and

Flavonoids exhibit prooxidant cytotoxicity in mamma-lian cells due to the formation of free radicals and oxidation products possessing quinone or quinomethide structure. However, it is unclear how the cytotoxicity of flavonoids depends on the ease of their single-electron oxidation in aqueous medium, i.e., the redox potential of the phenoxyl radical/phenol couple. We verified the previously calculated redox potentials for several flavonoids according to their rates of reduction of cy-tochrome c and ferricyanide, and proposed experimentally based values of redox potentials for myricetin, fisetin, morin, kaempferol, galangin, and naringenin. We found that the cytotoxicity of flavonoids (n=10) in bovine leukemia virus-transformed lamb kidney fibro-blasts (line FLK) and murine hepatoma (line MH-22a) increases with a decrease in their redox potential of the phenoxyl radical/phenol couple and an increase in their lipophilicity. Their cytotoxicity was decreased by antioxidants and inhibitors of cytochromes P-450, α-naphthoflavone and isoniazide, and increased by an inhibitor of catechol-O-methyltransferase, 3,5-dinitro-catechol. It shows that although the prooxidant action of flavonoids may be the main factor in their cytotoxic-ity, the hydroxylation and oxidative demethylation by cytochromes P-450 and O-methylation by catechol-O-methyltransferase can significantly modulate the cyto-toxicity of the parent compounds.


INTRODUCTION
Flavonoids are universally recognized antioxidants which can protect the cell from the oxidative stress, i.e., neutralise the damaging effect of reactive oxygen species (ROS).However, at high concentrations, flavonoids and other polyphenols may be cytotoxic, causing an increase in mitochondrial permeability, cytochrome c release, the activation of caspases, an increase in levels of p53 and p21, the suppression of Bcl-2, apoptosis induction, and necrotic cell death (Bolton et al., 2000;Inayat-Hussain et al., 2001;Morin et al., 2001;Salvi et al., 2002;Shen et al., 2004;Lee et al., 2011).
The structural factors determining the prooxidant cytotoxicity of flavonoids are not well understood.One may expect that the cytotoxicity of flavonoids, if determined by their prooxidant action, may increase with a decrease in the redox potential of the semiquinone/ hydroquinone (or phenoxyl radical/phenol) couple (E 7 (Q •-/QH 2 )).This relationship has been demonstrated for a related group of antioxidants, polyhydroxybenzenes (Nemeikaitė-Čėnienė et al., 2005;Grellier et al., 2008).It reflects the relative ease of formation of prooxidant oxidation products or ROS by hydroquinones and other polyphenols, because the rates of their oxidation by cytochromes, Fe 3+ , ferritin, or oxygen increase with a decrease in their E 7 (Q •-/QH 2 ) (Rich & Bendall, 1980;Rich, 1982;O'Brien, 1991;Hynes & Coincemainn, 2002).However, the values of E 7 (Q •-/QH 2 ) have been determined by pulse-radiolysis for a few flavonoids only, and for several other flavonoids redox potentials based on calculations are available (Jovanovic et al., 1998, and references cited therein).This hampers further studies in this direction.
In this study, we examined the reactivity of a series of flavonoids (Fig. 1) and model hydroxybenzenes possessing a broad range of E 7 (Q •-/QH 2 ) values towards the single-electron oxidants cytochrome c and ferricyanide.The obtained dependences enabled us to propose experimentally-based E 7 (Q •-/QH 2 ) values for several flavonoids.Further, we demonstrated an increase in the mammalian cell cytotoxicity of flavonoids upon a decrease in their potential of the phenoxyl radical/phenol redox couple.

MATERIALS AND METHODS
Chemicals.Flavonoids (Fig. 1), cytochrome c, other enzymes and chemicals were obtained from Sigma-Aldrich and used as received.
Kinetic studies.The kinetic experiments were carried out spectrophotometrically in 0.1 M K-phosphate buffer (pH 7.0), containing 1 mM EDTA, at 25°C.The reduction of cytochrome c by myricetin, fisetin, and morin was monitored using a DX.17MV stopped-flow spectrophotometer (Applied Photophysic) assuming ∆ε 550 = 20 mM - 1 cm -1 .The concentrations of flavonoids and cytochrome c after mixing were equal to 20-100 µM and 4.0 µM, respectively.The apparent first order rate constants (k app ) were obtained from the analysis of the kinetics of the absorbance increase according to a single-exponent fit, using the software supplied by Applied Photophysics.The second order rate constants (k) were calculated from the slopes of the linear dependences of k app on the reductant concentration.Several experiments were performed with 500 µM cytochrome c, and 20 µM of each flavonoid, using a 0.2-cm optical path cell.For other compounds, the bimolecular reaction rate constants were calculated from the initial rates of reduction of the excess cytochrome c (500-150 µM) by 20 µM polyphenol, using a Hitachi-557 spectrophotometer and a 0.2-cm optical path cell.When the initial reaction rates were measured, corrections were introduced in several cases for the absorbance of polyphenol oxidation products at 550 nm: ∆ε 550 =0.8 mM -1 cm -1 (catechin), ∆ε 550 =0.28 mM -1 cm -1 (taxifolin, caffeic acid), and ∆ε 550 =0.2 mM -1 cm -1 (ethylgallate, catechol).The absorbance of the oxidation products at 550 nm was determined after the oxidation of the polyphenols by excess ferricyanide.
The reduction of ferricyanide by flavonoids and hydroxybenzenes was monitored at 420 nm using excess ferricyanide (final concentrations, 0.4-1.8mM) over the reductant (final concentrations, 0.05-0.2mM).Additional measurements were performed by monitoring the kinetics of the decrease of the absorbance of the reductant at 370 nm (myricetin, quercetin, morin), 365 nm (kaempferol, fisetin), and 320 nm (hesperetin).The reaction rate constants above 10 M -1 s -1 were determined using a stopped-flow spectrophotometer.Because the oxidation of polyphenols is sometimes accompanied by the slower formation of secondary products (Terland et al., 2006), only the initial stage of the processes was analyzed according to a single-exponent fit.For the slowly reacting compounds, the reaction rates were measured using a Hitachi-557 spectrophotometer.In those cases, the k app values were determined according to the Guggenheim method from the plots ln ∆A vs. t, where ∆A is the absorbance changes at equal time intervals (Connors, 1990).The second order rate constants were calculated from the plots k app vs. ferricyanide concentration.
Cell culture cytotoxicity studies.Cultures of bovine leukemia virus-transformed lamb kidney fibroblasts (line FLK) and murine hepatoma (line MH-22) were grown and maintained at 37°C in Eagle's medium or in DMEM medium, respectively, supplemented with 10% fetal bovine serum and antibiotics, as described (Nemeikaitė-Čėnienė et al., 2005;Grellier et al., 2008).In the cytotoxicity experiments, cells (2.5×10 4 /ml FLK, and 3.0×10 4 /ml MH-22a) were seeded on 18×18 mm glass slides in 5-ml flasks either in the presence or in the absence of compounds, and were grown for 24 h.Then, the slides were rinsed 3-4 times with phosphate buffered saline and stained with Trypan blue.The cells adherrent to the slides were counted under a light microscope.Typically, they did not accumulate Trypan blue and their viability was 98.5-99.3%.Stock solutions of poorly soluble compounds were prepared in dimethyl sulfoxide.Its concentration in cultivation media did not exceed 0.2%, and did not affect cell viability.The experiments were conducted in triplicate.
Log P calculation, statistical analysis.The octanol/ water partition coefficients of flavonoids (log P) were calculated using the ACD/ChemSketch software (version 4.02, Advanced Chemistry Development, Toronto, Ontario, Canada), while statistical and multiparameter regression analysis was performed using Statistica (version 4.3, Statsoft Inc., 1993).

Kinetics of reduction of cytochrome c and ferricyanide by flavonoids and hydroxybenzenes
Among the flavonoids examined (Fig. 1), the redox potentials for quercetin, taxifolin, catechin, and hesperetin were obtained directly from pulse-radiolysis studies, whereas the values of E 7 (Q •-/QH 2 ) for other flavonoids (Table 1) are based on calculations (Jovanovic et al., 1998).In order to verify the calculated E 7 (Q •-/QH 2 ) values, we examined the reactions of the flavonoids and a number of model polyhydroxybenzenes with the E 7 (Q •-/QH 2 ) values from 0.33 V to 0.73 V (Table 1), with two single-electron oxidants, cytochrome c and ferricyanide.According to the model of an ‚outer-sphere' electron-transfer, one may expect a linear dependence of log (rate constant) on the difference between the redox potential of a single-electron oxidant and a series of homologous reductants, e.g., polyphenols, if the reactions are endothermic or modestly exothermic (Marcus & Su-tin, 1985).A similar approach, based on linear log (rate constant) vs. redox potential dependences was used in the determination of unknown values of single-electron reduction potentials for nitroaromatic compounds (Čėnas et al., 2009;Uchimiya et al., 2010, and references therein).
decrease at 420 nm.However, the absorbance changes at 420 nm during ferricyanide reduction by myricetin, quercetin, morin, kaempferol, fisetin, and ethylgallate followed the first order kinetics for more than 6-8 reaction half-times.In this time scale, the kinetics of oxidation of myricetin, quercetin, morin, kaempferol, and fisetin monitored at their λ max , 365-370 nm, followed the first order as well.Their k app values were identical within the experimental error to those obtained at 420 nm.On the other hand, the biphasic absorbance changes at 420 nm during the reduction of ferricyanide by taxifolin, catechin, and catechol were more pronounced (not shown).In those cases, only the initial stages of the process, 3-4 reaction half-times, were analyzed according to a single-exponent fit.The oxidation of other slowly reacting polyphenols also followed the first order kinetics.The data of Table 1 show that the reactivity of hydroxybenzenes and flavonoids towards ferricyanide increased with a decrease in their experimentally determined E 7 (Q •-/QH 2 ) values with ∆log k/∆E 7 (Q •-/QH 2 )=-7.47±1.11V -1 (r 2 =0.835,F(1,9)=45.41)(Fig. 2B).Again, like in the reduction of cytochrome c (Fig. 2A), the reactivities of morin and kaempferol were much higher than expected (Fig. 2B).It shows that these deviations are not caused by the specificity of the particular flavonoids towards cytochrome c, but most probably by the overestimation of calculated E 7 (Q •-/QH 2 ) values for morin and kaempferol (Jovanovic et al., 1998).
Concerning the obtained E 7 (Q •-/QH 2 ) (calc.) values for flavonoids, the calculated redox potentals for naringenin and galangin (Table 1) are close to that of resorcinol, 0.81 V, which reflects the ease of oxidation of the resorcinol group in the A-ring of flavonoids (Jovanovic et al., 1998).The E 7 (Q •-/QH 2 ) (calc.) for myricetin and fisetin (Table 1) are close to their previously calculated values (Jovanovic et al., 1998).In contrast, the E 7 (Q •-/QH 2 ) (calc.) values for morin and kaempferol (Table 1) are much more negative than those suggested previously (Jovanovic et al., 1998).In our opinion, the previous calculations (Jovanovic et al., 1998) could have underestimated the effects of charge delocalization in flavonol anionradicals, which may result in their stabilization and a decrease in their E 7 (Q •-/QH 2 ) values (Scheme 1).The quinone/quinomethide tautomerisation of flavonoid oxidation products was experimentally demonstrated in quercetin oxidation (Boersma et al., 2000;Awad et al., 2002).This way of stabilization is not characteristic for other groups of flavonoids.The obtained low E 7 (Q •-/QH 2 ) values for morin and kaempferol are in line with their voltammetric characteristics.Although morin and kaempferol do not possess an electrochemically reversible catechol group in B-ring, their electrochemical oxidation is reversible, with the voltammetric midpoint potentials at pH 7.4 being equal to 0.34 V and 0.39 V, respectively, which is close to the midpoint potential of quercetin, 0.29 V (Jorgensen & Skibstedt, 1998).This may be explained by the formation of quinomethides as two-electron oxidation products (Scheme 1).In contrast, an analogue of kaempferol, the flavone apigenin, which lacks 3-OH group in C-ring, is oxidized irreversibly with the peak potential at 0.71 V (Jorgensen & Skibstedt, 1998).Thus, the values of E 7 (Q •-/QH 2 ) of flavonoids obtained in this work may be considered as reasonable approximations.
ing that the cytotoxicity of flavonoids in the three cell lines increases with a decrease in their E 7 (Q •-/QH 2 ) (calc.) values (Eqns.2-4, Fig. 4A, B).Taken together with the protective effects of the antioxidants (Fig. 3A, B), this indicates that the oxidative stress may be a key factor for flavonoid cytotoxicity.An increase in the cytotoxicity of flavonoids with an increase in their lipophilicity (Eqsn.2-4) points to the importance of their intracellular accumulation.The obtained QSARs and the effects of antioxidants, prooxidants, and inhibitors of cytochromes P-450 and COMT (Figs. 3,4) for flavonoids are similar to those observed in the action of polyhydroxybenzenes in the same cell lines (Nemeikaitė-Čėnienė et al., 2005;Grellier et al., 2008).It shows that both groups of polyphenolic antioxidants may share the same main mechanism(s) of prooxidant cytotoxicity.
On the other hand, the dependence of flavonoid cytotoxicity on their oxidation potential is not strongly expressed, because the coefficients ∆log cL 50 /∆E 7 (Q •-/ QH 2 ) (calc.) in Eqns.2-4, 1.8-2.2V -1 , are lower than those describing the cytotoxicity of polyhydroxybenzenes in the same cell lines, 5.1-6.9V -1 (Nemeikaitė- Čėnienė et al., 2005;Grellier et al., 2008).The latter coefficients closely match the order of reactivity of polyphenols towards single-electron oxidants, ∆log k/∆E 7 (Q •-/QH 2 )=~ -8.5 V -1 (Rich & Bendall, 1980;Rich, 1982).Thus, apart from the ease of formation of prooxidant oxidation products or ROS, the cytotoxicity of flavonoids may be affected by other factors.A possible explanation is the interconversion of flavonoids under the action of cytochromes P-450 and COMT (Duarte Silva et al., 1997;Nielsen et al., 1998;Lautala et al., 2002;Lee et al., 2005), which may attenuate the expected dependence of cytotoxicity of flavonoids on their E 7 (Q •-/QH 2 ).Typically, O-methylation of catechols decreases their cytotoxicity, evidently due to an increase in their E 7 (Q •-/QH 2 ), whereas their hydroxylation increases the autooxidation rate and cytotoxicity (Moridani et al., 2002a;2002b).Although myricetin is not a substrate for cytochromes P-450 (Nielsen et al., 1998), and galangin does not possess hydroxy groups in B-ring, the data of Fig. 3A, B show that their cytotoxicity is modulated by inhibitors of cytochromes P-450 and COMT to a similar extent.Thus, O-methylation of myricetin by COMT (Lee et al., 2005) may be followed by the subsequent oxidative demethylation of the reaction products by cytochromes P-450.In turn, cytochromes P-450 may convert galangin into kaempferol, and, subsequently, into quercetin (Fig. 1) (Duarte Silva et al., 1997), which may be followed by O-methylation of quercetin by COMT (Lee et al., 2005).This is in line with the sufficiently close cL 50 values for galangin, kaempferol, and quercetin in FLK cells (Table 2).Hesperetin and naringenin (Fig. 1) may also undergo the cytochrome P-450-catalyzed conversion into eriodictyol (Hodek et al., 2002), which may be also partly responsible for the similar cytotoxicity of the above flavanones (Table 2).Because flavonoids are O-methylated much faster than catechols (Lautala et al., 2002), the action of COMT may decrease their cytotoxicity more efficiently.This may explain a less pronounced role of redox potential in the cytotoxicity of flavonoids as compared to that of polyhydroxybenzenes (Nemeikaitė-Čėnienė et al., 2005;Grellier et al., 2008).

CONCLUSIONS
Our studies have shown that the reactivity of flavonoids and related polyphenolic antioxidants with model single-electron oxidants may be a useful tool to characterize their E 7 (Q •-/QH 2 ).Subsequently, this parameter may be used in the prediction of the cytotoxicity of flavonoids to the mammalian cell, which seems to be caused mainly by their prooxidant action.Another important factor to be considered is the reactions of flavonoids with cytochromes P-450 and COMT, which may significantly attenuate the dependence of the cytotoxicity of flavonoids on their E 7 (Q •-/QH 2 ).The values of redox potentials of phenoxyl radical/phenol couples of flavonoids at pH 7.0 calculated according to Eqn. (1) (E 7 (Q •-/QH 2 ) (calc.) ), the octanol/water partition coeficients of flavonoids (log P), and their concentrations for 50% survival (cL 50 ) of FLK, MH-22a, and HL-60 cells.

Compound
E 7 (Q
are shown in parentheses.b From Nemeikaitė-Čėnienė et al., 2005.c Determined from the absorbance changes at 420 nm.