1521-009X/43/5/713–724$25.00 http://dx.doi.org/10.1124/dmd.114.060905DRUG METABOLISM AND DISPOSITION Drug Metab Dispos 43:713–724, May 2015Copyright ª 2015 by The American Society for Pharmacology and Experimental Therapeutics

Guanfu Base A, an Antiarrhythmic Alkaloid of Aconitum coreanum, Isa CYP2D6 Inhibitor of Human, Monkey, and Dog Isoforms s

Jianguo Sun, Ying Peng, Hui Wu, Xueyuan Zhang, Yunxi Zhong, Yanan Xiao, Fengyi Zhang,Huanhuan Qi, Lili Shang, Jianping Zhu, Yue Sun, Ke Liu, Jinghan Liu, Jiye A, Rodney J. Y. Ho,

and Guangji Wang

Key Laboratory of Drug Metabolism and Pharmacokinetics, State Key Laboratory of Natural Medicines (J.S., Y.P., H.W., X.Z., Y.Z.,Y.X., F.Z., H.Q., L.S., J.Z., Y.S., K.L., J.A., G.W.), and Department of Natural Medicinal Chemistry (J.L.), China Pharmaceutical

University, Nanjing, China; and Department of Pharmaceutics, University of Washington, Seattle, Washington (R.J.Y.H.)

Received September 3, 2014; accepted February 13, 2015

ABSTRACT

Guanfu base A (GFA) is a novel heterocyclic antiarrhythmic drugisolated from Aconitum coreanum (Lèvl.) rapaics and is currently ina phase IV clinical trial in China. However, no study has investigated theinfluence of GFA on cytochrome P450 (P450) drug metabolism. Wecharacterized the potency and specificity of GFA CYP2D inhibitionbased on dextromethorphan O-demethylation, a CYP2D6 probe sub-strate of activity in human, mouse, rat, dog, and monkey livermicrosomes. In addition, (+)-bufuralol 19-hydroxylation was used asa CYP2D6 probe for the recombinant form (rCYP2D6), 2D1 (rCYP2D1),and 2D2 (rCYP2D2) activities. Results show that GFA is a potentnoncompetitive inhibitor of CYP2D6, with inhibition constant Ki = 1.206

0.33mM in human liver microsomes (HLMs) and Ki = 0.376 0.16mM forthe human recombinant form (rCYP2D6). GFA is also a potent

competitive inhibitor of CYP2D in monkey (Ki = 0.38 6 0.12 mM) anddog (Ki = 2.4 6 1.3 mM) microsomes. However, GFA has no inhibitoryactivity on mouse or rat CYP2Ds. GFA did not exhibit any inhibitionactivity on human recombinant CYP1A2, 2A6, 2C8, 2C19, 3A4, or 3A5,but showed slight inhibition of 2B6 and 2E1. Preincubation of HLMs andrCYP2D6 resulted in the inactivation of the enzyme, which wasattenuated by GFA or quinidine. Beagle dogs treated intravenouslywith dextromethorphan (2 mg/ml) after pretreatment with GFA injectionshowed reduced CYP2D metabolic activity, with the Cmax of dextro-rphan being one-third that of the saline-treated group and area underthe plasma concentration–time curve half that of the saline-treatedgroup. This study suggests that GFA is a specific CYP2D6 inhibitor thatmight play a role in CYP2D6 medicated drug-drug interaction.

Introduction

Guanfu base A (GFA; Fig. 1) (Liu and Gao, 1981) is a novelheterocyclic antiarrhythmic drug isolated from Aconitum coreanum(Lèvl.) rapaics (“Guanbaifu” in Chinese). In 2005, injectable GFA wasapproved in China for the treatment of ventricular arrhythmias (StateFood and Drug Administration, or China Food and Drug Administra-tion since 2013). In addition to being equally effective in controllingpremature ventricular contraction as propafenone (an active compara-tor), intravenous GFA was shown to be more effective and safer thanpropafenone in a 24-hour postinfusion in phase II/III clinical studies(Yang et al., 2006). As a product approved under the regulatorypathway for Chinese medicine, no information is available about GFA

effects on metabolic enzyme activities or its role in drug-drug in-teractions. GFA was demonstrated to convert to Guanfu base I, Guanfualcohol-amine, and GFA glucuronide and sulfate conjugates in rats(A et al., 2002). In human urine, an additional metabolite, but not theformer metabolites, GFA oxide, was also detectable (A et al., 2003).However, metabolism of GFA and its effects on metabolic enzymessuch as cytochrome P450 (P450) isoenzymes are yet to be fullyunderstood. In a preliminary P450 inhibition study, GFA appeared tobe a potent inhibitor of CYP2D6.As CYP2D6 is a key P450 isoenzyme that metabolizes multiple

antiarrhythmic drugs, including propafenone (Kroemer et al., 1991)and quinidine (Kobayashi et al., 1989; Bramer and Suri, 1999; Aiet al., 2009), it is essential to understand the selectivity and extent ofGFA inhibitory effects on P450 enzymes, particularly CYP2D6. CYP2D6is one of the best studied cytochrome P450 superfamilies initially reportedas the enzyme responsible for the debrisoquine/sparteine polymorphism.Since the discovery in 1977 that debrisoquine and sparteine producedunexpected adverse effects in several patients at the recommended doses(Mahgoub et al., 1977; Bertilsson et al., 1980), drug-drug interactions(DDIs) and the clinical impacts have been well documented. Some DDIsmay affect elimination, whereas others may impact overall drug exposure(de la Torre et al., 1999). Fatal toxicity could also occur due to DDIs.

This work was supported by the China “Creation of New Drugs” KeyTechnology Projects [Grants 2009ZX09103-345 and 2009ZX09502-004], theJiangsu Province Key Laboratory of Drug Metabolism and PharmacokineticsProjects [Grant BM2012012], Jiangsu Key Laboratory of Drug Design andOptimization, and the National Undergraduate Training Programs for Innovationand Entrepreneurship [Grant 201210316068].

J.S. and Y.P. contributed equally to this research.dx.doi.org/10.1124/dmd.114.060905.s This article has supplemental material available at dmd.aspetjournals.org.

ABBREVIATIONS: ADME, absorption, distribution, metabolism and excretion; arb, arbitrary units; AUC, area under the plasma concentration–timecurve; AUCi, AUC with inhibitor; BL, bufuralol; DDI, drug-drug interaction; DLM, dog liver microsome; DM, dextromethorphan; DXO, dextrorphan;GFA, Guanfu base A; HLM, human liver microsome; LC-MS/MS, liquid chromatography–tandem mass spectrometry; MKLM, monkey livermicrosome; P450, cytochrome P450; rCYP, recombinant cytochrome P450.

713

http://dmd.aspetjournals.org/content/suppl/2015/02/13/dmd.114.060905.DC1Supplemental material to this article can be found at:

at ASPE

T Journals on A

pril 7, 2018dm

d.aspetjournals.orgD

ownloaded from

CYP2D6-mediated venlafaxine-propafenone interaction may lead tohallucinations and psychomotor agitation (Gareri et al., 2008). Com-bination use of serotonergic agents that are substrates or metabolites ofCYP2D6 can also lead to severe adverse effects (Jagestedt and von Bahr,2004).As P450 metabolism of the antiarrhythmic drug GFA and its

potential to play a role in P450-mediated drug-drug interactions is notclear, we have characterized the interactions of GFA with P450enzymes. We found that GFA is a selective inhibitor for CYP2D6 inhuman, monkey, and dog microsomes, but not rats and mice. The invitro CYP2D6 selectivity data were confirmed with in vivo CYP2D6probe inhibition studies in beagle dogs, in which the Cmax of the mainmetabolite dextrorphan (DXO) was found to be one-third that of thesaline-treated group and the area under the plasma concentration–timecurve (AUC) was half that of the saline-treated group.Our article is the first to report that GFA is a potent and specific

noncompetitive inhibitor of CYP2D6 and has no effect on other P450isoenzme types. A predicted human drug-drug interaction model ofGFA with dextromethorphan is presented.

Materials and Methods

GFA (98.58%, acehytisine, Compound identification number 121446 or6917802) was provided by Liu Jinghan (Department of Phytochemistry, China Phar-maceutical University). Phenacetin, acetaminophen, coumarin, 7-hydroxycoumarin,paclitaxel, diclofenac, 4-hydroxydiclofenac, tolbutamide, 4-hydroxytolbutamide,S-mephenytoin, 49-hydroxymephenytoin, bufuralol, 1-hydroxybufuralol, dextro-rphan, chlorzoxazone, 6-hydroxychlorzoxazone, testosterone, 6b-hydroxytestos-terone, tryptamine, sertraline hydrochloride, quinidine, quercetin, sulfaphenazole,and nootkatone were purchased from Sigma-Aldrich (St. Louis, MO). Bupropionhydrochloride, hydroxybupropion, and 6a-hydroxypaclitaxel were obtained from

Toronto Research Chemicals Inc. (Ontario, Canada). Dextromethorphan hydro-bromide monohydrate was purchased from Adamas-beta (Shanghai, China).Ketoconazole and a-naphthoflavone were obtained from Tokyo Chemical In-dustry Co., Ltd. (Tokyo, Japan). NADP (N0505), glucose 6-phosphate (G7250),and glucose-6-phosphate dehydrogenase (G6378) were obtained from Sigma-Aldrich. Mephenamine was obtained from the National Institute for Control ofPharmaceutical and Biologic Products (Beijing, People’s Republic of China). Allother chemicals and solvents used were of the highest quality and analytical gradeor higher.

Human and Animal Liver Microsomes and Recombinant P450 Enzymes.Supersome preparations containing recombinant CYP2D6, CYP1A2, CYP2A6,CYP2B6, CYP2C8, CYP2C19, CYP2D1, CYP2D2, CYP2E1, CYP3A4, orCYP3A5 (expressed in baculovirus-infected insect cells) plus NADPH-P450reductase were purchased from BD Gentest (Woburn, MA). Pooled human andanimal microsomes were purchased from the Research Institute for LiverDiseases (Shanghai) Co. Ltd. (Shanghai, People’s Republic of China).

Enzyme Incubation. Incubation mixtures contained pooled liver microsomesfrom human or different animal species (0.2 mg of protein/ml) or recombinantsupersomes containing rCYP2D6 (10 pmol P450/ml) and various concentrationsof dextromethorphan (DM; 0–200 mM, pooled microsome systems) or bufuralol(BL; rCYP2D6 system) in 100 mM potassium phosphate buffer (pH 7.4)containing 10 or 3.3 mM magnesium chloride, respectively. After incubating at37�C for 5 minutes, the reactions were initiated by the addition of a NADPH-generating system (0.5 or 1.3 mM NADP, 10 or 3.3 mM glucose 6-phosphate,and 1 or 0.4 U/ml glucose-6-phosphate dehydrogenase for pooled liver micro-somes or rCYP2D6, respectively). The reaction mixtures (in 200 ml) wereincubated at 37�C for 20 minutes for pooled liver microsomes or 10 minutes forCYP2D6. The reaction was terminated by addition of 200 ml of acetonitrilecontaining 5 ng/ml mephenamine (internal standard). Samples were subse-quently centrifuged at 30,065 � g (Thermo Sovall Biofuge Stratos, Osterode,Germany) for 10 minutes to remove protein precipitates. The products insupernatant are analyzed by liquid chromatography–tandem mass spectrometry(LC-MS/MS).

Inhibition Studies. To evaluate the effects of GFA on P450 isozyme se-lectivity, a cocktail of probe substrates was incubated with the pooled micro-somes from different species, and simultaneous LC-MS/MS detection wasperformed (see LC-MS/MS Assay). To confirm the inhibitory effects, human livermicrosomes (HLMs), purified rCYP2D6 in supersomes, or other recombinantP450s (rCYP2D1, rCYP2D2) were incubated with DM or BL in the presence ofthe test compounds, including GFA or quinidine (up to 200 mM), as described inEnzyme Incubation. All compounds used were dissolved in acetonitrile andadded to the incubation mixture at a final acetonitrile concentration of #1%. Theacetonitrile in the mixtures exhibits less than 5% inhibition of the DMO-demethylation to DXO and BL hydroxylation activities. To determine the Ki

values for GFA and quinidine, the reactions were incubated with three substrateconcentrations of BL (2, 4, and 10 mM for rCYP2D6), DM (1, 5, and 20 mM) forthe pooled HLMs, DM (2, 6, and 15 mM) for the pooled monkey livermicrosomes (MKLMs), DM (2, 5, and 15 mM) for the pooled dog livermicrosomes (DLMs), and serial concentrations of GFA or quinidine.

Fig. 1. Chemical structure of GFA.

TABLE 1

Substrate and positive inhibitor of each P450 isoform

P450s Substrate (Concentration) Metabolite Probe Inhibitor (Concentration)

mM mM

1A2 Phenacetin (50 mM)a Acetaminophen a-naphthoflavone (0.1 mM)2A6 Coumarin (1 mM) 7-hydroxycoumarin Tryptamine (2 mM)2B6 Bupropion (5 mM) 2-hydroxybupropion Sertraline (1 mM)2C8 Paclitaxel (5 mM) 6a-hydroxypaclitaxel Quercetin (0.5 mM)2C9 Diclofenac (4 mM) 49-hydroxydiclofenac Sulfaphenazole (0.5 mM)

Tolbutamide (70 mM)a 4-hydroxytolbutamide Sulfaphenazole (0.5 mM)2C19 Mephenytoin (1 mM) 49-hydroxymephenytoin Nootkatone (0.5 mM)

Oxyomeprazole (20mM)a 5-hydroxyomeprazole Ticlopidine (1.0 mM)2E1 Chlorzoxazone (20 mM)a 6-hydroxychlorzoxazone Clomethiazole (10 mM)3A4/5 Midazolam (5 mM)a 19-hydroxymidazolam Ketoconazole (0.05 mM)

49hydroxymidazolamTestosterone (75 mM)a 6b-hydroxytestosterone Ketoconazole (0.05 mM)

aCocktail substrate

714 Sun et al.

at ASPE

T Journals on A

pril 7, 2018dm

d.aspetjournals.orgD

ownloaded from

The inhibition by two concentrations of GFA (10, 100 mM) was evaluated ina panel of other recombinant P450s, including CYP1A2, 2A6, 2B6, 2C8, 2C9,2C19, 2E1, 3A4, and 3A5. The incubation conditions for each enzyme wereaccording to the instructions of BD Gentest, based on their respective probesubstrates. The selective inhibitors used for each P450 isoform are as follows:a-naphthoflavone, 1A2; tryptamine, 2A6; sertraline, 2B6; quercetin, 2C8;sulfaphenazole, 2C9; nootkatone, 2C19; clomethiazole, 2E1, and ketoconazole,3A4/5. These were used as the positive controls. Table 1 summarizes thesubstrates and positive inhibitors of these enzymes.

Testing for Mechanism-Based Inhibition of CYP2D6 Activity by GFA.To evaluate the mechanism of GFA inhibition of CYP2D6, HLM andrCYP2D6 inhibition experiments were performed with varying concentrationsof GFA (0, 0.2, 0.4, 1.0, 2, or 4 mM) while fixing the HLM concentration at 0.2mg/ml. Other conditions remained the same. For the rCYP2D6 supersomesystem, varying GFA (0, 1, or 10 mM) concentrations were incubated witha fixed 10 pmol/ml rCYP2D6. For the HLM system, after preincubation withGFA for 0, 5, 10, 20, or 30 minutes, the 20-ml mixture was mixed with 180 ml

of DM in 100 mM potassium phosphate buffer (pH 7.4). For the rCYP2D6system, after preincubation with GFA for 0, 10, 20, or 30 minutes, the 50-mlmixture was mixed with 50 ml of BL containing the NADPH-generating system(as described in Enzyme Incubation) and 100 mM potassium phosphate buffer(pH 7.4). The final substrate concentrations were 50 mM for DM and 20 mMfor BL, and the inhibitor concentrations were 0, 0.02, 0.04, 0.1, 0.2, and 0.4mM for the HLM system and 0, 0.5, and 5 mM for the rCYP2D6 system.CYP2D6 activities are estimated based on degrees of dextromethorphanO-demethylation and bufuralol hydroxylation per minute. To evaluate whetherenzyme activity reduction is caused by the denaturation of the enzyme,phenacetin (70 mM) is added to the incubation system to probe the activity ofCYP1A2.

Drug-Drug Interaction Study in Beagle Dogs. An equal number of maleand female beagle dogs (Beijing Marshall Biotechnology Co. Ltd., Beijing,China) enrolled in this study were housed in a standard animal laboratory(temperature 22–25�C, humidity 30–70%) with a 12-hour light/dark cycle.All experiments were performed under a protocol approved by the China

TABLE 2

Optimized transitions and parameters of the analytes

Analytes MRM Transition Declustering Potential Collision Energy Polarity

m/z V eV

Acetaminophen 152.0 → 110.0 60 24 ESI +4-hydroxytolbutamide 287.0→188.0 60 18 ESI +7-hydroxycoumarin 163.0 → 107.0 90 30 ESI +2-hydroxybupropion 256.3 → 238.0 60 18 ESI +6a-hydroxypaclitaxel 870.5 → 286.2 90 20 ESI +49-hydroxydiclofenac 312.0 → 266.0 70 19 ESI +5-hydroxyomeprazole 362.0→196.0 60 42 ESI +49-hydroxymephenytoin 233.0 → 190.0 55 17 ESI 21’-hydroxybufuralol 278.0 → 186.0 100 26 ESI +Dextrorphan 258.5 → 157.0 180 50 ESI +6-hydroxychlorzoxazone 184.0 → 120.0 55 25 ESI 219-hydroxymidazolam 342.0 → 203.0 130 37 ESI +49-hydroxymidazolam 342.0 → 234.0 130 32 ESI +6b-hydroxytestosterone 305.0 → 269.0 60 21 ESI +6-hydroxychlorzoxazone 184.0 → 120.0 55 25 ESI 2Mephenamine (IS) 270.0 → 181.0 60 13 ESI +

ESI +, Electrospray ionization Positive; ESI 2, Electrospray ionization Negative; IS, internal standard; MRM, Multiple reactionmonitoring.

TABLE 3

Parameters used in physiologically based pharmacokinetics (PBPK) prediction

Physical/Chemical Properties Distribution Parameters Elimination Parameters Interaction Parameters

DMMW: 271.4 Model: minimal PBPK Enzyme kinetics: HLM Ki (mM) CYP2D6: 1.20/Log Po:w: 3.8 Vss: user input 14.3 CLint: CYP2D6:253Type: MB CYP3A4:4.3,CYP2B6:4.7 ml/min/mg proteinpKa 1:8.3B:P: 1.32Fraction unbound in plasma: 0.5

DXOMW: 257.37 Full PBPK model Enzyme kinetics: HLMLog Po:w: 3.53 Vss: predicted CLint: CYP3A4:60Type: MB ml/min/mg microsomal proteinpKa 1:9.66B:P: 0.55Fraction unbound in plasma: 0.5

GFAMW: 429.51 Minimal PBPK In vivo clearance Ki(mM)CYP2D6: 1.20Log Po:w: -0.49 Vss: 1.67 l/kg CV(%) 25% CLiv: 10.72 l/hType: DB CV(%) 28.7pKa 1:4.2pKa2:8.8B:P : 0.88Fraction unbound in plasma:Predicted

B:P, Blood-to-plasma partition ratio; CLint, In vitro intrinsic clearance; CLiv, in vivo clearance after intra venous injection; CV, coefficient of variation; DB, biprotic base; MB, monoprotic base;MW, molecular weight; Po:w, neutral species octanol: buffer partition coefficient; Vss, volume of distribution at steady state.

Guanfu Base A is a CYP2D6 Inhibitor 715

at ASPE

T Journals on A

pril 7, 2018dm

d.aspetjournals.orgD

ownloaded from

Pharmaceutical University Animal Ethics Committee in accordance with theGuidelines for Animal Experimentation of Southeast University (Nanjing, China).

DM was used as substrate in an in vivo study (VandenBrink et al., 2012).Blood samples were obtained before drug administration as a baseline measure.After the dogs were fasted overnight for at least 12 hours with free access towater, they were intravenously administered 10 mg/kg (1 ml/kg) GFA or sodiumsaline. After 30 minutes, the dogs were treated with 2 mg/ml (1 ml/kg) DM todetermine the change in CYP2D activity in vivo. Blood samples were collected at2, 5, 15, 20, 30, and 45 minutes and 1, 2, 4, 6, 8, 12, 24, and 36 hours. The bloodsamples were immediately centrifuged at 1485 � g for 5 minutes. The plasmasamples in polypropylene tubes were stored at –20�C until LC-MS/MS analysis.

LC-MS/MS Assay. The drugs in plasma and microsomes were analyzedwith a LC-MS/MS system fitted with a SPD-20A UFLC system (Shimadzu,Tokyo, Japan) coupled to an API 4000 tandem mass spectrometer (AppliedBiosystems, Foster City, CA) equipped with an electrospray ionization interface.Aliquots were injected onto a reversed-phase Luna C18 column (2.0 � 150 mm,5 mm; Phenomenex, Torrance, CA) at a flow rate of 0.5 ml/min. The mobilephase consisted of aqueous 0.01% acetic acid with 5 mM ammonium acetate (A)and 50:50 (v/v) methanol:acetonitrile (B). The following gradient elution wasapplied: 0–0.5 minute, 2% B; 0.5–4.0 minutes, increase B to 45%; 4.0–6.5minutes, increase B to 60%; 6.5–6.8 minutes, increase B to 80%; 6.8–7.2minutes, 80% B; 7.2–7.5 minutes, reduce B to 2% (for a run total time of 10.0minutes). The mass spectrometer conditions were as follows: 5500 V (positivemode) or –4500 V (negative mode); gas 1 (N2), 65 arbitrary units (arb); gas 2(N2), 70 arb; collision gas (N2), 10 arb; curtain gas (N2), 30 arb; entrance

potential, 10 V; collision cell exit potential, 12 V. Quantitation was performed inthe multiple reaction monitoring mode. The optimized transitions and parametersof the analytes are shown in Table 2. The assay is robust and reproducible, andday-to-day variation is validated to be less than 10% covariance (chromatogramsare shown in Supplemental Fig. 1, validation of cocktail incubation is shown inSupplemental Tables 1 and 2). To detect DXO and DM in the dog plasma, theLC-MS/MS method was similar to that used in vitro, with some modifications,and with m/z transitions of 272.1→215.0 for DM, 258.1→157.0 for DXO, and180.0→110.0 for internal standard.

Enzyme Kinetics Analysis. Based on the two concentrations of P450 enzymeproduct detected by LC-MS/MS assay, the enzyme activity was calculated andexpressed in picomoles of DXO or 1-OH-BL formed per picomole of CYP2D6per minute. The Michaelis-Menten kinetics parameters were calculated accordingto the relating reaction rate V to [S], the concentration of a substrate:

V ¼ Vmax× ½S�Km þ ½S� ð1Þ:

The velocity data for dextromethorphan O-demethylation or bufuralol19-hydroxylation in the absence and presence of GFA were estimated byderivative-free iterative nonlinear least-squares regression. IC50 values weredetermined graphically using curves of mean enzyme activity versus inhibitorconcentration by Microsoft Excel (Redmond, WA).

The modes of inhibition and the apparent Ki values were determined bynonlinear regression using the following equations for competitive (eq. 2),

Fig. 2. Inhibition of P450 isoforms by GFA in HLMs (0–100 mM) (A) and rCYP2D6 (0–20 mM) (B). The activities of each isoform were measured by isoform-specificsubstrate reaction probes at their approximate or lower, respectively, Km values (Supplemental Table 1). The data represent the averages of four (HLMs) or duplicate(rCYP2D6) experiments and are expressed as the percentages of the control activity remaining. CLZ, Chlorzoxazone; MDZ, Midazolam; OPZ, Omeprazole; PN-ACE,Phenacetin; TB, Tolbutamide; TS, Testosterone.

716 Sun et al.

at ASPE

T Journals on A

pril 7, 2018dm

d.aspetjournals.orgD

ownloaded from

noncompetitive (eq. 3), or mixed competitive (eq. 4) with GraphPad Prismversion 4.03 (GraphPad Software, Inc., San Diego, CA) (Webb, 1963):

V ¼ Vmax× ½S�Kmð1þ I=KiÞ þ ½S� ð2Þ

V ¼ Vmax× ½S�Kmð1þ I=KiÞ þ ½S�ð1þ I=KiÞ ð3Þ

V ¼ Vmax× ½S�Kmð1þ I=KiÞ þ ½S�ð1þ I=Ki

9Þ ð4Þ

where V, Vmax, [S], Km, Ki, Ki9, and [I] are the velocity, maximal velocity,substrate concentration, Michaelis constant, dissociation constant for theenzyme-inhibitor complex, dissociation constant for the enzyme-substrate-inhibitor complex, and inhibitor concentration, respectively.

Akaike’s information criterion was used as a measure of the goodness offit. The mode of inhibition was verified by visual inspection of Lineweaver-Burkplots, Dixon plot, and Cornish-Bowden plot of the enzyme kinetics data.

Modeling and Molecular Docking. The human CYP2D6 with the crystalstructures was downloaded from the Protein Data Bank (identifier: 3TBG, DOI:10.2210/pdb3tbg/pdb). Homology modeling and molecular docking were usedto predict the modes of GFA and quinidine binding to animal and humanCYP2D. The homology models of CYP2D from rat (CYP2D1,2) were pro-duced using SWISS-MODEL (Swiss Institute of Bioinformatics, Switzerland)workspace based on the template of 3TBGA (chain A) (Guex and Peitsch,1997; Schwede et al., 2003; Arnold et al., 2006). Active site of the protein waspredicted with SiteMap module (version 2.3; Schrödinger, LLC, New York,NY) using identity top-ranked potential receptor binding sites to generate 15possible binding orientations. Docking studies were performed using the

Fig. 3. The inhibitory potency of GFA (10 and 100 mM) for different recombinant P450s (all data were duplicated). The respective isoenzyme probe inhibitor was used asa positive control. The concentration used for specific probe inhibitor and substrate is listed in Table 1.

Fig. 4. Inhibition of P450 isoforms by GFA (0–100 mM) in MKLMs (A), DLMs (B), rat liver microsomes (C), and mouse liver microsomes (D). The activities of each isoform weremeasured by isoform-specific substrate reaction probes at their approximate respective Km values with no apparent drug-drug interaction. The data represent the averages of four repeatedexperiments and are expressed as the percentages of the control activity remaining. CLZ, Chlorzoxazone; MDZ, Midazolam; OPZ, Omeprazole; PN-ACE, Phenacetin; TS, Testosterone.

Guanfu Base A is a CYP2D6 Inhibitor 717

at ASPE

T Journals on A

pril 7, 2018dm

d.aspetjournals.orgD

ownloaded from

program CDOCKER (Discovery Studio 2.5.5, Accelrys Software Inc.) with thehighest SiteScore for GFA and quinidine.

Pharmacokinetic Prediction and Simcyp ADME Simulator. The pharmaco-kinetic profiles of GFA at 10 mg/kg were simulated to predict possible drug-druginteraction based on our previous study at 4 mg/kg in dogs and humans (Wang et al.,2000; Yang et al., 2000; Sun Jianguo, 2012). The pharmacokinetic parameters ofDM and DXO in dogs were calculated using WinNonlin version 6.1 (Pharsight,Mountain View, CA) according to a noncompartmental model. The possibility of

DDIs is roughly evaluated according to the predictive function [I]/Ki, where [I]/Ki. 1suggests a high DDI risk (Ito et al., 2004; Obach et al., 2006). According to themetabolism pathway of DM by CYP2D6, only one pathway subject to inhibitionand the DDI was predicted with the following equation (Ito et al., 2005):

AUCi

AUC¼ 1

fmCYP1þ½I�

Ki

þ ð12 fmCYPÞð5Þ:

Fig. 5. Recombinant rat CYP2D1-, CYP2D2-, and human CYP2D6-catalyzed (+)-bufuralol 19-hydroxylation. The 19-hydroxylated (+)-bufuralol formation rate (picomoleper minute per picomole of P450) from 25 mM (+)-bufuralol by CYP2D1(triangle), CYP2D2 (circle), and CYP2D6 (square) in the presence of GFA (closed) or quinidine(QND, open) (0–200 mM) is expressed as a percentage of the control velocity with no inhibitor present. The data represent the averages of duplicate experiments.

Fig. 6. Determination of Ki values and types of inhibition. Dixon plot (1/V against i) (A) and Cornish-Bowden plot (S/V against i) (B) of dextromethorphan O-demethylationvelocities mediated by HLM (0.2 mg of protein/ml) in the presence of various GFA concentrations. Each line represents a concentration of dextromethorphan (circles, 1 mM;squares, 5 mM; and diamonds, 20 mM) and their intersection points determine the Ki values.

718 Sun et al.

at ASPE

T Journals on A

pril 7, 2018dm

d.aspetjournals.orgD

ownloaded from

A direct measure of the importance of CYP2D6 (that is, the fraction of thedrug metabolized by CYP2D6, or fmCYP2D6) can be made by comparison of thepoor and extensive phenotypes (Schmid et al., 1985; Yeh et al., 2003), and inthis case, fmCYP2D6 is 0.96 for DM (Nakashima et al., 2007). Ki is theinhibition constant, and [I] is the maximum inhibitor concentration, and theunbound concentration was used here for prediction.

The commercially available physiologically based pharmacokinetics soft-ware package Simcyp (version 13.1.61.0; Simcyp Ltd., Sheffield, UK) waswidely used for drug DDI prediction (McGinnity et al., 2008; Youdim et al.,2008; Xu et al., 2009; Shardlow et al., 2013). In this study, the DDI potentialof GFA on substrate of CYP2D6 in humans was predicted based on aphysiologically based pharmacokinetics model simulated by Simcyp at 4 mg/kgaccording to clinical dose. Simcyp was used to simulate a drug interaction trialinvolving one trial of 10 subjects per trial, with 2 mg/kg DM i.v. coadministeredwith 4 or 10 mg/kg GFA by i.v. bolus or infusion. The major metabolite DXO

was also simulated in the trial. The parameters used in the simulation are listed inResults (Table 3).

Results

Effects of Guanfu Base A on Different P450 Isoenzymes. Toassess its interaction potential, the inhibitory effects of GFA on differentP450s in HLM or rCYP2D6 were measured. The formation of DXOfrom DM or the 1-hydroxylation of BL was determined with LC-MS/MS. As presented in Fig. 2, GFA is a potent inhibitor of CYP2D6, witha 50% inhibitory concentration (IC50) recorded at ;0.46 mM in HLM(DM 5 mM, Fig. 2A) and 0.12 mM in rCYP2D6 (BL 5 mM, Fig. 2B).However, GFA showed no inhibition on other P450s (Fig. 2A).To assess whether and to what extent GFA inhibits specific cyto-

chrome P450 isoenzyme activity, specific recombinant human P450isoenzymes were incubated further with either 10 or 100 mM. Thehuman recombinant P450 isoenzymes 1A2, 2A6, 2B6, 2C8, 2C9, 2C19,2E1, 3A4, and 3A5 were evaluated with respective probe substrates aswell as probe inhibitors to validate the specific P450 isoenzymeassay. As shown in Fig. 3, other than CYP2B6 and CYP2E1, none ofthe isoenzymes’ activity was inhibited by GFA. At a high concentration(100 mM GFA), 40% of CYP2B6 and CYP2E1 activity was inhibited(Fig. 3).Collectively, these data indicate that GFA is a potent CYP2D6

inhibitor and can partially inhibit 2B6 and 2E1 at 100 mM (highconcentration), and shows no significant inhibition at 10 mM (lowconcentration). Therefore, the following experiments focus on GFAinhibitory effects on the CYP2D6 isoenzyme.Characterization of GFA Inhibitory Effects on CYP2D6 Iso-

enzyme Activity. We next determined whether GFA inhibitory effectson CYP2D6 display species specificity. To do so, dose-dependent GFA

Fig. 7. Determination of Ki values and types of inhibition. Dixon plot (1/V against [I]) (A) and Cornish-Bowden plot (S/V against [I]) (B) of (+)-bufuralol 19-hydroxylationvelocities mediated by rCYP2D6 (2 pmol/ml) in the presence of various GFA concentrations. Each line represents a concentration of (+)-bufuralol (circles, 2 mM; squares,5 mM; and diamonds, 10 mM) and their intersection points determine the Ki values.

TABLE 4

Inhibition of CYP2D by GFA and quinidine

The inhibition constants (Ki values) were calculated from the appropriate nonlinear regressionenzyme inhibition model. The mechanism of inhibition was determined graphically.

Drug Ki Mechanism of Inhibition

mM

GFAHLM 1.20 6 0.33 NoncompetitiverCYP2D6 0.37 6 0.16 NoncompetitiveMKLM 0.38 6 0.12 CompetitiveDLM 2.4 6 1.3 Competitive

QuinidineHLM 0.19 6 0.079 CompetitiverCYP2D6 0.027 Competitive (Ching et al., 1995)MKLM 3.64 6 1.63 CompetitiveDLM 2.5 6 1.29 Noncompetitive

Guanfu Base A is a CYP2D6 Inhibitor 719

at ASPE

T Journals on A

pril 7, 2018dm

d.aspetjournals.orgD

ownloaded from

CYP2D6 inhibitory effects in liver microsomes from different animalspecies were also evaluated. As with human data, GFA potentlyinhibited the CYP2D6 orthologs in monkeys and dogs, with IC50

recorded at ;0.45 and ;9 mM, respectively. However, we found thatGFA did not inhibit rat or mouse orthologs of human CYP2D6 (Fig. 4),but showed slight inhibition of CYP2E1; this might be due to thespecies difference (Supplemental Fig. 2). The CYP2D6 IC50 values forGFA were further evaluated in purified recombinant proteins insupersomes. The IC50 of GFA for rCYP2D6 was recorded at 0.1 mMcompared with 0.028 mM for the probe drug quinidine (Fig. 5). RatCYP2D1 and CYP2D2 are considered to be orthologs of human

CYP2D6 that are expressed in the liver (Komori, 1993; Wyss et al., 1995;Hiroi et al., 1998). Unlike its potent inhibitory effect on human CYP2D6,GFA had no inhibitory effect on CYP2D1 or CYP2D2 (Fig. 5).Taken together, these data confirmed that GFA inhibitory effects are

mainly restricted to CYP2D6 in humans and show a similar degree ofpotency for the orthologs in dogs and monkeys, but not rats or mice.Kinetic Analysis of GFA Inhibitory Effects on CYP2D6 Activity.

To characterize the CYP2D6 enzyme kinetics of GFA inhibition, thereaction velocities were assessed in the HLM system at three substrateconcentrations (1, 5, and 20 mM DM) in the absence and presence ofGFA (0, 0.2, 0.5, 1, 2, or 5 mM). In the rCYP2D6 system, the reactions

Fig. 8. Time-dependent inactivation of CYP2D6 by GFA. HLMs (0.2 mg/ml) (A) or rCYP2D6 (10 pmol/ml) (B) was preincubated with buffer (control) or variousconcentrations of GFA (0.1, 0.2, 0.5, 1.0, or 20 mM) and a NADPH-generating system. Each point represents the mean of four measurements6 S.D. for HLMs and duplicatemeasurements for rCYP2D6. Refer to Testing for Mechanism-Based Inhibition of CYP2D6 Activity by GFA in the text for details on the sample processing.

720 Sun et al.

at ASPE

T Journals on A

pril 7, 2018dm

d.aspetjournals.orgD

ownloaded from

were performed with three substrate concentrations (2, 5, and 10 mMBL) in the absence and presence of GFA (0, 0.04, 0.2, 1, or 4 mM). Thedata obtained for the dextromethorphan O-demethylation velocitiesmediated by HLMs (0.2 mg of protein/ml) and the velocities of the 1-hydroxylation of BL mediated by rCYP2D6 in the presence of GFAboth best described a noncompetitive inhibition model. The initial rateof product formation in varying concentrations of GFA was analyzedwith a Dixon plot (Figs. 6A and 7A) and a Cornish-Bowden plot(Figs. 6B and 7B). Based on the Dixon and Cornish-Bowden analysis inFigs. 6 and 7, GFA is a noncompetitive inhibitor of CYP2D6, withKi estimates of 1.20 6 0.33 mM and 0.37 6 0.16 mM in HLMs andrCYP2D6, respectively. To compare the mode of GFA inhibition, the Ki

values were further evaluated in DLMs and MKLMs. The results inTable 4 show that GFA is a competitive inhibitor of CYP2D in DLMsand MKLMs, with Ki = 2.4 6 1.3 and 0.38 6 0.12 mM, respectively.As the positive control, quinidine was shown to be a competitiveinhibitor of HLMs (Ki = 0.19 6 0.079) and MKLMs (Ki = 3.64 61.63), and a noncompetitive inhibitor of DLMs (Ki = Ki9 at 2.5 61.29; Table 4).Evaluation of Whether GFA Inhibition of CYP2D6 Is Mecha-

nism-Based. To determine whether the inhibition of CYP2D6 by GFA

is mechanism-based, GFA was preincubated with HLMs or rCYP2D6in the presence or absence of NADPH. Compared with the negativecontrol, which was preincubated without GFA, preincubation of HLMsor rCYP2D6 with GFA did not increase the inhibitory potency of thereaction (Fig. 8). Therefore, GFA seems to be a specific noncompetitiveinhibitor of CYP2D6, and this inhibition is not mechanism-based.Characterization of GFA Inhibitory Effect on CYP2D6 with

Dextromethorphan, a Probe Drug Substrate, in Beagle Dogs. Tocharacterize the GFA inhibitory effects on drugs subjected toCYP2D6-dependent metabolic clearance, we evaluated the pharma-cokinetics of the CYP2D6 probe drug dextromethorphan in beagledogs in the presence and absence of GFA (10 mg/kg). Time-courseplasma drug concentration data of O-demethylation product of DM arepresented in Fig. 9, and the pharmacokinetic parameters are presentedin Table 5. The pharmacokinetic profile of the O-demethylationmetabolite (DXO) differed significantly between the GFA and placebo(saline) groups. After treatment with GFA, CYP2D activity wasinhibited and the metabolite formed by CYP2D decreased, with Cmax

being one-third that of the saline group and the AUC being half that ofthe saline group (Fig. 9; Table 5).CYP2D Docking of GFA and Quinidine. To illustrate the

mechanism underlying the observed species difference in inhibitionby GFA and quinidine, molecular docking studies for homologymodels of CYP2D were further performed to confirm the inhibitionmodes for each P450 isoform. As can be seen from the dockingresults, GFA and quinidine are well predicted to be located in thehydrophobic area of sitemap in CYP2D6, with the active-site residuesPhe120, Glu216, Arg221, ASP301, and Phe483 (Fig. 10, A and B). InCYP2D2, GFA and quinidine are located outside the sitemap of thehydrophobic area (Fig. 10, C and D). This partly showed the differentbinding style between human and rat CYP2D. However, it cannotdistinguish the difference between GFA and quinidine. Furtherdocking with CYP2D6, 2D1, and 2D2 showed that quinidine locatedin the same area in CYP2D1, 2D2, and 2D6, but GFA located inCYP2D6 quite differently from CYP2D1 and 2D2 (Fig. 10, E and F).Further studies will be carried out to identify the species difference ininhibition ability of GFA and quinidine on CYP2D.Prediction of GFA Effects on CYP2D6-Dependent DDI in

Humans. The concentration of GFA in dogs and humans at 10 mg/kgwas simulated based on a previous study, and the parameters used arelisted in Table 3. Compared with the observed data, Cmax is generallypredicted quite accurately, but the AUC is overestimated sincemetabolism of GFA is ignored. According to the prediction eq. 5 byIto (2005), the AUC ratio of the inhibited and controlled uninhibitedgroup was predicted by fmCYP, [I], and Ki, and the results are listed inTable 6. This equation only gives an average estimation of the parentsubstrate and gives no information on metabolite. With the help ofSimcyp, both parent substrate (DM) and metabolite (DXO) pharma-cokinetic profiles can be predicted after the interaction of inhibitor(Table 6). Simulation results showed that the parent substrate is notinfluenced much by GFA, but the AUC of metabolite DXO is reducedalmost 2 times, which is validated by experimental data in dogs (theexperimental data of AUCi/AUC in dogs is about 0.99 for DM and0.52 for DXO).

Discussion

GFA is a novel antiarrhythmia drug approved for human use inChina. It is currently under a phase IV postapproval clinical trial forsafety evaluation. GFA is not extensively metabolized by liver micro-somes in vitro (Supplemental Fig. 3). The GFA sulfate and glucuronideare major metabolites found in vivo (A et al., 2002). Previous reported

Fig. 9. The inhibition effects of GFA on the metabolism of DM in beagle dogs.Mean 6 S.D. concentrations of DXO after DM (2.0 mg/kg) was administeredintravenously to six healthy adult beagle dogs pretreated with GFA (10 mg/kg,closed diamonds) or saline (open squares).

TABLE 5

Noncompartmental pharmacokinetic parameters for DM and DXO in beagle dogplasma after a single i.v. dose of DM (n = 6), pretreated with GFA or sodium saline

Data are expressed as means 6 S.D.

Pharmacokinetic Parameters Saline Group GFA Group

DMCmax (ng/ml) 291.2 6 66.4 302.0 6 126.4t1/2 (h) 2.08 6 0.20 2.16 6 0.27MRT (h) 2.27 6 0.24 2.50 6 0.49AUC0-t (ng×h/ml) 548.7 6 128.8 541.7 6 64.7AUC0-‘ (ng×h/ml) 556.8 6 132.5 551.4 6 66.7

DXOCmax (pg/ml) 2500.6 6 1585.5 862.5 6 569.5*Tmax (h) 0.9 6 0.7 1.1 6 0.7t1/2 (h) 7.67 6 2.66 8.36 6 2.78MRT (h) 7.31 6 2.58 9.47 6 1.77AUC0-t (pg×h/ml) 9093.5 6 3091.3 4609.4 6 2388.3*AUC0-‘ (pg×h/ml) 9433.8 6 3149.1 4885.6 6 2730.0*

AUC0-t, AUC from time 0 to 36 h; AUC0-‘, AUC from time 0 to infinite.*P , 0.05

Guanfu Base A is a CYP2D6 Inhibitor 721

at ASPE

T Journals on A

pril 7, 2018dm

d.aspetjournals.orgD

ownloaded from

Fig. 10. Molecular modeling of GFA and CYP2D. (A) The docking model of GFA and CYP2D6, the key residues, and hydrophobic area of the sitemap (yellow grid) areshown. (B) The docking model of quinidine and CYP2D6. (C) The docking model of GFA and CYP2D2. (D) The docking model of quinidine and CYP2D2. (E) Thedocking model of GFA and CYP2D1, 2, and 6 (light blue for CYP2D6 and green for CYP2D1 and 2). (F) The docking model of quinidine and CYP2D1,2, and 6 (light bluefor CYP2D6 and green for CYP2D1 and 2).

722 Sun et al.

at ASPE

T Journals on A

pril 7, 2018dm

d.aspetjournals.orgD

ownloaded from

and the data presented confirmed that P450 is not the major enzyme thatregulates GFA metabolism.In this report, we have evaluated how GFA interacts with P450

enzymes. Our study was the first to find that GFA is a potent CYP2D6-specific noncompetitive inhibitor, with Ki recorded at 1.20 6 0.33 and0.37 6 0.16 mM in HLMs and rCYP2D6, respectively. GFA isa competitive inhibitor of CYP2D6 homolog in dog and monkey livermicrosomes (with Ki = 2.4 6 1.3 and 0.38 6 0.12 mM, respectively),but has no inhibitory effect on mouse and rat orthologs. However,another CYP2D6 inhibitor, quinidine, was shown to be able to inhibitrat CYP2D as demonstrated by a 2-fold increase of DM in rats (Marieret al., 2004). The mechanism underlying the observed species differencein inhibition by GFA and quinidine was further investigated byhomology modeling and molecular docking. According to the study byMcLaughlin et al. (2005), at least 100-fold decreased quinidineinhibition of bufuralol 1-hydroxylation and dextromethorphan O-demethylation was found after abolition of the negative charge ateither or both residues Glu216 and Asp301. Our docking results showeda different binding style between human and rat CYP2D. GFA alsoshowed a different position in CYP2D6 from that of CYP2D1 and 2D2,but quinidine showed a similar position in CYP2D6, 2D1, and 2D2.The docking results suggest that the binding style of quinidine in ratsmight be different from that of GFA, which partly illustrated thedifferent inhibition mechanism of GFA and quinidine on CYP2D.The effect of 2D6 inhibition by GFA was further evaluated in

beagle dogs with DM as a probe drug for drug-drug interaction. ACYP2D6 probe, DM can be metabolized into the O-demethylationmetabolite (DXO) as an indicator of enzyme activity (Barnhart, 1980;

Shou et al., 2003). Dosing of GFA 30 minutes prior to DM did notinfluence the AUC of the parent drug DM. However, the concentrationof the active metabolite DXO decreased significantly. A two-thirdreduction in Cmax and 50% AUC reduction were attributed to GFAinhibition of CYP2D. The inhibition effect could lead to a clinicalimpact for some drugs with a narrow therapeutic index.Since patients with heart disease commonly also suffer from other

syndromes such as hypertension and dysthymia, a combination of drugs isoften prescribed to manage the disease and symptoms. In such cases,DDIs can be a significant problem when the patient takes GFA with drugsthat are substrates of CYP2D6. As CYP2D6 is a polymorphic enzyme,additional precaution should be taken to avoid drug-drug interactions.Clinical DDIs between CYP2D6 substrate and inhibitors have beenreported previously. A significant inhibition in the O-demethylation ofcodeine to morphine in homozygous extensive metabolizer of CYP2D6treated with low-dose levomepromazine has been documented (Vevelstadet al., 2009). A reported clinical DDI between flecainide and paroxetinehas led to the suggestion of monitoring plasma flecainide concentrationsfor those who are on CYP2D6 inhibitors, such as paroxetine, quinidine,and possibly GFA (Tsao and Gugger, 2009).Prior to our report, there was no drug-drug interaction information of

GFA on substrate of CYP2D6 in humans. Although it is clear that GFAis a CYP2D6 inhibitor, it is important to understand whether, at a clinicaldose, it may significantly impact CYP2D6 enzyme activity. In healthyvolunteers at an i.v. dose of 4 mg/kg, the concentration at 5 mins C5min

was 6.71 6 1.51 mg/ml (15.6 mM). Because the protein binding ratio ofGFA is 78.6–84.6% in human plasma (unpublished data), in light of theKi of GFA on CYP2D6, the possibility of DDIs is quite high accordingto the predictive functions [I]/Ki, where [I]/Ki . 1 suggests a high DDIrisk (Ito et al., 2004; Obach et al., 2006). It is likely that, at clinical dose,GFA may have a significant impact on CYP2D6 substrate.To provide a tool to predict DDI, we evaluated two prediction

simulations based on eq. 5 and a dynamic computer-based predictionof DDI in humans by the Simcyp model. Both models assumed that anincrease in the AUC of a substrate in the presence of an inhibitor of thesubstrate’s elimination pathway is a function of the ratio of the inhibitorconcentration ([I]) to the inhibition constant (Ki) (Shou, 2005; Obachet al., 2006; Einolf, 2007). Equation 5 only gives an average estimationof the parent substrate and gives no information on metabolite. With theSimcyp computer simulation, both parent substrate (DM) and metabolite(DXO) pharmacokinetic profiles can be predicted with inhibitor.According to the dynamic simulation by Simcyp and in vivo data indogs, there was no apparent change in the parent drug (DM), but

TABLE 6

In vivo pharmacokinetic profile and AUCi/AUC predictions using GFA as theinhibitor in human

AUCi/AUCa

GFAExperimental Data Simcyp Prediction Equation 5 Prediction

DMb DXOb DM DXO DM DXO

4 mg/kg ND ND 1.063 (1.028–1.096) 0.583 5.31 NA10 mg/kg 0.99 0.518 1.061 (1.031–1.090) 0.407 9.48 NA

NA, not available; ND, not determined.aThe Simcyp and eq. 5 estimated ratios of AUC in the absence (AUC) or presence of GFA

inhibitor (AUCi) at the indicated GFA dose (milligrams per kilogram) were presented incomparison with the experimental data collected in beagle dogs.

bExperimental data from beagle dogs.

Fig. 11. Time-dependent inactivation of CYP2D6 andCYP1A2 in HLMs. HLMs were preincubated with buffer(negative control) or a NADPH-generating system andaliquots (50 ml) were removed at 5, 10, 20, and 30minutes after the initiation of the preincubation reactionand were added to reaction tubes containing DM (finalconcentration 5 mM) and the NADPH-generating system.The reactions (200 ml total volume) were incubatedfor another 15 minutes. Each point represents the mean6 S.D. of three measurements. PN, Phenacetin.

Guanfu Base A is a CYP2D6 Inhibitor 723

at ASPE

T Journals on A

pril 7, 2018dm

d.aspetjournals.orgD

ownloaded from

a significant reduction in the AUC of the metabolite (DXO) comparedwith that of the control (AUCi/AUC = 0.583 and 0.407 in humans at 4and 10 mg/kg, respectively; Table 6). Therefore, in some cases, thechange in the concentration of the specific metabolite might be a moreaccurate and sensitive index for evaluating DDI. As the metabolite fromCYP2D6 is decreased after coadministration with GFA, activity of drugswhich need to be activated by CYP2D6 will be compromised.CYP2D6 has been shown to be unstable in the presence of cofactor

NADPH, detectable at a 50% activity reduction after 30 minutes(Bertelsen et al., 2003; Madeira et al., 2004), which was confirmed inour experiment. We found that GFA reversed the NADPH-dependentCYP2D inactivity. GFA reduced NADPH-dependent enzyme in-activity as much as 80% after 30-minute preincubation. Cimetidinehas been reported to have the same effect on CYP2D6 activity(Madeira et al., 2004). This protective effect may be due to the drug’santioxidant properties (unpublished data). In general, it has no impacton CYP1A2 (Fig. 11).In conclusion, this study has demonstrated that GFA is a specific

inhibitor of CYP2D6, and that this inhibition is noncompetitive andnot mechanism-based. It has therefore added to the growing list ofspecific CYP2D6 inhibitors that have been documented. Based on therecommended dose of GFA and its plasma drug concentration, it islikely to induce drug-drug interaction for those that are CYP2D6substrates.

Acknowledgments

The authors thank Jake Kraft for help with English correction. Help with theprotein docking analysis by Dr. Yanming Zhang, Yadong Chen, and Dr.Haichun Liu is greatly appreciated.

Authorship ContributionsParticipated in research design: J. Sun, Peng, Wang.Conducted experiments: J. Sun, Peng, Wu, X. Zhang, Zhong, Xiao,

F. Zhang, Qi, Shang, Zhu, Y. Sun, K. Liu.Contributed new reagents or analytic tools: J. Liu.Performed data analysis: J. Sun, Peng, A, Ho.Wrote or contributed to the writing of the manuscript: J. Sun, X. Zhang, Ho.

References

A JY, Wang GJ, Liu XQ, Jiang DY, and Liu JH (2002) Study on the metabolites of guanfu base Ahydrochloride in rat urine by high performance liquid chromatograph-mass spectrum. Yao XueXue Bao 37:283–287.

A JY, Wang GJ, Sun JG, Gu YC, Wu MS, and Liu JH (2003) Identification of phase I and phaseII metabolites of Guanfu base A hydrochloride in human urine. Eur J Drug Metab Pharma-cokinet 28:265–272.

Ai C, Li Y, Wang Y, Chen Y, and Yang L (2009) Insight into the effects of chiral isomersquinidine and quinine on CYP2D6 inhibition. Bioorg Med Chem Lett 19:803–806.

Arnold K, Bordoli L, Kopp J, and Schwede T (2006) The SWISS-MODEL workspace: a web-based environment for protein structure homology modelling. Bioinformatics 22:195–201.

Barnhart JW (1980) The urinary excretion of dextromethorphan and three metabolites in dogs andhumans. Toxicol Appl Pharmacol 55:43–48.

Bertelsen KM, Venkatakrishnan K, Von Moltke LL, Obach RS, and Greenblatt DJ (2003) Ap-parent mechanism-based inhibition of human CYP2D6 in vitro by paroxetine: comparison withfluoxetine and quinidine. Drug Metab Dispos 31:289–293.

Bertilsson L, Dengler HJ, Eichelbaum M, and Schulz HU (1980) Pharmacogenetic covariation ofdefective N-oxidation of sparteine and 4-hydroxylation of debrisoquine. Eur J Clin Pharmacol17:153–155.

Bramer SL and Suri A (1999) Inhibition of CYP2D6 by quinidine and its effects on the me-tabolism of cilostazol. Clin Pharmacokinet 37 (Suppl 2):41–51.

Ching MS, Blake CL, Ghabrial H, Ellis SW, Lennard MS, Tucker GT, and Smallwood RA (1995)Potent inhibition of yeast-expressed CYP2D6 by dihydroquinidine, quinidine, and its metab-olites. Biochem Pharmacol 50:833–837.

de la Torre R, Ortuño J, Mas M, Farré M, and Segura J (1999) Fatal MDMA intoxication. Lancet353:593.

Einolf HJ (2007) Comparison of different approaches to predict metabolic drug-drug interactions.Xenobiotica 37:1257–1294.

Gareri P, De Fazio P, Gallelli L, De Fazio S, Davoli A, Seminara G, Cotroneo A, and De Sarro G(2008) Venlafaxine-propafenone interaction resulting in hallucinations and psychomotor agi-tation. Ann Pharmacother 42:434–438.

Guex N and Peitsch MC (1997) SWISS-MODEL and the Swiss-PdbViewer: an environment forcomparative protein modeling. Electrophoresis 18:2714–2723.

Hiroi T, Imaoka S, Chow T, and Funae Y (1998) Tissue distributions of CYP2D1, 2D2, 2D3 and2D4 mRNA in rats detected by RT-PCR. Biochim Biophys Acta 1380:305–312.

Ito K, Brown HS, and Houston JB (2004) Database analyses for the prediction of in vivo drug-drug interactions from in vitro data. Br J Clin Pharmacol 57:473–486.

Ito K, Hallifax D, Obach RS, and Houston JB (2005) Impact of parallel pathways of drugelimination and multiple cytochrome P450 involvement on drug-drug interactions: CYP2D6paradigm. Drug Metab Dispos 33:837–844.

Jägestedt M and von Bahr C (2004) [Combination of serotonergic agents resulted in severeadverse effects]. Lakartidningen 101:1618–1619.

Kobayashi S, Murray S, Watson D, Sesardic D, Davies DS, and Boobis AR (1989) The speci-ficity of inhibition of debrisoquine 4-hydroxylase activity by quinidine and quinine in the rat isthe inverse of that in man. Biochem Pharmacol 38:2795–2799.

Komori M (1993) A novel P450 expressed at the high level in rat brain. Biochem Biophys ResCommun 196:721–728.

Kroemer HK, Fischer C, Meese CO, and Eichelbaum M (1991) Enantiomer/enantiomer in-teraction of (S)- and (R)-propafenone for cytochrome P450IID6-catalyzed 5-hydroxylation: invitro evaluation of the mechanism. Mol Pharmacol 40:135–142.

Liu JHWH and Gao YL (1981) Studies on aconitum SPP-XVI. new alkaloide from guanbaifuge.Clin Trad Herb Drugs 12:1.

Madeira M, Levine M, Chang TK, Mirfazaelian A, and Bellward GD (2004) The effect ofcimetidine on dextromethorphan O-demethylase activity of human liver microsomes andrecombinant CYP2D6. Drug Metab Dispos 32:460–467.

Mahgoub A, Idle JR, Dring LG, Lancaster R, and Smith RL (1977) Polymorphic hydroxylation ofDebrisoquine in man. Lancet 2:584–586.

Marier JF, Pope LE, Yakatan GJ, Berg JE, Stiles M, and Vachon P (2004) Influence of con-comitant quinidine administration on dextromethorphan disposition in rats. J Vet PharmacolTher 27:111–114.

McGinnity DF, Waters NJ, Tucker J, and Riley RJ (2008) Integrated in vitro analysis for the invivo prediction of cytochrome P450-mediated drug-drug interactions. Drug Metab Dispos 36:1126–1134.

McLaughlin LA, Paine MJ, Kemp CA, Maréchal JD, Flanagan JU, Ward CJ, Sutcliffe MJ,Roberts GC, and Wolf CR (2005) Why is quinidine an inhibitor of cytochrome P450 2D6? Therole of key active-site residues in quinidine binding. J Biol Chem 280:38617–38624.

Nakashima D, Takama H, Ogasawara Y, Kawakami T, Nishitoba T, Hoshi S, Uchida E, and TanakaH (2007) Effect of cinacalcet hydrochloride, a new calcimimetic agent, on the pharmacokineticsof dextromethorphan: in vitro and clinical studies. J Clin Pharmacol 47:1311–1319.

Obach RS, Walsky RL, Venkatakrishnan K, Gaman EA, Houston JB, and Tremaine LM (2006)The utility of in vitro cytochrome P450 inhibition data in the prediction of drug-drug inter-actions. J Pharmacol Exp Ther 316:336–348.

Schmid B, Bircher J, Preisig R, and Küpfer A (1985) Polymorphic dextromethorphan metabo-lism: co-segregation of oxidative O-demethylation with debrisoquin hydroxylation. ClinPharmacol Ther 38:618–624.

Schwede T, Kopp J, Guex N, and Peitsch MC (2003) SWISS-MODEL: An automated proteinhomology-modeling server. Nucleic Acids Res 31:3381–3385.

Shardlow CE, Generaux GT, Patel AH, Tai G, Tran T, and Bloomer JC (2013) Impact ofphysiologically based pharmacokinetic modeling and simulation in drug development. DrugMetab Dispos 41:1994–2003.

Shou M (2005) Prediction of pharmacokinetics and drug-drug interactions from in vitro metab-olism data. Curr Opin Drug Discov Devel 8:66–77.

Shou M, Norcross R, Sandig G, Lu P, Li Y, Lin Y, Mei Q, Rodrigues AD, and Rushmore TH(2003) Substrate specificity and kinetic properties of seven heterologously expressed dogcytochromes p450. Drug Metab Dispos 31:1161–1169.

Sun JG, Xiao YN, Peng Y, Zhang XY, Wu H, Liu JH, Wu MS, AJ, Wang GJ (2012) Pharma-cokinetic and absolute bioavailability study of Guanfu Base A hydrochloride in Beagle dogs.Chin J Clin Pharmacol Ther 17:1378–1383.

Tsao YY and Gugger JJ (2009) Delirium in a patient with toxic flecainide plasma concentrations: therole of a pharmacokinetic drug interaction with paroxetine. Ann Pharmacother 43:1366–1369.

VandenBrink BM, Foti RS, Rock DA, Wienkers LC, and Wahlstrom JL (2012) Prediction ofCYP2D6 drug interactions from in vitro data: evidence for substrate-dependent inhibition.Drug Metab Dispos 40:47–53.

Vevelstad M, Pettersen S, Tallaksen C, and Brørs O (2009) O-demethylation of codeine tomorphine inhibited by low-dose levomepromazine. Eur J Clin Pharmacol 65:795–801.

Wang GG, Cui HD, Zhu J, Zhang Y, Tan HQ, Yang XJ, Qiu NY, and Wang GJ (2000)Evaluation of Pharmacokinetics of the Hydrochloride Guanfu Base A. Chin J Clin Pharmacol16:283–285.

Webb J (1963) General Principles of Inhibition, (Enzyme and Metabolic Inhibitors.), Vol. 1.Academic Press, New York.

Wyss A, Gustafsson JA, and Warner M (1995) Cytochromes P450 of the 2D subfamily in ratbrain. Mol Pharmacol 47:1148–1155.

Xu L, Chen Y, Pan Y, Skiles GL, and Shou M (2009) Prediction of human drug-drug interactionsfrom time-dependent inactivation of CYP3A4 in primary hepatocytes using a population-basedsimulator. Drug Metab Dispos 37:2330–2339.

Yang X, Wang G, Ling S, Qiu N, Wang G, Zhu J, and Liu J (2000) Determination of Guan-FuBase A, a new anti-arrhythmic, in human plasma by gas chromatography and electron-capturedetection. J Chromatogr B Biomed Sci Appl 740:273–279.

Yang YM, Zhu J, Gao X, Yin YL, Hao YX, and Liu JH; Guanfu Base A Hydrochloride InjectionPhase II Clinic Trial Group Investigators (2006) [Effect of Guanfu Base A in patients withventricular arrhythmias]. Zhonghua Xin Xue Guan Bing Za Zhi 34:329–332.

Yeh GC, Tao PL, Ho HO, Lee YJ, Chen JY, and Sheu MT (2003) Analysis of pharmacokineticparameters for assessment of dextromethorphan metabolic phenotypes. J Biomed Sci 10:552–564.

Youdim KA, Zayed A, Dickins M, Phipps A, Griffiths M, Darekar A, Hyland R, Fahmi O, HurstS, and Plowchalk DR, et al. (2008) Application of CYP3A4 in vitro data to predict clinicaldrug-drug interactions; predictions of compounds as objects of interaction. Br J Clin Phar-macol 65:680–692.

Address correspondence to: Guangji Wang, Key Laboratory of Drug Metabolismand Pharmacokinetics, State Key Laboratory of Natural Medicines, ChinaPharmaceutical University, 24 Tong Jia Xiang, Nanjing, Jiangsu, 210009, China.E-mail: guangjiwang@hotmail.com

724 Sun et al.

at ASPE

T Journals on A

pril 7, 2018dm

d.aspetjournals.orgD

ownloaded from