Specific Structural Determinants Are Responsible for the Antioxidant Activity and the Cell Cycle Effects of Resveratrol*

,

Resveratrol (3,4Ј,5-trihydroxystilbene) is synthesized by several plants in response to adverse conditions such as environmental stress or pathogenic attack. For this reason, it is classified as a phytoalexin, a class of antibiotics of plant origin (1)(2)(3). Resveratrol has been found in a multitude of dietary plants, such as peanuts, mulberries, and in grape skin (4). Thus, relatively high concentrations of this compound are present in grape juice and, especially, in red wine (5)(6)(7)(8). Growing evidence suggest that resveratrol plays a role in the prevention of human pathological processes, such as inflammation (9 -11), atherosclerosis (12)(13)(14), and carcinogenesis (4,15,16). The protective effect has been attributed to its antioxidant properties (11,17,18), to an anticyclooxygenase activity (4,20), and to a modulating activity of lipid and lipoprotein metabolism (9,21,22). Resveratrol also inhibits platelet aggregation (13,23) and exhibits antiestrogenic activity (24,25). However, these effects do not exhaustively explain the antiproliferative and anticarcinogenic properties of resveratrol. The proliferation of various human malignant cell lines is slowed down by resveratrol (15,16,26). The inhibition of cell growth, which has also been described in normal cells (27)(28)(29), is accompanied by the accumulation of cells in S and G 2 phases (16,26,30). Conflicting results have been reported on the induction of apoptosis by resveratrol (26,31,32).
A number of antioxidants, such as vitamin E, N-acetylcysteine, flavonoids, and carotenoids have been reported to interfere with cell cycle progression by inducing the expression of cdk inhibitors, like p21 waf1-cip1 , p16 ink4a , and p27 kip1 (33)(34)(35)(36). In the case of resveratrol, such a mechanism is still controversial (26,27). The effects on cell cycle progression may be also explained by the direct inhibition of ribonucleotide reductase (37) and DNA polymerase (38).
The structural determinants of these diverse properties of the resveratrol molecule are obscure, but the number and position of the hydroxylic groups have been suggested to play an important role in the antioxidant activity (14,19,39). The aim of this study was to extend these studies on the structural determinants of the activity of resveratrol, and in particular to establish whether the antioxidant and antiproliferative activities are dependent on (i) the stereoisomery, (ii) the position of the different phenolic hydroxyl groups, and (iii) the stilbenic double bond of the molecule. For this purpose, the cis form (II) was obtained by UV irradiation of trans-resveratrol; three different derivatives were synthesized in which the hydroxylic functions were selectively protected by methyl groups: 3,5dihydroxy-4Ј-methoxystilbene (III), 3,5-dimethoxy-4Ј-hydroxystilbene (IV), and 3,4Ј,5-trimethoxystilbene (V). Finally, the ␣,␤-dihydro-3,4Ј,5-trihydroxystilbene (VI) was obtained by reduction of the stilbenic double bond.
The biological properties of trans-resveratrol were compared with those of the above derivatives. In particular, the antioxidant activity was investigated in vitro by measuring the inhibition of citronellal thermo-oxidation or the radical scavenging ability using the free radical DPPH. 1 The protection against lipid peroxidation induced by Fe/ascorbate and tert-butylhydroperoxide TBHP was also assessed in rat liver microsomes, or in cultured human fibroblasts, respectively. The effects on cell proliferation were studied by analyzing the cell clonogenic efficiency and cell cycle progression. In addition, the recruitment of proliferating cell nuclear antigen (PCNA) and replication protein A (RPA) to the DNA replication sites were investigated. These proteins are required for the initiation and elongation steps of DNA replication, respectively. Finally, the ability of resveratrol and its derivatives to inhibit replicative DNA polymerases was also assessed with in vitro assays.
The results have shown that the hydroxyl group at the 4Ј position is not the sole determinant for the antioxidant activity. Similarly, the 4Ј-hydroxyl group is necessary for the antiproliferative activity and the DNA polymerase inhibition, but the trans conformation is absolutely required for these effects.
Poly(dA)/oligo(dT) 12-18 primer-template was prepared according to the manufacturer's protocol. Briefly, poly(dA) template oligonucleotide was mixed with the complementary oligo(dT) 12-18 oligonucleotide in a 10:1 molar ratio (w/w, nucleotides) in 20 mM Tris-HCl (pH 8.0), containing 20 mM KCl and 1 mM EDTA, heated at 90°C for 5 min then incubated at 65°C for 2 h and slowly cooled at room temperature.
Stock solutions of each substance were prepared in N,N-dimethylformamide and final dilution was performed in chlorobenzene for the in vitro oxidation test. For cell culture experiments, stock solutions were prepared in Me 2 SO and diluted directly in cell culture medium.

Cell Culture and Treatments
Normal human embryonic fibroblasts and HT1080 fibrosarcoma cells (Istituto Zooprofilattico, Brescia, Italy) were cultured in Earle's minimal essential medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (Life Technologies, Inc.), 100 IU/ml penicillin, and 100 g/ml streptomycin. Normal fibroblasts were used between the 5th and 20th passages. Cell treatments were performed by adding transresveratrol or its derivatives in culture medium at final concentrations ranging from 7 to 100 M. For resveratrol, these concentrations are comparable with doses found in red wine and grapes (39). Untreated cultures received the same amount of the solvent alone (Me 2 SO Ͻ 0.1%). Aphidicolin (60 nM) was used as positive control for cell cycle arrest in S phase. For the antioxidant activity determination, normal fibroblasts were incubated for 30 min with 60 M trans-resveratrol or its derivatives, before the oxidative treatment. For cell cycle analysis, normal fibroblasts and HT1080 cells were incubated for 24, 48, and 72 h with trans-resveratrol at the concentration of 15, 30, and 90 M, then cells were washed twice with phosphate-buffered saline (PBS), and detached with a standard trypsinization procedure.

Antioxidant Activity
The antioxidant activity of trans-resveratrol and its derivatives was evaluated in vitro both by the citronellal thermo-oxidation inhibition test (46), and the DPPH method (47,48). In addition, antioxidant activity was assessed in rat liver microsomes by measuring lipid peroxidation inhibition after Fe 2ϩ /ascorbate treatment, and in human fibroblast cultures after TBHP treatment.

Citronellal Thermo-oxidation Method
In this test, the aldehyde (Ϫ)-citronellal is used as the oxidation substrate: it is subjected to heating and intensive oxygenation in chlorobenzene, and its disappearance with the consequent formation of its degradation products are monitored by gas chromatography. Chlorobenzene was selected as the reaction solvent because of (i) its stability to oxidation; (ii) the ability to dissolve both polar and nonpolar compounds, better than dimethylformamide; and (iii) the boiling point is higher then 80°C (temperature test). Fifteen ml of a chlorobenzene solution, containing 150 l of dodecane (Aldrich) and 150 l of tridecane (Aldrich) as internal standards, were poured into a two-necked flask equipped with a condenser to prevent evaporation. Resveratrol, or its derivatives, dissolved in dimethylformamide, were added to the chlorobenzene solution to reach final concentrations ranging from 60 to 120 M. The mixture was then heated at 80°C and intensively oxygenated by bubbling in O 2 at a flow rate of 10 ml/min. At time 0, 300 l of (Ϫ)-citronellal (Fluka) were added to the reaction medium. Immediately and at periodic intervals, 0.1-l samples were withdrawn and analyzed by gas chromatography. The antioxidant power of resveratrol and the derivatives was measured by determining the efficient quantity (EQ), i.e. the concentration required for each compound to double the half-life with respect to control reaction (citronellal without antioxidant).

DPPH Reduction Method
Antioxidant solution in methanol (0.1 ml) was added to 3.9 ml of a 6 ϫ 10 Ϫ5 M DPPH solution in methanol (48). The exact initial DPPH concentration in the reaction medium was calculated from a calibration curve. The decrease in absorbance was determined at 515 nm at 0 min, every 5 min for 1 h, and every 60 min until the reaction reached a plateau (about 6 h). Antiradical activity was expressed as the EC 50 , i.e. the antioxidant concentration necessary to decrease the initial amount of DPPH by 50%.

Lipid Peroxidation in Rat Liver Microsomes
Rat liver microsomes were prepared from Wistar rats by tissue homogenization with 5 volumes of ice-cold 0.25 M sucrose containing 5 mM Hepes, 0.5 mM EDTA (pH 7.5). Briefly, microsomal fractions were isolated by removal of the nuclear fraction at 8,000 ϫ g for 10 min, removal of mitochondrial fraction at 18,000 ϫ g for 10 min, and sedimentation at 105,000 ϫ g for 60 min. Microsomal fractions were diluted in phosphate buffer, 0.1 M (pH 7.5), at the final protein concentration of 1 mg/ml. The microsomes were preincubated in a shaking water bath at 37°C for 10 min with varying concentrations (0.01-100 M) of each compound before starting lipid peroxidation with 100 M Fe 2ϩ , 500 M ascorbate. After 60 min incubation, the inhibitory effect on lipid peroxidation was assessed by measuring thiobarbituric acid reactive substances (TBARS) by the method of Yagi et al. (49). Briefly, 500 l of microsomal fraction were added to 500 l of 20% trichloroacetic acid to stop the lipid peroxidation reaction, and then 500 l of 0.74% thiobarbituric acid was added. The mixture was then heated in a boiling water bath for 15 min. After centrifugation, 200 l of supernatant was transferred to the microtiter plate, the absorbance was measured at 535 nm and compared with standards prepared from the acid hydrolysis of malonaldehyde tetraethylacetyl (Sigma). Inhibition of lipid peroxidation was expressed as percentage, and the effective concentration giving 50% inhibition (EC 50 ) was calculated from the inhibition curve.

Lipid Peroxidation in Cell Cultures
Lipid peroxidation was induced in cultured normal human fibroblasts by TBHP (Sigma). Cells were preincubated for 30 min with 60 M trans-resveratrol or its analogues in PBS, and then 250 M TBHP was added for 60 min. The production of TBARS was assessed, as above described. In both experimental models, microsomes and cells, TBHPtreated and untreated control samples received the same amount of the solvent (Me 2 SO Ͻ 0.02%).

Clonogenic Efficiency Assay
The clonogenic efficiency was determined after incubation of cells in culture medium containing trans-resveratrol or its derivatives. Briefly, the cells were diluted in complete Earle's minimal essential medium to ϳ4000 cells/ml, and volumes of this suspension containing 200 cells were transferred to 30-mm dishes. After 24 h of treatment with transresveratrol and the different derivatives, the cells were washed twice and 5 ml of fresh medium was added. After 10 days, the colonies were stained with crystal violet and counted, and the clonogenic efficiency was calculated as the mean percentage with respect to control cells. Clonogenic efficiency of untreated control cultures was about 35%.

Cell Cycle Analysis
Cell cycle distribution was assessed by determining BrdUrd incorporation versus DNA content. Normal fibroblasts and HT1080 cells were incubated with 30 M BrdUrd (Sigma) during the last hour of culture, harvested, and fixed in cold 70% ethanol. Fixed cells were washed in PBS, resuspended in 2 N HCl for 30 min at room temperature, pelletted, and then resuspended in 0.1 N sodium tetraborate for 15 min. The samples were then washed in PBS, incubated for 15 min in PBS containing 1% bovine serum albumin and 0.2% Tween 20 (PBT), and then for 60 min in 100 l of anti-BrdUrd monoclonal antibody diluted 1:20 in PBT. After two washes with PBT, cells were incubated for 30 min with 100 l of FITC-conjugated anti-mouse antibody diluted 1:50 in PBT, then washed twice, and resuspended in PBS containing 5 g/ml propidium iodide and 2 mg/ml RNase A. Cells were analyzed with a Coulter Epics XL (Coulter Corp.) flow cytometer. Ten thousand cells were measured for each sample. Computer statistical analysis of mean fluorescence intensity (MFI) and graphic representation were performed with the XL2 software (Coulter).

Immunofluorescence Determination of Nuclear-bound PCNA and RPA (32 kDa)
In order to determine the nuclear-bound (detergent-insoluble) fraction of protein involved in the DNA replication complex, a hypotonic lysis was performed according to the following protocol (50). Briefly, cells were resuspended in hypotonic lysis buffer 10 mM Tris-HCl (pH 7.4), containing 2.5 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride (BDH), and 0.5% Nonidet P-40 (Sigma). After the lysis was completed, cells were resuspended in saline, fixed in cold ethanol to a final 70% concentration, and stored at Ϫ20°C until analysis. . All reactions were incubated for 15 min at 37°C unless otherwise stated and the DNA precipitated with 10% trichloroacetic acid. Insoluble radioactive material was determined as described (51).
Inhibition Assays-Assays were performed under the conditions described above. Different concentrations of the inhibitors to be tested were added as indicated in figure legends to the reaction mixture in the absence of DNA template and nucleotides. After 5 min of incubation at room temperature, the reaction was started by addition of the missing reagents and incubation was as described above. K i values were calculated by Dixon plot of the experimental data.

Electronic Structure and Thermodynamic Stability of the Phenoxyl Radicals from Resveratrol
The electronic structure and the formation enthalpy of the three different phenoxyl radicals arising from the loss of hydrogens at the 3-, 5-, and 4Ј-OH groups in resveratrol were determined by semiempirical Molecular Orbital calculations using to the MNDO-PM3 semiempirical method. Full optimization of molecular geometries were achieved using a gradient best fit procedure with an energy convergence criterium of 10 Ϫ4 kcal/mol. The imput geometries were determined by best fit using a Molecular Mechanics method (all methods implemented in the Hyperchem package, release 5.1).

Statistics
All experiments were performed at least three times. Experimental data were analyzed with one-way analysis of variance (ANOVA) followed by Tukey's multiple range test for significant differences. Fig. 1 shows the chemical structures of trans (I) and cisresveratrol (II), of three trans-form derivatives in which each hydroxylic group is protected by methyl groups (III, IV, and V), and a derivative with reduced double bond (VI).

RESULTS
Antioxidant Activity of trans-Resveratrol and Derivatives II-VI-The comparison of antioxidant activity of trans-resveratrol and the derivatives, as estimated in in vitro tests, is reported in Table I. The results are expressed as the EQ (citronellal test) or the EC 50 (microsomes and DPPH assays) of each compound used. The stronger the antioxidant, the smaller the EQ or EC 50 value. In all three tests, trans-resveratrol (I) showed the highest antioxidant activity, whereas compound V did not exert any significant effect. Increasing values of EQ and EC 50 were observed for derivatives II, IV, and VI, and compound III reaching values about 5 and 3 times higher (p Ͻ 0.01) than transresveratrol, in the citronellal and microsome test, respectively. With the DPPH assay, the trans (I) and the cis (II) forms showed a similar EC 50 , and derivative IV gave a comparable value. Derivatives III and VI were less efficient, showing EC 50 values about 2 (p Ͻ 0.05) and 4 times (p Ͻ 0.01) higher than trans-resveratrol, respectively. Fig. 2 shows the effects of transresveratrol (I) and the derivatives on TBARS production induced by TBHP in normal human fibroblasts. Incubation of the cells for 1 h in the presence of 250 M TBHP significantly increased membrane lipid peroxidation, raising the TBARS production to 4.04 nmol/5 ϫ 10 5 cells, from the level of 0.92 nmol/5 ϫ 10 5 cells measured in untreated control samples. trans-Resveratrol (I) inhibited the TBARS production by about 67% (p Ͻ 0.01). Among the derivatives, compounds II, IV, and VI exhibited significant antioxidant activity, by reducing TBARS production by about 42% (p Ͻ 0.01), 61% (p Ͻ 0.01), and 33% (p Ͻ 0.01), respectively. Derivatives III and V did not exert any statistically significant activity in this assay. The concentration of Me 2 SO used (Ͻ0.02%) did not induce any significant protective effect against lipid peroxidation (data not shown). Bybenzil (compound without double bond and OH groups) and trans-stilbene (compound with double bond, without OH groups) did not show any detectable antioxidant activity, with any of the three in vitro methods (results not shown).

Effect of trans-Resveratrol and Derivatives II-VI on the Clonogenic Efficiency of Normal Fibroblasts-The clonogenic efficiency was studied in normal fibroblasts treated for 24 h with
trans-resveratrol or the derivatives at concentrations ranging from 0 to 90 M. Fig. 3 shows that only trans-resveratrol and derivative IV induced a dose-dependent reduction in clonogenic efficiency, with an estimated IC 50 of about 60 M for both. All other derivatives failed to induce a significant inhibition in cell growth.

Effects of trans-Resveratrol and Derivatives II-VI on Cell Cycle Progression of Normal and Tumor Cell Lines-To further
investigate the effect of the various derivatives on cell proliferation, the distribution in each phase of the cell cycle was analyzed by determining the DNA content with flow cytometry. Fig. 4 compares the effects in normal fibroblasts treated for 24 h with trans-resveratrol or derivatives II-VI, at the 30 M concentration, because significant effects could be already observed in these conditions. A significant accumulation (p Ͻ 0.01) in S phase, and a consequent reduction in the number of cells in G 1 phase, was observed with trans-resveratrol (G 1 ϭ 44.5%, S ϭ 42.5%, G 2 ϩ M ϭ 13.0%), and with derivative IV (G 1 ϭ 48.0%, S ϭ 40.1%, G 2 ϩ M ϭ 11.9%), with respect to the cell cycle distribution of control cells (G 1 ϭ 61.3%, S ϭ 22.3%, G 2 ϩ  M ϭ 16.4%). The cell cycle distribution of samples treated with the other derivatives (II, III, V, and VI) was comparable to that observed in the controls. Aphidicolin, a well known inhibitor of DNA polymerases ␣ and ␦, used as a positive control, induced a significant accumulation of cells in S phase (G 1 ϭ 34.5%, S ϭ 58%, G 2 ϩ M ϭ 7.5%), comparable to that induced by compounds I and IV. Cells incubated with the solvent alone (Me 2 SO Ͻ 0.1%) did not show any alteration of cell cycle progression. The inhibitory effect on cell growth induced by transresveratrol and derivative IV was reversible, since 48 h after removal of the compounds, the percentage of cells in S phase returned to the control value (not shown).
To study whether the observed cell cycle imbalance induced by trans-resveratrol and derivative IV was consequent upon a change in DNA synthesis, DNA replication was assessed by BrdUrd incorporation and determined with immunofluorescence and flow cytometric analysis. Fig. 5 shows the dot plots of BrdUrd immunofluorescence versus DNA content in control cells and in fibroblasts treated with 30 M trans-resveratrol for 24, 48, and 72 h. The results showed that cells progressed through S phase at a slower rate than control cells and incorporated significantly lower amounts of BrdUrd incorporation. Quantitative analysis of BrdUrd immunofluorescence in the region corresponding to S-phase cells indicated that transresveratrol inhibited BrdUrd incorporation by about 55% at 24 h (see Figs. 5B and 7A), and by about 70 and 80% at 48 and 72 h (Fig. 5, C and D), respectively. Similar results were obtained in HT1080 fibrosarcoma cells (not shown).
Effects of trans-Resveratrol on the Recruitment of PCNA and RPA Proteins to DNA Replication Sites-To explore the basis underlying the inhibition of DNA synthesis, the recruitment of PCNA and RPA (32-kDa subunit) proteins to DNA replication sites was investigated next. To this aim, cells were lysed in hypotonic buffer to separate the detergent-insoluble forms of the two proteins. Fig. 6A shows the dot plots of PCNA immunofluorescence versus DNA content in control cells (a) and in fibroblasts treated for 24 h with 15 (b), 30 (c), or 90 M (d) trans-resveratrol. Accumulation of cells in S phase was evident at the lowest concentrations, while with 90 M trans-resveratrol, cells were blocked at the G 1 /S phase transition, as indicated by the high PCNA immunofluorescence levels typical of cells entering S phase (52). Quantitative analysis of immunofluorescence intensity in S phase indicated that the amount of PCNA assembled in replication foci in treated samples, was about 12% higher (not significant) than that of control cells. The amount of RPA protein (32 kDa) assembled at the replication foci was also not significantly modified after treatment with trans-resveratrol (Fig. 6B).
Effects of trans-Resveratrol and Derivatives II-VI on DNA Synthesis-The above results indicated that BrdUrd incorporation was inhibited at a step following to the recruitment of RPA and PCNA to the replication foci, thus suggesting that the inhibition of DNA synthesis occurred at the level of DNA polymerase activity, as also suggested by the reduction in BrdUrd incorporation. For this purpose, the ability of trans-resveratrol to inhibit DNA synthesis was compared with the other derivatives, both in cells and in in vitro assays testing the activity of replicative DNA pol ␣ and pol ␦. The results reported in Fig. 7A show that among the derivatives, only compound IV inhibited BrdUrd incorporation to a similar extent of trans-resveratrol. Comparable results were obtained in the in vitro assays on DNA polymerase activity (Fig. 7B). trans-Resveratrol inhibited both pol ␣ and pol ␦ with similar potencies, whereas the derivative IV showed a 2-fold preference for pol ␣ (p Ͻ 0.01) with respect to pol ␦ (p Ͻ 0.05). It must be noted that this inhibition was enhanced by the preincubation with the enzyme, suggesting a slow-binding mode of inhibition. All the other compounds were 3-10-fold less active than the trans-resveratrol. The cisform (II) of resveratrol was also tested and found to be inactive, confirming that for DNA synthesis inhibition, the active configuration of the compound was in the trans-form. The corresponding K i values for the active compounds are listed in Table  II. None of the derivatives inhibited Escherichia coli pol I (Klenow fragment), HIV-1 reverse transcriptase, and HSV-1 DNA polymerase (data not shown), confirming the specificity of the above inhibition observed.
Electronic Structure and Thermodynamic Stability of the Phenoxyl Radicals from trans/cis-Resveratrol- Fig. 8 shows the limit surface diagrams for the singly occupied molecular orbitals in the three phenoxyl-type radicals of both the trans and cis conformations, as were obtained from PM3. In all cases a delocalization of the unpaired spin is observed. However, in the trans configuration, the 4Ј-phenoxyl (picture A) is extended also to the adjacent ring through the stilbene double bond, whereas in the 3-and 5-phenoxyl species (picture B) the delocalization is confined to the aromatic ring bound to the oxygen radical center. This different electronic structure leads to a greater resonance stabilization energy for the 4Ј-phenoxyl rad-ical, the formation of which is accordingly predicted to be more exothermic, as indicated by ⌬H f°r eported in Fig. 8.
The rationale for these results is obtained by considering that in mononuclear phenoxyl radicals the maximum unpaired spin densities are at the ring meta and para positions. In the resveratrol 4Ј-phenoxyl radical the spin density can flow to the adjacent ring since the stilbene double bond acts as a bridge being bound to a maximum spin density center. This is not the case for the other two radical species where the double bond is bound at positions of minimum spin density. As a consequence of the greater resonance stabilization energy, the hydrogen abstraction from the 4Ј-OH bond is expected to be favored being more exothermic with respect to the analogue reactions at the 3-OH and 5-OH positions. According to the above arguments, the absence of the olefinic double bond in ␣,␤-dihydro-3,4Ј,5trihydroxystilbene is expected to cause a decrease in the 4Јphenoxyl radical resonance stabilization energy and, consequently, a decrease in the overall inhibition efficiency.
Based on molecular orbitals calculations, all the phenoxyl radicals of resveratrol in the cis conformation ( Fig. 8C and D) show an intrinsic thermodynamic stability lower than that of corresponding radicals in the trans-form (Fig. 8, A and B), since the formation enthalpies are less exothermic by 7.1 kcal/mol for the 4Ј-phenoxyl radical (⌬H f°t rans (Ϫ55.8 kcal/mol) Ϫ ⌬H f°c is form (Ϫ48.7 kcal/mol)), and by 6.4 kcal/mol, for the 5-and 3-phenoxyl radicals. Furthermore, within the cis configuration framework, the 3-and 5-phenoxyl radicals (Fig. 8D) are also found to be less stable (⌬H f°ϭ Ϫ46.77 kcal/mol) than the 4Ј-phenoxyl analogue (Fig. 8C, ⌬H f°ϭ Ϫ48.71 kcal/mol). In fact, in the cis 4Ј-phenoxyl radical, the spin delocalization to the phenyl ring via the double bond is partially hindered by the lack of coplanarity of the system. This effect is witnessed by the higher resonance stabilization energy of the trans, with respect to the cis analogue, which is obtained by this calculation (⌬H f°A Ϫ⌬H f°B ϭ Ϫ2.7 kcal/mol; ⌬H f°C Ϫ ⌬H f°D ϭ Ϫ1.94 kcal/ mol). The difference between these values leads to 0.76 kcal/mol, which represents a 30% decrease in the resonance stability.
For the sake of comparing the radical reactivities of cis-and trans-resveratrol, the reaction enthalpies for the hydrogen abstraction reaction have been calculated using the molecular orbitals method (Table III). The hydrogen abstraction reactions from trans-resveratrol are predicted to be more exothermic by 2-3 kcal/mol. DISCUSSION In agreement with abundant evidence obtained on other systems in vitro (14,17,39,53,54) the work presented here has documented a significant antioxidant activity of resveratrol. The results have shown that the hydroxyl group in the 4Ј position is required for the antioxidant activity, but acts synergistically with the 3-and 5-OH groups. In fact, the derivative IV, which has a free hydroxyl group in the 4Ј position, exert a presence of trans-resveratrol and derivatives II-VI. Assays were performed as described under "Experimental Procedures" in the presence of 5 M of each inhibitor to be tested. DNA polymerase activity was expressed as percentage of the control reaction without inhibitor which was 0.8 pmol ϫ min Ϫ1 for pol ␣ and 0.6 pmol ϫ min Ϫ1 for pol ␦ and taken as 100%. *, p Ͻ 0.05; **, p Ͻ 0.01 significantly different from to control by one way ANOVA-Tukey's test. significant antioxidant activity in all the tests used. The transisomery and the double bond in the stilbenic skeleton also play a role, at least as indicated by the citronellal and microsome tests, since cis-resveratrol and compound VI, in which the double bond is reduced, are significantly less effective than the trans-resveratrol. A similar conclusion is suggested by the cell culture assay, based on the inhibition of TBHP-induced lipid peroxidation. The major role played by the 4Ј-OH group in the radical scavenging and antioxidant activity of trans-resveratrol could be related to the electronic structure and the formation enthalpy of the three different phenoxyl radicals arising from the loss of hydrogens at the 3-, 5-, and 4Ј-OH groups in res-veratrol (Fig. 8). The hydrogen abstraction from 4Ј-OH bond is expected to be favored, being more exothermic than the analogue reactions at the 3-OH and 5-OH positions, as a consequence of the greater resonance stabilization energy. In addition, the absence of the double bond is expected to cause a decrease in the 4Ј-phenoxyl radical resonance stabilization energy. This would induce a decrease in the antioxidant activity of compound VI, as observed in our experimental models. Finally, on comparing the radical reactivity of trans-and cisresveratrol (Table III), the hydrogen abstraction reactions from the trans-form are predicted to be more exothermic. This provides a thermodynamic rationale for the experimental observation in two in vitro methods and in cell culture that the antioxidant power of cis-resveratrol is lower than that of the trans-analogue. DPPH measurements differ from the above results in that trans-and cis-form have the same effect, while again both methylation and, more markedly, reduction of the olefinic double bond, decrease the antioxidant activity. Thus, in this assay, hydrogen transfer to the radical is equally efficient from both isomers. It may be inferred from these results that the calculated decrease in the stabilization energy for the phenoxyl radical in the cis configuration does not significantly influence the reaction kinetic activation parameters. This in turn might be considered diagnostic of the activated complex along the reaction path being closer to reactants than the products. In agreement with studies by others (15,16,(25)(26)(27)(28)30), trans-resveratrol has been found to inhibit in a dose-and time-dependent manner cell growth both in normal fibroblasts and in fibrosarcoma cells. An antiproliferative activity comparable to that of trans-resveratrol has been observed only for derivative IV. Importantly, their effect was cytostatic and reversible since no evidence of cell death was obtained by a number of tests such as visual inspection for detached cells, trypan blue exclusion, and annexin V staining (data not shown).
The cytostatic effect may be attributed to decreased DNA synthesis given that a significant inhibition of BrdUrd incorporation was previously found in other cell lines (19,29,30,37), and also observed in our study on normal fibroblasts and in fibrosarcoma cells. The results described here suggest that inhibition of DNA synthesis occurred at the level of DNA polymerase activity, since the recruitment of PCNA and RPA (32 kDa) proteins to DNA replication sites was not affected by trans-resveratrol.
The in vitro assays have demonstrated that only trans-resveratrol and derivative IV inhibited significantly DNA polymerases ␣ and ␦. Interestingly, there was an increase in specificity for the inhibition of pol ␣ with respect to pol ␦, from trans-resveratrol to compound IV suggesting a possible role of the -OH groups in positions 3 and 5 in the binding to the different DNA polymerases. The inhibition by resveratrol was found to be strictly specific for the B-type DNA polymerases, since neither pol I (member of the A-type family of DNA polymerases), nor HIV-1 RT (belonging to the reverse transcriptases/RNA-dependent RNA polymerases/telomerases family) were inhibited. Moreover, within the B-type family, trans-resveratrol discriminated between eukaryotic and viral enzymes, i.e. HSV-1 pol was not inhibited. These data suggest that the FIG. 8. Electronic structure and thermodynamic stability of the phenoxyl radicals from resveratrol. Picture of the limit surface diagrams for the singly occupied molecular orbitals in the 4Ј-phenoxyl radical (A), and 5-or 3-phenoxyl radicals (B), from trans-resveratrol and from cis-resveratrol (C and D), respectively. Values of formation enthalpy (⌬H f°) for each phenoxyl radical are also shown. interaction of resveratrol with the eukaryotic replicative DNA polymerases ␣ and ␦ is highly specific.
In conclusion, the results of this study indicate that (i) 4Јhydroxyl group in trans-conformation (hydroxystyryl moiety) is not the sole determinant for antioxidant properties, while it is absolutely required for antiproliferative activity. (ii) There is a direct correlation, from a structural point of view, between the antiproliferative effect and the ability to inhibit DNA pol ␣ and ␦. Thus, a mechanism underlying the inhibition of cell cycle progression is the interaction between the 4Ј-hydroxystyryl moiety of trans-resveratrol and DNA polymerases.
The structure-activity relationship revealed by this study should be taken in account in studies aimed at synthesizing resveratrol derivatives with more selective antioxidant and/or antiproliferative activity. To this respect, the observation that the cis conformation of resveratrol still showed antioxidant activity but was totally inactive against DNA polymerases, is very interesting and deserves further investigation.