Abstract
Methylophiopogonanone A (MOA), an abundant homoisoflavonoid bearing a methylenedioxyphenyl moiety, is one of the major constituents in the Chinese herb Ophiopogon japonicas. This work aims to assess the inhibitory potentials of MOA against cytochrome P450 enzymes and to decipher the molecular mechanisms for P450 inhibition by MOA. The results showed that MOA concentration-dependently inhibited CYP1A, 2C8, 2C9, 2C19, and 3A in human liver microsomes (HLMs) in a reversible way, with IC50 values varying from 1.06 to 3.43 μM. By contrast, MOA time-, concentration-, and NADPH-dependently inhibited CYP2D6 and CYP2E1, along with KI and kinact values of 207 µM and 0.07 minute−1 for CYP2D6, as well as 20.9 µM and 0.03 minutes−1 for CYP2E1. Further investigations demonstrated that a quinone metabolite of MOA could be trapped by glutathione in an HLM incubation system, and CYP2D6, 1A2, and 2E1 were the major contributors to catalyze the metabolic activation of MOA to the corresponding O-quinone intermediate. Additionally, the potential risks of herb-drug interactions triggered by MOA or MOA-related products were also predicted. Collectively, our findings verify that MOA is a reversible inhibitor of CYP1A, 2C8, 2C9, 2C19, and 3A but acts as an inactivator of CYP2D6 and CYP2E1.
Significance Statement Methylophiopogonanone A (MOA), an abundant homoisoflavonoid isolated from the Chinese herb Ophiopogon japonicas, is a reversible inhibitor of CYP1A, 2C8, 2C9, 2C19, and 3A but acts as an inactivator of CYP2D6 and CYP2E1. Further investigations demonstrated that a quinone metabolite of MOA could be trapped by glutathione in a human liver microsome incubation system, and CYP2D6, 1A2, and 2E1 were the major contributors to catalyze the metabolic activation of MOA to the corresponding O-quinone intermediate.
Introduction
Ophiopogonis radix (also named mai-dong in Chinese) is a commonly used edible herb in East Asia. Currently, dozens of mai-dong prepared herbal products or health foods (such as sheng-mai-yin decoction, shen-mai injection, and mai-dong soup) have been marketed for preventing and treating a variety of disorders, including cardiovascular disease and various types of cancer (Chen et al., 2016a). In clinical settings, to treat the diseases in a more efficient way, mai-dong prepared herbal products have been widely used in combination with a wide range of Western drugs, in which most therapeutic drugs are the substrates of human cytochrome P450 enzymes (P450s). It has been reported that some mai-dong prepared herbal products (such as shen-mai injection and dengzhan shengmai capsule) can inhibit a panel of hepatic P450s in liver microsomes from both rat and human (Xia et al., 2010; Chen et al., 2016b). However, the key constituents in mai-dong responsible for P450 inhibition and the related inhibitory mechanism(s) still remain unknown.
Bygone investigations have discovered that the Chinese herb mai-dong contains several classes of ingredients, including homoisoflavonoids, steroidal saponins, and polysaccharides (Chen et al., 2016a; Zhao et al., 2017; Fang et al., 2018; Chen et al., 2019; Tan et al., 2019; Tian et al., 2019; Zhan et al., 2019). Among all reported ingredients, the homoisoflavonoids (such as methylophiopogonanone A and its analogs) have been identified as the principal bioactive compounds and marker compounds of mai-dong, which have been found to have broad biologic activities, including anticancer, anti-inflammatory, and antibacterial activities (Zhu et al., 2004; Lin et al., 2010; Li et al., 2012; Wang et al., 2017, 2019). As one of the most abundant homoisoflavonoid constituents occurring in Ophiopogon japonicas (up to 0.19 mg/g dried materials), methylophiopogonanone A (MOA) has drawn increasing attention from both phytochemists and pharmacologists (Ma et al., 2009; Wang et al., 2013; He et al., 2016) owing to its diverse pharmacological functions, such as anticancer, anti-inflammatory, and antibacterial effects (Li et al., 2012; Liang et al., 2012; He et al., 2016). Notably, MOA is a representative compound of the naturally occurring homoisoflavonoids, and all of these compounds bear a methylenedioxyphenyl (MDP) moiety at the C4′-C5′ sites (Fig. 1). It is well known that the MDP moiety can be bioactivated by mammalian cytochrome P450 enzymes. Such biotransformation may form a reactive metabolite and bring some undesirable effects to human beings, such as idiosyncratic toxicity or mechanism-based inactivation of P450s (Kalgutkar and Dalvie, 2015). Inactivation of human P450s may result in clinically relevant herb-drug interactions (HDIs)/drug-drug interactions, particularly for those herbs or drugs containing time-dependent inactivators of P450s. Currently, the inhibition/inactivation potency of MOA on human P450s and related molecular mechanisms have not been reported. Thus, it is crucial to assay the potential inhibition and inactivation potentials of MOA against human P450s and to decipher the inhibitory mechanism of this homoisoflavonoid.
The major purposes of this study were to investigate reversible and time-dependent inhibition of human P450s by MOA and to decipher the molecular mechanisms for P450 inhibition or inactivation by MOA. For these purposes, the inhibition and inactivation potentials of MOA against human P450s were measured via performing a series of P450 inhibition assays. Meanwhile, the reactive metabolite(s) of MOA, and the key enzymes responsible for bioactivation of MOA, were identified by a panel of state-of-the-art techniques. The findings presented here provide key information for deep understanding of the interactions between MOA and human P450s, which would be very useful for the clinical applications of MOA and MOA-containing herbal products.
Materials and Methods
Chemicals, Reagents, and Biologic Samples
MOA (Fig. 1) and 5,7-dihydroxycoumarin were provided by Heowns Biochem Technologies (Tianjin, China). Chlorzoxazone, d-glucose-6-phosphate (G-6-P), testosterone, ketoconazole, glucose-6-phosphatedehydrogenase (G-6-PDH), propranolol, glutathione (GSH), β-NADP+, and NADPH were all supplied by Sigma-Aldrich (St. Louis, MI). Phenacetin, MgCl2, coumarin, omeprazole, paclitaxel, dextromethorphan, acetaminophen, potassium ferricyanide [K3Fe(CN)6], superoxide dismutase (SOD), and catalase (CAT) were ordered from Dalian Meilun Biotech (Dalian, China). The source of diclofenac was from Ark Pharm (Wuhan, China). 5-Hydroxyomeprazole and 6β-hydroxytestosterone were offered by TLC Pharmaceutical Standards (South Aurora, Canada). 6-Hydroxychlorzoxazone and 6α-hydroxytaxol were obtained from Torabto Research Chemicals (ON, Canada), and 4′-hydroxydiclofenac was ordered from Cayman Chemicals (Ann Arbor, Michigan). Lansoprazole was ordered from Hairong Pharmaceutical (Chengdu, China), and 7-hydroxycoumarin was supplied by Alfa Aesar (Heysham, UK). The purities of all compounds used in this study were greater than 98%. The recombinant human P450s and pooled human liver microsomes (HLMs, lot X008067) were all purchased from Bioreclamation IVT (Baltimore, MD). All biologic samples were stored at −80°C until use. The stock solution of each compound was dissolved in methanol and stored at 4°C until use. HPLC-grade acetonitrile and formic acid were acquired from Fisher (St. Louis, MI). Ultrapurified water, which was prepared by a Millipore purification system, was used throughout.
P450 Inhibition Assays
The procedure and in vitro incubation system (200 μl) for P450 inhibition assays were identical to the published literature (He et al., 2015; Wei et al., 2018; Fang et al., 2020), and the concentrations of the probe substrates as well as the incubation times for each P450 reaction are shown in Supplemental Materials. Briefly, the reaction was conducted in a mixed system containing PBS (100 mM, pH 7.4), NADPH regeneration system (1.0 mM β-NADP+, 10 mM G-6-P, 1.0 U/ml G-6-PDH, and 4.0 mM MgCl2), HLMs, and MOA. The reaction mixture was vortexed and prewarmed for 3 minutes at 37°C, and the oxidative reaction was started by adding each probe substrate. All incubation samples were incubated at 37°C for an additional 10–30 minutes (please see Supplemental Table 1 for details). The reactions were quenched by adding ice-cold acetonitrile (200 μl) containing internal standard lansoprazole or 5,7-dihydroxycoumarin [internal standard (IS)]. The mixtures were then centrifuged at 20,000g for 30 minutes at 4°C, the suspension liquid (100 μl) was diluted with Millipore water (100 μl), and then data were possessed by LC-MS/MS as depicted in Supplemental Table 2.
Time-dependent inhibition tests were conducted according to the previously reported procedure with slight modification (Obach et al., 2007; Fang et al., 2020). Briefly, a mixture containing PBS buffer, NADPH regeneration system (1.0 mM β-NADP+, 10 mM G-6-P, 1.0 U/ml G-6-PDH, and 4.0 mM MgCl2), HLMs, and MOA was vortexed for 10 seconds. After preliminary incubation at 37°C for 30 minutes, the reaction was subsequently initiated by addition of individual probe substrate. The reaction was performed at 37°C for an additional 10–30 minutes (the final concentrations of each P450 substrate and the incubation time are listed in Supplemental Table 1) before quenching the oxidative reaction. The supernatants were prepared in the same way as mentioned above and then analyzed by LC-MS/MS as shown in Supplemental Table 2.
Kinetic Analyses for Reversible Inhibition
To determine the types of inhibition and the values of Ki, MOA (0–10 μM) was incubated with HLMs supplemented with the NADPH-generating system, and each probe substrate was added in reaction mixture for 10–30 minutes at 37°C. The final concentrations of MOA and substrate for target P450 enzyme are shown in Supplemental Table 3. The residual activities of each enzymes were measured as described above. The kinetic data were then put into the equations below for competitive (a), noncompetitive (b), and mixed inhibition (c) mode, respectively.
Here, V represents the reaction velocity, Vmax is the maximum reaction velocity, and S and I are the substrate and inhibitor (MOA) concentrations, respectively. Km is the Michaelis constant of each P450 substrate, and Ki is the inhibitory constant of MOA against a particular P450 enzyme.
Kinetic Analyses for Time-Dependent Inhibition
Inactivation kinetics of MOA against CYP2E1 and CYP2D6 were also performed by a dilution method, and the inhibition constant (KI) and the maximum inactivation rate constant (kinact) of MOA against CYP2E1-mediated chlorzoxazone hydroxylation and CYP2D6-mediated dextromethorphan O-demethylation were determined as follows. First, the incubation system (200 μl) containing PBS buffer, HLMs, G-6-P, G-6-PDH, MgCl2, and MOA was vortexed, equilibrated, and then initiated by adding β-NADP+. At 0, 5, 10, 20, or 30 minutes, respectively, 20 μl of aliquot was taken from the primary incubation mixture and then added to the secondary incubation mixture (total volume: 200 μl) containing PBS buffer, the NADPH regeneration system, and each probe substrate for a target P450. After incubation for another 20 minutes (CYP2E1) or 30 minutes (CYP2D6), the reaction was terminated by adding ice-cold acetonitrile (200 µl) containing lansoprazole or 5,7-dihydroxycoumarin (IS). Preparation of the samples for LC-MS/MS analysis was performed as mentioned above.
The inactivation kinetic parameters of both CYP2D6 and CYP2E1 by MOA were determined using the following equation:
Here, [I] is the final concentration of MOA (inhibitor), kinact represents the maximal rate constant of inactivation, and KI means the inhibitor concentration required for 50% the maximal rate of inactivation.
Determination of NADPH-Dependent Inhibition of CYP2E1 by MOA
To determine the NADPH dependence of CYP2E1 inactivation, the samples were divided into three groups (A, B, and C). Group A was the mixture of PBS buffer, HLMs (1.0 mg/ml), and NADPH-generating system (as mentioned above), whereas group B was MOA (10 μM) mixed with PBS buffer, HLMs (1.0 mg/ml), and NADPH-generating system (as mentioned above). Group C contained MOA (10 μM) in PBS buffer (pH 7.4) containing G-6-P (10 mM), G-6-PDH (1.0 U/ml), and MgCl2 (4.0 mM) but without β-NADP+. At the time points of 0, 5, 10, 20, or 30 minutes, respectively, 20 μl of the aliquot was taken from the primary incubation mixture and then transferred to the secondary incubation mixture (total volume: 200 μl) that contained PBS, the NADPH-generating system, and chlorzoxazone. After a 20-minute incubation, refrigerated ice-cold acetonitrile (200 µl) containing 5,7-dihydroxycoumarin (IS) was added to completely end the reaction. The samples were prepared and analyzed as described above.
Determination Reversibility of MOA-Mediated CYP2E1 Inactivation
To investigate the mechanism of MOA-mediated CYP2E1 inactivation, the reversibility of MOA-mediated CYP2E1 inhibition was determined via testing the changes in residual activities with or without treatment of potassium ferricyanide [K3Fe(CN)6], as reported previously (Watanabe et al., 2007). First, the primary mixtures were incubated with HLMs (1.0 mg/ml), MOA (0 or 100 μM), and the NADPH-generating system in PBS buffer (pH 7.4). At the time intervals of 0, 5, 10, 20, or 30 minutes, 20 μl of the aliquot was transferred into the secondary solutions (20 μl) containing 100 mM PBS buffer with or without potassium ferricyanide (2.0 mM, final concentration). After incubating for another 10 minutes, the secondary incubation was diluted 5-fold with the final incubation solutions that contained PBS, the NADPH-generating system, and chlorzoxazone. The final incubation was continued for an additional 20 minutes before quenching the oxidation reaction by adding ice-cold acetonitrile (200 µl) containing 5,7-dihydroxycoumarin (IS).
The Effects of SOD/CAT and GSH on CYP2E1 Inactivation by MOA
The primary mixtures contained MOA (100 μM), HLMs, the NADPH-generating system and PBS in the presence of a mixture of SOD/CAT (800 U for each), GSH, or vehicle. After a 30-minute incubation under physiologically relevant conditions (pH 7.4, 37°C), 20 μl of aliquot was transferred into the secondary incubation mixture (total volume: 200 μl) that contained PBS, the NADPH-generating system, and chlorzoxazone. After another 20-minute incubation, the reactions were terminated by adding ice-cold acetonitrile (200 µl) containing 5,7-dihydroxycoumarin (IS).
Trapping the O-Quinone Metabolite by LC-MS/MS
To identify the potential O-quinone metabolite of MOA, GSH was used as a trapping agent. In brief, the incubation mixture containing PBS buffer, MOA (100 μM), MgCl2 (3.2 mM), GSH (10 mM), and HLMs (1.0 mg protein/ml) was vortexed for 10 seconds and then placed at 37°C for a 3-minute preincubation. The oxidative reaction was initiated via adding NADPH (1.0 mM). After a 30-minute incubation, the reaction was terminated by ice-cold acetonitrile (equal volume) and then vortexed and centrifuged. The supernatants (5.0 μl) obtained from this reaction were injected into another LC-MS/MS system and then analyzed (Shenyang Laboratory).
Determination of P450 Enzymes Involved in MOA Bioactivation
To assign the key enzymes responsible for MOA bioactivation, MOA together with individual human recombinant P450s (100 nM), including CYP1A2, 2A6, 2C8, 2C9, 2C19, 2D6, 2E1, and 3A4, was incubated. The incubation conditions and sample processing were identical with those for the characterization of metabolites as above. The enzyme with the most effective catalysis was normalized to 100%.
Docking Simulations
To gain insight into the binding of MOA to the active pockets of three key enzymes (CYP2D6, 1A2, and 2E1) responsible for MOA bioactivation, AutoDock Vina (version 1.1.2) utilizing the Lamarckian genetic algorithm was employed to molecular docking calculation (Trott and Olson, 2010). The crystal structures of CYP2D6 (PDB: 3TDA), CYP1A2 (PDB: 2HI4), and CYP2E1 (PDB: 3KOH) were downloaded from Protein Data Bank (http://www.rcsb.org/pdb), followed by inputting MOA generated in ChemOffice (version 14.0). Water molecules in crystal structures were eliminated, and hydrogen atoms were added to the protein. Subsequently, Kollman charges were assigned. The grid box was set to 70 × 70 × 70 Å3, with the spacing of 0.375. Then, MOA was docked within the active site of each P450 three-dimensional structure. The optimal conformers obtained on the basis of taking binding energy into consideration were held for further analyses.
Prediction of Herb-Drug Interaction Risks Triggered by MOA
To forecast the risks of potential herb-drug interactions triggered by MOA, we assessed the changes of the area under the plasma concentration-time curve (AUC) in the presence or absence of MOA. The interaction risk resulting from reversible inhibition was predicted using eq. 1. The risk caused by time-dependent inhibition of CYP2E1 and CYP2D6 was predicted using eq. 2 (Fang et al., 2010).
The two primary parameters of eq. 1 were the fraction of hepatic clearance of the drugs mediated by P450s (fhep) and the hepatic extraction ratio (Eh), respectively. I reflects the predicted maximum concentration of MOA in human plasma, and Ki or KI (micromolar) is the determined inhibitory constant of MOA against P450s. KI,u is the unbound KI. The Kdeg values of CYP2D6 and CYP2E1 are 0.000226 and 0.000192 minutes−1, respectively (Obach et al., 2007). fu,m is free fraction of MOA in HLMs, and Cmic is microsomal protein concentration used in preincubation; Cmic for CYP2E1 was 1.0 mg/ml, and for CYP2D6 it was 2.0 mg/ml. ADMET laboratory was used to calculate the log P value of MOA as 2.87.
Data Analysis
The inhibition constants (including IC50 values, Ki, KI, and kinact values) were calculated by nonlinear regression, and all data are expressed as means ± S.D. of triplicate assays.
Results
Inhibitory Effects of MOA on Eight Human Hepatic P450s
To assess the inhibitory effects of MOA on eight key hepatic P450s, a preliminary assay was performed in HLMs fortified with MOA at final concentrations of 1.0, 10, and 100 μM. As shown in Supplemental Fig. 1, MOA displayed concentration-dependent inhibition on CYP1A, 2C9, 2C19, 3A, 2C8, 2E1, and 2D6 but not CYP2A6. After that, concentration-response curves of MOA on the formation rates of each oxidative metabolite in HLMs were plotted. As shown in Fig. 2, MOA concentration-dependently inhibited the catalytic activities of CYP1A, 2C8, 2C9, 2C19, 2D6, 2E1, and 3A, with IC50 values of 1.06, 3.43, 1.07, 2.63, 16.38, 9.13, and 2.75 µM, respectively (Table 1).
Time-Dependent Inhibition of MOA against Seven Human P450s
Next, time-course changes in the remaining P450 activities were determined in human microsomal incubations containing MOA and the NADPH-generating system. We found that a 30-minute preincubation of MOA with HLMs in the presence of NADPH could result in significant losses of CYP2D6 (from 16.38 to 8.35 µM of IC50) and CYP2E1 (9.13 µM to 1.09 µM of IC50) activities. By contrast, no such enzyme inhibitions were observed for CYP1A, 2C8, 2C9, 2C19, or 3A (the changes were less than 15%) (Fig. 2; Supplemental Table 4). These findings clearly demonstrated that MOA was a reversible inhibitor of CYP1A, 2C8, 2C9, 2C19, and 3A and that this agent was a time-dependent inhibitor of CYP2D6 and 2E1.
Inhibition Kinetics of MOA against Human P450s
Next, the inhibition kinetics of MOA against eight human hepatic P450s were investigated. As shown in Fig. 3, the Lineweaver-Burk plots clearly showed that MOA was a competitive inhibitor of CYP1A, CYP2C8, CYP2C9, and CYP2C19 in HLMs, whereas MOA acted as a noncompetitive inhibitor of CYP3A in HLMs. The Ki values for CYP1A, 2C8, 2C9, 2C19, and 3A were found to be 1.71, 1.64, 2.87, 5.78, and 4.31 µM, respectively.
To assess the time-dependent inhibition of CYP2D6 and CYP2E1 by MOA, inactivation parameters KI and kinact values were determined in HLMs. As calculated from the plots of Fig. 4, the KI, kinact, and kinact/KI values of MOA for CYP2D6 were found to be 207 µM, 0.07 minutes−1, and 3.38 × 10−4 minutes−1µM−1, respectively, whereas the KI, kinact, and kinact/KI values for CYP2E1 were 20.9 µM, 0.03 minutes−1, and 14.4 × 10−4 minutes−1µM−1, respectively. Clearly, MOA inactivated CYP2E1 more potently than CYP2D6.
NADPH-Dependent Inhibition of CYP2E1 by MOA
Next, NADPH-dependent inhibition assay was conducted to determine whether the time-dependent inhibition of MOA against CYP2E1 was NADPH-dependent. As shown in Fig. 5A, coincubation with CYP2E1 and MOA (10 µM) without NADPH did not decrease the catalytic activity of CYP2E1. In sharp contrast, a 30-minute coincubation of HLMs with MOA (10 µM) in the NADPH-generating system resulted in significant loss of CYP2E1 activity (35%). These findings clearly suggested that the observed CYP2E1 inhibition was NADPH-dependent and that such a process required the metabolic activation by the host P450 enzyme(s).
Effects of Potassium Ferricyanide on MOA-Mediated CYP2E1 Inactivation
Potential involvement of carbene in the P450 inactivation was probed by coincubation with potassium ferricyanide (Watanabe et al., 2007), which would reverse the inhibition that resulted from carbene-Fe2+ chelation (Fig. 10) by oxidation of ferrous ion of the host enzyme. Herein, we investigated the effects of potassium ferricyanide on MOA-mediated CYP2E1 inactivation. As shown in Fig. 5B, potassium ferricyanide failed to show the protective effect against the inactivation of CYP2E1 (less than 20%) induced by MOA. This finding suggested that carbene intermediate (Fig. 10A) did not participate in the inactivation of CYP2E1.
Identification of O-Quinone Metabolite of MOA
Next, the reactive metabolite(s) of MOA in HLMs in the NADPH-generating system was trapped by GSH. As shown in Fig. 6, a single product peak (retention time = 7.64 minutes) with a pseudomolecular ion [M + H]+ at m/z 331 (12 Da less than that of the parent drug) was observed from the incubation mixture, indicating that MOA experienced a loss of carbon during oxidation by microsomal proteins. The mass spectrum of the product revealed product ions m/z 123 and m/z 209, suggesting that an O-demethylenation took place to form the corresponding catechol metabolite (Fig. 10B). Additionally, we succeeded in detection of a GSH conjugate–derived ([M + H]+ at m/z = 636, retention time = 5.84 minutes, Fig. 7) O-quinone intermediate (Fig. 10D). The product ion of MOA-GSH conjugate possessed a characteristic loss of GSH moiety, including 75 Da (C2H5O2N, loss of glycinyl moiety), 129 Da (C5H7O3N, neutral loss of γ-glutamyl moiety), and 275 Da (C10H16O6N3, cleavage of C-S bond moiety of GSH). Both the catechol metabolite and MOA-GSH-adduct shared the same product ion at m/z 209, revealing that the A-C ring system retained intact. By contrast, trace amount of MOA-GSH conjugate was found in the negative incubations without addition of NADPH (Fig. 7A). The detection of MOA-GSH conjugate suggests that quinone metabolite (Fig. 10D) could be formed and that this reactive intermediate might inactivate the host P450 enzymes via modification of the cystines in the host enzyme.
Effects of SOD/Catalase and GSH on MOA-Mediated CYP2E1 Inactivation
Subsequently, the effects of GSH and SOD/CAT on MOA-mediated CYP2E1 inactivation were also investigated. As shown in Table 2, after a 30-minute incubation, the residual activities of CYP2E1 with GSH (15.6% ± 0.35%) were much closed to that without GSH (18.85% ± 0.51%). This finding suggested that GSH could not block CYP2E1 inactivation, implying that CYP2E1 inactivation possibly happened in the active site of CYP2E1. Additionally, the mixture of SOD and CAT, as reactive oxygen species scavenger, displayed weak effects on MOA-mediated CYP2E1 inactivation. This finding indicated that reactive oxygen species made limited contribution to the enzyme inactivation.
P450s Participating in MOA Bioactivation
To recognize the key P450 enzyme(s) responsible for the bioactivation of MOA, eight major hepatic human recombinant P450s were individually incubated with MOA in the presence of both GSH and NADPH, followed by quantification of the formation rates of the MOA-GSH conjugate. CYP2D6, 1A2, and 2E1 were found to be the major contributors to catalyze the metabolic activation of MOA to form a catechol metabolite that could be captured by GSH, whereas CYP3A4, 2A6, 2C8, 2C9, and 2C19 contributed to a lesser extent (Fig. 8). Clearly, multiple human hepatic P450s could catalyze the metabolic activation of MOA.
Docking Simulations of MOA in CYP1A2, CYP2D6, and CYP2E1
To better decipher the interactions between MOA and CYP1A2, 2D6, and 2E1 at molecular levels, docking simulations of MOA into these three key hepatic P450s that catalyzed the bioactivation of this agent were carried out. As drawn in Figs. 9 and 10, MOA could be tightly docked into the catalytic cavity of CYP1A2, 2D6, and 2E1, whereas the carbon atom of MDP moiety in MOA was positioned toward the heme, with the distance between the carbon atom of MDP and the heme iron being 3.4, 4.1, and 4.4 Å in CYP2D6, 1A2, and 2E1, respectively. Meanwhile, the binding energies of the lowest-energy conformations of MOA in these three P450s were also predicted as −9.8, −9.3, and −8.8 kcal/mol for CYP2D6, 1A2, and 2E1, respectively (Supplemental Table 5). In addition, the key interactions between MOA and these three P450s were also investigated. As depicted in Supplemental Figs. 3–5, rings A and B of MOA could interact with some key residuals and heme in the catalytic cavity of CYP1A2, 2D6, and 2E1 mainly via hydrophobic interactions, such as π-π stacking and π-alkyl interactions. These findings suggested that MOA could enter into the catalytic cavities of CYP1A2, 2D6, and 2E1 and that these enzymes could activate this agent via oxidation of the carbon atom in the MDP moiety of MOA.
Quantitative Predictions of the HDI Risks Triggered by MOA
Finally, the risks of HDIs triggered by MOA were predicted by using a set of in vitro inhibition data. The magnitudes of the potential HDI between P450 substrate drugs and MOA-related products were estimated by increase of AUC in the presence of the MOA. As shown in Table 3, MOA hardly triggered a significant increase in AUC of the substrate drugs that were predominately metabolized by human hepatic CYP1A2, 2C8, 2C9, 2C19, and 3A4. By contrast, MOA, the irreversible inactivator of CYP2D6 and CYP2E1, could trigger dramatic increase in the AUC of the substrate drugs that were predominately metabolized by human CYP2D6 and CYP2E1, especially the substrate drugs of CYP2E1. These findings suggested that MOA-related herbal medicines might trigger HDI when MOA-related products were coadministered with CYP2D6 and CYP2E1 substrate drugs.
Discussion
In view of the fact that O. radix–related herbal products are commonly used in combination with some synthetic pharmaceutical agents for treating angiocardiopathy and different types of cancer (Chen et al., 2016a; Liu et al., 2017; Duan et al., 2018; Fan et al., 2020), the potential risks of the HDIs should be carefully examined. Although the molecular mechanisms of HDIs are diverse, most cases of the clinically relevant HDIs are related with pharmacokinetic alterations (Qin et al., 2017; Sun et al., 2017; Dunkoksung et al., 2019; Ge, 2019) in which coadministered herbal products alter the function or expression of drug-metabolizing enzymes or transporters that are responsible for the elimination of the therapeutic agents. Unfortunately, to date, the interactions between the major constituents in O. radix and drug-metabolizing enzymes in humans are rarely investigated.
In this study, MOA, one of the most abundant constituents in the Chinese herb mai-dong, was selected as a representative compound of naturally occurring homoisoflavonoids to assay its inhibitory effects against human P450s and to decipher the inhibitory mechanism of this homoisoflavonoid. The results obtained from this work clearly demonstrated that MOA was a reversible inhibitor of CYP1A, 2C8, 2C9, 2C19, and 3A but that this agent acted as an irreversible inhibitor of CYP2D6 and CYP2E1. In these cases, the potential risks of MOA to trigger HDI via P450 inhibition or inactivation were routinely predicted by using in vitro inhibition constants and their pharmacokinetic parameters. As shown in Table 3, MOA hardly triggered significant increases in the AUC of the substrate drugs that were predominately metabolized by human hepatic CYP1A, 2C8, 2C9, 2C19, and 3A. By contrast, MOA may result in an obvious increase in the AUC of the substrate drugs that were predominately metabolized by human CYP2D6 and CYP2E1. Considering that the MOA-related herbal products are always used long-term in some aged patients, enhanced attention should be paid on the potential risks of herb-drug interactions between MOA-related products and CYP2D6/CYP2E1 substrate drugs. It also should be noted that most of CYP2E1 substrate drugs (such as chlorzoxazone) have relatively wide therapeutic windows and that these agents are rarely used in combination with O. radix–related products; it is unlikely to trigger clinically relevant HDIs when O. radix–related products are coadministered with CYP2E1 substrate drugs. For CYP2D6, it is well known that this enzyme participates in the metabolic clearance of over 25% of therapeutic agents (Carrão et al., 2019), such as antiarrhythmics (encainide, etc.), antihypertensives (nicergoline, etc.), and antidepressants (clomipramine, etc.). In these cases, the combination use of MOA-related products with the drugs prominently metabolized by CYP2D6 should be used with caution to avoid the occurrence of clinically relevant HDIs, especially those CYP2D6-substrate drugs possessing narrow therapeutic index (such as dextromethorphan, metoprolol, and venlafaxine) (Zhou, 2009).
It should also be noted that inhibition or dysfunction of human P450s is a double-edged sword and, in some cases, may result in beneficial effects for human health. For example, inactivation of CYP2E1 and CYP2D6 could block the metabolic activation of some drugs or toxins, such as acetaminophen, isoniazid, tamoxifen, and styrene (Fan and Bolton, 2001; Chen et al., 2019). Inactivation of CYP2E1 and CYP2D6 by MOA may block the formation of the reactive metabolites of these agents in part, which in turn reduces the adduct of biomolecules, such as protein and DNA, derived from these reactive metabolites (Fan and Bolton, 2001; Zembutsu et al., 2011; Chen et al., 2019). Therefore, it is easily conceivable that MOA may be beneficial for alleviating the idiosyncratic toxicity or hepatoxicity of acetaminophen, isoniazid tamoxifen, and styrene by considering that these agents could be activated by CYP2E1 or CYP2D6.
Furthermore, both CYP1A and CYP2E1 have been validated as the crucial enzymes responsible for the activation of some known procarcinogens (such as benzopyrene, nitrate, and aflatoxin) to their carcinogenic metabolites (Zhou et al., 2009; Chen et al., 2019); thus, these two P450s have been considered as key targets for preventing various types of cancer. Meanwhile, CYP2E1 also plays a crucial role in oxidation of ethanol, which converted ethanol to acetate in the human body and might result in immunogenic reactions and alcoholic liver injury (Chen et al., 2019). Increasing evidence has demonstrated that CYP1A or CYP2E1 inhibitor therapy may partially block CYP1A- or CYP2E1-mediated activation of procarcinogens and ethanol, which in turn reduces the risks of certain cancers and alcoholic hepatotoxicity. In the present study, our findings clearly demonstrate that MOA strongly inhibits CYP1A (IC50 = 1.06 μM, Ki = 1.71 μM) and potently inactivates CYP2E1 (KI = 20.9 µM; kinact = 0.03 minutes−1), which partially explains the cancer-preventive effects of MOA and suggests that MOA is a drug with great promise for the development of dual inhibitors against both CYP1A and CYP2E1. As one of the most abundant constituents in O. radix, MOA could be isolated from this herb, and its chemical structure could be easily modified by the chemists to generate a series of MOA derivatives. Therefore, in the future, a series of MOA derivatives should be obtained to study the structure-P450 inhibition activity relationships, which will be very useful for designing and developing novel cancer-preventive agents via targeting procarcinogen-activating P450s.
Authorship Contributions
Participated in research design: Zheng, Ge.
Conducted experiments: Tu, Mao.
Contributed new reagents or analytic tools: Zhang, He, J.-J. Wu, Zhao.
Performed data analysis: Tu, Mao.
Wrote or contributed to the writing of the manuscript: Tu, Mao, Y. Wu, Zheng, Ge.
Footnotes
- Received December 4, 2020.
- Accepted March 9, 2021.
↵1 D.-Z.T. and X.M. contributed equally to this work.
This study was financially supported by the National Natural Science Foundation of China [81773810, 81922070, 81973286, 81773687], the National Key Research and Development Program of China [2020YFC0845400, 2017YFC1700200], Program of Shanghai Academic/Technology Research Leader [18XD1403600], Shanghai Talent Development Fund [2019093], Shuguang Program [18SG40], and the Shanghai Education Development Foundation and Shanghai Municipal Education Commission, the Key R&D and Transformation Science and Technology Cooperation Project of Qinghai Province [2019-HZ-819] and Program for Innovative Leading Talents of Qinghai Province [2018 and 2019].
The authors declare no competing interests.
↵This article has supplemental material available at dmd.aspetjournals.org.
Abbreviations
- AUC
- area under the plasma concentration-time curve
- CAT
- catalase
- G-6-P
- d-glucose-6-phosphate
- G-6-PDH
- glucose-6-phosphate dehydrogenase
- GSH
- glutathione
- HDI
- herb-drug interaction
- HLM
- human liver microsome
- HP LC
- High performance liquid chromatography
- IS
- internal standard
- LC-MS
- liquid chromatograph-mass spectrometer
- m/z
- mass-to-charge ratio
- MDP
- methylenedioxyphenyl
- MOA
- methylophiopogonanone A
- P450
- cytochrome P450 enzyme
- PDB
- protein data bank
- PBS
- potassium phosphate buffer
- SOD
- superoxide dismutase
- tR
- retention time
- Copyright © 2021 by The American Society for Pharmacology and Experimental Therapeutics