Toxicology and Applied Pharmacology 261 (2012) 50–58

Contents lists available at SciVerse ScienceDirect

Toxicology and Applied Pharmacology

j ourna l homepage: www.e lsev ie r .com/ locate /ytaap

Metabolism of bilirubin by human cytochrome P450 2A6

A'edah Abu-Bakar a,⁎, Dionne M. Arthur a,b, Anna S. Wikman a,c, Minna Rahnasto d, Risto O. Juvonen d,Jouko Vepsäläinen d, Hannu Raunio d, Jack C. Ng a,b, Matti A. Lang a

a The University of Queensland, National Research Centre for Environmental Toxicology (Entox), 4072 Brisbane, Queensland, Australiab Cooperative Research Centre for Contamination Assessment & Remediation of the Environment, Adelaide, Australiac Department of Pharmaceutical Biosciences, Uppsala University, SE-75123 Uppsala, Swedend School of Pharmacy, Faculty of Health Sciences, University of Eastern Finland, POB 1627, 70211 Kuopio, Finland

Abbreviations: BR, bilirubin; BV, biliverdin; ROS, reaUDP-glucuronosyltransferase 1A1; CYP, cytochrome P42A5; CYP2A6, cytochrome P450 2A6; HMOX1, haem oxerythroid 2-like 2.⁎ Corresponding author at: Entox, The University of

Coopers Plains, 4108 Queensland, Australia. Fax: +61 7E-mail address: a.abubakar@uq.edu.au (A. Abu-Baka

0041-008X/$ – see front matter. Crown Copyright © 20doi:10.1016/j.taap.2012.03.010

a b s t r a c t

a r t i c l e i n f o

Article history:Received 1 February 2012Revised 13 March 2012Accepted 16 March 2012Available online 24 March 2012

Keywords:CYP2A6Bilirubin oxidationHPLC/ESI-MSBilirubin oxidaseMolecular dockingOxidative stress

The mouse cytochrome P450 (CYP) 2A5 has recently been shown to function as hepatic “Bilirubin Oxidase”(Abu-Bakar, A., et al., 2011. Toxicol. Appl. Pharmacol. 257, 14–22). To date, no information is available onhuman CYP isoforms involvement in bilirubin metabolism. In this paper we provide novel evidence for humanCYP2A6 metabolising the tetrapyrrole bilirubin. Incubation of bilirubin with recombinant yeast microsomesexpressing the CYP2A6 showed that bilirubin inhibited CYP2A6-dependent coumarin 7-hydroxylase activity toalmost 100% with an estimated Ki of 2.23 μM. Metabolite screening by a high-performance liquid chromatogra-phy/electrospray ionisation mass spectrometry indicated that CYP2A6 oxidised bilirubin to biliverdin and tothree other smaller products withm/z values of 301, 315 and 333. Molecular docking analyses indicated that bil-irubin and its positively charged intermediate interacted with key amino acid residues at the enzyme's activesite. They were stabilised at the site in a conformation favouring biliverdin formation. By contrast, the end prod-uct, biliverdin was less fitting to the active site with the critical central methylene bridge distanced from theCYP2A6 haem iron facilitating its release. Furthermore, bilirubin treatment of HepG2 cells increased theCYP2A6 protein and activity levelswith no effect on the correspondingmRNA. Co-treatmentwith cycloheximide(CHX), a protein synthesis inhibitor, resulted in increased half-life of the CYP2A6 compared to cells treated onlywith CHX. Collectively, the observations indicate that the CYP2A6 may function as human “Bilirubin Oxidase”where bilirubin is potentially a substrate and a regulator of the enzyme.

Crown Copyright © 2012 Published by Elsevier Inc. All rights reserved.

Introduction

Bilirubin (BR), the lipophilic end product of haem breakdown is apotent antioxidant that may serve to protect cells against oxidativefree radicals (Stocker, 2004; Stocker and Ames, 1987). As BR is cyto-toxic at intracellular concentrations above 20 μM (Cashore, 1990;Keshavan et al., 2004; Menken et al., 1966; Noir et al., 1972) and aneffective antioxidant at concentrations ranging from ~0.01 to 10 μM(Dore et al., 1999; Jansen et al., 2010; Stocker and Ames, 1987) themaximum benefit from BR can only be achieved with strict regulationof its cellular levels. It is well established that BR is cleared from thebody by conjugation with glucuronic acid and elimination in thebile (Crawford et al., 1992). In humans, BR glucuronidation is cata-lysed by hepatic UDP-glucuronosyltransferase 1A1 (UGT1A1)

ctive oxygen species; UGT1A1,50; CYP2A5, cytochrome P450ygenase-1; Nrf2, nuclear factor

Queensland, 39 Kessels Road,3274 9003.r).

12 Published by Elsevier Inc. All rig

(Tukey and Strassburg, 2000). However, when BR glucuronidation ca-pacity is reduced, as in neonatal jaundice and in hereditary forms ofcongenital jaundice, cytochrome P450 (CYP) driven BR oxidationhas been suggested to contribute significantly to the elimination ofBR (Schmid and Hammaker, 1963).

The proposition that CYP enzymes are involved in BR metabolismis based on observations that inducers of CYP1A1 and CYP1A2may re-duce blood plasma BR and increase biliary excretion of hydroxylatedBR metabolites in the Gunn rats (Cohen et al., 1986; Kapitulnik andOstrow, 1978). It has also been shown that BR oxidation takes placewhen the CYP enzymes are uncoupled from the electron transferchain by polyhalogenated substrate analogues (De Matteis et al.,1991, 1993, 2002, 2006). These conditions create a futile cyclewhere a large proportion of reducing equivalents from the cofactorNADPH is released as radical oxygen species (ROS). In turn, ROS inter-act with BR to produce BR oxidative metabolites (BOMs) (De Matteiset al., 1991, 2006). These studies however, have not demonstrated BRas a substrate of a CYP enzyme interacting with its catalytic site.

By contrast, we have reported that BR is a high affinity substrateand a regulator of the mouse CYP2A5 (Abu-Bakar et al., 2011). Wefurther demonstrated that CYP2A5 is a major catalyst of BR degrada-tion when cellular BR levels are elevated due to induction of haem

hts reserved.

51A. Abu-Bakar et al. / Toxicology and Applied Pharmacology 261 (2012) 50–58

oxygenase-1 (HMOX1) (Abu-Bakar et al., 2005). Until now, little isknown about the role of human CYP enzymes in BR oxidation. Inthis paper, we present evidence indicating that the human orthologueof CYP2A5, the CYP2A6, is potentially a “BR Oxidase”. By using recom-binant CYP2A6 we show that BR interacts with the protein with ahigh affinity and is metabolised to biliverdin (BV) and to threeminor products. Molecular docking analyses of BR with the crystalstructure of CYP2A6 (2FDW) (Yano et al., 2006) supported the exper-imental findings. Additionally, BR up-regulates the CYP2A6 proteinand activity through protein stabilisation in HepG2 cells. Collectively,the observations indicate that BR is potentially a substrate and a reg-ulator of CYP2A6.

Materials and methods

Chemicals and reagents. β-NADPH, EDTA, glucose, dimethyl sulfox-ide (DMSO), sorbitol, histidine, cycloheximide, dithiothreitol (DTT),phenylmethylsulfonyl fluoride (PMSF), and coumarin were obtainedfrom Sigma-Aldrich (Sydney, Australia). Bilirubin and biliverdinwere sourced from Frontier Scientific, Inc. (Utah, USA). Acetonitrile(HPLC grade) and ammonium acetate were from MERCK (Australia)and UNIVAR (USA), respectively. Zymolyase 100T was purchasedfrom Seikagaku Corporation (Japan).

Isolation of yeast microsomes. Recombinant Saccharomyces cerevisiaeAH22 expressing human CYP2A6 (provided by IARC, Lyon, France),were cultured and microsomes extracted in accordance with publishedmethod (Yabusaki, 1998) with some modifications. Briefly, yeast cellswere cultured in yeast nitrogen base solution containing 0.2 g/ml glu-cose and 2 mg/ml histidine until OD600 of 1.3–1.6 was reached. Cellsfrom 4×1 L cultures were collected by centrifugation at 2000 ×g for15 min at 4 °C. The yeast cells were mixed with Zymolyase 100T at aconcentration of 43 mg Zymolyase per 80 ml Zymolyase buffer [10 mMTris–HCl (pH 7.5), 2 M sorbitol, 0.1 mM EDTA, and 0.1 mM DTT]. Themixture was incubated at 30 °C for 90 min and centrifuged at 2000 ×gfor 12 min at 4 °C. The pellet was rinsed twice with 60 ml Zymolyasebuffer and centrifuged at 2000 ×g for 12 min at 4 °C. The seroplasts(yeast particles obtained after Zymolyase digestion) were sonicated in60 ml of sonication buffer [10 mM Tris–HCl (pH 7.5), 0.65 M sorbitol,0.1 mM EDTA, 0.1 mM DTT, and 1 mM PMSF] for 5×1min at 100W.The mixture was then centrifuged at 10000 ×g for 20 min at 4 °C. Thesupernatant was further centrifuged at 100000 ×g for 60 min at 4 °C.The microsomal pellet was resuspended in 3 ml of 0.1 M potassiumphosphate buffer (pH 7.4) containing 20% glycerol and 1 mM DTT andstored at −80 °C until used.

Culture and treatment of HepG2 cells. HepG2 hepatocellular carci-noma cells were purchased from the American Type Culture Collec-tion (Manassas, VA). The cells were propagated in 6 cm plates(Corning, Palo Alto, CA) in minimal essential medium (MEM) con-taining 10% foetal calf serum, 2 mM L-glutamine, 1 mM sodium pyru-vate, MEM nonessential amino acids, and 10 μg/ml gentamicin. Thecells were kept in an atmosphere of 5% CO2 at 37 °C in a humidifiedincubator and were subcultured at least two times per week. All cellculture products were purchased from Invitrogen (Carlsbad, CA).

Cells in culture (5×105 cells/6 cm plate) were treated with 5 μMBR (dissolved in DMSO) for 1, 3, 9, or 12 h (N=6 per group). Controlcells were treated with DMSO only (final DMSO concentration lessthan 0.5%). To investigate the effects of BR on the half-life ofCYP2A6 protein, cells were treated with 400 μM cycloheximide(CHX) or with 5 μM BR+400 μM CHX for 0, 1, 2 or 4 h (N=6 pergroup).

Isolation of S9 fraction and total RNA. S9 fraction from the controland treated HepG2 cells were prepared as described previously(Geneste et al., 1996). The cells were washed and resuspended in

phosphate-buffered saline. The cell suspension was centrifuged at2000 ×g for 30 s. The resulting pellet was resuspended in buffer A[10 mM HEPES–KOH (pH 7.9), 10 mM KCl, 1.5 mM MgCl2, 0.5 mMDTT, 0.2 mM PMSF, 1 g/ml leupeptin, and 0.4% Igepal] and kept onice for 1 h. The cell suspension was vortexed, homogenised, and cen-trifuged at 12000 ×g at 4 °C for 10 min. The supernatant containingS9 proteins was aliquoted and stored at −80 °C. Total RNA wasextracted from the control and treated cells using the RNeasy MiniKit (QIAGEN GmbH, Hilden, Germany). The S9 protein and RNA con-centrations were determined spectrometrically.

Protein and mRNA analysis. The CYP2A6 protein levels were deter-mined by Western Immunoblotting. Microsomal or S9 proteins wereseparated by NuPAGE 4–12% w/v Bis–Tris 1.0 mm mini-gel (Invi-trogen, Australia) and blotted onto a PVDF membrane (Pierce Bio-technology, USA) and blocked for 4 h in phosphate-buffered salinecontaining 0.1% Tween 20 (TBS-T) and 5% non-fat milk. Blots were in-cubated with a mouse monoclonal anti-CYP2A6 antibody (Abcam,Cambridge, UK) in a 1:500 dilution for 2 h. The membranes werethen incubated with HRP-conjugated goat anti-mouse IgG (Sigma,Australia) in a 1:10000 dilution for 2 h. After further washing withTBS-T, blots were incubated in chemiluminescence reagents (Amer-sham ECL Prime Western blotting agent, GE Healthcare Ltd, UK).The luminescent signal was detected with Versadoc4000 gel Imager(BioRad Pte Ltd, CA, USA). After detection, the membranes werestripped in stripping buffer (1 M Tris pH 6.7, 10% SDS, β-MeOH) andincubated with a polyclonal β-actin antibody (Abcam, Cambridge,UK) (1:5000 dilution) to assess equal loading of the samples. The in-tensity of the bands was analysed with Image LabTM (BioRad Pte Ltd,CA, USA) software.

The catalytic activity of the recombinant CYP2A6was determined bymeasuring the rate of coumarin 7-hydroxylation (COH) with 10 μMfinal concentration of coumarin spectrofluorometrically (Abu-Bakar etal., 2004). For the enzyme kinetic study, the final concentration of cou-marin used was 0.5–10 μM and the final BR concentration range usedwas 1.25–100 μM. In all experiments 100 μl of microsomal protein(0.2 mg/ml), corresponding to a concentration of cytochrome P450ranging from 2 to 5 pmol/ml was used.

Messenger RNA levels were determined by quantitative real timeRT-PCR (iQ SYBR Green Supermix, BioRad, CA, USA). One microgramof total RNA from the control and treated HepG2 cells was used tosynthesise first-strand cDNA with the BluePrint™ 1st strand cDNAsynthesis kit (Takara Bio Inc., Japan). Two microlitres of cDNA solu-tion was used as template in 20 μl PCR reactions containing 10 μl ofthe 2× master mix and each PCR primer at 0.2 μM (final concentra-tion) (see Table 1 for primer's sequence). The samples were incubat-ed at 95 °C for 3 min, followed by 30 cycles of 95 °C for 20 s and67.6 °C for 1 min in Corbett RotorGene 3000 (QIAGEN GmbH, Hilden,Germany). The specificity of the PCR products was confirmed bymelting curve analysis and size (2.5% agarose gel electrophoresis).

Bilirubin disappearance activity and spectrometry analysis of oxidativemetabolites. The rate of microsomal bilirubin degradationwas deter-mined as described previously (Abu-Bakar et al., 2005). Briefly, incuba-tion mixture [0.1 M Tris–HCl (pH 8.2), 26 mM KCl, 2 mM EDTA, and100 μl of yeast microsomes (corresponding to a concentration ofcytochrome P450 ranging from 10 to 100 pmol/ml) in a total volumeof 2 ml] was placed in plastic disposable cuvettes. After 5 min pre-incubation, β-NADPH (1.5 mM) was added to both sample and refer-ence cuvettes and absorbance was zeroed against reference cuvetteat 440 nm (Cary 3E UV–Visible Spectrophotometer, Varian Australia,Pty. Ltd.). Immediately, BR in DMSO (final BR concentration=10 μM;DMSO final concentration=0.1%) was added to the sample cuvetteand absorbance at 440 nm was recorded for 20 cycles at 1 min/cycle.The rate of BR disappearance is expressed as pmol bilirubin dis-appearing/min/nmol P450, using a εmM=41.15 cm−1, which was

Table 1Primers used for the qualitative real-time RT-PCR.

Gene name GenBankaccession no.

Primer sequence (5′→3′) Ampliconsize (bp)

CYP2A6 NM_007812 F: TGGTCCTGTATTCACCATCTACCR: ACTACGCCATAGCCTTTGAAAA

150

GAPDH NM_008084 F: CATGTTCCAGTATGACTCCACTCR: GGCCTCACCCCATTTGATGT

136

52 A. Abu-Bakar et al. / Toxicology and Applied Pharmacology 261 (2012) 50–58

obtained experimentally under the conditions of the assay. All experi-ments were done in the dark and 2 mM (final concentration) of EDTAwas added in all incubation mixtures.

Metabolite screening was conducted as previously described(Abu-Bakar et al., 2011). Briefly, the reaction mixtures were incubat-ed for 1 h after which time samples were filtered using 0.45 μmmem-brane (Millipore) and immediately injected to a HPLC-MS/MS (AB/Sciex API4000Q mass spectrometer, USA) equipped with an electro-spray (TurboV) interface coupled to a HPLC system (ProminenceModel XYZ, Shimadzu Corp., Kyoto, Japan). In the absence of stan-dards for the putative dipyrroles, we used bilirubin as internal stan-dard to monitor extraction efficiency. We have achieved on average80% recovery. Separation was achieved using a 2.5 μm 50×2 mmSynergi MAX RP column (Phenomenex, Torrance, CA, USA) run at40 °C, and mobile phase flow rate of 0.5 ml/min with a linear gradientstarting at 5% B for 1 min, ramped to 80% B in 10 min, held for 1 min,to 5% B in 0.2 min and equilibrated for 3 min (mobile phase A=1%acetonitrile/HPLC grade water, mobile phase B=95% acetonitrile/HPLC grade water, both containing 5 mM ammonium acetate). Themass spectrometer was operated in the positive ion, multiple reac-tions monitoring mode using nitrogen as the collision gas under thefollowing conditions: declustering potential was 75 or 110 (biliverdinonly), and the collision energy was between 25 and 83 eV. To confirmmetabolite structure by mass, samples were run on an Agilent 6530Q-TOF with Agilent 1290 HPLC under the same HPLC conditionsused on the Shimadzu HPLC AB/Sciex 4000Q triple quadrupole massspectrometer.

Kinetics and statistical analysis. Enzyme kinetic parameters weredetermined by GraphPad Prism 5 kinetics programme from GraphPadSoftware Inc. (La Jolla, CA, USA). It uses a non-linear regression meth-od of curve fitting and the Runs test of residuals to determine statis-tically whether experimental data are randomly distributed aroundthe curve with 95% confidence. The Ki value was determined by plot-ting observed Km (Y-axis) vs bilirubin concentrations (X-axis). Fordata other than enzyme kinetic parameters Student's t-test wasused for comparisons between two groups. Differences were consid-ered significant when pb0.05.

Refinement of the CYP2A6 crystal structure and construction of BR andconjugated intermediate. The crystal structure of CYP2A6 (2FDW)(Yano et al., 2006) was obtained from the PDB database. Solvent mol-ecules were deleted from the structure and hydrogen atoms wereadded followed by visual inspection. Structure refinement wasobtained by minimization with the Powell Programme (Fletcher,1970; Rahnasto et al., 2008; Stahl and Höltje, 2005) and by simulationof molecular dynamics (MD) with the GROMACS 3.3.3. Programme(Lindahl et al., 2001). Construction of BR, BR-BV intermediate andBV was carried out using the SYBYL (Tripos Associates Inc., St. Louis,MO) modelling package. Atomic point charges were calculated usingthe MMFF94 method (Halgren, 1996).

Molecular docking. The binding features of BR and the positivelycharged BR-BV conjugated intermediate were analysed by predictingthe preferred orientation of the molecules in the active site of CYP2A6

protein, a method known as molecular docking. Genetic optimizationfor ligand docking (GOLD) (Jones et al., 1997; Verdonk et al., 2003) isa programme that evaluates solutions of conformers at the active siteusing scoring functions to estimate free energies (ΔG) during thebinding. The GOLD programme produced several possible dockingpositions, which were then ranked based on a scoring function(ChemScore). The radius for docking was set at 20 Å around the por-phyrin ring of the CYP2A6 haem. The results were analysed using theDiscovery Studio modelling package (Accerylys, San Diego, CA).

Results

BR an inhibitor of CYP2A6 catalysed coumarin 7-hydroxylation

As expected, the CYP2A6 protein was expressed in the recombi-nant yeast but not in thewild type yeast as shownbyWestern blot anal-ysis (Fig. 1A). The CYP2A6-dependent coumarin 7-hydroxylation wasdetected only in the recombinant yeast microsomes and the forma-tion of 7-hydroxycoumarin followed simple Michaelis–Menten ki-netics with a Km=0.64±0.06 μM for coumarin (Fig. 1B). The Km issimilar to that obtained with human liver microsomes and withcDNA-expressed CYP2A6, which ranged from 0.5 to 2.0 μM (Draperet al., 1997; Kinonen et al., 1995).

Bilirubin increased the Km but did not change the Vmax for theCYP2A6-catalysed coumarin 7-hydroxylation (Fig. 1B). The plotKm observed vs BR concentrations (Fig. 1C) yielded a linear graph,which confirmed that BR is a competitive inhibitor of the reaction.Bilirubin inhibited coumarin 7-hydroxylase activity to almost 100%with an IC50 of 8.58±0.18 μM (Fig. 1D) and estimated Ki=2.23 μM(Fig. 1C).

Metabolism of BR by the CYP2A6

The metabolic rate of BR with the recombinant CYP2A6 micro-somes is 60 times faster (69.12±5.9 pmol/min/nmol CYP) thanwith the wild type microsomes (1.12±0.07 pmol/min/nmol CYP).This suggests that in the system used, BR degradation is predomi-nantly driven by the CYP2A6 enzyme.

We have previously shown that chemical oxidation of BR by Fe-EDTA/H2O2 system generated ions with m/z values 301, 315, and333 (Abu-Bakar et al., 2011). They were purported to be dipyrrolesoriginated from cleavage of the central methylene bridge of BR(Abu-Bakar et al., 2011; De Matteis et al., 2006). After one hour incu-bation of BR with the recombinant CYP2A6 enzyme, these productsand biliverdin (BV) were formed (Fig. 2B). Negligible amounts of BVand products with m/z 301 were seen after incubation of BR withthe wild type yeast microsomes (Fig. 2C) and in the absence of micro-somes (Fig. 2D).

The relative signal intensity of the various ions shown in Fig. 2 wascalculated from the sum counts recorded for each group of peaks. Anaverage of 91% loss of BR was observed after incubation with the re-combinant CYP2A6 microsomes, with 39% appearing in BV, and 52%in the ion products. The percentages of the ion products formedwere ion m/z 333 (32%), ion m/z 301 (12%), and ion m/z 315 (8%)(Table 2). By contrast, only 20% of BR was lost in the incubationwith the wild type yeast microsomes and only 2% through auto-oxidation (Table 2). These observations confirmed that the CYP2A6is catalysing BR oxidation into the various metabolites.

An average of 98% loss in BR was observed in the chemical oxida-tion, with 19% of the counts appearing in BV, 32% in ion m/z 333, 26%in ion m/z 315, and 21% in ion m/z 301 (Table 2). Although enzymaticand chemical oxidations produce identical metabolites, the CYP2A6catalysed reaction gives proportionally more of BV, and the chemicaloxidation more of ionsm/z 301 and 315. This suggests that conditionsin the enzymatic reaction favour BV formation over the smallermetabolites.

Fig. 1. Effect of BR on 7-hydroxylation of coumarin by recombinant yeast expressingCYP2A6. (A) Western immunoblotting of microsomal fractions from wild type and re-combinant yeast expressing CYP2A6 using mouse monoclonal anti-CYP2A6 antibody.Lanes 1 and 3 contained 10 μg microsomal proteins, and lanes 2 and 4 contained 50 μgmicrosomal proteins. The band at 50 kDa corresponds to CYP2A6 protein. β-Actin pro-tein levels are shown as control for protein loading. For β-actin detection 5 μg of proteinwas loaded per well. WT denotes microsomes from the wild type yeast and + denotespositive control (pyrazole-induced mouse liver microsomes). (B) Recombinant CYP2A6microsomes were incubated with various concentrations of coumarin in the presence ofincreasing amounts of BR. The rates of coumarin 7-hydroxylation were determined byfluorescence spectroscopy. The formation of 7-hydroxycoumarin by CYP2A6 followedsimple Michaelis–Menten kinetics. BR increased the Km and did not change the Vmax

of coumarin 7-hydroxylation. (C) Km observed vs BR concentrations plot indicates thatBR is a competitive inhibitor with Ki=2.23 μM. (D) At substrate concentration of10 μM, BR inhibits formation of 7-hydroxycoumarin in a concentration dependent man-ner with an IC50 of 8.58±0.18 μM.

Fig. 2. HPLC/ESI-MS profile of bilirubin incubated with yeast microsomal fractions. (A)Incubation at 0 h in the presence or absence of yeast microsomal fractions. (B) After 1 hincubation in the presence of yeast microsomes expressing human CYP2A6. (C) After1 h incubation in the presence of wild type yeast microsomes. (D) After 1 h incubationin the absence of yeast microsomes (control for spontaneous oxidation of BR). The var-ious ions detected are indicated in the order of their elution time: ions m/z 301 (2.6–3.2 min); ions m/z 333 (3.4–3.6 min); ions m/z 315 (5.6–5.7 min); ions BV m/z 583(5.8–6.0); and ions BR m/z 585 (9.3–10.3).

53A. Abu-Bakar et al. / Toxicology and Applied Pharmacology 261 (2012) 50–58

Accurate mass measurements with Q-TOF revealed at least four sepa-rate peaks with a nominal mass (M+H+) of 301; accurate mass was of301.1187 (calculated monoisotopic mass for C16H17N2O4 is 301.1188).Two peaks with and accurate mass of 333.1449 (M+H+) were alsofound (calculated monoisotopic mass for C17H21N2O5 is 333.1450).These accurate masses were found to be much closer to the theoreticalvalues (Table 3) than those previously reported (De Matteis et al., 2006)confirming the dipyrrole structures based on mass of ions m/z 301 and333 proposed by these authors. Ion m/z 315 was not detected in ouranalysis.

Table 2Summary of the percentage of oxidative products of bilirubin formed following onehour incubation.

Metabolites Percentage of initial bilirubin (%)

CYP2A6 WT microsomes Auto-oxidationa Chemicaloxidationb

Biliverdin 39.0±2.1c 10.8±1.2 1.5±0.1 19.0±0.98Ion m/z 301 11.5±1.2c 4.1±0.3 0.3±0.02 21.0±1.03Ion m/z 333 32.1±2.3c ND ND 32.0±2.36Ion m/z 315 8.3±0.6c ND ND 26.0±3.45

Results are given as mean±SD of four observations. ND, not detected.a Incubation in the absence of NADPH, recombinant and wild type yeast microsomes.b Incubation with bilirubin in the presence of Fe-EDTA/H2O2.c Mean difference is significant from wild type at pb0.05 (Student's t-test).

54 A. Abu-Bakar et al. / Toxicology and Applied Pharmacology 261 (2012) 50–58

These observations and previous finding that the purported dipyr-roles originated from cleavage of the central methylene bridge(Abu-Bakar et al., 2011) suggest that BV could be formed in a twostep oxidation reaction from BR. It is possible that the first step isabstraction of an H− anion from the central methylene bridge ofBR to form a positively charged intermediate, where the basic ni-trogen atom of the pyrrole ring next to the central methylene bridgeis transformed to positively charged acidic nitrogen (Fig. 3). Thisstructure is stabilised by the double bond conjugation system of thetetrapyrrole. In the subsequent step the positively charged intermediatedonates a proton in water environment to form BV (Fig. 3). The pro-posed reaction mechanism may take place at the active site of CYP2A6.To confirm this proposition, docking analyses using the CYP2A6 crystalstructure were conducted.

Docking simulation

The crystal structures of CYP2A6 (2FDW) (Yano et al., 2006) wereused to simulate molecular docking of BR and the positively chargedintermediate at the CYP2A6's active site. We observed that the

Table 3Accurate mass values for each of the three ion products detected in in vitro enzymic oxida

Theoretical dipyrrole structure (De Matteis et al., 2006) Theoretical mass Accurate

301.1188 301.1187

333.1450 333.1449

315.1345 ND

Me=CH3; Pr=−CH2\CH2\COOH; Vi=−CH2_CH2.

orientation of BR (Fig. 4A) differs from that of the intermediate(Fig. 4B). The latter is more extended due to the double bond conju-gation system of the molecule. However, they shared common fea-tures of interaction with the active site and the haem iron, wherethe central CH2-carbon of BR and the CH-carbon of the intermediatecoordinated towards the CYP2A6 haem iron (square and circles inFig. 4). Additionally, both molecules interact with amino acid residuesPhe107, Phe111, Val117, Phe118, Asn297, Ile300, Thr305, Leu370,Ala371, and Phe480. The propionic acid groups of these moleculesformed hydrogen bonds with Asn297 and Thr305. The conjugated in-termediate has a more favourable interaction energy with CYP2A6protein than BR as: (i) it forms an additional H-bond with Ile300,Gly301, Thr305, and Thr309; (ii) it has a more complementary inter-action with the hydrophobic residues of the CYP2A6 protein (lowerΔG); and (iii) it forms a conjugated system where the pi electronsare delocalised. By contrast, the orientation of BV is different fromBR or the intermediate as its docking position is further away fromthe haem and it interacts with different amino acid residues (Fig. 5).

The docking analysis supports the experimental evidence and sug-gests that BR is potentially a high affinity substrate for the CYP2A6.Orientation of BR and the conjugated intermediate in the active sitesupports the hypothesis that the abstraction of an H− anion fromthe central methylene bridge can initiate oxidation of BR to BV.

BR increases the levels of CYP2A6 and coumarin 7-hydroxylase activity inHepG2 cells with no effect on the corresponding mRNA levels

Treatment of HepG2 cells with 5 μM BR resulted in a significantincrease of the CYP2A6 protein but not the corresponding mRNAlevels (Figs. 6A and C). The CYP2A6 apoprotein levels in BR treatedcells increased by 2–2.5 fold during the course of 12 h (Figs. 6A andC). This effect was also observed in coumarin 7-hydroxylase activity,known to be catalysed by the CYP2A6 (Fig. 6C). As BR treatmentdid not alter the CYP2A6 mRNA levels (Fig. 6C), the observationsof persistently high CYP2A6 apoprotein and activity levels suggestthat BR regulates the CYP2A6 expression through a post-transcriptionalmechanism, possibly by stabilising the protein in its active conformation.

tion of bilirubin.

mass found confirmed by Q-TOF Reported accurate mass (De Matteis et al., 2006)

301.1171

333.1468

315.1324

Fig. 3. Hypothetical mechanism for oxidation of BR. When BR is oxidised to BV the initial step is abstraction of a hydride anion H− from the central methylene bridge of BR to form apositively charged conjugated intermediate, where the basic nitrogen atom of the pyrrole ring next to the central methylene bridge is transformed to acid nitrogen. In the subse-quent step the conjugated intermediate donates a proton in water environment and BV is formed. Direct interaction of superoxide anion with BR will form BR peroxyradical and issuggested to be fragmented to products shown in Table 3.

55A. Abu-Bakar et al. / Toxicology and Applied Pharmacology 261 (2012) 50–58

BR regulates CYP2A6 expression through protein stabilisation

To confirm that BR regulates the expression of CYP2A6 by pre-venting degradation of the protein, HepG2 cells were treated witha protein synthesis inhibitor, cycloheximide (CHX) only, or in combi-nation with 5 μM BR. Treatment of cells with 400 μM CHX resulted ina time-dependent disappearance of the CYP2A6 protein, with 50% ofthe protein disappeared 4 h after treatment (Fig. 7). Bilirubin treat-ment delayed the disappearance significantly.

Discussion

Findings of the present study suggest a novel function for CYP2A6as human “BR Oxidase”. The CYP2A6 enzyme interacts with BR with ahigh affinity to be broken down to BV (a major product) and smallerdipyrroles. The observations were supported by docking analysis ofBR to the CYP2A6 active site, which indicated an orientation for oxi-dation of BR primarily to BV.

The products of CYP2A6-catalysed BR oxidation were also foundin oxidation by ROS generated in Fenton reaction. However, in this re-action formation of the smaller metabolites predominates that of BV.Similarly, BV is a minor metabolite of BR oxidation by intracellularROS (Cuypers et al., 1983; Reed et al., 1985), and by uncoupled CYPenzymes (De Matteis et al., 1993, 2002; Pons et al., 2003). These obser-vations indicate that the CYP2A6-catalysed BR oxidation is different

from other oxidation processes in that it favours BV production overthe smaller products.

We hypothesised that the first step in CYP2A6-catalysed BV for-mation is abstraction of an H− anion from the central methylenebridge of BR to form a double bond conjugated intermediate(Fig. 3). This is in agreement with the molecular docking analysesthat indicated favourable orientation of the central methylene bridgeto the CYP2A6 haem iron to facilitate abstraction of H− anion. This inturn, transformed the basic nitrogen atom of the pyrrole ring next tothe methylene bridge to acidic nitrogen to form a positively chargeddouble bond conjugated intermediate. It is noted that the hypo-thetical sequence of events fits the sequence of CYP reaction cycle(Guengerich, 2001).

In our opinion, this process is less random than other reported ox-idation reactions because both BR and the conjugated intermediatestably docked at the CYP2A6 active site through their interactionswith the critical amino acid residues. Furthermore, the intermediateis stabilised at the active site by its inherent conjugated system andby its lower free energy than that of BR radical. The observation thatBV docking orientation is further away from the haem iron and thatit does not interact with the critical amino acid residues indicates aless accurate binding of BV. This, in turn, may facilitate BV dissocia-tion from the active site. Collectively, the docking simulation suggestsa mechanism where the intermediate formation is the key step in theCYP2A6-catalysed oxidation of BR to BV. The final conversion of the

Fig. 4. Comparison of BR and its conjugated intermediate binding to the CYP2A6 active site. (A) Docking simulation of BR; (B) docking simulation of the conjugated intermediate.Both molecules interact with amino acid residues Phe107, Phe111, Val117, Phe118, Asn297, Ile300, Thr305, Leu370, Ala371, and Phe480. The propionic acid groups of these mol-ecules formed hydrogen bonds with Asn297 and Thr305. The conjugated intermediate forms an additional hydrogen bond with Ile300, Gly301, Thr305, and Thr309. The centralmethylene bridge of BR (A) and the positive charged acidic nitrogen atom of the pyrrole ring in the intermediate (B) coordinated towards haem iron (black square). Moleculesin red represent BR and the conjugated intermediate.

56 A. Abu-Bakar et al. / Toxicology and Applied Pharmacology 261 (2012) 50–58

intermediate to BV takes place by proton donation in the CYP2A6active site before the product is finally released from the site.

In addition to BV, oxidation of BR produces at least three smallerbreakdown products withm/z 301, 315, and 333 (prominent productsobserved in positive ionisation mode). These products can be foundin oxidation driven by Fenton reaction, CYP uncouplers in livermicrosomes system (De Matteis et al., 2006), and our recombinantCYP2A6 system. These products could be originated from fragmenta-tion of BR peroxyradical (De Matteis et al., 2006), which could beformed by a superoxide anion attacking BR directly (Fig. 3).

The observation that bilirubin treatment delayed degradationof CYP2A6 protein (Fig. 7) suggests that BR regulates the CYP2A6

Fig. 5. Docking simulation of BV to the CYP2A6 active site. Structure in red representsBV.

expression through protein stabilisation. This finding is in agreementwith our previous observation that the CYP2A5 (the mouse ortholo-gue of CYP2A6) is a labile protein that is stabilised by the binding ofhigh affinity substrate, coumarin, to its active site (Juvonen et al.,1985). Similarly, we recently observed that BR delayed degradationof CYP2A5 protein (Abu-Bakar et al., 2011). These observationstogether with the data obtained from kinetics study, docking sim-ulation, and metabolite screening indicate that BR is potentially asubstrate and a regulator of CYP2A6.

The physiological significance of CYP2A6 as BR oxidase is not ap-parent from this study. However, our recent results obtained fromstudies in mice supported the proposition that CYP2A5 is part of themachinery that controls intracellular BR levels during oxidative stress(Abu-Bakar et al., 2005, 2007, 2011). We demonstrated that duringoxidative stress the temporal induction of haem oxygenase-1(HMOX1), an enzyme that produces BR, preceded that of CYP2A5 byseveral hours (Abu-Bakar et al., 2005). At elevated concentrations,BR up-regulates the CYP2A5 through protein stabilisation mechanism(Abu-Bakar et al., 2011).

These observations support a scenario where the elevated BR pro-duced by HMOX1 would primarily be oxidised by ROS and eliminatedas waste products. The remaining BR (should such situation occur)would bind to the CYP2A5, leading to stabilisation of the proteinand acceleration of BR oxidation to BV, which is reduced back to BRby biliverdin reductase, permitting BR to act as an antioxidant manytimes over (Baranano et al., 2002; Sedlak and Snyder, 2004). Ourlatest results using arsenite as an oxidant stressor in DBA/2 miceshowed that coordinated induction of HMOX1 and CYP2A5 is asso-ciated with increased excretion of BOMs in the urine (Arthur et al.,in press), thus further supporting our hypothesis.

The present observation that BR upregulates the CYP2A6 protein byprotein stabilisation suggests that the human enzyme acts like itsmouse orthologue. In this regard, it is important to note that the affinityof BR to CYP2A6 (see Fig. 1) seems optimal for the enzyme to work

Fig. 6. Time-course effect of BR on CYP2A6 protein and mRNA expression. HepG2 cellswere treated with 5 μM bilirubin (BR) or DMSO (Ctr) for 1, 3, 9, or 12 h. (A) Westernimmunoblotting of S9 fractions (15 μg protein) from control and BR treated cellsprobed with anti-CYP2A6 or anti-β-actin antibody. β-Actin protein levels are shownas control for protein loading. Microsomes from recombinant yeast expressingCYP2A6 were used as positive control (+). The band at 54 and 47 kDa correspondsto CYP2A6 and β-actin protein, respectively. (B) Migration of PCR products in 2.5% aga-rose gel. Lanes 1 and 5=50 bp DNA ladder; lanes 2 and 6 are products from treatedcells; lanes 3 and 7 are products from control cells; and lanes 4 and 8 are non-template control (NTC). (C) Densitometric quantification of Western blot and time-dependent changes in CYP2A6-dependent coumarin 7-hydroxylase (COH) activity,and CYP2A6 mRNA expression relative to GAPDH mRNA expression in control and BRtreated cells. Each data point represents the mean±S.D. of six samples normalisedagainst loading control (β-actin). ⁎Mean difference is significant from control group,pb0.005 (Student's t-test). #Mean difference is significant from 1 h control group,pb0.005 (Student's t-test).

Fig. 7. Effect of BR on half-life of CYP2A6 protein. HepG2 cells were treated with 400 μMcycloheximide (CHX) or 5 μM bilirubin (BR)+400 μM CHX for 0, 1, 2, or 4 h. (A) Fifteenmicrograms of S9 proteins was loaded, electrophoresed, transferred to PVDF membraneand probed with anti-CYP2A6 or anti-β-actin antibody. β-Actin protein levels are shownas control for protein loading. Microsomes from recombinant yeast expressing CYP2A6were used as positive control (+). The band at 54 and 47 kDa corresponds to CYP2A6and β-actin protein, respectively. (B) Densitometric quantification of Western blot. Eachdata point represents the mean±S.D. of six samples. ⁎Mean difference is significantfrom CHX treatment group, pb0.005 (Student's t-test).

57A. Abu-Bakar et al. / Toxicology and Applied Pharmacology 261 (2012) 50–58

efficiently as a BR oxidase in subtoxic concentrations. Given that theCYP2A6 gene, is also regulated by the oxidative stress responsive tran-scription factor, nuclear factor erythroid 2-like 2 (Nrf2) (Yokota et al.,2011), and that the enzyme activity is elevated when the hepaticantioxidant capacity is reduced (Jin et al., 2011; Niemela et al.,2000; Raunio et al., 1998), it is plausible that the human CYP2A6has the same biological function as the mouse CYP2A5.

Conflict of interest

The authors declare that they have no conflict of interest in this work.

Acknowledgments

We thank Mr. Geoff Eaglesham (Queensland Health Forensic andScientific Services, Coopers Plains, Australia) for his assistance withHPLC/ESI-MS analysis of the oxidative metabolites. This project is

partly funded by the Cooperative Research Centre for ContaminationAssessment & Remediation of the Environment (CRC CARE: projectnumber 1-3-04-06/7), Australia, and the University of QueenslandNew Staff Research Start-Up Fund (UQRSF 2007000471). The Nation-al Research Centre for Environmental Toxicology is a partnership be-tween Queensland Health and the University of Queensland. AAB andDMA contributed equally to this work.

References

Abu-Bakar, A., Satarug, S., Marks, G.C., Lang, M.A., Moore, M.R., 2004. Acute cadmium chlo-ride administration induces hepatic and renal CYP2A5 mRNA, protein and activity inthe mouse: involvement of transcription factor NRF2. Toxicol. Lett. 148, 199–210.

Abu-Bakar, A., Moore, M.R., Lang, M.A., 2005. Evidence for induced microsomal biliru-bin degradation by cytochrome P450 2A5. Biochem. Pharmacol. 70, 1527–1535.

Abu-Bakar, A., Lamsa, V., Arpiainen, S., Moore, M.R., Lang, M.A., Hakkola, J., 2007. Regulationof CYP2A5 gene by the transcription factor nuclear factor (erythroid-derived 2)-like 2.Drug Metab. Dispos. 35, 787–794.

Abu-Bakar, A., Arthur, D.M., Aganovic, S., Ng, J.C., Lang, M.A., 2011. Inducible bilirubinoxidase: a novel function for the mouse cytochrome P450 2A5. Toxicol. Appl. Phar-macol. 257, 14–22.

Arthur, D.M., Ng, J.C., Lang, M.A., Abu-Bakar, A., in press. Urinary excretion of bilirubinoxidative metabolites in arsenite-treated mice. J. Toxicol. Sci.

Baranano, D.E., Rao, M., Ferris, C.D., Snyder, S.H., 2002. Biliverdin reductase: a majorphysiologic cytoprotectant. PNAS 99, 16093–16098.

Cashore, W.J., 1990. The neurotoxicity of bilirubin. Clin. Perinatol. 17, 437–447.Cohen, M., Kapitulnik, J., Ostrow, J.D., Webster, C., 1986. Effect of combined treatment

with 2,3,7,8-tetrachlorodibenzo-p-dioxin and phototherapy on bilirubin metabo-lism in the jaundiced Gunn rat. Hepatology 6, 490–494.

Crawford, J.M., Ransil, B.J., Narciso, J.P., Gollan, J.L., 1992. Hepatic microsomal bilirubinUDP-glucuronosyltransferase: the kinetics of bilirubin mono- and diglucuronidesynthesis. J. Biol. Chem. 267, 16943–16950.

Cuypers, H.T.M., Ter Haar, E.M., Jansen, P.L.M., 1983. Microsomal conjugation and oxida-tion of bilirubin. Biochim. Biophys. Acta 758, 135–143.

De Matteis, F., Dawson, S.J., Boobis, A.R., Comoglio, A., 1991. Inducible bilirubin-degrading system of rat liver microsomes: role of cytochrome P4501A1. Mol.Pharmacol. 40, 686–691.

58 A. Abu-Bakar et al. / Toxicology and Applied Pharmacology 261 (2012) 50–58

De Matteis, F., Dawson, S.J., Gibbs, A.H., 1993. Two pathways of iron-catalyzed oxida-tion of bilirubin: effect of desferrioxamine and trolox, and comparison with micro-somal oxidation. Free Radic. Biol. Med. 15, 301–309.

De Matteis, F., Dawson, S.J., Pons, N., Pipino, S., 2002. Bilirubin and uroporphyrinogenoxidation by induced cytochrome P4501A and cytochrome P4502B. Role ofpolyhalogenated biphenyls of different configuration. Biochem. Pharmacol. 63,615–624.

De Matteis, F., Lord, G.A., Lim, C.K., Pons, N., 2006. Bilirubin degradation byuncoupled cytochrome P450. Comparison with a chemical oxidation systemand characterization of the products by high-performance liquid chromatogra-phy/electrospray ionization mass spectrometry. Rapid Commun. Mass Spectrom.20, 1209–1217.

Dore, S., Takahashi, M., Ferris, C.D., Hester, L.D., Guastella, D., Snyder, S.H., 1999. Biliru-bin, formed by activation of haem oxygenase-2, protects neurons against oxidativestress injury. PNAS 96, 2445–2450.

Draper, A.J., Madan, A., Parkinson, A., 1997. Inhibition of coumarin 7-hydroxylase activityin human liver microsomes. Arch. Biochem. Biophys. 341, 47–61.

Fletcher, A.A., 1970. New approach to variable metric algorithms. Comput. J. 13, 317–322.Geneste, O., Raffalli, F., Lang, M.A., 1996. Identification and characterization of a 44 kDa

protein that binds specifically to the 3′-untranslated region of CYP2A5 mRNA:inducibility, subcellular distribution and possible role in mRNA stabilization.Biochem. J. 313, 1029–1037.

Guengerich, F.P., 2001. Common and uncommon cytochrome P450 reactions related tometabolism and chemical toxicity. Chem. Res. Toxicol. 14, 611–650.

Halgren, T., 1996. Merck molecular force field. II. MMFF94 van der Waals and electro-static parameters for intermolecular interactions. J. Comput. Chem. 17, 520–552.

Jansen, T., Hortmann, M., Oleze, M., Opitz, B., Steven, S., Schell, R., Knorr, M., Karbach, S.,Schuhmacher, S., Wenzel, P., Münzel, T., Daiber, A., 2010. Conversion of biliverdinto bilirubin by biliverdin reductase contributes to endothelial cell protectionby heme oxygenase-1-evidence for direct and indirect antioxidant actions ofbilirubin. J. Mol. Cell. Cardiol. 49, 186–195.

Jin, M., Arya, P., Patel, K., Singh, B., Silverstein, P.S., Bhat, H.K., Kumar, A., Kumar, S.,2011. Effect of alcohol on drug efflux protein and drug metabolic enzyme inU937 macrophages. Alcohol. Clin. Exp. Res. 35, 132–139.

Jones, G., Willett, P., Glen, R.C., Leach, A.R., Taylor, R., 1997. Development and validationof a genetic algorithm for flexible ligand docking. Abstr. Pap. Am. Chem. Soc. 214(154-COMP).

Juvonen, R., Kaipainen, P., Lang, M.A., 1985. Selective induction of coumarin 7-hydroxylaseby pyrazole in D2 mice. Eur. J. Biochem. 152, 3–8.

Kapitulnik, J., Ostrow, J.D., 1978. Stimulation of bilirubin catabolism in jaundiced Gunnrats by an inducer of microsomal mixed-function monooxygenases. PNAS 75,682–685.

Keshavan, P., Schwemberger, S.J., Smith, D.L., Babcock, G.F., Zucker, S.D., 2004. Uncon-jugated bilirubin induces apoptosis in colon cancer cells by triggering mitochon-drial depolarization. Int. J. Cancer 112, 433–445.

Kinonen, T., Pasanen, M., Gynther, J., Poso, A., Jarvinen, T., Alhava, E., Juvonen, R.O., 1995.Competitive inhibition of coumarin 7-hydroxylation by pilocarpine and its interac-tion with mouse CYP2As and human CYP2A6. Br. J. Pharmacol. 116, 2625–2630.

Lindahl, E., Hess, B., van der Spoel, D., 2001. GROMACS 3.0: a package for molecularsimulation and trajectory analysis. J. Mol. Model. 7, 306–317.

Menken, M., Waggoner, J.G., Berlin, N.I., 1966. The influence of bilirubin in oxidativephosphorylation and related reactions in brain and liver mitochondria: effects ofprotein-binding. J. Neurochem. 13, 1241–1248.

Niemela, O., Parkkila, S., Juvenon, R.O., Viitala, K., Gelbion, H.V., Pasanen, M., 2000.Cytochromes P450 2A6, 2E1, and 3A and production of protein-aldehyde adductsin the liver of patients with alcoholic and non-alcoholic liver diseases. J. Hepatol.33, 893–901.

Noir, B.A., Boveris, A., Garaza Pereira, A.M., Stoppani, A.O., 1972. Bilirubin: a multi-siteinhibitor of mitochondrial respiration. FEBS Lett. 27, 270–274.

Pons, N., Pipino, S., De Matteis, F., 2003. Interaction of polyhalogenated compounds ofappropriate configuration with mammalian or bacterial CYP enzymes increasedbilirubin and uroporphyrinogen oxidation in vitro. Biochem. Pharmacol. 66,405–414.

Rahnasto, M., Wittekindt, C., Juvonen, R.O., Turpeinen, M., Petsalo, A., Pelkonen, O.,Poso, A., Stahl, G., Höltje, H.D., Raunio, H., 2008. Identification of inhibitors of thenicotine metabolising CYP2A6 enzyme— an in silico approach. PharmacogenomicsJ. 8, 328–338.

Raunio, H., Juvenon, R., Pasanen, M., Pelkonen, O., Paakko, P., Soini, Y., 1998. CytochromeP4502A6 (CYP2A6) expression in human hepatocellular carcinoma. Hepatology 27,427–432.

Reed, G.A., Lasker, J.M., Eling, T.E., Sivarajah, K., 1985. Peroxidative oxidation ofbilirubin during prostaglandin biosynthesis. Prostaglandins 30, 153–165.

Schmid, R., Hammaker, L., 1963. Metabolism and disposition of 14C-bilirubin in congen-ital non-hemolytic jaundice. J. Clin. Invest. 42, 1720–1734.

Sedlak, T.W., Snyder, S.H., 2004. Bilirubin benefits: cellular protection by a biliverdinreductase antioxidant cycle. Pediatrics 113, 1776–1782.

Stahl, G.R., Höltje, H.D., 2005. Development of models for cytochrome P450 2A5 as wellas two of its mutants. Pharmazie 60, 247–253.

Stocker, R., 2004. Antioxidant activities of bile pigments. Antioxid. Redox Signal. 6,841–849.

Stocker, R., Ames, B.N., 1987. Bilirubin is an antioxidant of possible physiological im-portance. PNAS 84, 8130–8134.

Tukey, R.H., Strassburg, C.P., 2000. Human UDP-glucuronosyltransferases: metabolism,expression, and disease. Annu. Rev. Pharmacol. Toxicol. 40, 581–616.

Verdonk, M.L., Cole, J.C., Hartshorn, M.J., Murray, C.W., Taylor, R.D., 2003. Improvedprotein–ligand docking using GOLD. Proteins 52, 609–623.

Yabusaki, Y., 1998. Expression of mammalian cytochromes P450 in yeast. In: Phillips,I.R., Shephard, E.A. (Eds.), Cytochrome P450 Protocols. Humana Press, New Jersey,pp. 195–202.

Yano, J.K., Denton, T.T., Cerny, M.A., Zhang, X., Johnson, E.F., Cashman, J.R., 2006. Syn-thetic inhibitors of cytochrome P-450 2A6: inhibitory activity, difference spectra,mechanism of inhibition, and protein co-crystallization. J. Med. Chem. 49,6987–7001.

Yokota, S., Higashi, E., Fukami, T., Yokoi, T., Nakajima, M., 2011. Human CYP2A6 isregulated by nuclear factor-erythroid 2 related factor 2. Biochem. Pharmacol.81, 289–294.