data are presented as absorbance readings at 405 nm. Statistical Analysis — The data are expressed as the means (cid:7) S.D. Statistical comparisons were made using an unpaired, two-tailed Stu- dent ’ s t test with a confidence level of 95%. The significance level was set at p (cid:4) 0.05.

Much recent interest has focused on the potential of flavonoids to interact with intracellular signaling pathways such as with the mitogen-activated protein kinase cascade. We have investigated whether the observed strong neurotoxic potential of quercetin in primary cortical neurons may occur via specific and sensitive interactions within neuronal mitogen-activated protein kinase and Akt/protein kinase B (PKB) signaling cascades, both implicated in neuronal apoptosis. Quercetin induced potent inhibition of both Akt/PKB and ERK phosphorylation, resulting in reduced phosphorylation of BAD and a strong activation of caspase-3. High quercetin concentrations (30 M) led to sustained loss of Akt phosphorylation and subsequent Akt cleavage by caspase-3, whereas at lower concentrations (<10 M) the inhibition of Akt phosphorylation was transient and eventually returned to basal levels. Lower levels of quercetin also induced strong activation of the pro-survival transcription factor cAMP-responsive element-binding protein, although this did not prevent neuronal damage. O-Methylated quercetin metabolites inhibited Akt/PKB to lesser extent and did not induce such strong activation of caspase-3, which was reflected in the lower amount of damage they inflicted on neurons. In contrast, neither quercetin nor its O-methylated metabolites had any measurable effect on c-Jun N-terminal kinase phosphorylation. The glucuronide of quercetin was not toxic and did not evoke any alterations in neuronal signaling, probably reflecting its inability to enter neurons. Together these data suggest that quercetin and to a lesser extent its O-methylated metabolites may induce neuronal death via a mechanism involving an inhibition of neuronal survival signaling through the inhibition of both Akt/PKB and ERK rather than by an activation of the c-Jun N-terminal kinase-mediated death pathway.
Recent epidemiological and dietary intervention studies in both humans and animals have suggested that diet-derived phenolics, in particular the flavonoids, may play a beneficial role in the prevention of neurodegeneration, age-related cognitive and motor decline (1,2), and brain ischemia/reperfusion injury (3). For example, the neuroprotective action of one of the green tea flavonoids, epigallocatechin gallate, has been shown in both oxidative stress-(4) and A␤-induced (5) neuronal death models. Protective effects in both systems were linked to a modulation in signaling through protein kinase C and/or modulation of cell survival/cell cycle genes (4,6). Much evidence also exists to support the potential beneficial and neuromodulatory effects of flavonoid-rich ginkgo biloba extracts, such as EGb 761, in the central nervous system (7)(8)(9). Clinical trials with EGb 761, which contains kaempferol and quercetin, have indicated beneficial effects on brain function, particularly in connection with age-related dementias and Alzheimer's disease (9,10). However, although there is growing evidence in favor of the beneficial affects of flavonoids, there is still uncertainty about their actions in vivo and concern about their potential toxic side effects at higher concentrations. For instance, epigallocatechin gallate has also been shown to exert pro-apoptotic as well as anti-apoptotic effects in neuroblastoma cells (11), whereas quercetin has been observed to express no protection against A␤-induced neurotoxicity, whereas the structurally related flavonols, apigenin and kaempferol, expressed significant protection (12).
Although the flavonol quercetin is one of the most frequently researched flavonoids, with evidence for both its beneficial (13,14) and deleterious effects (15,16) on different cell types, its mechanism of action remains unclear. Furthermore, recent evidence has shown that quercetin is extensively metabolized to O-methylated and glucuronide metabolites during absorption in the small intestine and in the liver (17,18), and such metabolites should be taken into consideration to provide in vivo relevance for any mechanism. Quercetin itself has been thoroughly investigated for its abilities to express anti-proliferative effects (19,20) and induce death predominantly by an apoptotic mechanism in cancer cell lines (21)(22)(23). For example, it has been observed to induce caspase-3 activation in the malignant cell line HPB-ALL (24), activate caspase-3 and caspase-9, release cytochrome c in HL-60 cells (25), and induce chromatin and nuclear fragmentation in colonic cancer cells (20). On the other hand, quercetin treatment has been shown to suppress the c-Jun N-terminal kinase (JNK) 1 activity and apoptosis induced by hydrogen peroxide (26) and 4-hydroxy-2nonenal (27). Furthermore, quercetin may evoke anti-apoptotic effects via the suppression of the peroxide-induced JyNK-c-Jun/AP-1 pathway and the ERK-c-Fos/AP-1 pathway in cultured mesangial cells (28). The ability of quercetin to inhibit both AP-1 activation and the JNK pathway (29) has been shown to have relevance in both phorbol 12-myristate 13-ace-tate-and tumor necrosis factor-␣-induced intercellular adhesion molecule 1 (ICAM-1) expression.
The toxic effects of quercetin, or indeed other flavonoids, are likely to involve an apoptotic mode of death in which members of the mitogen-activated protein kinase (MAPK) family, such as JNK, may play a role (30). Alternatively, the modulation of signaling through the serine/threonine kinase, Akt/PKB, one of the main downstream effectors of phosphatidylinositol (PI) 3-kinase and a pivotal kinase in neuronal survival (31)(32)(33)(34), may also be important. There is considerable evidence linking the activation of the JNK pathway to neuronal apoptosis (35,36), and stress signaling through ERK has also been observed to be detrimental to neurons (37), although the latter has also been shown to be pro-survival partly through the activation of the cAMP-responsive element-binding protein (CREB) (38,39). Generally, an activation of JNK is considered pro-apoptotic, whereas activation of ERK and Akt are viewed as being prosurvival in neurons (4,31).
In the study described here, we have investigated the mechanisms underlying the neurotoxic effects of quercetin and its major in vivo metabolites, 3Ј-O-methyl quercetin, 4Ј-O-methyl quercetin, and quercetin-7-O-␤-D-glucuronide, on primary cortical neurons in terms of the balance of influence between pro-survival pathways, such as the Akt/PKB cascade, and prodeath pathways, namely apoptotic signaling through JNK and caspase-3.
Cell Culture and Quercetin Exposure-Primary cultures of mouse cortical neurons were prepared as described previously (41,42). The neurons were plated onto 6-and 24-well Nunc multiwell plates that had been precoated overnight with poly-L-ornithine and then with 10% heat-inactivated fetal bovine serum (Invitrogen) for 2 h. Following removal of the final coating solution, the cells were plated (10 6 /ml) in a serum-free medium composed of a mixture of Dulbecco's modified Eagle's medium and F-12 nutrient (1:1 v/v) supplemented with 33 mM glucose, 2 mM glutamine, 6.5 mM sodium bicarbonate, 5 mM HEPES, pH 7.4, 100 g/ml streptomycin, and 60 g/ml penicillin (all from Invitrogen). A mixture of hormones and salts composed of 25 g/ml insulin, 100 g/ml transferrin, 60 g/ml putrescine, 20 nM progesterone, and 30 nM sodium selenate (all from Sigma) was also added to the culture medium. The cells were cultured at 37°C in a humidified atmosphere of 95% air and 5% CO 2 , and after 5-7 days the vast majority of cells were neuronal (Ͼ98%) with Ͻ2% astrocytes as determined by ␤-tubulin and glial fibrillary acidic protein immunocytochemistry, respectively (not shown).
Primary cortical neurons (10 6 cells/ml; 24-well plates) were exposed to quercetin, 3Ј-O-methyl-and 4Ј-O-methyl-quercetin and quercetin-7-O-␤-D-glucuronide (0.3-10 M) for 6 h. The time dependence was also investigated by the addition of quercetin, 3Ј-O-methyl-and 4Ј-O-methyl-quercetin and quercetin-7-O-␤-D-glucuronide (3-30 M) for 0.5, 2, or 6 h. Following all exposures, the medium was removed and replaced with fresh conditioned medium, and the neurons were incubated for up to a total of 24 h (exposure and reincubation for 6 and 18 h, respectively) before assessment of neuronal viability. In addition, the assays for neuronal damage were also performed immediately following the exposures (i.e. 0.5, 2, or 6 h). For analysis of signaling proteins, the neurons (10 6 cells/ml) were grown on 6-well Nunc plates. The neurons were exposed to quercetin, 3Ј-O-methyl-quercetin, 4Ј-O-methyl-quercetin, and quercetin-7-O-␤-D-glucuronide (0.3-30 M) for 0.5, 2, or 6 h. Directly following exposure, the neurons were washed twice with ice-cold PBS, pH 7.4, containing EGTA/EDTA (200 M) and lysed on ice before preparation of samples for immunoblotting.
Assessment of Neuronal Damage-Cellular damage elicited by treatments was evaluated by measuring MTT reduction (41,43), by utilizing the Dead or Alive assay (Molecular Probes, Eugene, OR), and by morphological examination (38,41). Following exposure of neuronal cultures to quercetin and its metabolites, the cultures were washed twice with sterile PBS before the addition of MTT (0.5 mg/ml) in HEPESbuffered medium (5 mM HEPES, 154 mM NaCl, 4.6 mM KCl, 2.3 mM CaCl 2 , 33 mM glucose, 5 mM NaHCO 3 , 1.1 mM MgCl 2 , 1.2 mM Na 2 HPO 4 ). Following incubation (60 min; 37°C), MTT solutions were removed, and the formazan product was solubilized in Me 2 SO, and the absorbance read at 505 nm. The results were expressed as the percentages of protection against the decline in MTT reduction.
Cell survival was further assessed using calcein AM and ethidium homodimer (Molecular Probes). Calcein AM is taken up into living cells and is metabolized in the cytoplasm by esterases to the green fluorescent product, calcein, so that viable cells show a uniform green fluorescence (emission, 530 nm) under appropriate excitation (495 nm). Ethidium homodimer is excluded from living cells but can cross the compromised plasma membrane of dying cells and interact with nucleic acids to give a strong red fluorescence (emission, 530 nm) under appropriate excitation (495 nm). At the end of the experiment exposures, the neurons were washed, calcein AM (4 M) and ethidium homodimer 2 M) were added, and the neurons were incubated for 30 min at 37°C. The fluorescence was measured using a SPECTRAmax® Gemini microplate spectrofluorometer (Molecular Devices). The temperature was maintained at 25°C, and the emission was recorded at 530 and 645 nm for calcein and ethidium, respectively, after exciting at 495 nm. Each well was scanned in the well scan mode accumulating data from 21 independent points/well, which were then transformed in an average signal expressed in relative light units. All data were calculated and normalized with respect to the increase of fluorescence of a control. The percentages of living and dead cells were determined and analyzed statistically by Student's t test.
The morphological assessment of neuronal damage was made by analysis of phenotypic markers under light microscopy using a Nikon eclipse T5100 at 40ϫ magnification. The images were captured using a Nikon E995 digital camera fitted with a Coolpix MDC lens adaptor.
Caspase-3 Activity-The neurons (10 6 /ml) were pretreated with quercetin, 3Ј-O-methyl-quercetin, 4Ј-O-methyl-quercetin, and quercetin-7-␤-D-glucuronide (0.3-30 M) for 6 h. In a separate series of experiments, the time course of activation was investigated using a fixed 30 M exposure for 0.5, 2, and 6 h. Following exposures, the medium was removed, and the neurons were incubated with fresh conditioned medium. After a total of 12 h (exposure and reincubation for 6 and 6 h, respectively), the cells were washed twice with ice-cold PBS (ϩ EGTA 200 M) and lysed on ice as described above. The lysates were scraped from the plates and incubated on ice for 45 min before centrifugation (microcentrifuge, 1000 ϫ g), and the supernatant was collected. Caspase-3-like protease activity in neuronal lysates was assessed based on the hydrolysis of the peptide substrate acetyl-Asp-Glu-Val-Asp pnitroanilide by caspase-3, resulting in the release of the p-nitroaniline moiety, which absorbs at 405 nm. Quantification of the absorbance at 405 nm was carried out using a Versa Max UV microplate reader (Molecular Devices). Suitable control experiments were performed using the caspase-3 inhibitor acetyl-Asp-Glu-Val-Asp-al. All of the data are presented as absorbance readings at 405 nm.
Statistical Analysis-The data are expressed as the means Ϯ S.D. Statistical comparisons were made using an unpaired, two-tailed Student's t test with a confidence level of 95%. The significance level was set at p Ͻ 0.05.

Neurotoxicity of Quercetin and Its in Vivo
Metabolites-Exposure of neurons to quercetin (0.3-30 M) for 6 h resulted in a concentration-dependent increase in neuronal damage as evidenced by a decreased ability of neurons to reduce MTT when measured 24 h following the exposure (6-h exposure and 18-h reincubation) (Fig. 1A). Significant neuronal injury was observed at quercetin exposures as low as 1 and 3 M (p Ͻ 0.01), although this was increased substantially following treatment with 10 and 30 M concentrations. The two O-methylated metabolites of quercetin also induced an impairment of neuronal function that was concentration-dependent (Fig. 1A). However, the amount of damage induced was much lower than that seen with quercetin and only reached significance after exposure to 10 and 30 M (p Ͻ 0.01). Notably, the neuronal injury observed following exposure to the flavonols for 6 h was observed only 18 h after the exposure and not immediately following the 6-h exposure: quercetin, 95.7% Ϯ 3.4%; 3OmeQ, 96.3% Ϯ 2.6%; 4OmeQ, 97.9% Ϯ 1.6%; Q glucuronide, 98.3% Ϯ 1.2%; mean Ϯ S.D., n ϭ 4). Quercetin-7-O-␤-D-glucuronide did not induce toxicity at any concentration. The neurotoxicity of quercetin was also apparent at shorter exposure times. For example, quercetin (10 and 30 M) expressed significant toxicity following exposure for 0.5, 2, and 6 h, whereas quercetin (3 M) only induced significant neuronal injury following 2-and 6-h exposures (Fig. 1B). Again this toxicity was only apparent 24 h post-stress, and no damage was apparent immediately following quercetin exposure. 3Ј-O-methyl quercetin (30 M) and 4Ј-O-methyl quercetin (30 M) also induced significant neuronal injury at 2 h (p Ͻ 0.01), although this was small in comparison with quercetin (Fig. 1C).
To establish further the neurotoxic properties of quercetin, the percentage of dead and alive cells were measured following exposure. The percent of "alive" neurons was established based on the cytoplasmic esterase conversion of calcein AM to the green fluorescent product, calcein, by living cells. The percentage of "dead" neurons was estimated on the basis that ethidium homodimer is excluded from living cells and will only cross compromised plasma membranes of dying cells and interact with nucleic acids to give a strong red fluorescence. Quercetin exposure (0.3-30 M) for 6 h resulted in a concentration-dependent reduction in the percentage of viable neurons and a corresponding concentration-dependent increase in the percentage of dead cells at 24 h ( Fig. 2A). The increase in dead cells indicates the loss of membrane integrity after exposure to quercetin. However, as with the MTT measurements, damage was only observed 18 h after the completion of the flavonoid treatment with no increase apparent immediately following the quercetin 6-h exposure ( Fig. 2A), indicating that at this time there has been no alteration in neuronal membrane integrity. Morphological analyses using light microscopy also highlighted this delayed neuronal damage in that the morphological ap-pearance of cortical neurons was dramatically affected 24 h post-quercetin exposure (30 M; 6-and 18-h reincubation) but not immediately following the addition (Fig. 2B). Control cultures of cortical neurons exhibit a large intact cell body as well as a finely developed dendritic network (38,42) (Fig. 2B, panel I). Following exposure to quercetin (10 or 30 M) for 6 h and reincubation for 18 h, the neuronal cell bodies appeared shrunken, and there was a significant loss of dendrites (Fig.  2B, panels III and IV). In contrast, no significant loss of the dendritic network was observable immediately after the 6-h quercetin exposure, and similarly there was no alteration in cell body morphology (Fig. 2B, panel II).
Effects of Quercetin on Akt/PKB Signaling-To begin exploring the mechanism of quercetin-induced neurotoxicity, we examined the effect of quercetin on Akt/PKB, an essential kinase in neuronal survival and a downstream effecter of PI 3-kinase (31,32). Activation of the PI 3-kinase/Akt signaling pathway in neurons has been strongly implicated in the regulation of neuronal survival and/or protection (4,45). We investigated whether quercetin could influence Akt phosphorylation in cortical neurons by immunoblotting neuronal homogenates with an anti-phospho-Akt polyclonal antibody that detects Akt when it is phosphorylated at Ser 473 , an event known to be essential for full activation of the kinase. Exposure of cortical neurons to quercetin (10 and 30 M) for 0.5, 2, or 6 h resulted in a marked decrease in Akt phosphorylation relative to control neurons, as demonstrated by a robust decrease in the relative intensity of the immunodetectable band relating to phosphoryated Akt/ PKB (ϳ60 kDa) (Fig. 3, A and D). The changes in Akt phosphorylation were both time-and concentration-dependent. Levels of phosphoryated Akt in neurons exposed to quercetin (10 M) were greatly reduced at early exposure times (0.5 and 2 h) but showed some recovery at the 6-h time point with levels significantly higher (p Ͻ 0.01) at this time than at the earlier exposure times. In contrast, exposure to quercetin (30 M) resulted in a concentration-dependent reduction in phospho-Akt levels with pAkt undetectable following a 6-h exposure. A similar decrease in the relative intensity of the immunodetectable band relating to phosphoryated threonine 308 (within the catalytic domain of Akt) was also observed following a 2-h quercetin exposure (Fig. 3C), which matched the decrease in pAkt (Ser 473 ). Parallel immunoblots with an antibody that detects total Akt protein levels (nonphosphorylated and phosphorylated Akt) were performed. There was no change in total Akt at 0.5 h of exposure. Interestingly, there were changes in total levels of Akt in response to high quercetin concentrations and at longer exposure times (Fig. 3B), most dramatically in neurons exposed to quercetin (30 M) for 6 h.
Effect of Quercetin on Phosphorylation of ERK1/2 and JNK1/2-There was a clear concentration-dependent inhibition of pAkt (Ser 473 ) following exposure of cortical neurons to quercetin for 0.5, 2, and 6 h (Fig. 4, A and B). However, at the 6-h exposure time there was significant recovery (p Ͻ 0.01) in Akt/PKB phosphorylation toward control levels in neurons exposed to 3 and 10 M of quercetin (Fig. 4B) compared with the earlier exposure times. ERK has also been implicated in neuronal death/survival signaling (37,39). Therefore it was of relevance to investigate the effect of quercetin and its metabolites on this kinase. The phosphorylation state of ERK1/2 was probed using a phospho-specific antibody, which recognizes the dually phosphorylated motif pTEpY within activated ERK1/2. Incubation of cortical neurons with quercetin (0.3-30 M; 0.5 h) resulted in a strong dose-dependent decrease in the phosphorylation below basal levels in the two bands corresponding to ERK1 (44 kDa) and ERK2 (42 kDa) (Fig. 4C).
JNK has been implicated in pro-apoptotic signaling in neu- rons (35), and consequently we investigated the ability of quercetin and its metabolites to modulate the phosphorylation of this kinase. The phosphorylation state of JNK was investigated by immunoblotting of neuronal homogenates with anti-active JNK1/2 antibody. This antibody detects JNK when it is dually phosphoryated within the Thr 138 -Pro-Tyr 185 motif (pTPpY) in the catalytic core of active JNK. Treatment of cortical neurons with quercetin (0.3-30 M) for 2 h resulted in no measurable alteration in the relative intensities of the two immunodetectable bands, corresponding to the activated JNK isoforms, JNK (54 kDa) and JNK (46 kDa), as compared with basal levels (Fig. 4D).
The Effects of Quercetin Metabolites on the Phosphorylation of Akt and JNK1/2-When considering the effects of flavonoids in culture models, it is vital to investigate the effects of its actual in vivo metabolite forms. In contrast to the dramatic inhibition of phospho-Akt (Ser 473 ) induced by quercetin (30 M; 2 h), the two O-methylated metabolites of quercetin induced a much smaller reduction in the basal phosphorylation of Akt (Ser 473 ) (Fig. 5A). Exposure of cortical neurons to either 3Ј-Omethyl quercetin or 4Ј-O-methyl quercetin (30 M) for 2 h resulted in 67.2 and 62.4% reductions in Akt phosphorylation, respectively, relative to control cells, as demonstrated by the decrease in the relative intensity of the immunodetectable band relating to phosphoryated Akt/PKB (Ser 473 ) (Fig. 5, A and  D). In contrast, quercetin-7-␤-D-glucuronide had no significant effect on the basal state of Akt (Ser 473 ) phosphorylation in cortical neurons. In addition, the metabolites did not significantly alter the levels of total Akt in neurons following exposure (Fig. 5B), and neither quercetin nor any of its metabolites induced a change in basal levels of phospho-JNK1/2 (Fig. 5C).
Effect on the Phosphorylation of BAD and CREB-Akt maintains neuronal viability by regulating the actions of downstream effectors such as BAD and CREB (46). The pro-apo-ptotic Bcl-2 family member, BAD, was the first cell death component to be identified as a regulatory target of survival signaling (47). It has been shown to be under direct regulatory control by Akt/PKB, which specifically phosphorylates BAD at the Ser 136 (46,48) an event essential for the inhibition of apoptosis. We were interested to investigate the effect of quercetin on this important anti-apoptotic molecule, especially because a marked inhibition of Akt/PKB had been observed by quercetin. BAD phosphorylation in cortical neurons was undertaken by immunoblotting neuronal homogenates with antiphospho-BAD polyclonal antibody that detects BAD when phosphorylated at Ser 136 . Exposure of cortical neurons for 6 h to quercetin (10 or 30 M) led to a significant reduction in BAD phosphorylation (Fig. 6A). Interestingly, the inhibition of BAD phosphorylation by quercetin paralleled the changes seen in the phosphorylation state of Akt/PKB at 6 h (Fig. 3A). Parallel blots were run and probed with an antibody that detects total levels of BAD. The exposure to quercetin did not alter the total levels of BAD (Fig. 6A), which also confirmed equal protein loading. In contrast to experiments with quercetin, there were no observable changes in the BAD phosphorylation state when cortical neurons were exposed to either 3Ј-O-methyl quercetin, 4Ј-O-methyl quercetin or the glucuronidated quercetin (30 M; 6h) (data not shown). This lack of alteration in the BAD phosphorylation state in the presence of Akt/PKB inhibition by the O-methylated metabolites may reflect a requirement for a threshold of Akt deactivation to occur prior to BAD inhibition.
Another possible downstream target of Akt/PKB, and indeed ERK, is CREB, a pro-survival transcription factor of importance in neurons (39,49). We were interested in whether quercetin-induced inhibition of both Akt/PKB and ERK could transduce downstream to affect CREB phosphorylation/activation. However, exposure of cortical neurons to quercetin (3, 10, and 30 M) for 0.5 h resulted in an increase in CREB phosphorylation FIG. 3. Inhibition of Akt/PKB phosphorylation by quercetin in primary cortical neurons exposed to quercetin. A, crude lysates (20 g) prepared from cultured cortical neurons exposed to vehicle (MeOH) or quercetin (10 or 30 M) for 0.5, 2, or 6 h were immunoblotted with an antibody that specifically recognizes phosphorylated Akt (Ser 473 ). B, the same crude lysates (20 g) immunoblotted with an antibody that recognizes total levels of Akt (Total Akt). C, crude lysates (20 g) prepared from cultured cortical neurons exposed to vehicle (MeOH) or quercetin (10 or 30 M) for 2 h were immunoblotted with an antibody that specifically recognizes phosphorylated Akt (Thr 308 ). Blots B and C are representative blots of three independent experiments on different cultures that yielded similar results. D, data obtained from immunoblot experiments represented in A were analyzed using Bioimage Intelligent Quantifier software. Each column represents the mean Ϯ S.D. of four independent experiments. *, p Ͻ 0.01. relative to control neurons, as demonstrated by a significant increase in the relative intensity of the immunodetectable band relating to phosphoryated CREB (Fig. 6, B and C). The activation of CREB was most significant in neurons exposed to quercetin (3 or 10 M) for 2 h. The increase in CREB activation was unexpected and did not reflect upstream quercetin-induced Akt/PKB and ERK1/2 inhibition, although neuronal exposure to quercetin (30 M) for 2 h did result in a strong inhibition of CREB.
Caspase-3 Activation-Both Akt/PKB (31) and MAPK (35) signaling have been implicated in the activation of CED-3/ICElike proteases. Although we observed no increases in JNK activation in response to quercetin or its metabolites, it is possible that caspase-activation may occur following the reduction of Akt and BAD phosphorylation, which lead to the release of cytochrome c, the formation of the apoptosome and subsequent activation of effecter caspases such as caspase-3 and caspase-7 (reviewed in Refs. 31 and 32). We therefore investigated the potential activation of caspase-3 by the quercetin compounds. Exposure of neurons to quercetin (0.3-30 M) for 6 h led to an enhanced activity of caspase-3 in neurons, which was apparent as an increase in absorbance at 405 nm reflecting the increased cleavage of the caspase-3 specific substrate Ac-DEVD-p-nitroanalide (Fig. 7A). Levels of caspase-3 activity were increased markedly in neurons exposed to quercetin, in particular those exposed to concentrations of Ͼ3 M. In contrast, exposure of neurons to 3Ј-O-methyl quercetin and 4Ј-Omethyl quercetin led only to increases in caspase-3 activity at concentrations of 3 M and above, whereas quercetin-7-O-␤-Dglucuronide failed to stimulate significant caspase-3 activation at any concentration. Measurable increases in caspase-3 activity were detected after a 30-min exposure to quercetin (30 M), and this level increased markedly after 2 h but did not significantly differ from that measured at 6h (Fig. 7B). In contrast, the two O-methylated metabolites of quercetin induced a timedependent increase in caspase-3 activity. The glucuronide of quercetin was ineffective at inducing the activity of caspase-3 at any time point.
Effects of Caspase-3 Activation on Loss of Total Akt-Apo- FIG. 4. Phosphorylation of Akt, ERK1/2, and JNK1/2 in cortical neurons exposed to quercetin (0.3-30 M). Crude lysates (20 g) prepared from cultured cortical neurons exposed to vehicle (MeOH) or quercetin (0.3, 1, 3, 10, or 30 M) for 0.5, 2, or 6 h were immunoblotted with an antibody that specifically recognizes phosphorylated Akt (Ser 473 ). B, data obtained from immunoblot experiments represented in A were analyzed using Bioimage Intelligent Quantifier software. Each column represents the mean Ϯ S.D. of four independent experiments. *, p Ͻ 0.01. C and D, the same lysates (20 g) (quercetin 30 M; 2 h) immunoblotted with an antibody that specifically recognizes the dually phosphorylated region of the active form of ERK1 and ERK2 (pERK1/2) (C) and the dually phosphorylated region of the active form of JNK1 and JNK2 (pJNK) (D). ptotic proteases have been observed to cleave and inactivate survival signaling molecules such as Akt/PKB (50,51), phospholipase C␥1 (52), and Bcl-2. To investigate a possible link between the increases in caspase-3 activity in neurons seen in our experiments and the observed loss of total Akt in neurons exposed to quercetin (30 M), a specific caspase-3 inhibitor was employed. Pretreatment of neurons with the caspase-3 inhibitor for 15 min prior to the addition of quercetin (30 M) significantly reduced the depletion in total Akt after 2 or 6 h of exposure (Fig. 8A). This was particularly striking at 6 h, where Akt loss was most apparent. This prevention of total Akt loss was paralleled by a decrease in the activity of caspase-3 in neuronal lysates treated with the inhibitor (Fig. 8B). Furthermore, neurons pretreated with the inhibitor prior to quercetin exposure (3, 10, and 30 M; 6 h) suffered significantly less neuronal damage, as measured by their ability to reduce MTT, compared with neurons treated with quercetin alone (Fig. 8C), which was most significant at the lower quercetin exposures. DISCUSSION Much effort to characterize the potential activities of flavonoids in vivo has centered on the flavonol quercetin, found widely in the human diet and generally regarded be the flavonoid with the greatest antioxidant potential (53). However, although quercetin is a highly effective radical scavenger, its low redox potential has also been used to ascribe to its occasional pro-oxidative (54) and cytotoxic (40) behavior. We have shown that quercetin and its O-methylated metab-olites are toxic to cortical neurons via inhibition of pro-survival protein kinase cascades. Low concentrations (Ͻ10 M) were observed to induce a reversible inhibition of Akt phosphorylation and an enhancement of a potential pro-survival response in the form of increased CREB phosphorylation. In contrast, higher concentrations (30 M) induced a sustained deactivation of Akt/PKB, extensive caspase-3 activation, and subsequent caspase-dependent cleavage of anti-apoptotic Akt/PKB. This concentration-dependent effect on neuronal signaling pathways is important because the levels accumulated in vivo are unclear.
The toxicity of quercetin at higher concentrations (50 M to 100 mM) toward cells in culture, in particular cancer cells, is well established and has been linked to an apoptotic mode of death (15,16,(21)(22)(23). However, little is known regarding the precise mechanism of quercetin-induced toxicity in other cell systems, and nothing is known of its effects on neurons. In addition, it is difficult to relate this to what occurs in vivo because during absorption from the gastrointestinal tract to the circulation quercetin and its glycosides are metabolized to 3Ј-O-methyl quercetin, 4Ј-O-methyl quercetin, and quercetin-7-␤-D-glucuronide (as well as sulfates) by the action of phase I and II enzymes present in the small intestine and liver (17). These metabolic processes lead to relatively low amounts of quercetin and higher levels of the metabolites in the circulation. Furthermore, the cells in culture readily accumulate relatively high amounts of quercetin and its O-methylated metabolites, which result in further intracellular oxidative metabolism, glutathionylation, and conversion of O-methylated metabolites to free quercetin (40). Potential toxicity of flavonoids such as quercetin toward neuronal cells in vivo may be limited by the action of the blood-brain barrier preventing the entry of such compounds to the brain. However, recent evidence has demonstrated that flavonoids and some of their metabolites are able to traverse the blood-brain barrier and that the potential for permeation is consistent with compound lipophilicity (55). Indeed quercetin, and especially its O-methylated metabolites, have relatively high lipophilicity, indicating the potential brain access of these compounds.
Whereas earlier investigations into flavonoid bioactivity in vitro and in vivo concentrated mainly on their ability to act as either as free radical scavengers and antioxidants or as prooxidants, much recent interest has concentrated on their potential to interact with intracellular signaling pathways such as the MAPK cascade (29,41,56). For example, epicatechin and its 3Ј-O-methylated metabolite protect primary striatal neurons against oxidized low density lipoprotein-induced death by potently inhibiting oxidized low density lipoprotein-induced activation of JNK1/2, c-Jun, and caspase-3 (41). These data suggest that flavonoids might exert effects on neurons via direct interactions with signaling proteins in reactions independent of their antioxidant ability. The speculation that their actions may occur via interactions with neuronal signaling cascades implicated in apoptosis led us to investigate whether the observed strong neurotoxic actions of quercetin and its metabolites toward primary cortical neurons reflect their sen-sitive interactions within pro-survival MAPK and Akt/PKB signaling cascades.
These investigations provide evidence that low micromolar concentrations of quercetin, and to a lesser extent its O-methylated metabolites, are potently neurotoxic toward cortical neurons, whereas quercetin-7-O-␤-D-glucuronide has no effect. The toxicity caused by quercetin and its O-methylated metabolites was characterized both by reductions in neuronal viability and loss of membrane integrity, although both markers of neuronal damage were only observed 24 h post-exposure and not directly following the treatments. Prior to measurable losses of neuronal viability and membrane integrity, quercetin stimulated a strong inhibition of basal Akt phosphorylation in cortical neurons that was both time-and concentration-dependent. This inhibition of Akt phosphorylation was apparent at both the regulatory serine 473 and catalytic threonine 308 sites, rendering it inactive. High quercetin concentrations (30 M) led to rapid inhibition of Akt phosphorylation, which was sustained up to 6 h and was accompanied by reductions in total Akt protein levels. In contrast, lower concentrations of quercetin resulted in a transient inhibition of pAkt, with levels of pAkt (Ser 473 ) and total Akt not significantly different from control levels at 6 h, despite a strong initial inhibition of pAkt. This transient effect could be due to reversible inhibition and/or metabolism of quercetin intracellularly, as we have shown previously (40). Although the two O-methylated metabolites of quercetin also induced reductions in basal phospho-Akt (Ser 473 ), the level of inhibition was less than that caused by FIG. 6. Effect of quercetin exposure on the downstream partners of Akt and ERK. A, effect of quercetin on BAD phosphorylation. Phosphorylation of BAD in cortical neurons exposed to quercetin (10 or 30 M) for 6 h. Crude homogenates (30 g) prepared from cultured cortical neurons exposed to vehicle (MeOH), quercetin (10 M), or quercetin (30 M) for 6 h were immunoblotted with an antibody that specifically recognizes phosphorylated BAD (Ser 136 ). B, CREB phosphorylation in cortical neurons exposed to quercetin (3, 10, or 30 M) for 0.5 or 2 h. Crude homogenates (20 g) prepared from cultured cortical neurons exposed to vehicle (MeOH) or quercetin (3, 10, or 30 M) for 0.5, 2, or 6 h were immunoblotted with an antibody that specifically recognizes phosphorylated CREB (Ser 133 ). C, data obtained from immunoblot experiments represented in B were analyzed using Bioimage Intelligent Quantifier software. Each column represents the mean Ϯ S.D. of four independent experiments. *, p Ͻ 0.01. quercetin and was not accompanied by changes in total Akt. The inhibition of Akt phosphorylation by quercetin and its O-methylated metabolites paralleled the neuronal toxicity data, with the O-methylated metabolites being less neurotoxic than quercetin. This was further highlighted by the fact that the glucuronide of quercetin, too polar to enter cells, expressed no significant toxicity and no ability to inhibit the phosphorylation of Akt/PKB. The inhibition of Akt/PKB phosphorylation by quercetin seen in our neuronal system may reflect potential inhibition of its upstream partner PI 3-kinase, as has previously been described (57). This inhibition appears to be directed toward the ATP-binding site of the kinase, and analogues of quercetin such as LY294002 have been developed as potent PI 3-kinase inhibitors (58). However, the inhibition of PI 3-kinase by quercetin in neurons may be influenced by potential intracellular metabolism known to occur in other cell systems (40) and also observed in cortical neurons. 2 Oxidative metabolism and glutathionylation of quercetin and its O-methylated metabolites may act to hinder or enhance the potency of PI 3-kinase inhibition, and consequently our data cannot directly be compared with those studies conducted in a cell-free environ-ment (57,58). The potential inhibition of neuronal PI 3-kinase by quercetin and its O-methylated metabolites and by the novel intracellular metabolite 2Ј-glutathionyl quercetin (40) is the subject of further investigation.
Akt/PKB has been strongly implicated in cell survival pathways (31,32,39,59) and along with ERK has been proposed to be central to neuronal survival responses and neuroprotection (4,34). Activation of Akt in some neuronal types has been shown to lead to an inhibition of proteins central to the cell death machinery, such as the pro-apoptotic Bcl-2 family member, BAD (47), and members of the caspase-family (31, 39) that specifically cleave poly(ADP-ribose) polymerase (32,60), thus promoting cell survival. BAD is regulated by phosphorylation of two serine residues, Ser 112 and Ser 136 (47), and several studies have revealed that the Ser 136 site can be specifically phosphorylated by Akt/PKB (46,48). In our studies, quercetininduced inhibition of Akt phosphorylation was coupled with a significant loss of BAD (Ser 136 ) phosphorylation at 6 h. This inhibition of BAD phosphorylation through Akt may underlie quercetin-induced cortical neuron damage because active or un-phosphorylated BAD is known to induce apoptosis by inhibiting anti-apoptotic Bcl-2 family members such as Bcl-X L , allowing pro-apoptotic proteins such as BAX and BAK to aggregate and initiate cytochrome c release and subsequent caspaseactivation (61,62).
Marked activation of caspase-3 was also observed in cortical neurons exposed to quercetin. As well as contributing directly to the apoptotic mechanism, it was postulated that this distinct activation of caspase-3 by quercetin might result in the processing of Akt/PKB seen during high quercetin exposures (30 M). Apoptotic proteases have been observed to cleave and inactivate survival-signaling molecules such as Akt/PKB (50,51), phospholipase C␥1 (52), and Bcl-2. In our experiments, cortical neurons exposed to quercetin in the presence of a specific caspase-3 inhibitor did not suffer the same reduction in total Akt/PKB protein levels at 6 h. Concurrently, there was a reduction in the overall level of caspase-3 activity in neurons exposed to quercetin and the inhibitor compared with neurons exposed to quercetin alone. Furthermore, the toxicity induced by quercetin was reduced in the presence of the caspase-3 inhibitor.
The activation of Akt/PKB and ERK1/2 in neurons has also been linked at the transcriptional level to the phosphorylation of CREB (Ser 133 ) (39, 49) a transcription factor linked to prosurvival through the up-regulation of genes such as BDNF and Bcl-2. An inhibition of Akt activation by quercetin might therefore be expected to lead to a reduction in the phosphorylation of CREB at Ser 133 , a site critical for its activation. Indeed at the higher concentration of quercetin (30 M) at 2 h, we observed inhibition of CREB phosphorylation. However, lower level quercetin exposures (3 and 10 M) resulted in an increased phosphorylation of CREB, indicating that CREB activation by quercetin may occur by a separate mechanism or that high concentrations trigger dephosphorylation. ERK has also been shown to be pro-survival partly through the activation of CREB (38,39). The fact that we observe an activation of CREB at concentrations of quercetin where we see potent inhibition of ERK1/2 strengthens the concept that quercetin-induced activation of CREB is likely to be through a pathway independent of PI3K/Akt or ERK cascades, for example through calcium/ calmodulin kinase or protein kinase A. However, inhibition of ERK phosphorylation by quercetin may contribute to the observed neuronal damage and may also reflect upstream inhibition of PI 3-kinase, which is known to signal to ERK in neurons (38,63).
In contrast to the marked effects of quercetin on survival signaling through Akt/PKB and ERK, neither quercetin nor any of its metabolites affected the pro-apoptotic JNK pathway at any concentration. This is interesting because there is considerable evidence linking the activation of the JNK pathway to neuronal apoptosis (35,36), and consequently the lack of activation of JNK by quercetin suggests that this pathway is not a potential route to the observed cortical neuron damage observed in our system. This strengthens the prospect of a mechanism that acts through inhibition of survival signaling rather than stimulation of death signaling. Quercetin has been observed to suppress JNK activity induced by hydrogen peroxide (26, 28) and 4-hydroxy-2-nonenal (27), suggesting that quercetin is able to prevent stress-induced activation of JNK but not basal activation. This agrees with previous investigations where the flavonoids epicatechin, 3Ј-O-methyl epicatechin, and kaempferol inhibited JNK activation by oxidized low density lipoprotein but had no effect on basal JNK phosphorylation (41).
Together these data suggest that quercetin, and to a lesser extent its O-methylated metabolites, may induce neuronal death via a mechanism involving direct inhibition of survival signaling through Akt/PKB and ERK rather than by an induction of the JNK-mediated death pathway. The observation of CREB activation in neurons, where we also observe potent inhibition of Akt and ERK and inactivation of BAD, indicates that both pro-apoptotic and potentially anti-apoptotic path- FIG. 8. Effect of the specific caspase-3 inhibitor on levels of total Akt, caspase-3 activation and neuronal damage in neurons exposed to quercetin. A, crude homogenates (20 g) prepared from cultured cortical neurons exposed to vehicle (MeOH), quercetin (30 M), and quercetin (30 M) and inhibitor for 2 or 6 h were immunoblotted with an antibody that recognizes total levels of Akt (Total Akt). B, increased activity of caspase-3-like proteases at 2 and 6 h induced by quercetin (30 M) and inhibition of caspase-3 activation in the presence of the specific caspase-3 inhibitor. Treatment with compounds was for 6 h, and caspase-like protease activity was assessed 12 h after exposure. Capsase-3 was assayed as described in the legend for Fig. 6C. ways are activated in neurons in response to quercetin stimulus. However, because overall neuronal death results, it appears that quercetin-induced inhibition of the Akt/BAD survival pathway is dominant here in determining the fate of the neurons. We propose that in our cells, high concentrations of quercetin produce a sustained deactivation of Akt/PKB, which leads to extensive caspase-3 activation and subsequent caspase-dependent cleavage of anti-apoptotic Akt/PKB, an event that effectively turns off the major survival signal and results in the acceleration of apoptotic neuronal death. However, at lower concentrations reversible inhibition of Akt phosphorylation is observed, and there is evidence of an attempted survival response reflected in the increase in CREB phosphorylation. Thus, not all flavonoids should be simply regarded as being potentially beneficial in the treatment of neurological disease, and due consideration must be given to the in vivo concentrations of flavonoids and to the bioactivity of relevant in vivo metabolites when studying the effects of dietary flavonoids in cell culture models.