Comprehensive Investigation of Stereoselective Food Drug Interaction Potential of Resveratrol on Nine P450 and Six UGT Isoforms in Human Liver Microsomes

The stereoselectivity of the food drug inhibition potential of resveratrol on cytochrome P450s and uridine 5′-diphosphoglucuronosyl transferases was investigated in human liver microsomes. Resveratrol enantiomers showed stereoselective inhibition of CYP2C9, CYP3A, and UGT1A1. The inhibitions of CYP1A2, CYP2B6, and CYP2C19 by resveratrol were stereo-nonselective. The estimated Ki values determined for CYP1A2 were 13.8 and 9.2 μM for trans- and cis-resveratrol, respectively. Trans-resveratrol noncompetitively inhibited CYP3A and UGT1A1 activities with Ki values of 23.8 and 27.4 μM, respectively. Trans-resveratrol inhibited CYP1A2, CYP2C19, CYP2E1, and CYP3A in a time-dependent manner with Ki shift values >2.0, while cis-resveratrol time-dependently inhibited CYP2C19 and CYP2E1. The time-dependent inhibition of trans-resveratrol against CYP3A4, CYP2E1, CYP2C19, and CYP1A2 was elucidated using glutathione as a trapping reagent. This information helped the prediction of food drug interaction potentials between resveratrol and co-administered drugs which are mainly metabolized by UGT1A1, CYP1A2, CYP2C19, CYP2E1, and CYP3A.


Introduction
Resveratrol (3,5,4 -Trihydroxystilbene, RVT), a polyphenolic phytoalexin, is a common constituent found in many plants or beverages such as berries, grapes, peanuts, soy, and in wine, cranberry, and grape juices, as well as in many other plant products [1,2]. It is considered to be one of the anti-inflammatory and anti-oxidant constituents in grape juice and red wine [3]. Owing to various pharmacological effects of RVT [4][5][6], this molecule has been consumed in the forms of dietary supplements.
As the consumption of dietary supplements for health benefits has been increasing every year, food drug interactions (FDI) have been a focus of intensive research. In 2019, the global dietary supplement market was valued at 163 billion [7]. Most of these FDI might be attributed to the induction or inhibition of drug transporters and drug-metabolizing enzymes [8]. Grapefruit juice is a well-known example of such FDI, and bergamottin is considered to be the major inhibitor of CYP3A [9]. Schisandra extracts also markedly increased the plasma concentration of tacrolimus by inhibiting the CYP3A4 isoform in liver transplant patients [10].
mottin is considered to be the major inhibitor of CYP3A [9]. Schisandra extracts also m edly increased the plasma concentration of tacrolimus by inhibiting the CYP3A4 iso in liver transplant patients [10].
Resveratrol naturally occurs in two isomeric forms, trans-and cis-resveratrol (cR ( Figure 1). RVT can isomerize from the trans-to the cis-isomer under ultraviolet-ligh posure [21], therefore, cRVT is also mainly present in red wine and grape juice tog with tRVT. tRVT concentrations in red wine ranged from 0.22 to 47.8 μM (0.05 μg/mL), whereas cRVT ranged from 0.18 to 52.6 μM (0.04~12.0 μg/mL) [22][23][24]. De the wide observation of cRVT in commercial wine and grape juice products, studie the inhibitory potential of cRVT against drug-metabolizing enzymes have been lim cRVT inhibited CYP2C19-mediated S-mephenytoin 6-hydroxylation and CYP3A-m ated testosterone 6β-hydroxylation with IC50 values of 41.1 and 4.5 μM, respecti whereas it had negligible effects on other P450s (IC50 > 100 μM) [1]. Though the tim pendent P450 inhibitory effects of tRVT are well known, no reports have been repo on cRVT. In addition, no detailed studies have been performed to elucidate the inhib mode and potency of P450s and UGTs by cRVT. A chiral phytochemical can stereoselectively modulate P450-catalyzed biotran mation. For example, R-naringenin has a 2-fold more potent inhibition for CYP2C9 CYP3A inhibition compared with that of the S-enantiomer [25]; (-)-tetrahydropalm significantly inhibits CYP2D6-mediated dextromethorphan O-demethylase activit times more extensively than (+)-tetrahydropalmatine [26], and quinidine stereoselect A chiral phytochemical can stereoselectively modulate P450-catalyzed biotransformation. For example, R-naringenin has a 2-fold more potent inhibition for CYP2C9 and CYP3A inhibition compared with that of the S-enantiomer [25]; (-)-tetrahydropalmatine significantly inhibits CYP2D6-mediated dextromethorphan O-demethylase activity 15 times more extensively than (+)-tetrahydropalmatine [26], and quinidine stereoselectively inhibits CYP2D6-mediated debrisoquine 4-hydroxylation as much as quinine [27,28]. Until now, little information is available on the stereoselective inhibition of RVT on the UGT and P450 activities. The four objectives of this study were: (1) to investigate the inhibition mode and kinetics of cRVT against six UGTs and nine P450s in HLMs; (2) to elucidate the timedependent inhibition (TDI) of cRVT against P450s; (3) to compare the inhibitory potential of cRVT and tRVT against these enzymes, and (4) to elucidate the TDI mechanism of RVT in human recombinant P450 isoforms (rP450s).

Stereoselective Inhibition of Resveratrol against Cytochrome P450 Activity
The stereoselective inhibition of RVT on the activity of nine cytochrome P450 (P450) isoforms were evaluated as previously described with minor modifications [29,30]. RVT and each P450 substrates were dissolved in methanol. The microsomal incubation was performed in 1.5 mL amber tubes using two substrate cocktail sets (set A: amodiaquine, bupropion, dextromethorphan, diclofenac, S-mephenytoin, and phenacetin; set B: chlorzoxazone, coumarin, midazolam, and nifedipine) ( Table 1). These substrate cocktails were validated by comparison of the inhibition obtained from incubation of each individual index substrate alone and the substrate cocktail in our previous study [29,30]. The incubation mixture included RVT (0, 0.5, 2, 5, 20, and 50 µM), P450 probe substrate cocktail set, 0.1 M potassium phosphate buffer, and 0.25 mg/mL HLMs. After preincubation (5 min, 37 • C), the reaction initiated upon the addition of an NADPH-generating system (1.3 mM β-NADP + , 3.3 mM MgCl 2 , 3.3 mM G6P, and 1.0 unit/mL G6PDH). Following further incubation for 10 min, acetonitrile containing internal standard (IS, 7 nM trimipramine) was added to terminate the reaction. After centrifugation (4 • C, 5 min, 14,000 rpm), the aliquots of supernatants were assayed by liquid chromatography-tandem mass spectrometry (LC-MS/MS). The experiment of RVT was conducted in amber tubes protected from light in order to avoid isomerization. All microsomal incubations were performed in triplicate.
The TDI of RVT against nine P450s was examined using an IC 50 shift assay [31]. Each RVT (0~50 µM) was pre-incubated with HLMs for 30 min at 37 • C in the presence of an NADPH-generating system. After preincubation, P450 probe substrate cocktail set was added to start the reaction prior further incubation for 10 min. Afterwards, it was carried out the same as the above experimental method.

Kinetic Characterization of trans-Resveratrol on UGT1A1 in HLMs
We used HLMs to elucidate the constants and mechanisms for tRVT inhibition of UGT1A1 which had IC 50 values less than 20 µM. tRVT (0~50 µM) was added into the reaction solutions, each of which contained concentrations of SN-38 (0.5, 2, and 10 µM). The concentrations of SN-38 glucuronide were measured by LC-MS/MS as previously described [32]. The lower limits of quantification for SN-38 glucuronide was 5 nM, respectively. The inter-assay precision values for all of the samples were less than 10.7%.

Characterization of Glutathione Conjugates of Resveratrol in Recombinant Cytochrome P450 Isoforms
tRVT (20 µM) was incubated (37 • C, 45 min) with rP450s (20 pmol/mL) in 100 mM phosphate buffer in the presence of glutathione (10 mM) and NADPH (1 mM). Control incubations in the absence of glutathione and NADPH were performed. Incubations were stopped by the addition of acetonitrile. The supernatants were concentrated and reconstituted with methanol (100 µL). Samples were assayed by LC-high resolution mass spectrometry (LC-HRMS) [34].

LC-MS/MS Analysis
All samples were assayed using a Shimadzu LC-MS 8060 tandem mass spectrometer combined with a Nexera X2 liquid chromatography equipped with an electrospray ionization device (Shimadzu, Kyoto, Japan). Nine P450-and six UGT-isoform specific metabolites were separated on a Kinetex XB-C18 column (100 Å, 2.6 µm, 100 × 2.1 mm; Phenomenex, Torrance, CA, USA). The mobile phase consisted of 0.1% formic acid (FA) containing water (A) and 0.1% FA containing acetonitrile (B), and elution condition was set as follows: the gradient was maintained with 8% B for 0.5 min and then linearly increased to 60% in 5 min and held for 1 min, after linearly reduced to 8% in 6.1 min; finally, the original gradient was applied and maintained in 9 min [30]. Gradient elution conditions were set as follows: 0% B, 30% B (0-1 min), 50% B (1-5 min), and 0% B (5.1-8 min) [32]. The flow rate was 0.2 mL/min. Ionization voltages in positive and negative ionization modes were 4000 V and −3500 V, respectively. Quantitation was conducted in selected reaction monitoring (SRM) modes for each metabolite (Table 1).
Using a Vanquish HPLC combined with a QExactive Focus Orbitrap mass spectrometer (Thermo Fisher Scientific Inc., Waltham, MA, USA), we determined the glutathione adducts produced by rP450s. A Kinetex C18 column was also used to separate the analytes. The mobile phase was set as 80% acetonitrile in water containing 0.1% FA. The flow rate was set at 0.2 mL/min. Data acquisition was carried in the total ion scan mode (m/z 100-700) with a resolution of 70,000, and MS/MS spectra were acquired in the product ion scan mode (m/z 50-580) at a resolution of 17,500. Parallel reaction monitoring (PRM) conversion m/z 550.1495 was used for the identification of GSH conjugate.

Data Analysis
All results were acquired from three replicates in different microsomal incubations. IC 50 values were determined by nonlinear regression analysis using WinNonlin (Pharsight, Mountain View, CA, USA). We used WinNonlin to estimate the apparent kinetic parameters of inhibitory activity (K i ) and the type of inhibitory activity by several criteria, including visual inspection of Lineweaver-Burk double reciprocal plots, Dixon plots, and secondary plots of Lineweaver-Burk plots versus RVT concentrations in each inhibitory model [35,36].

Stereoselective Inhibition of Uridine 5 -Diphosphoglucuronosyl Transferase Isoform Activities by Resveratrol
UGT inhibition by phytochemicals is known to be one of the most important factors for FDI. For example, pretreatment with UGT1A1 inhibitor psoralidin, a natural phenolic component found in the seeds of Psoralea corylifolia, increased the toxicity of irinotecan, an UGT1A1 substrate, as indicated by the severe colon histology damage in mice [40]. The plasma concentrations of emodin were significantly increased by pretreatment with stilbene glucoside, indicating that stilbene glucoside significantly affected the pharmacokinetics of emodin through the inhibition of UGT1A8 mRNA expression [41]. Pretreatment with soybean induces the induction of UGT, resulting in a decrease of the bioavailability of valproic acid [42]. However, there is still very limited data on the UGT-mediated drug interaction potential of resveratrol. Therefore, we investigated the stereoselective inhibitory potential of RVT for six UGTs' isoform activities using HLMs (Table 3). tRVT inhibited UGT1A1-mediated SN-38 glucuronidation with an IC 50 value of 9.57 µM, similarly to a previous finding (K i = 6.2 µM) [20] in a stereoselective manner. The inhibition of the other five UGTs by tRVT and cRVT was considered negligible (IC 50 > 50.0 µM).

Characterization of Glutathione Conjugates of trans-Resveratrol in Recombinant Cytochrome P450 Isoforms
RVT is known to be converted to reactive quinone methides by CYP3A4 [43,44] or CYP1A2 [34,45]. These quinone methide intermediates might react with CYP3A4 or CYP1A2 through covalent modification with these P450s [34,43]. The formation of a reactive intermediate-P450 complex has been reported to play an essential role in the TDI of P450 by RVT. Glutathione might be used as a trapping agent to identify quinone methide because the latter is unstable and cannot be directly identified [34]. The glutathione conjugates with the oxidized intermediate of RVT were identified in human liver microsomal incubation samples [34]. In this study, tRVT showed TDI of CYP1A2, CYP2C19, CYP2E1, and CYP3A activities with a K i shifting > 5.0 in HLMs (Table 4).
To elucidate the TDI mechanism of tRVT against CYP1A2, CYP2C19, CYP2E1, and CYP3A, tRVT was incubated with rP450s in the presence of NADPH and glutathione.  [46,47]. The glutathione conjugate was also observed in the incubation samples with rCYP1A2, rCYP2C19, and rCYP2E1 ( Figure 5). Our results showed that CYP1A2, CYP2C19, CYP2E1, and CYP3A4 were involved in the formation of reactive quinone methide of tRVT ( Figure 5).

Characterization of Glutathione Conjugates of trans-Resveratrol in Recombinant Cytochrome P450 Isoforms
RVT is known to be converted to reactive quinone methides by CYP3A4 [43,44] or CYP1A2 [34,45]. These quinone methide intermediates might react with CYP3A4 or CYP1A2 through covalent modification with these P450s [34,43]. The formation of a reactive intermediate-P450 complex has been reported to play an essential role in the TDI of P450 by RVT. Glutathione might be used as a trapping agent to identify quinone methide because the latter is unstable and cannot be directly identified [34]. The glutathione conjugates with the oxidized intermediate of RVT were identified in human liver microsomal incubation samples [34]. In this study, tRVT showed TDI of CYP1A2, CYP2C19, CYP2E1, and CYP3A activities with a Ki shifting > 5.0 in HLMs (Table 4).
To elucidate the TDI mechanism of tRVT against CYP1A2, CYP2C19, CYP2E1, and CYP3A, tRVT was incubated with rP450s in the presence of NADPH and glutathione. LC-HRMS analyses indicated that there was one glutathione conjugate ([M+H] -, m/z 550.1495, tR = 0.89 min) formed in rCYP3A4. UPLC-HRMS analyses of the peak responsible for this GSH conjugate displayed a protonated molecule [M+H] -at m/z 550.1495 (mass error < 2 ppm), 323 Da higher than that of tRVT. This suggested that tRVT first oxidizes before it is being conjugated with one molecule of glutathione (MW = 307.3). The product ion scan spectrum of the glutathione conjugate by fragmenting m/z 550.1495 through collision generated characteristic daughter ions at m/z 428.1128 suggests the loss of a methyl dihydroxybenzene residue (−122 Da) ( Figure 5). The fragment ion of m/z 306.0762 and 272.0885 was produced by glutathione moiety and cleavage of the cysteinyl C-S bond, respectively. The fragment ions observed from the glutathione moiety and cleavage of the cysteinyl C-S bond are the most typical ions found in glutathione conjugates [46,47]. The glutathione conjugate was also observed in the incubation samples with rCYP1A2, rCYP2C19, and rCYP2E1 ( Figure 5). Our results showed that CYP1A2, CYP2C19, CYP2E1, and CYP3A4 were involved in the formation of reactive quinone methide of tRVT ( Figure 5).

Evaluation of Food Drug Interaction Potential of Resveratrol
In previous studies, red wine and tRVT were found to change the pharmacokinetics of drugs which are substrates of CYP3A [48]. Zhan et al. (2015) reported that the multiple dose of RVT (100 or 200 mg/kg) significantly increased the AUC and C max of aripiprazole (oral, 3 mg/kg) in rats [49]. Animal studies also showed that a seven-day co-administration with RVT (100 mg/kg) significantly increased oral pharmacokinetics of alogliptin, saxagliptin, and sitagliptin in rats through CYP3A inhibition [50]. In addition, four weeks of RVT dosing (1 g, once daily) inhibited the phenotypic indices of CYP2C9, CYP2D6, and CYP3A, and increased the metabolic ratio (caffeine/paraxanthine ratio) of CYP1A2 in healthy subjects [51].
In contrast to the extensive studies on FDI with CYP3A substrates, there are limited data on FDI with other P450s. We anticipated the clinical FDI risk induced by RVT based on each of the IC 50 values. tRVT inhibited CYP2C19 and CYP2E1 activities with IC 50 values of 8.1 and 9.8 µM, similar to CYP3A inhibition (IC 50 = 5.6~16.6 µM) in a concentrationand time-dependent manner. Considering that tRVT participates in the pharmacokinetic intervention of dipeptidyl peptidase-4 inhibitors [50] and diltiazem [52] by inhibiting CYP3A-mediated biotransformation in rats, tRVT might interact with CYP2E1 or CYP2C19 substrate drugs such as clopidogrel [53], omeprazole [54], acetaminophen [55], and theophylline [56]. cRVT may also interact with CYP2E1 or CYP2C19 substrate drugs, because their inhibitory potential (IC 50 = 10.8~19.8 µM) is stronger than CYP2C9 inhibition (IC 50 = 23.8 µM). A dose of RVT received from food supplements may reach up to 2-5g/day, and RVT concentrations in plasma, after oral administration, may achieve micromolar concentrations [13,57]. In rats, oral administration of tRVT (100 mg/kg) resulted in a significant increase of S-warfarin plasma concentration and international normalized ratio through CYP2C9 inhibition [58]. In addition, we again predicted in vivo FDI potential of RVT using the volume per dose index (VDI (L) = RDI/IC 50 , RDI means recommended daily intake), volume in which the daily dose would be dissolved to reach the corresponding IC 50 concentration [59]. The VDI cut-off value for enzymes present in the liver is 5.0 L [60], and the recommended daily intake value for RVT is 450 mg per day [61]. VDI values of tRVT for inhibiting CYP2C19 and CYP2E1 were 243 and 202 L, respectively, while those of cRVT were 183 and 100 L, respectively. The calculated VDI values of RVT exceeded 100 L per unit dose with S-mephenytoin and chlorzoxazone, indicating RVT might have potential to inhibit biotransformation by CYP2C19 and CYP2E1 enzyme in vivo. However, clinical studies are needed to assess whether RVT affects drug metabolism by CYP2C19 and CYP2E1 enzymes in vivo.
An in vivo FDI via the inhibition of a drug-metabolizing enzyme might occur if the ratio of the maximum plasma concentration (C max )/K i exceeded 1.0 [35,62]. After oral administration of RVT (5 g, single dose) to healthy subjects, the plasma C max value of RVT was approximately 2.36 µM [63]. The plasma C max value of RVT increased to 8.78 µM after a single oral dose of micronized RVT formulation (5 g) in humans [57]. The calculated values of C max /K i were 1.48~5.49, 0.51~8.78, 0.26~0.96, and 0.24~0.90 from the inhibition constant of tRVT against CYP1A2, CYP3A, CYP2E1, and CYP2C19 (K i = 1.6 µM, 1.0~4.6 µM, 9.1 µM, and 9.8 µM, respectively), suggesting that RVT might have FDI potential with these P450 substrate drugs at high doses.
Here, we demonstrated the stereoselective inhibitory potential of RVT against both P450s and UGTs in HLMs. We also determined their TDI against P450s, and indirectly confirmed the formation of reactive quinone methide intermediate using glutathione as a trapping agent. tRVT stereoselectively inhibited CYP2C9, CYP3A, and UGT1A1 activities whereas cRVT had negligible inhibition on them (IC 50 > 50 µM). In addition, tRVT time-dependently inhibited CYP1A2, CYP2C19, CYP2E1, and CYP3A with K i shift values >2.0, while cis-resveratrol time-dependently inhibited CYP2C19 and CYP2E1. From the inhibition potential (IC 50 , K i , and VDI values) of RVT against P450s and UGTs, RVT at high doses might cause significant pharmacokinetic FDI with co-administered drugs predominantly metabolized by UGT1A1, CYP3A, CYP2E1, CYP2C19, and CYP1A2 in humans.