Abstract
The purpose of this study was to investigate the sulfation of resveratrol (3,5,4′-trihydroxystilbene) and its potential to exhibit drug-drug interactions via sulfation. The possible interaction of resveratrol with 17β-estradiol (E2), a major estrogen hormone and prototypic substrate for sulfate conjugation, was studied. Resveratrol and E2 are both known to undergo sulfate conjugation catalyzed by human sulfotransferases (SULTs). Resveratrol is a phytoestrogen with mixed estrogen agonist/antagonist properties that is being developed as a chemopreventive agent. The sulfate conjugation of E2 and resveratrol were studied individually using S9 fractions from human liver and jejunum as well as recombinant human SULT isoforms. The sulfation of E2 (3–20 nM) was then investigated in the presence of various concentrations (0, 0.5, 1, and 2 μM) of resveratrol using the two S9 preparations as well as recombinant SULT1E1, the major isoform responsible for E2 sulfation. Resveratrol inhibited E2 sulfation with estimated Ki values of 1.1 μM (liver), 0.6 μM (jejunum), and 2.3 μM (SULT1E1), concentrations that could be pharmacologically relevant. The results suggest that these phytoestrogens can potentially alter the homeostasis of estrogen levels. These findings also imply that resveratrol may inhibit the metabolism of other estrogen analogs or therapeutic agents such as ethinylestradiol or dietary components that are also substrates for SULT1E1.
Resveratrol (trans-3,5,4′-trihydroxystilbene) (Fig. 1) is a polyphenolic phytoestrogen found in high concentrations in red wine (Kopp, 1998) and a variety of plant sources such as grape skin, berries, pomegranates, and peanuts. It has generated considerable research interest because it has a wide range of biological effects that include antioxidant, chemopreventive, cardioprotective, neuroprotective, antiinflammatory, and antiviral activity (Bhat et al., 2001; Aziz et al., 2003; Baur and Sinclair, 2006). Resveratrol has also been shown to have antiaging effects on the basis of recent reports of a significant improvement in health and lifespan after chronic administration of this phytoestrogen to mice fed with a high-calorie diet (Baur et al., 2006). The estrogenic potential of resveratrol is under active investigation. It has been shown to bind to estrogen receptors (ERα and ERβ) and appears to have antiestrogenic effects at high doses (Gehm et al., 1997). Recent studies have shown an induction of apoptosis and a reduction in extracellular levels of angiogenic vascular endothelial growth factor by resveratrol in human breast cancer xenografts (Garvin et al., 2006).
The metabolism of resveratrol in humans involves sulfation and glucuronidation. The oral bioavailability of free resveratrol is low because of its rapid phase II conjugation to sulfates and glucuronides (Walle et al., 2004). Resveratrol 3-O-sulfate is the major reported sulfated metabolite of resveratrol in humans and rats, with other sulfate conjugates such as resveratrol 4′-sulfate, 3,5-disulfate, 3,4′-disulfate, and 3,4′,5-trisulfate having been identified in in vivo studies (De Santi et al., 2000a,b; Walle et al., 2004; Wenzel and Somoza, 2005). SULT1A1 and SULT1E1 have been implicated in the sulfation of resveratrol (Miksits et al., 2005). Our previous work (Brill et al., 2006) as well as other studies (Aumont et al., 2001) have reported the roles of uridine diphosphate glucuronosyltransferase (UGT) isoforms 1A1 and 1A9 in the 3-O-glucuronidation and 4′-O-glucuronidation of trans-resveratrol, respectively. Piceatannol (3,3′, 4′,5–tetrahydroxystilbene) is another polyphenol that has been shown to be formed from resveratrol by CYP1B1 and CYP1A2 during in vitro studies (Potter et al., 2002; Nath et al., 2004; Piver et al., 2004), although cytochrome P450 metabolites of resveratrol have not been identified in clinical studies (Walle et al., 2004; Boocock et al., 2007). Piceatannol by itself has anticancer properties with the ability to inhibit tyrosine kinase (Thakkar et al., 1993).
The current study was performed to determine whether resveratrol can interfere with the sulfation of other compounds. E2 (Fig. 1) was selected for these studies because it is a prototypic endogenous estrogen hormone that also undergoes sulfate conjugation. E2 mediates its physiological effects by binding to ERs and is important for the growth and development of normal breast tissue. Sulfate conjugation is mediated by SULTs, a superfamily of enzymes that includes at least three families (SULT1, SULT2, and SULT4) in humans with 13 members identified to date (Gamage et al., 2006). This pathway is responsible for the metabolism of a variety of xenobiotics and endogenous compounds, resulting usually in the formation of more polar, hydrophilic, and pharmacologically inactive compounds or occasionally in the generation of biologically active or procarcinogenic compounds (Coughtrie, 2002; Gamage et al., 2006). E2 is conjugated to its 3-sulfate (E2 sulfate) by estrogen sulfotransferase (EST, SULT1E1) (Adjei and Weinshilboum, 2002). In the present study, we have performed detailed in vitro investigations using S9 fractions from human liver and intestine (jejunum) as well as recombinant human SULT isoforms and have demonstrated that resveratrol is capable of significantly inhibiting the sulfation of E2.
Materials and Methods
Materials. [3H]E2, Ultima Gold scintillation cocktail, and high-performance glass vials were purchased from PerkinElmer Life and Analytical Sciences (Shelton, CT). Trans-resveratrol (manufactured by Pharmascience, Montreal, ON, Canada) was supplied by the National Cancer Institute. Trans-resveratrol was selected for all these studies, as this conformation (4′-hydroxystyryl moiety) is required for in vitro activity, along with the presence of the 4′-OH group (Matsuoka et al., 2002) and is under development as a chemopreventive agent. Human hepatic S9 (pooled from seven females and eight males) and human intestinal (jejunal) S9 (pooled from eight males and two females) were purchased from Tissue Transformation Technologies (Edison, NJ). Sf-9 insect cell-expressed human SULT isoforms were purchased from Invitrogen (Carlsbad, CA). Dithiothreitol (DTT), adenosine 3′-phosphate 5′-phosphosulfate (PAPS) (78% purity), and Trizma base were purchased from Sigma-Aldrich (St. Louis, MO). Bovine serum albumin (BSA) and organic solvents were purchased from Mallinckrodt Baker, Inc. (Paris, KY). All procedures with human liver and jejunal S9 were approved by the Human Subjects Committee of SRI International.
E2 Sulfation Assay. Sulfation of E2 was performed with radiolabeled E2 ([3H]E2) using methods modified from previous studies (Kester et al., 1999; Otake et al., 2000). Individual conditions of incubation such as incubation time and concentrations of S9 or SULT1E1 protein, PAPS, and organic solvent (ethanol) were selected on the basis of preliminary experiments (Furimsky et al., 2006). The final assay conditions used included hepatic or jejunal S9 (protein, 30 μg/ml) or recombinant SULT1E1 (protein, 1 μg/ml) with [3H]E2 (10 nM in ethanol, 0.14% v/v), PAPS (25 μM), DTT (8 mM), and BSA (0.0625% w/v) in a 50 mM Tris-HCl buffer (pH 7.4) for 45 min in a volume of 200 μl. A range of [3H]E2 concentrations (0–50 nM) was used in the incubations performed to determine kinetic parameters for the sulfation of E2. The reaction was started by the addition of S9 or recombinant SULT1E1 and stopped with the addition of chilled chloroform (3 ml) and deionized water (250 μl). The samples were vortexed for 2 min and centrifuged at 1500 rpm for 10 min after which the upper aqueous phase (200 μl) was transferred to vials containing 10 ml of cocktail to measure the radioactivity using a scintillation counter (Packard Tri-Carb 2900TR; PerkinElmer, Inc., Wellesley, MA). A blank sample with all ingredients except PAPS was included with each test sample.
The sulfation of E2 was also investigated using Sf-9 insect cell-expressed human SULT isoforms, SULT1A1*2, SULT1A2, SULT1A3, SULT1E1, and SULT2A1. Incubation conditions and sample analysis were similar to those described above, with a substrate concentration of 10 nM and individual SULT protein concentrations of 1 μg/ml.
Resveratrol Sulfation Assay. Sulfation of resveratrol was studied using S9 fractions from human liver and jejunum as well as individual cDNA-expressed human SULT isoforms (SULT1A1*2, SULT1A2, SULT1A3, SULT1E1, and SULT2A1) using methods modified from Miksits et al. (2005). Conditions of incubation were optimized as described for E2 sulfation. Final incubation conditions included hepatic or jejunal S9 (protein, 1 mg/ml), resveratrol (0–10 μM in methanol, 0.5% v/v), PAPS (25 μM), DTT (1 mM), and EDTA (0.1 mM) at 37°C in 50 mM Tris-HCl buffer (pH 7.4) with an incubation time of 60 min. In the experiments with SULT isoforms, resveratrol was incubated at 0.2, 2, or 10 μM with individual SULT protein (0.2 mg/ml). The reaction was stopped with the addition of 200 μl of cold methanol, samples were centrifuged at 14,000 rpm for 5 min, and 80 μl of supernatant was transferred to an HPLC vial for analysis using a Waters 2695 high-performance liquid chromatograph with a Waters 996 photodiode array detector (PDA) set at 307 nm (Waters Corp, Milford, MA). The HPLC column was a Luna C18(2), 250 mm × 4.6 mm, 5 μm i.d. (Phenomenex Corp., Torrance, CA). The mobile phase consisted of 5 mM ammonium acetate (pH 5.0, solvent A) and methanol (solvent B) used at a flow rate of 1 ml/min under the following gradient conditions: 0 min (90% A and 10% B), 10 min (80% A and 20% B), 22 min (65% A and 35% B), 25 min (40% A and 60% B), 30 min (40% A and 60% B), 32 min (90% A and 10% B), and 40 min (90% A and 10% B). As pure standards of resveratrol sulfates were not available, the formation of the sulfates was expressed on the basis of their respective peak areas from chromatograms.
Identification of Sulfated Metabolites of Resveratrol. Identification of resveratrol sulfates in samples generated from the in vitro incubations was performed by LC-MS using the same LC method as described by Miksits et al. (2005). The LC-MS system was comprised of a Waters 2795 high-performance liquid chromatograph with an inline Waters 996 PDA detector, coupled to a Micomass Quattro Ultima triple quadrupole mass spectrometer (Waters). A 30:70 fixed-ratio flow splitter was used (model 620, Analytical Scientific Instruments, El Sobrante, CA) so that 30% of the postcolumn, post-PDA flow was diverted to the MS detector. The mass spectrometer was operated in electrospray ionization negative ion mode and at unit mass resolution. Nitrogen was used as the nebulization and desolvation gas at flow rates of 60 and 680 liters/h. Source and desolvation temperatures were 115 and 310°C. Capillary and sample cone voltages were 1.5 kV and 25 V. Product-ion scans were collected using argon as the collision gas. Collision energies varied from 15 to 40 eV, depending on the experiment. Precursor-ion spectra were recorded from 60 to 1000 m/z with a dwell time of 1 s. Product-ion scan spectra were recorded from 60 to 450 m/z with a dwell time of 0.2 to 1 s per channel and a cycle time of <1.5 s.
Inhibition of E2 Sulfation by Resveratrol. The sulfation of E2 was studied by coincubating resveratrol (0, 0.5, 1, and 2 μM) with [3H]E2 and S9 from human liver or jejunal samples (protein, 30 μg/ml) or recombinant human SULT1E1 (1 μg/ml). The concentration ranges of [3H]E2 were selected from preliminary studies and varied with each source of SULT used, i.e., 5, 10, 25, and 50 nM (hepatic S9); 5, 7.5, 10, and 20 nM (jejunal S9); and 3, 7, 10, and 15 nM (SULT1E1). [3H]E2 and resveratrol were dissolved in ethanol at a final concentration of 0.3% v/v, which was not found to significantly influence the sulfation of E2 in the initial studies (Furimsky et al., 2006).
Western Blot Analyses of SULT1A1 and SULT1E1 in Hepatic and Jejunal S9 Fractions. The content of SULT proteins (SULT1A1 and SULT1E1) was measured in hepatic S9 and jejunal S9 preparations using quantitative Western blot analysis. A rabbit polyclonal antibody directed against SULT1A1 (wild-type) amino acids 81 to 98 was used to perform the SULT1A1 immunoreactive protein study. For SULT1E1, the polyclonal antibody used was directed against amino acids 1 to 13 (Her et al., 1996; Adjei et al., 2003). Aliquots of hepatic or jejunal S9 preparation as well as SULT1A1 or SULT1E1 COS-1 recombinant protein were loaded on a 12% SDS-polyacrylamide gel (SDS-PAGE mini-gel; Bio-Rad, Hercules, CA) for Western blot analysis. Proteins were separated by electrophoresis before transfer to nitrocellulose membranes (Schleicher and Schuell, Keene, NH). SULT1A1 and SULT1E1 proteins were detected with the ECL Western Blotting System (GE Healthcare, Piscataway, NJ), and the Ambis Radioanalytical Imaging System, Quant Probe (version 4.31, Ambis Inc., San Diego, CA) was used to quantitate the levels of immunoreactive protein. The resulting data were expressed as milligrams of the specific recombinant SULT protein in COS-1 cell high-speed supernatant.
Data and Statistical Analysis. All incubations for E2 and resveratrol sulfation were performed in triplicate with the formation of each metabolite being expressed as mean ± S.D. The data for enzyme kinetic analysis were fit to Michaelis-Menten (hyperbolic) and Hill (sigmoidal) models. Standard parameters such as coefficient of determination (R2), standard deviation of the parameter estimates, and visual inspection were used to determine the quality of fit to a specific model. The apparent enzyme kinetic parameters of Km and n (for the Hill model) were calculated by nonlinear regression analysis (Enzyme Kinetics Module 1.1 of Sigma Plot 2001, version 7.101; SPSS Inc., Chicago, IL). Inhibitory constants (Ki) values from the drug interaction studies with resveratrol were calculated from the mean data of three replicates in each experiment using Lineweaver-Burk analysis (Enzyme Kinetics Module 1.1 of Sigma Plot 2001).
Results
Estradiol Sulfation by S9 Fractions and Recombinant SULTs. The formation of E2 sulfate by liver and jejunum was linear with respect to S9 protein concentrations up to 40 μg/ml and incubation times up to 60 min, as shown previously in preliminary studies (Furimsky et al., 2006). Variations in PAPS concentrations (5–60 μM) and ethanol concentrations (0.1–1% v/v) in the incubation mix did not influence the sulfation of E2 (Furimsky et al., 2006). E2 sulfate was formed to a greater extent in hepatic tissue compared with jejunal tissue. There was a 2-fold increase in the rate of formation of E2 sulfate observed with hepatic S9 (4.08 ± 0.17 pmol/min/mg) compared with jejunal S9 (1.95 ± 0.14 pmol/min/mg), at a protein concentration of 30 μg/ml. The rate of formation of E2 sulfate by individual recombinant SULT isoforms is shown in Fig. 2. Among the human recombinant SULT isoforms screened, E2 sulfation was most predominant with cDNA-expressed SULT1E1 (4.88 ± 0.13 pmol/min/mg) followed by SULT1A1*2 (2.26 ± 0.18 pmol/min/mg), SULT2A1 (0.87 ± 0.02 pmol/min/mg), SULT1A2 (0.81 ± 0.00 pmol/min/mg), and SULT1A3 (0.15 ± 0.01 pmol/min/mg) (Fig. 2). It should be noted that our SULT activity assays used the *2 variant of SULT1A1, which is defined by Arg to His substitution in position 213 (G to A conversion at nucleotide 638) of its sequence and has been reported to be associated with low enzyme activity (Nagar et al., 2006).
The kinetics of E2 sulfation was studied with S9 fractions from pooled human hepatic and jejunal samples as well as with cDNA-expressed human SULT1E1, as this was the major human SULT isoform responsible for E2 sulfation. The formation of E2 sulfate by all three sources followed atypical enzyme kinetics that was best described using the Hill model. The following kinetic parameters were determined for E2 sulfation: Km = 11.5 nM, Vmax = 3.8 pmol/min/mg, and n = 1.6 with hepatic S9 (R2 = 0.98); Km = 4.5 nM, Vmax = 0.91 pmol/min/mg, and n = 2.2 with jejunal S9 (R2 = 0.93); and Km = 5.6 nM, Vmax = 112.7 pmol/min/mg, and n = 2.2 with SULT1E1 (R2 = 0.95) (Table 1). The maximal velocity of formation (Vmax) of E2 sulfate was 4-fold higher with S9 fractions from the liver versus jejunum. The estimated Vmax was highest with SULT1E1 (112.7 pmol/min/mg), which was 30- and 125-fold faster than that observed with hepatic and jejunal S9, respectively. The protein concentration (1 μg/ml) used for recombinant human SULT1E1 in these experiments was selected from preliminary results in which the formation of E2 sulfate was linear between 0.1 and 5 μg/ml (data not shown). The affinity between E2 and SULT(s) was higher with jejunum (Km = 4.5 nM) and SULT1E1 (Km = 5.6 nM) compared with liver (Km = 11.5 nM) (Table 1).
Resveratrol Sulfation by S9 Fractions and Recombinant SULTs. Three sulfated metabolites of resveratrol (designated as M1, M2, and M3) were formed during in vitro incubations with S9 fractions from liver and jejunal tissue. These metabolites were confirmed to be sulfates by LC-MS analysis and by the lack of their respective peaks in HPLC chromatograms from samples without PAPS.
The involvement of specific SULT isoforms in the formation of M1, M2, and M3 at resveratrol concentrations of 0.2, 2, and 10 μM is shown in Fig. 3. M3 was found to be the major sulfated metabolite at all substrate concentrations studied. SULT1A3 and SULT1A2 were mainly responsible for the formation of M3, which increased with substrate concentrations of 0.2 to 10 μM (Fig. 3, A–C). Other SULTs were also responsible for the formation of M3 with a rank order of SULT1A3 > SULT1A2 > SULT1E1 > SULT1A1*2, at a resveratrol concentration of 10 μM (Fig. 3C). M1 was formed predominantly by SULT1A1*2 and SULT1E1, with minor involvement of SULT1A2. M2 was formed exclusively by SULT1E1 and at the higher concentrations of substrate (2 and 10 μM). SULT2A1 was not found to be involved in the formation of M1, M2, or M3 in preliminary experiments at a substrate concentration as high as 50 μM (data not shown) and hence, was not included in these experiments. The formation of the three metabolites at a substrate concentration of 10 μM (Fig. 3C) follows the same trend as that observed by Miksits et al. (2005). Our studies have been designed to also include substrate concentrations of 0.2 and 2 μM, which are likely to be pharmacologically more relevant.
The formation of the major sulfated metabolite (M3) was used to study the kinetics of resveratrol sulfation, using S9 fractions from human liver and jejunum as well as the major SULT isoforms that can catalyze the formation of this metabolite. The formation of M3 by hepatic S9 was best described by the Hill equation, with Km = 1.3 μM, Vmax = 65.4 pmol/min/mg, and n = 2 (Table 2). The formation of M3 by jejunal S9 followed Michaelis-Menten kinetics with Km = 5.9 μM and Vmax = 186 pmol/min/mg (Table 2). The kinetic parameters for M3 formation by SULT1A1*2, SULT1A2, SULT1A3, and SULT1E1 which followed Hill kinetics (R2 = 0.98, 0.95, 0.94, and 0.96, respectively), are also listed in Table 2.
Identification of Sulfated Metabolites of Resveratrol. The primary approach to confirm the identities of the sulfated metabolites of resveratrol was to use LC-MS. Samples were chromatographed, and the metabolites detected by PDA and MS using the same LC conditions as those of Miksits et al. (2005), who had reported their identities by nuclear magnetic resonance as resveratrol 3-O-4′-O-disulfate (M1), resveratrol 4′-O-sulfate (M2), and resveratrol 3-O-sulfate (M3). The samples that were analyzed by LC-MS/MS included those generated using recombinant SULT1E1 at a substrate concentration of 10 μM (Fig. 3C). SULT1E1 was found to be capable of generating all three sulfates of resveratrol in our studies as well as by Miksits et al. (2005) at a substrate concentration of 10 μM. These samples had previously been analyzed by HPLC with UV detection at 307 nm and contained different relative levels of the metabolites. It was found that the metabolite peaks in our studies eluted in the same order as those of Miksits et al. (2005).
The full-scan LC-MS spectra and LC-MS/MS product-ion spectra of the metabolites generated in our studies agreed with the metabolite identities proposed by Miksits et al. (2005). For M1, precursor-ion scans revealed masses at 307, 387, 409, and 425 m/z, which are logical molecular ions for the proposed structure of M1 as resveratrol 3-O-4′-O-disulfate. Specifically, 387, 409, and 425 m/z correlate to the disulfate ([M - H]-), disulfate plus sodium adduct ([M - 2H + Na]-), and disulfate plus potassium adduct ([M - 2H + K]-), respectively; 307 m/z corresponds to [M - sulfate - H]-, formed due to capillary/cone fragmentation of one of the sulfates of the disulfate metabolite upon ionization. The LC-MS/MS product-ion spectra are also consistent with the proposed structure of M1 as resveratrol 3-O-4′-O-disulfate; product-ion scans of 387 m/z yielded a base peak of 227 m/z, which corresponds to the loss of both sulfates and product-ion scans of 307 m/z yielded a base peak of 227 m/z, which corresponds to the loss of the one remaining sulfate. For M2 and M3, precursor-ion scans revealed a base peak of 307 m/z, which corresponds to a monosulfate form of resveratrol. Product-ion spectra of M2 and M3 yielded equivalent fragment ions and relative intensities, regardless of the collision energies used, and thus structural elucidation of trans-resveratrol-4′-O-sulfate (M2) versus trans-resveratrol-3-O-sulfate (M3) could not be positively confirmed by LC-MS/MS. M3, which had a longer retention time than M2 in our studies and others (Miksits et al., 2005), was designated as resveratrol 3-O-sulfate, as it is less polar than resveratrol 4′-O-sulfate (Yu et al., 2002). Other samples with hepatic S9, jejunal S9, and other SULT isoforms were also analyzed by LC-MS/MS and were confirmatory in terms of the identities of the three sulfated metabolites of resveratrol. The M1, M2, M3, and resveratrol peaks also had characteristic PDA profiles that are consistent with the proposed structures; specifically, the λmax was ∼307 nm and a second absorbance band was present at ∼240 nm.
Inhibition of E2 Sulfation by Resveratrol. There was a significant inhibition of E2 sulfation by resveratrol, using all three sources of SULT, i.e., liver S9, jejunal S9, and recombinant SULT1E1. The estimated Ki values for the inhibition were in the low micromolar range, i.e., 1.1 μM (liver), 0.6 μM (jejunum), and 2.3 μM (SULT1E1) (Table 3). Lineweaver-Burk double reciprocal plots of the inhibition of E2 sulfation by resveratrol using liver S9, jejunal S9, and recombinant SULT1E1 are shown in Figs. 4, 5, and 6, respectively. The position of the intersection points of the control plots (without resveratrol) and the “plus resveratrol” plots in the graphs is above (Figs. 4 and 5) or below (Fig. 6) the x-axis and on the left of the y-axis. Evaluation of these plots as well as the plots of rate of reaction versus substrate concentration (plots not shown) indicated a hyperbolic mixed (partial) inhibition of E2 sulfation by both resveratrol and picetannol using S9 from liver and jejunum and SULT1E1, i.e., both Km and Vmax values are altered in the presence of resveratrol.
Western Blot Analysis of SULT1A1 and SULT1E1 in Hepatic and Jejunal S9 Fractions.Fig. 7 shows the Western blots for SULT1A1 and SULT1E1 in the same batches of S9 fractions from human liver and jejunum as those used in the E2 and resveratrol sulfation experiments. The recombinant SULT1A1 and SULT1E1 proteins used in Western blots were prepared using COS-1 cells. SULT1A1 protein was found to be present in greater amounts in hepatic S9 than in jejunal S9 preparations (Fig. 7A). When the immunoreactive proteins for the hepatic and jejunal S9 fractions were determined in terms of COS-1 recombinant SULT1A1 protein, only 6% of the SULT1A1 immunoreactive protein present in the hepatic cytosol was observed in the jejunal preparation. On the other hand, the jejunal S9 fractions expressed SULT1E1 to a greater extent than in the hepatic S9 preparations (Fig. 7B). Approximately 10% of the SULT1E1 protein expressed in jejunal S9 was present in hepatic S9.
Discussion
Resveratrol is a well known polyphenolic compound with mixed estrogen agonist/antagonist properties that has gained wide popularity in complementary and alternative medicine because of its broad range of biological activities. It is found in high concentrations in red wine (2–40 μM) and may contribute to the “French paradox” (Kopp, 1998), the relatively low incidence of heart disease in the French population despite a diet high in fat content. Resveratrol is easily available as a nutritional supplement in health food stores and has the potential to be consumed on a chronic basis. Therefore, it is a likely candidate for interactions with concomitantly administered therapeutic agents, dietary components, or endogenous substances. We have previously shown that resveratrol significantly decreases the glucuronidation of other substrates of UGT1A1 and UGT1A9 (Brill et al., 2006). The current study focused on the potential of resveratrol to exhibit drug interactions via sulfation. Resveratrol was investigated for its ability to inhibit E2 sulfation. E2 was selected for the interaction studies as it is also known to be sulfated by SULT1E1 and SULT1A1 and is a prototypic estrogen hormone (Adjei and Weinshilboum, 2002; Miksits et al., 2005).
In this study, the sulfation of E2 was first studied in detail and our results, especially the role of SULT1E1, were in agreement with previous studies (Adjei and Weinshilboum, 2002; Nagar et al., 2006). A clear substrate inhibition of E2 sulfation was not observed in our studies as only one substrate concentration (50 nM) was included beyond 25 nM. It is to be noted that E2 has been reported to partially inhibit EST at concentrations greater than 20 nM via an allosteric binding site in mechanistic studies with purified EST and E2 (Zhang et al., 1998). Although the sulfation of resveratrol has been reported in other studies (Otake et al., 2000; Miksits et al., 2005), we believe it was important to investigate this topic further using pharmacologically relevant concentrations of resveratrol, i.e., 0.2 and 2 μM (besides 10 μM). Our studies indicate a more prominent possible role of SULT1A2 and SULT1A3 in the sulfation of resveratrol than previously reported, although we should emphasize that ability to catalyze a reaction is only one facet of relative in vivo importance. That is especially true for SULT1A2, which is not highly expressed in most human tissues (Nowell et al., 2005; Teubner et al., 2007). We have also determined the kinetics of resveratrol sulfation using four isoforms, SULT1A1*2, SULT1A2, SULT1A3, and SULT1E1.
In addition to human SULT1E1, the interaction experiments in our study were performed using subcellular fractions from intestine and liver, the primary tissues in which resveratrol is expected to be metabolized. The interaction studies indicated an inhibition of E2 sulfation by resveratrol using hepatic and jejunal tissue and recombinant SULT1E1. The pharmacological significance of this interaction needs to be determined. One of the major challenges with resveratrol has been the apparent lack of correlation between the low in vivo exposure of free resveratrol and the relatively high concentrations (>5 μM) necessary for in vitro activity (Gescher and Steward, 2003). Clinical studies have reported free resveratrol concentrations in the range of 21 nM after an oral dose of 25 mg (Walle et al., 2004) and of 0.5 to 2 μM after oral doses of 1 to 3 g (Boocock et al., 2007). The low Ki values (0.6, 1.1, and 2.3 μM) determined in our study for the inhibition of E2 sulfation by resveratrol are within this range of concentrations. However, a “therapeutic” dose of resveratrol for chemoprevention or cardioprotection remains to be established. The gastrointestinal tract may be a potential site of interaction of resveratrol with estrogen-containing therapeutic agents. Our studies have demonstrated significant sulfation activity using both E2 and resveratrol as substrates as well as a major interaction between the two compounds using S9 from intestinal samples. The synthetic oral contraceptive, ethinyl estradiol (EE), has been shown to be metabolized by SULT1E1 and SULT1A3 with low Km values of 7 and 19 nM, respectively (Schrag et al., 2004). As EE is an analog of E2 and as SULT1A3 and SULT1E1 are known to mediate resveratrol sulfation, it is possible that resveratrol may interact with oral contraceptives containing EE. We have previously shown that piceatannol, the cytochrome P450-mediated metabolite of resveratrol, is also capable of inhibiting E2 sulfation with low Ki values of 1.6 μM (liver), 0.4 μM (jejunum), and 1.2 μM (SULT1E1) (Furimsky et al., 2006). However, the clinical significance of this finding is not clear. Although piceatannol is known to be formed from resveratrol in in vitro studies, it has not been found in humans (Walle et al., 2004; Boocock et al., 2007) and is mainly found as glucuronide and sulfate conjugates in mice (unpublished data from our laboratory). The inhibitory effect of resveratrol may also be due to possible inhibition of E2 sulfation by resveratrol sulfates, which might themselves be substrates for SULT1E1, although it is not clear if these metabolites are substrates for SULT1E1.
Our experiments with hepatic and jejunal S9 provided additional insight into the sulfation of E2 and its inhibition by resveratrol, as it represents the whole tissue versus use of only recombinant SULT1E1 alone. These experiments were especially important because E2 is conjugated to its sulfate by several SULT isoforms besides SULT1E1. As sulfation of E2 in liver S9 was about twice that observed in jejunal S9, the expression levels of SULT1E1 and SULT1A1 were measured in the same batches of S9 used to determine SULT enzyme activity. The Western blots indicated a greater quantity of SULT1A1 in hepatic versus jejunal tissue, whereas the reverse was true with SULT1E1; i.e., protein levels of SULT1E1 were greater in jejunum versus liver. This finding was surprising because E2 sulfation rates, predominantly mediated by SULT1E1, were greater with S9 from liver samples in our studies and may be explained by the fact that SULT1A1 is also known to participate in E2 sulfation, with reported Km values of 0.59 μM (Nagar et al., 2006) and 31 μM (Adjei and Weinshilboum, 2002). Our results on E2 sulfation with individual SULT isoforms indicated a 2.5-fold higher formation of E2 sulfate by SULT1E1 compared with that for the *2 variant of SULT1A1 (i.e., SULT1A1*2). Given that this variant has been reported to have very low SULT1A1 activity (Adjei and Weinshilboum, 2002; Nagar et al., 2006), it is likely that the wild-type *1 variant of SULT1A1 may have exhibited E2 sulfation rates closer to those observed with SULT1E1. Hence, the observed higher rates of E2 sulfation in liver S9 may be due to the contributions of both SULT1A1 and SULT1E1. An evaluation of extrahepatic SULTs by Chen et al. (2003) showed no major differences in E2 sulfation between liver and small intestinal samples (Chen et al., 2003). However, that study also reported significant differences in intestinal SULT activities between individuals and between different sections of the intestine in the same individuals.
Sulfate conjugation plays a significant role in the metabolism of estrogens and contributes to maintain circulating levels of E1 and E2. E1 is known to be conjugated by SULT2A1 and SULT1E1 (Adjei and Weinshilboum, 2002). E2, as shown in several studies (including this one) is metabolized by SULT1E1 and SULT1A1. Factors that can alter the metabolism and intracellular concentration of estrogens are thought to play a role in the growth of human breast cancer cells and breast tumors (Miller and O'Neill, 1990). High estrogen levels can contribute to the pathophysiology of breast cancer because of ER-mediated events that stimulate cell proliferation. In addition, E1 and E2 may be converted to catechol estrogens, which can form quinones that can bind to DNA and form potentially carcinogenic complexes (Adjei and Weinshilboum, 2002; Cavalieri et al., 2006). Breast cancer cells such as MCF-7 lack the high-affinity SULT1E1-mediated conjugation of E2, which is considered to be a significant event in the transformation of normal breast cells to tumor cells (Falany et al., 2002). Therapeutic drugs or dietary components that can interfere with the sulfate conjugation of E2 can theoretically increase circulating levels of E2. Phytoestrogens and dietary components such as soy isoflavones (daidzein and genistein) and quercetin have been shown to inhibit E2 sulfation by SULT1E1 with reported Ki values of 14 μM (daidzein), 7 μM (genistein) and 0.61 μM (quercetin) (Otake et al., 2000; Nishiyama et al., 2002). Components of oral contraceptives such as 19-norethindrone acetate, ethynodiol diacetate, and mifepristone inhibit E2 sulfation by SULT1A1 with IC50 values of 0.193, 1.84, and 2.98 mM, respectively (Yasuda et al., 2005). Resveratrol has been reported to potently inhibit E2 sulfation in normal human mammary epithelial cells as well as recombinant SULT1E1 (Otake et al., 2000). However, these results (including the results from this study) have been generated during in vitro studies, and the physiological significance of such interactions remains to be established. In addition, estrogen metabolism could be shunted to other pathways (e.g., glucuronidation) in the event of inhibition of its sulfation by resveratrol. It should also be noted that resveratrol has been reported to induce phase 2 metabolizing enzymes (Baur and Sinclair, 2006). Therefore, the effect of an increase in estrogen levels could potentially be offset by antiestrogenic effects of resveratrol. On the other hand, it is also possible that estradiol glucuronidation may be inhibited by resveratrol, especially because UGT1A1 has been implicated in the glucuronidation of both resveratrol (Brill et al., 2006) and estradiol (Senafi et al., 1994). Another key enzyme that regulates estrogen homeostasis is estrone sulfatase, which hydrolyzes estrone sulfate (a major circulating form of plasma estrogen) to estrone (E1). Steroid sulfatases are known to be overexpressed in human breast carcinoma (Suzuki et al., 2003). It is not clear whether resveratrol can induce or inhibit steroid sulfatases or whether resveratrol mono- or disulfates are substrates for steroid sulfatases.
In conclusion, the results from this investigation indicate that resveratrol can inhibit the sulfate conjugation of E2 with Ki values in the low micromolar ranges. Hence, resveratrol can potentially influence the disposition of endogenous and exogenous compounds whose elimination is largely dependent on sulfation.
Footnotes
-
This study was supported by funds from the National Cancer Institute ((N01-CN-43305; to C.E.G.), the National Institutes of General Medical Sciences (R01 GM35720 and U01 GM61388), The Pharmacogenetics Research Network (to A.A. and R.W.) and the PhRMA Foundation (A Center of Excellence in Clinical Pharmacology Award to R.W.).
-
Preliminary results from this study were presented in “Inhibition of human hepatic and jejunal estradiol sulfation by resveratrol and piceatannol” (Furimsky AM, Sharp LE, Bettridge A, Kapetanovic IM, Green CE, Iyer LV) at Experimental Biology 2006 in San Francisco, CA, April 1–5, 2006.
-
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
-
doi:10.1124/dmd.107.016725.
-
ABBREVIATIONS: ER, estrogen receptor; SULT, sulfotransferase; UGT, uridine diphosphate glucuronosyltransferase; E2, 17β-estradiol; EST, estrogen sulfotransferase, SULT1E1; DTT, dithiothreitol; PAPS, adenosine 3′-phosphate 5′-phosphosulfate; BSA, bovine serum albumin; HPLC, high-performance liquid chromatography; PDA, photodiode array detector; LC, liquid chromatography; MS, mass spectrometry, MS/MS, tandem mass spectrometry.
- Received June 1, 2007.
- Accepted October 17, 2007.
- The American Society for Pharmacology and Experimental Therapeutics