Natural Products Inhibition of Cytochrome P450 2B6 Activity and Methadone Metabolism

Abstract

Methadone is cleared predominately by hepatic cytochrome P450 (CYP) 2B6-catalyzed metabolism to inactive metabolites. CYP2B6 also catalyzes the metabolism of several other drugs. Methadone and CYP2B6 are susceptible to pharmacokinetic drug–drug interactions. Use of natural products such as herbals and other botanicals is substantial and growing, and concomitant use of prescription medicines and non-prescription herbals is common and may result in interactions, often precipitated by CYP inhibition. Little is known about herbal product effects on CYP2B6 activity, and CYP2B6-catalyzed methadone metabolism. We screened a family of natural product compounds used in traditional medicines, herbal teas, and synthetic analogs of compounds found in plants, including kavalactones, flavokavains, chalcones and gambogic acid, for inhibition of expressed CYP2B6 activity and specifically inhibition of CYP2B6-mediated methadone metabolism. An initial screen evaluated inhibition of CYP2B6-catalyzed 7-ethoxy-4-(trifluoromethyl) coumarin O-deethylation. Hits were further evaluated for inhibition of racemic methadone metabolism, including mechanism of inhibition and kinetic constants. In order of decreasing potency, the most effective inhibitors of methadone metabolism were dihydromethysticin (competitive, K_i 0.074 µM), gambogic acid (noncompetitive, K_i 6 µM), and 2,2’-dihydroxychalcone (noncompetitive, K_i 16 µM). Molecular modeling of CYP2B6-methadone and inhibitor binding showed substrate and inhibitor binding position and orientation and their interactions with CYP2B6 residues. These results show that CYP2B6 and CYP2B6-catalyzed methadone metabolism are inhibited by certain natural products, at concentrations which may be clinically relevant.

SIGNIFICANCE STATEMENT

This investigation identified several natural product constituents which inhibit in vitro human recombinant CYP2B6 and CYP2B6-catalyzed N-demethylation of the opioid methadone. The most potent inhibitors (K_i) were dihydromethysticin (0.074 µM), gambogic acid (6 µM) and 2,2’-dihydroxychalcone (16 µM). Molecular modeling of ligand interactions with CYP2B6 found that dihydromethysticin and 2,2’-dihydroxychalcone bound at the active site, while gambogic acid interacted with an allosteric site on the CYP2B6 surface. Natural product constituents may inhibit CYP2B6 and methadone metabolism at clinically relevant concentrations.

Introduction

Methadone is a synthetic mu receptor opioid used clinically for relief of acute, chronic, and cancer pain, and for treatment of opioid use disorder (Kharasch, 2011; Oesterle et al., 2019; Hanna and Senderovich, 2021). Methadone is chiral and generally used clinically as a racemic mixture of R- and S- enantiomers, although R-methadone alone is used in some countries and S-methadone is in Phase 3 investigation for major depressive disorder (Fava et al., 2023). R-methadone binding affinity for mu-receptors and analgesic potency are 30- to 50-fold greater than the S enantiomer (Scott et al., 1948; Fava et al., 2023). Thus, R-methadone is responsible for the majority of racemic methadone analgesia. Methadone is metabolized predominantly by N-demethylation to the inactive metabolite 2-ethyl-1,5-dimethyl-3,3-diphenylpyrrolidine (EDDP), as a major route of systemic clearance. In vitro, both human liver microsomal and cDNA-expressed cytochrome P450 (CYP) 2B6 and CYP3A4 efficiently catalyze methadone N-demethylation, and CYP2B6 metabolism is stereoselective, while CYP3A4 is not (Gerber et al., 2004; Kharasch et al., 2004; Totah et al., 2007, 2008; Chang et al., 2011). In contrast, CYP2B6, rather than CYP3A4, is the predominant CYP isoform responsible for human methadone metabolism in vivo (Kharasch and Stubbert, 2013; Kharasch, 2017; Miano et al., 2022). Methadone is susceptible to CYP2B6-mediated inhibitory and inductive pharmacokinetic drug interactions (Kharasch, 2017; Younis et al., 2019). For example, ticlopidine, a mechanism-based CYP2B6 inhibitor, reduced methadone metabolism and clearance (Kharasch and Stubbert, 2013). Similar effects occurred with the CYP2B6 inhibitor voriconazole (Liu et al., 2007). CYP2B6 also catalyzes the metabolism of several other drugs, including efavirenz, nevirapine, ketamine, propofol, bupropion, nicotine, artemesinin, cyclophosphamide, and ifosfamide.

Use of natural products such as herbals and other botanicals (Paine, 2020) is substantial and the market has grown rapidly, with sales of herbal dietary supplements in the United States exceeding $11 billion in 2020 (Smith et al., 2021). Concomitant use of prescription medicines and non-prescription herbal products is common (Harris et al., 2022) and may result in reduced treatment effect or adverse effects (Paine et al., 2018). Unregulated use of natural products coupled with their unknown effects on conventional drug pharmacokinetics and pharmacodynamics may lead to unwanted or potentially harmful herb-drug interactions, and the identification of such interactions has been highlighted as a clinically significant information gap (Birer-Williams et al., 2020; Paine, 2020; Weber and Hopp, 2020). Inhibition of CYP is a common mechanism by which natural products may precipitate natural product-drug interactions (Johnson et al., 2018; Paine et al., 2018; Rombolà et al., 2020; Cox et al., 2021). Although CYPs 1A2, 2C9, 2C19, 2D6, and 3A4 are more commonly evaluated in screens for inhibition by natural products, CYP2B6 is comparatively infrequently assessed (Anke and Ramzan, 2004; Birer-Williams et al., 2020).

To understand the potential for natural product-methadone interactions emanating from inhibition of CYP2B6-mediated metabolism, we screened a family of natural product compounds used in traditional medicines, herbal teas, and synthetic analogs of compounds found in plants, for inhibition of CYP2B6 activity and specifically inhibition of CYP2B6-mediated methadone metabolism. We screened kavalactones and flavokavains isolated from kava plants (Piper methysticum) native to the South Pacific Islands (Dharmaratne et al., 2002), the active ingredients extracted from Asiatic toads Bufo gargarizans (bufalin, resibufogenin, cinobufotalin and cinobufagin) (Yang et al., 2008; Emam et al., 2012; Lan et al., 2019), active ingredients extracted from plants Euphorbia lathyris (euphorbia steroid) (Park et al., 2015; Xiao et al., 2022), Garcinia hanburyi (gambogic acid) (Hatami et al., 2020), Wikstroemia indica (daphnoretin) (Ko et al., 1993), as well as chalconoids (naturally occurring or synthetic analogs), for CYP2B6 inhibition. The major ingredients of kava with reported biologic activities are kavalactones (Fig. 1) and flavokavains (Table 1). Kava contains 19 kavalactones (of which 6 account for 96% of pharmacologic activity-kavain, dihydrokavain, methysticin, dihydromethysticin, yangonin, desmethoxyyangonin) and the minor constituent chalcones (flavokavains A, B, C) (Olsen et al., 2011; Soares et al., 2022). Chalcone compounds, also known as chalconoids, are a large family of plant-derived polyphenolic flavonoids with a common chemical scaffold of α,β-unsaturated ketone consisting of two aromatic rings (1,3-diaryl-2-propen-1-one). Chalcones exhibit a variety of biologic activities and are chemical source for developing new pharmaceutical agents (Nowakowska, 2007; Zhuang et al., 2017). Gamboge, extracted from a tree resin, has been used for centuries to treat infections and tumors in traditional Chinese medicine. The active ingredient gambogic acid has antitumor, antiangiogenic, and antimetastatic activities (Li et al., 2022).

Fig. 1.

Structures of kavalactones (1–6), gambogic acid (7), euphorbiasteroid (8), daphnoretin (9), and cardiotonic steroids (10–13).

TABLE 1.

Chalconoids screened in the present study, including flavokavains, other naturally occurring chalcones, and synthetic derivatives

	Chalconoids	R2	R3	R4	R5	R2’	R4’	R5′	R6’
14	2,2’-dihydroxychalcone	OH				OH
15	3,2’-dihydroxychalcone		OH			OH
16	3,4-dimethoxychalcone		OCH3	OCH3
17	2',4'-dimethoxychalcone					OCH3	OCH3
18	4-methoxy-4'-(methylthio)chalcone			OCH3			SCH3
19	4,2',5′-trihydroxychalcone			OH		OH		OH
20	4'-hydroxy-4-methoxy-2'-methylchalcone			OCH3		CH3	OH
21	4,4'-dimethoxy-2'-hydroxychalcone			OCH3		OH	OCH3
22	Flavokavain B					OH	OCH3		OCH3
23	2',4',6'-trimethoxychalcone					OCH3	OCH3		OCH3
24	2,2'-dihydroxy-4',6'-dimethoxychalcone	OH				OH	OCH3		OCH3
25	2'-hydroxy-2,3,4'-trimethoxychalcone	OCH3	OCH3			OH	OCH3
26	2'-hydroxy-2,3,5′-trimethoxychalcone	OCH3	OCH3			OH		OCH3
27	2'-hydroxy-2,4,5′-trimethoxychalcone	OCH3		OCH3		OH		OCH3
28	2',4'-dihydroxy-3,4-dimethoxychalcone		OCH3	OCH3		OH	OH
29	3-methoxy-4,2',5′-trihydroxychalcone		OCH3	OH		OH		OH
30	2'-hydroxy-3,4,5′-trimethoxychalcone		OCH3	OCH3		OH		OCH3
31	3,4-dimethoxy-2'-hydroxy-5′-methylchalcone		OCH3	OCH3		OH		CH3
32	3,4,2',5′-tetramethoxychalcone		OCH3	OCH3		OCH3		OCH3
33	3-nitro-2',4',6'-trimethoxychalcone		NO3			OCH3	OCH3		OCH3
34	4',6'-dimethoxy-2'-hydroxy-4-methylchalcone			CH3		OH	OCH3		OCH3
35	Flavokavain A			OCH3		OH	OCH3		OCH3
36	4-methyl-2',4',6'-trimethoxychalcone			CH3		OCH3	OCH3		OCH3
37	4-dimethylamino-2',4',6'-trimethoxychalcone			N(CH3)2		OCH3	OCH3		OCH3
38	3,2'-dihydroxy-4,4',6'-trimethoxychalcone		OH	OCH3		OH	OCH3		OCH3
39	3-hydroxy-2',4,4',6'-tetramethoxychalcone		OH	OCH3		OCH3	OCH3		OCH3
40	4-hydroxy-2',3,4',6'-tetramethoxychalcone		OCH3	OH		OCH3	OCH3		OCH3
41	4'-hydroxy-2'-methyl-3,4,5-trimethoxychalcone		OCH3	OCH3	OCH3	CH3	OH
42	4-benzyloxy-2'-hydroxy-3,4',6'-trimethoxychalcone		OCH3	OBz		OH	OCH3		OCH3
43	4-benzyloxy-2',3,4',6'-tetramethoxychalcone		OCH3	OBz		OCH3	OCH3		OCH3

These compounds were chosen for study due to their potential interactions with cytochrome P450 enzymes. For example, kava extract inhibited CYPs 1A2, 2C9, 2C19, 2D6, and 3A4 in human liver microsomes (Mathews et al., 2002). Daphnoretin is a bis-coumarin composed of two coumarin moieties, and binding to CYP2B6 might be expected as coumarin is an excellent substrate for multiple CYPs including CYP2B6. Licochalcone A, a chalconoid isolated from liquorice, inhibited activities of CYPs 1A2, 2C19, 2C8, 2C9, and 3A4 in human liver microsomes (He et al., 2015). Structures of kavalactones and the other natural products studied are shown in Fig. 1. The 30 chalcones screened are listed in Table 1.

Initial screening was performed using 7-ethoxy-4-(trifluoromethyl) coumarin (7-ETFMC) as the CYP2B6 substrate. 7-ETFMC is not a clinically relevant drug but the assay is a quick and cost-effective method for screening large numbers compounds. This initial screen was followed by another screen using racemic methadone. Relevant interactions with racemic methadone were further characterized to determine mechanism of inhibition and kinetic constants. Structural modeling was also performed for docking the ligands into the structure of CYP2B6 to assess receptor-ligand interactions.

Materials and Methods

Materials

Chalcones were obtained from Indofine Chemical (Hillsborough Township, NJ). Natural products, including flavokavains, gambogic acid, cardiotonic steroids (bufalin, resibufogenin, cinobufotalin, and cinobufagin, originally from the National Institute for Food and Drug Control, Beijing, China), daphnoretin, euphorbiasteroid, and kavalactones (originally from Dr Jianqiao Gu at the University of Arizona College of Pharmacy (Tucson, AZ), were all generous gifts from Dr. Sheng-Xiang (Samuel) Qiu at Washington University St Louis. 7-Ethoxy-4-trifluoromethylcoumarin (7-ETFMC) was obtained from AnaSpec (Fremont, California). Racemic methadone hydrochloride and 7-hydroxy-4-trifluoromethylcoumarin were purchased from Sigma Aldrich (St. Louis, MO). rac-EDDP perchlorate and EDDP-d3 were from Cerilliant (Round Rock, Texas). All other reagents were from Sigma Aldrich (St. Louis, MO).

Construction of plasmids for CYP2B6, P450 reductase (POR) and cytochrome b5, and generation of recombinant baculoviruses were carried out as described previously (Gadel et al., 2015). Expression of recombinant proteins of CYP2B6, POR, and cytochrome b5 were carried out in insect cells by triple infection (Gadel et al., 2015; Wang et al., 2018a). Protein contents of CYP2B6, POR, and b5 were measured as described previously (Wang et al., 2018a). The ratio of CYP2B6:POR:b5 was 1:2:4.

Natural products were dissolved in DMSO and were diluted in 100 mM potassium phosphate buffer pH 7.4 before addition to the reaction mixture. During screening, DMSO levels varied based on concentration of inhibitor up to 0.2% (v/v) in the assay mixture. When using methadone as a substrate, DMSO was kept at a constant 0.1% (v/v) of final volume at all inhibitor concentrations tested.

Primary Screening with 7-Ethoxy-4-Trifluoromethylcoumarin Assay

The initial screen of natural products was carried out using 7- ETFMC as the substrate of CYP2B6 following procedures outlined previously with modifications (Chang et al., 2006). CYP2B6 catalyzes O-deethylation of 7-ETFMC to form the metabolite 7-hydroxy-4-trifluoromethylcoumarin which can be followed by spectrofluorometric analysis. A typical assay mixture of 180 µl containing7-ETFMC, CYP2B6/POR/b5 and natural products in 100 mM potassium phosphate buffer (pH 7.4) was placed in a 96-well plate. After preincubation for 5 minutes at 37°C, the reaction was initiated by adding 20µl NADPH regenerating system (final concentrations: 10 mM glucose 6-phosphate, 1 mM β-NADP, 1 U/ml glucose-6-phosphate dehydrogenase, and 5 mM magnesium chloride, preincubated at 37°C for 10 minutes). The total reaction volume was 200 µl, and the final concentrations were 50 µM 7-ETFMC and 25 pmol/ml CYP2B6. The reaction was terminated after 15 minutes incubation at 37°C by adding 40 µl 20% trichloroacetic acid. The plate was centrifuged at 2500 rpm for 5 minutes to remove precipitated proteins. Fifty microliters of supernatant was transferred to a new 96 deep-well plate, and mixed with 950 μl of 100 mM Tris-HCl buffer at pH 9.0. 200 µl samples from each well of the deep-well plate were transferred to a 96-well clear bottom plate for fluorescence measurement, along with the standard samples. 7-hydroxy-4-trifluoromethylcoumarin (7-HTFMC) was used as the standard to generate a calibration curve. The calibration samples were prepared in 100 mM Tris-HCl buffer pH 9.0 at the concentrations of 0, 0.01, 0.025, 0.05, 0.075, 0.1, 0.25, 0.5, 0.75, and 1 µM. Two hundred microliters of calibration samples were transferred to the 96-well clear bottom plate containing the incubation samples, and processed for fluorescence measurement at an emission wavelength of 510 nm and excitation wavelength of 410 nm using a BioTek Synergy MX Microplate Reader. Initial screenings of natural products were done with inhibitor concentrations of 0.1, 1, and 10 µM. Samples that displayed inhibitory effects in the initial screening were further tested with inhibitor concentration expanded to a wide range of 0.1, 0.5, 1, 5, 10, 50, and 100 µM. Preliminary experiments showed that under the assay conditions O-deethylation of 7-ETFMC was linear with time and CYP2B6 concentration.

Inhibition of Methadone Metabolism

Compounds that passed the initial ETFMC screen underwent further screening with racemic methadone as the substrate of CYP2B6. Incubations with racemic methadone were performed as previously described (Gadel et al., 2015), with modifications. A typical assay mixture of 180 µl containing racemic methadone, CYP2B6/POR/b5 and natural products in 100 mM potassium phosphate buffer (pH 7.4) was placed in a 96-well plate. After preincubation for 5 minutes at 37°C, the reaction was initiated by adding 20µl NADPH regenerating system (final concentrations: 10 mM glucose 6-phosphate, 1 mM β-NADP, 1 U/ml glucose-6-phosphate dehydrogenase, and 5 mM magnesium chloride, preincubated at 37°C for 10 minutes). The total reaction volume was 200 µl, and the final concentrations were 2 µM racemic methadone and 25 pmol/ml CYP2B6. For each inhibitor, an IC₅₀ value was estimated from the initial ETFMC screening. The inhibitor concentrations in the methadone incubations covers a range of 0.25, 0.5, 1, 1.5, and 2-fold of its IC₅₀. The reaction was terminated after 10 minutes incubation at 37°C by adding 40 µl 20% trichloroacetic acid containing internal standard EDDP-d3 with a final concentration of 1.6 ng/ml. The plate was centrifuged at 2500 rpm for 5 minutes to remove precipitated proteins, and the supernatant was subjected to solid phase extraction and liquid chromatography-tandem mass spectrometry analysis. Preliminary experiments showed that under the assay conditions methadone demethylation was linear with time and CYP2B6 concentration.

Analysis of Methadone Demethylation by HPLC/Tandem Mass Spectrometry

Calibration samples were prepared using standard rac-EDDP perchlorate. Aqueous solutions of calibration standards were prepared at 0.25, 0.5, 2.5, 5, 10, 25, 50, 100, 250, 500, 1000, and 2500 ng/ml EDDP by diluting from the certified methanolic stock solution (1 mg/ml) in deionized water. Calibration standards were treated with 20% TCA containing internal standard EDDP-d3 as described above for incubation samples. Both incubation samples and calibration samples were processed by solid-phase extraction using Strata-X-C-33 µm polymeric strong cation 96-well plate (Phenomenex, Torrance, CA) as described (Gadel et al., 2015). To 150 µl samples in a 96-well 2.2 ml deep-well plate was added 0.4 ml 5% phosphoric acid, and mixed on a titer plate shaker (Laboratory-Line Instruments, Melrose Park, IL). Solid-phase extraction plate was conditioned with 1 ml of methanol, and then 1 ml of 0.1 N HCl. Samples were loaded under soft vacuum (5 mmHg) at 0.5 ml/min, and the plate was washed with 1 ml of 0.1 N HCl and 1 ml of methanol. The plate was then dried under high vacuum (10 to 15 mmHg) for 1 minute. Analytes were eluted with 2 × 0.5 ml of 5% ammonium hydroxide in methanol under soft vacuum, and then evaporated to dryness under nitrogen at 40°C. Samples were reconstituted with 150 µl 0.1% formic acid. EDDP achiral analysis was performed on an ultra-fast liquid chromatography system (Shimadzu Scientific Instruments, Columbia, MD) composed of two LC-20AD XR pumps, DGU20A5R degasser, CBM-20A system controller, CTO-20AC column oven, FCV-11AL solvent selection valve, a SIL-20AC XR autosampler, and a Sunfire C18 column coupled to an API 6500 triple quadrupole tandem mass spectrometer (Applied Biosystems/MDS Sciex, Foster City, CA) operated with Analyst 1.6.2. MultiQuant 3.0.1(AB Sciex) was used for peak integration, generation of calibration curves, and data analysis. The chromatographic separation method and operating conditions for analysis on mass spectrometer were as previously described (Gadel et al., 2015).

Kinetic Characterization of Inhibitors

The compounds that displayed significant inhibition of racemic methadone demethylation were further studied to measure kinetic parameters and to evaluate possible mechanisms of their inhibition on methadone metabolism. The assay method is the same as described above for methadone incubation, except a matrix of variable methadone substrate concentrations and variable inhibitor concentrations were used for each inhibitor. Methadone concentration varied over a range of 5, 10, 25, 50, 100, 250, and 500 µM. Inhibitor concentrations generally covers a range of approximately as low as 0.2-fold and up to 10-fold of IC₅₀ values estimated from the methadone incubation assays performed with fixed methadone concentration at 2 µM for each inhibitor.

Pre-Incubation Versus Co-Incubation

Methadone incubations were carried out to compare pre-incubation and co-incubation for inhibitors to explore possible time-dependent inhibitions. Single inhibitor concentrations that can achieve approximately 50–70% inhibition of EDDP formation were used for these assays. Co-incubation followed the same procedure as described above for methadone (2 µM) incubation. The pre-incubations were performed similarly with following modification. The assay mixture of 180 μl containing CYP2B6/POR/b5, inhibitor and NADPH regenerating system in 100 mM potassium phosphate buffer (pH 7.4) was incubated at 37°C for 30 minutes before addition of 20 μl of 20 μM methadone to initiate the reaction.

Data Analysis

Incubation results are the mean ± standard deviation with three replicates unless otherwise indicated. For estimation of IC₅₀ values, the incubation results from assays with fixed substrate concentration and variable inhibitor concentrations were analyzed by nonlinear regression analysis (SigmaPlot 13.0; Systat, San Jose, CA) using a four-parameter logistic nonlinear regression model (eq. 1), which is a modification from the original equation developed by Hill to quantify the binding of oxygen to hemoglobin.

where Y represents the activity observed at inhibitor concentration [I], A is the activity with minimal inhibition (no inhibition), B is the activity with maximal inhibition (complete inhibition), IC₅₀ is the inhibitor concentration that causes 50% inhibition of the catalytic activity with specific substrate at fixed concentration, and h is the slope at the steepest part of the curve, also known as the Hill slope.

EDDP formation versus methadone concentration with variable inhibitor concentrations was analyzed by nonlinear regression analysis (SigmaPlot 13.0; Systat, San Jose, CA). The kinetic parameters K_i, K_m, V_max, and K_s were derived from nonlinear regression analysis of the rate of EDDP formation as a function of both methadone and inhibitor concentrations using models described for competitive inhibition (eq. 2), or non-competitive inhibition (eq.3). In cases where substrate (or product) inhibition was observed, data were analyzed using a model for non-competitive inhibition and substrate inhibition (eq.4).

where v is the rate observed at substrate concentration [S] and inhibitor concentration [I], K_m is the Michaelis constant, K_i is the inhibition constant which represents the inhibitor concentration required to produce half maximum inhibition, and K_s is the substrate inhibition constant.

Modeling of CYP2B6–Ligand Complexes

The crystal structure of CYP2B6 (3IBD) with inhibitor 4-(4-chlorophenyl)imidazole was used as the receptor template (Gay et al., 2010c). The 4-(4-chlorophenyl)imidazole and cymal-5 detergents and water molecules were removed from the 3IBD structure using PyMOL (The PyMOL Molecular Graphics System, Version 2.6.0a0. Schrödinger, LLC). Three-dimensional conformers of R- and S-methadone, and all natural products (1–43) were downloaded from PubChem (pubchem.ncbi.nlm.nih.gov) and converted to pdb files using PyMOL. Coordinates of clopidogrel were extracted from 3ME6 pdb file (CYP2B4 in complex with clopidogrel) (Gay et al., 2010b). The pdb files of receptor and ligands were converted to pdbqt files, which is an Autodock-specific coordinate file format, by using AutoDockTools (part of MGLTools package) with default settings (Morris et al., 2009). Docking of ligands into CYP2B6 were carried out using AutoDock Vina with default settings (Trott and Olson, 2010; Eberhardt et al., 2021). For docking into the substrate binding pocket in the active site, the center of the crystallized inhibitor 4-(4-chlorophenyl)imidazole was used as an initiation point (x = 23.77, y = 13.79, z = 25.91) and a search space was created with dimension of 26 × 26 × 26 Å. To carry out a blind search for an alternative binding site other than the active site, the center of the whole CYP2B6 molecule was used as the center of the search space (x = 20.14, y = 10.93, z = 24.49), and the dimensions were increased to 74 × 76 × 80 Å, which covers the entire protein molecule. Docking results with top nine binding poses were saved, analyzed and visualized using PMV (the Python Molecular Viewer, part of MGLTools package) and PyMOL.

Results

In a primary screen, natural products were tested for inhibitory effects on CYP2B6-catalyzed O-deethylation of 7-ETFMC. The exception was yangonin (3), which was fluorescent under the assay conditions and could not be evaluated with the ETFMC assay.

Results of the primary screen of kavalactones (1, 2, 4–6), flavokavains A and B (35, 22), gambogic acid (7), euphorbiasteroid (8), daphnoretin (9) and cardiotonic steroids (10–13) with the ETFMC assay are shown in Supplemental Fig. 1, and compared with clopidogrel which is a known CYP2B6 inhibitor. With inhibitor concentration fixed at 10 µM, dihydromethysticin (6) and gambogic acid (7) displayed strong (∼75%) inhibition of CYP2B6 activity. Desmethoxyyangonin (4) and flavokavain A (35) exhibited moderate (∼40%) inhibition. Methysticin (5), daphnoretin (9), resibufogenin (13), and dihydrokavain (2) showed weak effects (17% to 28% inhibition).

Results of the primary screen for 29 chalcone compounds (14-34, 36–43) is shown in Supplemental Fig. 2A. Assays were performed with fixed substrate concentration (50 µM 7-ETFMC) and three inhibitor concentrations (0.1, 1, and 10 µM). Due to the large number of compounds, screening was performed with single measurements. Most chalcones showed weak or no inhibition. However, three compounds, 2,2’-dihydroxychalcone (14), 3,4,2’,5′-tetramethoxychalcone (32), and 4-dimethylamino-2’,4’,6’-trimethoxychalcone (37), showed concentration (0.1–10 µM)-dependent inhibition. To verify the inhibitory effects of these three chalcones (14, 32 and 37), the ETFMC assay was repeated with triplicate measurements. Results confirmed concentration-dependent inhibition (Supplemental Fig. 2B). The chalcones 2,2’-dihydroxychalcone (14), and 4-dimethylamino-2’,4’,6’-trimethoxychalcone (37) showed moderate inhibition at 10 µM, while 3,4,2’,5′-tetramethoxychalcone (32) was a weaker inhibitor.

Eleven inhibitory compounds identified from the primary screen (Supplemental Figs. 1 and 2) were further evaluated with expressed CYP2B6 and the ETFMC assay over a wider substrate concentration range (0.1 to 100 µM 7-ETFMC, Fig. 2), and IC₅₀ values were calculated using a four-parameter logistic model (Table 2). Dihydromethysticin (6), gambogic acid (7), and 2,2'-dihydroxychalcone (14) displayed strong inhibitory activities with IC₅₀ values of 0.52, 9.8 and 4.0 µM for inhibition of 7-ETFMC deethylation. Methysticin (5) showed moderate inhibitory activities with clear concentration dependency. Three compounds, including resibufogenin (13), flavokavain A (35) and 4-dimethylamino-2',4',6'-trimethoxychalcone (37) showed poor concentration dependency and were not studied further. The other four compounds, dihydrokavain (2), desmethoxyyangonin (4), daphnoretin (9), and 3,4,2',5′-tetramethoxychalcone (32) displayed weak inhibitory activities. Dihydrokavain (2) and desmethoxyyangonin (4), though weakly inhibitory, are important structurally for comparison with other kavalactones and were therefore advanced to the next round, while daphnoretin (9) and 3,4,2',5′-tetramethoxychalcone (32) were placed in a separate study with lower priority for further characterization.

Fig. 2.

Concentration-dependent inhibition of 7-ETFMC O-deethylation by twelve compounds that passed the initial ETFMC assay screening. Substrate concentration was fixed (50 µM 7-ETFMC). Results are the mean ± S.D. of triplicate determinations. Solid lines represent predicted concentrations based on parameters obtained by nonlinear regression analysis of measured concentrations using the four-parameter logistic nonlinear regression model as described in the Methods section.

TABLE 2.

IC₅₀ values for inhibition of 7-ETFMC and methadone metabolism by CYP2B6

Inhibitor	IC50 (µM) 7-ETFMC assay	IC50 (µM) Methadone assay
Gambogic acid (7)	9.8 ± 1.0	2.0 ± 0.1
Dihydromethysticin (6)	0.52 ± 0.09	0.044 ± 0.007
Methysticin (5)	68 ± 84	56 ± 12
Dihydrokavain (2)	52 ± 258	50 ± 5
Desmethoxyyangonin (4)	132 ± 174	54 ± 6
2,2’-Dihydroxychalcone (14)	4.0 ± 0.7	1.6 ± 0.6
Yangonin (3)		5.6 ± 1.1
3,4,2’,5′-Tetramethoxychalcone (32)	45 ± 299
4-Dimethylamino-2’,4’,6’-trimethoxychalcone (37)	4.6 ± 1.4
Resibufogenin (13)	92 ± 754
Daphnoretin (9)	75 ± 75

IC₅₀ values are the parameter estimate and standard error of the estimate determined by nonlinear regression analysis.

The six compounds which were identified as concentration-dependent CYP2B6 inhibitors in the ETFMC assay screen (Fig. 2), along with yangonin (3) which was not screened with the ETFMC assay due to fluorescence interference, were next evaluated using racemic methadone as the clinically relevant CYP2B6 substrate. Results of that initial evaluation are shown in Fig. 3 and Table 2. Dihydromethysticin (6) and gambogic acid (7) exhibited the most potent inhibition of methadone metabolism, with IC₅₀ values of 0.044 and 2.0 µM, respectively. Yangonin (3) displayed weak inhibitory activity.

Fig. 3.

Inhibition of racemic methadone demethylation by seven compounds identified as inhibitory in the coumarin assay screen and yangonin (3). (A) dihydromethysticin, methysticin; (B) gambogic acid, dihydrokavain; (C) 2,2'-dihydroxychalcone, desmethoxyyangonin; (D) yangonin. Substrate concentration was fixed at 2 µM racemic methadone. Results are the mean ± S.D. of triplicate determinations. Solid lines represent predicted concentrations based on parameters obtained by nonlinear regression analysis of measured concentrations using the four-parameter logistic nonlinear regression model as described in the Methods section.

After both the ETFMC and methadone screens, six compounds (2, 4, 5, 6, 7, 14) were identified as CYP2B6 and methadone N-demethylation inhibitors and evaluated further. To explore possible inhibition mechanisms involved, the six compounds were kinetically characterized by measuring inhibition of methadone N-demethylation at variable inhibitor and methadone concentrations. EDDP formation as a function of racemic methadone concentration and different inhibitor concentratins is shown in Fig. 4. Data were analyzed with different inhibition models as described in the Methods section and the best fit selected. Mechanism of inhibition and kinetic parameters are provided in Table 3. Dihydromethysticin (6) inhibition was unambigously competitive (Fig. 4A). Gambogic acid (7) was a typical non-competitive inhibitor (Fig. 4B). Dihydrokavain (2) and desmethoxyyangonin (4) were best fit with models of competitive and non-competitive inhibition, respectively (Fig. 4, D and E). Methysticin (5) and 2,2’-dihydroxychalcone (14) showed characteristics of substrate or product inhibition, which was observed previously for methadone metabolism (Gadel et al., 2015). The data of these two inhibitors were best fit with a model of non-competitive inhibition plus substrate inhibition (Fig. 4, C and F).

Fig. 4.

Inhibition of racemic methadone demethylation by (A) dihydromethysticin, (B) gambogic acid, (C) methysticin, (D) dihydrokavain, (E) desmethoxyyangonin, and (F) 2,2’-dihydroxychalcone. Results are the mean ± S.D. of triplicate determinations. Lines are predicted concentrations based on kinetic parameters obtained by nonlinear regression analysis of measured concentrations using the models of competitive inhibition (dihydromethysticin, dihydrokavain), non-competitive inhibition (gambogic acid, desmethoxyyangonin), and non-competitive inhibition with substrate inhibition (methysticin, 2,2’-dihydroxychalcone).

TABLE 3.

Estimated kinetic parameters for inhibition of racemic methadone demethylation by CYP2B6

Inhibitor	Inhibition model	Vmax (pmol/min/pmol CYP)	Km (µM)	Ki (µM)	Ks (µM)
Gambogic acid (7)	non-competitive	9.0 ± 0.2	102 ± 5	6.0 ± 0.2
Dihydromethysticin (6)	competitive	6.0 ± 0.1	60 ± 2	0.074 ± 0.003
Methysticin (5)	non-competitive with substrate inhibition	7.8 ± 1.3	104 ± 29	81 ± 10	520 ± 191
Dihydrokavain (2)	competitive	7.1 ± 0.1	53 ± 3	48 ± 3
Desmethoxyyangonin (4)	non-competitive	6.0 ± 0.1	70 ± 4	422 ± 42
2,2’-Dihydroxychalcone (14)	non-competitive with substrate inhibition	12 ± 3	255 ± 92	16 ± 2	354 ± 169

Results (V_max, K_m, K_i, and K_s) are the parameter estimate and standard error of the estimate determined by nonlinear regression analysis.

Based on the K_i values shown in Table 3, dihydromethysticin (6) was the most potent inhibitor with a K_i value of 0.074 µM, followed by gambogic acid (7) and 2,2’-dihydroxychalcone (14) with K_i values of 6 and 16 µM, respectively. The other three, dihydrokavain (2), methysticin (5), and desmethoxyyangonin (4), were weaker inhibitors with K_i values of 48 to 422 µM.

To explore possible time-dependent inhibition and influence of pre-incubation on inhibition kinetics, we carried out pre-incubation and co-incubation experiments for the seven compounds subjected to the initial methadone screen. For the pre-incubation assays, inhibitors were incubated with CYP2B6/NADPH for 30 minutes, followed by addition of methadone to initiate the reaction, while the 30-minute incubation step was skipped for the co-incubation assays. Supplemental Fig. 3 shows EDDP formation activity remaining in the presence of inhibitor as percentage of control, and comparison of pre-incubation and co-incubation. Inhibitory effects of methysticin appeared enhanced after pre-incubation (74 versus 40% inhibition). For other compounds, pre-incubation differed little from co-incubation.

To better understand the molecular basis of natural products inhibition of CYP2B6 activity and methadone metabolism, we modeled CYP2B6–ligand complexes at the CYP2B6 active-site and/or alternative binding site. Methadone N-demethylation by CYP2B6 is stereoselective, and methadone binding was explored to understand stereoselective metabolism and inhibitor binding. Ligand binding affinity derived from Autodock Vina docking is listed in Table 4, and the binding poses shown for R- and S-methadone (Fig. 5), clopidogrel (Fig. 6), dihydromethysticin and 2,2’-dihydroxychalcone (Fig. 7). Methadone metabolism involves oxidation of one of the N-methyl groups via an iron-oxo intermediate (compound I), hence expected binding of methadone should position the dimethylamino group close to the heme iron. Although the binding poses with best affinity (pose #1, Table 4) for both R- and S-methadone appeared to put the dimethylamino group distant from the heme, one of the poses with slightly lower affinity (pose #4 for R-methadone and pose #3 for S-methadone, Table 4) displayed an orientation that favors demethylation (Fig. 5). Distances from the N-methyl groups to the heme iron are 4.0 and 4.4 Å for R-methadone, and 4.3 and 6.5 Å for S-methadone. The phenyl rings of Phe206 and Phe297 interact with the two aromatic rings of R-methadone at 3.3–3.5 Å. Other residues involved in hydrophobic interactions with R-methadone include Ile101, Val104, Ile114, Ile209, Ala298, and Val477. Interactions involved in binding of S-methadone include aromatic-aromatic interaction with Phe206, and hydrophobic interactions from Ile114. Phe115, Ile209, Thr302, Leu363, Val367, and Val477. It has been reported that ten residues of CYP2B6 appear within 5 Å of the bound ligand in almost all of the existing crystal structures (Angle and Cox, 2023). These residues are mostly hydrophobic and include Ile101, Ile114, Phe115, Phe206, Phe297, Ala298, Thr302, Leu363, Val367, and Val477 (Angle and Cox, 2023). Our CYP2B6-methadone models (R- and S-methadone) showed interactions with all ten residues.

TABLE 4.

Binding affinity values for docking each ligand into the substrate binding pocket of CYP2B6 using AutoDock Vina

No.	Ligand	Affinity (kcal/mol)	No.	Ligand	Affinity (kcal/mol)
	R-methadone, pose #1	−5.41	20	4'-hydroxy-4-methoxy-2'-methylchalcone	−8.79
	R-methadone, pose #4*	−5.03	21	4,4'-dimethoxy-2'-hydroxychalcone	−8.47
	S-methadone, pose #1	−6.76	22	flavokavain B	−8.16
	S-methadone, pose #3*	−6.21	23	2',4',6'-trimethoxychalcone	−7.80
	clopidogrel, pose #1**	−6.85	24	2,2'-dihydroxy-4',6'-dimethoxychalcone	−7.88
	clopidogrel, pose #2**	−6.55	25	2'-hydroxy-2,3,4'-trimethoxychalcone	−9.17
1	kavain	−8.96	26	2'-hydroxy-2,3,5′-trimethoxychalcone	−9.45
2	dihydrokavain	−8.74	27	2'-hydroxy-2,4,5′- trimethoxychalcone	−8.57
3	yangonin	−8.79	28	2',4'-dihydroxy-3,4-dimethoxychalcone	−8.85
4	desmethoxyyangonin	−9.19	29	3-methoxy-4,2',5′-trihydroxychalcone	−9.37
5	methysticin, pose #1¶ methysticin, pose #4¶	−9.38 −8.99	30	2'-hydroxy-3,4,5′-trimethoxychalcone	−9.29
6	dihydromethysticin, pose #1§ dihydromethysticin, pose #2§	−9.08 −9.05	31	3,4-dimethoxy-2'-hydroxy-5′-methylchalcone	−9.41
7	gambogic acid	8.47	32	3,4,2',5′-tetramethoxychalcone	−8.42
8	euphorbiasteroid	1.56	33	3-nitro-2',4',6'-trimethoxychalcone	−9.65
9	daphnoretin	−9.93	34	4',6'-dimethoxy-2'-hydroxy-4-methylchalcone	−8.52
10	cinobufagin	−1.65	35	flavokavain A	−8.18
11	cinobufotalin	−0.50	36	4-methyl-2',4',6'-trimethoxychalcone	−8.36
12	bufalin	−4.05	37	4-dimethylamino-2',4',6'-trimethoxychalcone	−7.49
13	resibufogenin	−3.81	38	3,2'-dihydroxy-4,4',6'-trimethoxychalcone	−9.38
14	2,2’-dihydroxychalcone	−9.15	39	3-hydroxy-2',4,4',6'-tetramethoxychalcone	−7.77
15	3,2’-dihydroxychalcone	−9.23	40	4-hydroxy-2',3,4',6'-tetramethoxychalcone	−9.12
16	3,4-dimethoxychalcone	−9.15	41	4'-hydroxy-2'-methyl-3,4,5-trimethoxychalcone	−8.07
17	2',4'-dimethoxychalcone	−8.35	42	4-benzyloxy-2'-hydroxy-3,4',6'-trimethoxychalcone	−8.49
18	4-methoxy-4'- (methylthio)chalcone	−8.10	43	4-benzyloxy-2',3,4',6'-tetramethoxychalcone	−8.59
19	4,2',5′-trihydroxychalcone	−9.37

* Pose #4 for R-methadone and pose #3 for S-methadone are orientated with the dimethylamino group approaching the heme.

** Pose #1 and #2 for clopidogrel are oriented ‘thiophene down’ and ‘chlorophenyl down’, respectively.

¶ Pose #1 for methysticin is oriented with lactone ring approaching the heme, and pose #4 is oriented with methylenedioxy approaching the heme.

§ Pose #1 for dihydromethysticin is oriented with lactone ring approaching the heme, and pose #2 is oriented with methylenedioxy approaching the heme.

Fig. 5.

Binding poses of (A) R-methadone and (B) S-methadone at the substrate binding pocket of CYP2B6. The backbone of methadone is orange. The heme is purple. The CYP residues interacting with methadone are dark cyan.

Fig. 6.

Binding poses of clopidogrel at the substrate binding pocket of CYP2B6 in two orientations: (A) ‘thiophene down’ orientation and (B) ‘chlorophenyl down’ orientation. The backbone of clopidogrel is orange. The heme is purple. The CYP2B6 residues interacting with clopidogrel are dark cyan.

Fig. 7.

Binding poses of (A) dihydromethysticin and (B) 2,2’-dihydroxychalcone at the CYP2B6 substrate binding pocket. The backbone of the inhibitor ligand is orange. The heme is purple. The CYP2B6 residues interacting with the inhibitors are dark cyan.

Clopidogrel and ticlopidine are mechanism-based inhibitors of CYP2B6. Thiophene ring oxidation initiates covalent modification and CYP2B6 inactivation. These inhibitors can bind oriented ‘thiophene down’ with the thiophene ring close to the heme or ‘chlorophenyl down’ with the chlorophenyl group closest to the heme (Richter et al., 2004; Gay et al., 2010b). The present investigation found clopidogrel docking in the substrate CYP2B6 binding pocket in an orientation (pose #1, Fig. 6A) similar to the ‘thiophene down’ and in another orientation (pose #2, Fig. 6B) similar to the ‘chlorophenyl down’ orientation. Clopidogrel interactions include hydrogen bonding from Thr302, halogen-π interaction from Phe297 for the ‘thiophene down’ pose (Fig. 6A) and halogen-π interaction from Phe115/Phe297, aromatic-aromatic interaction between the phenyl ring of Phe206, and thiophene ring for the ‘chlorophenyl down’ orientation (Fig. 6B).

Binding poses of two selected natural products with CYP2B6 are shown in Fig. 7. One possible binding orientation of dihydromethysticin is shown (Fig. 7A), in which the methylenedioxy group is approaching the heme and the methylene is 3.1 Å from the heme iron. In this orientation, Ser210 forms a hydrogen bond with the oxygen of the lactone carbonyl group (2.9 Å). Thr302 forms hydrogen bond with either oxygen of methylenedioxy (3.2 Å) or oxygen of lactone methoxy group (3.1 Å). Phe206 and Phe297 interact with the lactone (3.2 Å) and phenyl rings (4.0 Å). We note, however, that this orientation (pose #2) had slightly decreased binding affinity (-9.05 kcal/mol) compared with the better ranked pose (pose #1, binding affinity -9.08 kcal/mol, Table 4), in which the whole dihydromethysticin molecule is flipped and the lactone ring is approaching the heme (not shown). In pose #1, Thr305 forms a hydrogen bond with one oxygen atom of the 11,12-methylenedioxy group. A cluster of three phenylalanine residues (Phe206, Phe297, and Phe115) are involved in binding of dihydromethysticin, including aromatic-aromatic interaction (Phe206 with the phenyl ring) and hydrophobic interactions (Phe297 and Phe115 with the lactone ring). Although not best ranked, pose #2 is shown (Fig. 7A) due to a potential interaction of the methylenedioxy group with the heme iron. A specifically favored orientation is not identifiable from binding energies alone. 2,2’-dihydroxychalcone has two aromatic rings. Each is involved in aromatic edge-to-face interaction, with residues Phe206 and Phe297 (Fig. 7B). Binding of 2,2’-dihydroxychalcone exhibited a hydrogen bonding network involving the sidechain of Thr302 and two oxygen atoms from 2,2’-dihydroxychalcone (the carbonyl oxygen and 2-hydroxyl group).

Docking of gambogic acid into the CYP2B6 substrate binding pocket resulted in a binding affinity of 8.47 kcal/mol, indicating no binding. A blind global search for alternative gambogic acid binding sites over the entire CYP2B6 protein was explored. A groove between helix C and helix H on the surface of CYP2B6 appeared to be a binding site for gambogic acid with binding affinity of -8.73 kcal/mol (Fig. 8). Interactions with surrounding residues include hydrogen bond with Asp134, hydrogen bond with amide nitrogen from one of three residues (Gly136, Gly138, Lys139), and hydrophobic interactions with residues Phe135, Pro261, Thr267, Leu270, and His271 (Fig. 8A). Gambogic acid shown in ball and stick form and as spheres on the protein surface is presented in Fig. 8, B and C, respectively.

Fig. 8.

Docking of gambogic acid into an alternative binding site of CYP2B6 in the groove between helix C and helix H. (A) Details of the interactions with surrounding residues (The backbone of gambogic acid is shown in blue). (B) Gambogic acid (backbone in cyan) shown as ball and stick on protein surface. (C) Gambogic acid (cyan) shown as spheres on the protein surface.

Discussion

Screening of 43 natural compounds and synthetic derivatives with great diversity in chemical structure identified six final hit compounds possessing different strength, potency, and mechanism of CYP2B6 and methadone N-demethylation inhibition. The six compounds, in order of decreasing potency (with K_i values for inhibition of methadone metabolism) were dihydromethysticin (0.074 µM), gambogic acid (6.0 µM), 2,2’-dihydroxychalcone (16 µM), dihydrokavain (48 µM), methysticin (81 µM), and desmethoxyyangonin (422 µM). Dihydromethysticin, dihydrokavain, methysticin, and desmethoxyyangonin are kavalactones (Soares et al., 2022). 2,2’-dihydroxychalcone is one of the simplest chalcones (Nowakowska, 2007; Zhuang et al., 2017).

CYP2B6 inhibition by kavalactones is a novel observation. Previous studies evaluated kava extracts and kavalactone constituents (desmethoxyyangonin, dihydromethysticin, methysticin, kavain, dihydrokavain, yangonin) effects on other human recombinant or liver microsomal CYPs (1A2, 2A6, 2C8, 2C9, 2C19, 2D6, 2E1, 3A4) (Mathews et al., 2002; Zou et al., 2002, 2004; Mathews et al., 2005). The most inhibited recombinant enzyme was CYP2C19, and the most potent inhibitors were dihydromethysticin, desmethoxyyangonin, and methysticin, with IC₅₀ values of 0.43, 0.51, and 0.93 µM, respectively; IC₅₀ values for kavain (4.9 µM) and dihydrokavain (10 µM) were greater (Zou et al., 2002). CYP2C9 and CYP2C19 in human liver microsomes were both inhibited, mostly by dihydromethysticin, desmethoxyyangonin and methysticin, with K_i values of 5–10 µM (Mathews et al., 2005). Dihydromethysticin was the strongest inhibitor (Mathews et al., 2002). CYP2B6 now appears to be the CYP isoform most susceptible to inhibition by dihydromethysticin, and dihydromethysticin is the most potent of the CYP2B6 natural product inhibitors identified.

Structure-activity evaluations reveal features important for CYP2B6 inhibition by kavalactones. Dihydromethysticin and dihydrokavain were competitive, while methysticin and desmethoxyyangonin were non-competitive. The only difference between dihydromethysticin and methysticin is the C₇–C₈ single versus double bond. This difference conferred substantial difference in inhibitory activity, with >1000- fold difference in K_i (0.074 versus 81 µM), as well as inhibition mechanism. Assuming dihydromethysticin competitive interaction at the CYP2B6 substrate binding site, free rotation about the C₇–C₈ single bond appears essential for a conformation that can fit into the substrate binding site, while restriction by the C₇=C₈ double bond of methysticin maintains a more planar structure, making it unable to interact at the substrate binding site. Similarly, the only difference between dihydrokavain and kavain, and a major difference between dihydrokavain and desmethoxyyangonin, is the C₇–C₈ single versus double bond, whereby the double bond of kavain and desmethoxyyangonin conferred much less CYP2B6 inhibition. Previous studies revealed nonplanarity to be an important feature for CYP2B6 substrates, and CYP2B6 has difficulty metabolizing planar substrates (Angle and Cox, 2023). Our results appear to be in agreement with the nonplanarity rule. Dihydromethysticin is more non-planar compared with methysticin. The more planar methysticin may confer poor binding affinity at the substrate binding pocket.

Comparing dihydromethysticin versus dihydrokavain and methysticin versus kavain revealed that the 11,12-methylenedioxy group on the aromatic ring also influences CYP2B6 inhibition. Dihydromethysticin and dihydrokavain differ only in the 11,12-methylenedioxy group. Both were competitive inhibitors, but absence of the methylenedioxy group in dihydrokavain, present in dihydromethysticin, caused a 650-fold increase in K_i (48 versus 0.074 µM). Similarly, methysticin and kavain differ only in the 11,12-methylenedioxy group. Absence of this group in kavain resulted in substantial loss of CYP2B6 inhibition compared with methysticin. Influence of the 11,12-methylenedioxy group on kavalactones inhibition of CYPs other than CYP2B6 (Mathews et al., 2002; Zhang et al., 2022), and on bioactivation more broadly (Puppala et al., 2017), has been noted previously. For example, dihydromethysticin, with a methylenedioxy group, potently and selectively inhibited nitrosamine-induced lung tumorigenesis, while dihydrokavain and analogs lacking a methylenedioxy group were completely ineffective (Puppala et al., 2017). The known lung carcinogen 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) and its reduced metabolite 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL) both undergo CYP-mediated hydroxylation and metabolic activation to ultimate alkylating carcinogens. Dihydromethysticin but not dihydrokavain, and methysticin but not kavain, inhibited NNK-induced DNA methyl adduct formation and lung adenoma formation, demonstrating that the methylenedioxy functional group in dihydromethysticin and methysticin was essential for cancer chemoprevention. More generally, the methylenedioxyphenyl group is well-known to confer mechanism-based drug inhibition of CYPs (e.g., paroxetine, tadalafil) (Murray, 2000; Orr et al., 2012). While methysticin was less potent than other inhibitors of CYP2B6 in the present experiment, it did show evidence of greater inhibition with preincubation. Whether it is a time-dependent, mechanism-based inhibitor, and the identity of an active metabolite-enzyme complex, would require further study. Other structure–activity relationships are also notable. Comparison of desmethoxyyangonin versus yangonin shows that the presence of the methoxy substituent results in less CYP2B6 inhibition. Kavain and desmethoxyyangonin differ only in the presence of an additional unsaturated bond in the lactone ring, which conferred slightly greater CYP2B6 inhibition.

Inhibition of CYP2B6 by gambogic acid is another novel observation. Methadone N-demethylation was inhibited potently and non-competitively, suggesting binding to an alternative (allosteric) site other than the substrate binding site. Some CYPs are well known for allosteric regulation and cooperative behaviors (Denisov et al., 2009). Multiple-site substrate binding and cooperativity has been reported for CYP2B6 (Wang et al., 2019), though it has a 50% smaller substrate binding pocket than the active site of the larger and more accommodating CYP3A4 (Gay et al., 2010a). Ekins et al., analyzed all known CYP2B6 substrates and found 28 which were generally small hydrophobic molecules (molecular weight mean 254 ± 74, range 106-407) (Ekins et al., 2008). Gambogic acid (molecular weight 629) is much larger than typical CYP2B6 substrates. Active site binding may thus be less likely, and could be another reason for allosteric binding.

CYP2B6 inhibition by 2,2’-dihydroxychalcone is a third novel observation. This was the only inhibitor identified from screening 30 chalcones. Two other chalcones, 3,2’-dihydroxychalcone and 2,2'-dihydroxy-4',6'-dimethoxychalcone are structurally similar to 2,2’-dihydroxychalcone (Table 1). 2-hydroxyphenyl replaced by 3-hydroxyphenyl, or extra methoxy substituents at 4’,6’ positions resulted in loss of inhibitory activity. Reports of P450 inhibition by chalcones are rare. Licochalcone A (2-methoxy-4,4’-dihydroxy-5-(2-methylbut-3-en-2-yl)chalcone) inhibited the activity of recombinant CYPs 1A2, 2C19, 2C8, 2C9, and 3A4, with K_i values of 0.2–2 µM (He et al., 2015).

Compared with other CYPs, there are fewer studies of natural product effects on CYP2B6, and fewer compounds inhibit CYP2B6 than other CYPs (Salminen et al., 2011). A few in vitro CYP2B6 natural product inhibitors have been identified, including borneol and isoborneol, which are constituents of the traditional Chinese medicine woohwangcheongsimwon (competitive, K_i 10 and 6 µM, respectively) (Kim et al., 2008), the Ginko biloba component kaempferol (competitive, K_i 35 µM) (Lau and Chang, 2009), green tea catechin (-)-epigallocatechin-3-gallate (competitive, K_i 4 µM) (Misaka et al., 2013), medicinal herb honokiol (competitive, K_i 18 µM) (Jeong et al., 2013), the bioactive sesquiterpenes cedrol, β-cedrene and thujopsene (competitive, K_i 1-2 µM) (Jeong et al., 2014), α-terpinyl acetate (competitive, K_i 7.6 µM) (Lee et al., 2018), and extracts of licorice species and the components licoricidin and glycycoumarin (Li et al., 2017). In contrast, in volunteers in vivo, neither oral Ginko biloba extract (Lei et al., 2009) nor woohwangcheongsimwon suspension affected bupropion and/or hydroxybupropion pharmacokinetics (Kim et al., 2010).

Inhibition of recombinant CYP2B6-catalyzed methadone N-demethyation by natural products is a fourth novel observation. There are scant data on methadone-natural products interactions (one case report) (Izzo and Ernst, 2009). Any clinical implications for natural products inhibition of methadone metabolism remain unknown. Next steps, in general, for evaluating such natural products inhibition potential have been articulated, and include modeling, human liver microsomal, and clinical studies to determine how inhibitor dose and exposure may affect a clinical interaction (Paine et al., 2018; Paine, 2020; Weber and Hopp, 2020; Cox et al., 2021). Some information is available. Plasma total kavalactone concentrations averaged 20–40 µg/ml (80–160 µM) assuming an oral dose of 240mg kava (Anke and Ramzan, 2004). However various commercial kava products differ in kavalactone content (Mamallapalli et al., 2022). Plasma dihydromethysticin concentrations were 0.2–0.3 µM in volunteers given 225 mg/d total kavalactones (containing 4.7 mg dihydromethysticin) for 1 week, but hepatic concentrations may be considerably higher (Wang et al., 2018b; Kanumuri et al., 2022). With K_i values of 0.074, 81, and 48 µM for dihydromethysticin, methysticin, and dihydrokavain, one or more of these three kavalactones might affect CYP2B6 and methadone metabolism at clinically relevant concentrations. These kavalactone K_i values are similar to grapefruit juice components which cause clinically significant CYP3A4 inhibition (Zou et al., 2002), hence there is plausible potential for CYP2B6-kava interactions in vivo.

Docking of ligands into the CYP2B6 active site substrate binding pocket was modeled for R- and S-methadone, clopidogrel, and all natural products. The crystal structures of CYP2B6 (including 3IBD) were initially solved using CYP2B6dH (N-terminal-deleted and C-terminal His-tagged) mutants (Y226H/K262R) due to protein thermal stability (Gay et al., 2010c; Roberts et al., 2023), which is actually the polymorphic variant CYP2B6.4 plus Y226H mutation (Gay et al., 2010c; Zanger and Klein, 2013). The CYP2B6–ligand model in the present work, built from 3IBD, may be closer to CYP2B6.4 rather than wild-type CYP2B6.1. Previous studies showed that methadone N-demethylation by CYP2B6.4 (with coexpressed cytochrome b₅) had V_max values 2-fold greater for R- than S-methadone and K_i values 4-fold less for S- than R-methadone (Gadel et al., 2015). In our present model, the two N-methyl groups of R-methadone were positioned close to, and at similar distances from, the heme iron, making both available for oxidation. For S-methadone, only one N-methyl is available for oxidation while the other is more distant. Two oxidation sites for R-methadone versus one for S-methadone may contribute to the 2-fold V_max difference. Binding affinity derived from docking is -5.03 and -6.21 kcal/mol for R- and S-methadone, respectively; more negative values indicate greater binding affinity, which may contribute to the lower K_m for S-methadone. Modeling of methadone binding into the CYP2B6 active site had been reported (Kamal et al., 2013). Homology modeling generated a wild-type CYP2B6 structure from 3IBD, and models of 10 CYP2B6 genetic variants (but not 2B6.4) (Kamal et al., 2013). Docking overlaid all binding poses of R- and S-methadone in wild-type CYP2B6 and all variants together in a single plot; hence, individual binding poses were not identifiable and methadone enantiomers orientation and details of ligand-receptor interactions were not described (Kamal et al., 2013). Therefore, comparison with current results is not possible.

To complement the docking results, we modeled CYP2B6–ligand complexes with the known inhibitor clopidogrel. The two binding poses identified, ‘thiophene down’ and ‘chlorophenyl down’, resemble the ligand orientations in the crystal structures of CYP2B4 (which shares 88% sequence similarity and 77% identity with CYP2B6) in complex with ticlopidine, and the structure of the CYP2B4-clopidogrel complex, which showed a ‘chlorophenyl down’ orientation only (Gay et al., 2010b). In a previous study, clopidogrel docking was reported using a homology model of CYP2B6, as the crystal structure of CYP2B6 was not yet available (Richter et al., 2004). In this model, clopidogrel was oriented ‘thiophene down’ (Richter et al., 2004).

CYP2B6 modeling was next performed with the natural products. Dihydromethysticin binding orientation (Fig. 7A) featured two hydrogen bonds (Ser210 and Thr302), interactions with two phenylalanine residues Phe206 and Phe297 (including aromatic edge-to-face interaction), potential interactions of methylenedioxy with heme iron, and hydrophobic interactions with the other ten residues. Combination of all these interactions makes dihydromethysticin an excellent ligand to compete with methadone for binding at the active site. The network of hydrogen bonding formed between 2,2’-dihydroxychalcone and CYP2B6 could be an important driving force for binding of this inhibitor at the substrate binding site.

Binding affinity derived from docking the natural products as an evaluation for a virtual screen, a process often used in drug discovery (Jaghoori et al., 2016), yielded mixed results. Our natural products binding did have some parallels with inhibitory activity including: (a) docking energy of gambogic acid into the active site implied no binding, in agreement with observed non-competitive inhibition (suggesting binding to a site other than the substrate binding pocket) and (b) euphorbiasteroid and cinobufotalin do not bind to the active site, and cinobufagin, bufalin, and resibufogenin have weak affinity, and did not inhibit CYP2B6. However, virtual screening failed for the remaining natural products, and was unable to differentiate inhibitors from non-inhibitors. The chalconoids have good binding affinity but most were not inhibitory. This may represent the limitation of the modeling method. Autodock Vina treats ligands as flexible molecules while the receptor kept rigid. Docking with receptor flexibility might help to make the binding affinity more accurate.

An alternative binding site for gambogic acid was identified through a blind search over the entire CYP2B6 molecule. An allosteric site is located in the groove between helix C and helix H. An allosteric site on CYP2D6 has also been reported, which is involved in regulating ligand access (Fischer and Smieško, 2021). This CYP2D6 allosteric site is denoted as hotspot1 (H1), is located among helices C, E, and H, and appears to be conserved in the CYP2 family, including CYP2B6, based on molecular dynamics simulations. Ligand association at CYP2D6 hotspot1 facilitates active site access of substrates and may promote their metabolism, leading to positive allosteric regulation (Fischer and Smieško, 2021). For CYP2B6, the allosteric site for gambogic acid we identified is located analogous to the CYP2D6 hotspot1. Binding of gambogic acid to this CYP2B6 site may trigger allosterism, but unlike CYP2D6, may lead to inhibition of catalytic activity. Allosterism for gambogic acid inhibition of CYP2B6 may occur by a different mechanism than allosterism at hotspot1 of CYP2D6.

In summary, this investigation identified several natural product constituents which inhibit in vitro human recombinant CYP2B6 activity and CYP2B6-catalyzed methadone N-demethylation. The most effective inhibitors were dihydromethysticin, gambogic acid, and 2,2’-dihydroxychalcone.

Abbreviations

7-ETFMC

7-ethoxy-4-(trifluoromethyl) coumarin

CYP

cytochrome P450

EDDP

2-ethyl-1,5-dimethyl-3,3-diphenylpyrrolidine

K _i

inhibition constant

Km

Michaelis constant

K _s

substrate inhibition constant

POR

P450 reductase

V _max

maximal velocity

Authorship Contributions

Participated in research design: Wang, Kharasch.

Conducted experiments: Wang, Yang, Patel, Neiner.

Performed data analysis: Wang, Yang.

Wrote or contributed to the writing of the manuscript: Wang, Yang, Kharasch.