1521-009X/44/2/250–261$25.00 http://dx.doi.org/10.1124/dmd.115.067173DRUG METABOLISM AND DISPOSITION Drug Metab Dispos 44:250–261, February 2016Copyright ª 2016 by The American Society for Pharmacology and Experimental Therapeutics

Application of Micropatterned Cocultured Hepatocytes to Evaluatethe Inductive Potential and Degradation Rate of Major Xenobiotic

Metabolizing Enzymes s

Vaishali Dixit, Amanda Moore, Hong Tsao, and Niresh Hariparsad

Drug Metabolism and Pharmacokinetics, Vertex Pharmaceuticals Incorporated, Boston, Massachusetts

Received September 11, 2015; accepted December 8, 2015

ABSTRACT

Long-term coculture models of hepatocytes are promising tools tostudy drug transport, clearance, and hepatoxicity. In this report wecompare the basal expression of drug disposition genes and theinductive response of prototypical inducers (rifampin, phenobarbi-tal, phenytoin) in hepatocyte two-dimensional monocultures and thelong-term coculture model (HepatoPac). All the inducers used in thestudy increased the expression and activity of CYP3A4, CYP2B6 andCYP2C enzymes in the HepatoPac cultures. The coculture modelshowed a consistent and higher induction of CYP2C enzymescompared with the monocultures. The EC50 of rifampin for CYP3A4and CYP2C9 was up to 10-fold lower in HepatoPac than themonocultures. The EC50 of rifampin calculated from the clinical drug

interaction studies correlated well with the EC50 observed in theHepatoPac cultures. Owing to the long-term stability of theHepatoPac cultures, we were able to directly measure a half-life(t1/2) for both CYP3A4 and CYP2B6 using the depletion kinetics ofmRNA and functional activity. The t1/2 for CYP3A4 mRNA was 26hours and that for the functional protein was 49 hours. The t1/2 ofCYP2B6was 38 hours (mRNA) and 68 hours (activity), which is longerthan CYP3A4 and shows the differential turnover of these twoproteins. This is the first study to our knowledge to report theturnover rate of CYP2B6 in human hepatocytes. The data presentedhere demonstrate that the HepatoPac cultures have the potential tobe used in long-term culture to mimic complex clinical scenarios.

Introduction

The current landscape of in vitro models for predicting inductionrelies on primary human hepatocytes in conventional monoculturelayers without an overlay [two-dimensional (2D)] or in a sandwich-culture format. Several publications have highlighted the utility of 2Dor sandwich-cultured models using cryopreserved human hepatocytesto assess the CYP3A4-induction–based drug-drug interaction (DDI)risk of new chemical entities (NCEs) by conducting retrospectiveanalyses on compounds that are known to be CYP3A4 inducers ornoninducers in the clinic (Fahmi et al., 2010; Einolf et al., 2014; Zhanget al., 2014). Although these models have proven to be invaluable inunderstanding the CYP3A4-induction–based DDI risk of NCEs in botha practical and cost-effective way, the 2D substratum forces hepatocytesto alter their cytoskeleton toward a flattened morphology (Buxboimet al., 2010). This change in cell shape and form limits cell–cell andcell–matrix interactions that consequently leads to reduced polarization,reduced bile canaliculi formation, and a loss of important signalingpathways necessary for normal hepatocyte function (Berthiaume et al.,1996). Not surprisingly, hepatocytes cultured in this manner remainviable for only 4–5 days and rapidly deviate from their differentiatedphenotype (Gómez-Lechón et al., 1998).

In addition to CYP3A4, it is also important to identify in vitrosystems for the evaluation of CYP2C induction. In a recent publicationby Yajima et al. (2014), in which the standard 2D hepatocyte culturemodel was used, of the eight hepatocyte lots treated with rifampin,CYP2C8 and CYP2C9 mRNA were not induced in five and twohepatocyte lots, respectively. This can be problematic given thatboth the European Medicines Agency and the Food and DrugAdministration Regulatory Guidances now recommend that theCYP2C-induction–based DDI risk of NCEs need to be evaluated giventhat CYP2C comprises approximately 20% of total P450 content inhuman livers (Inoue et al., 1994) and that CYP2Cs are responsible formetabolizing approximately 20% of all prescribed drugs, includingtolbutamide, phenytoin, warfarin, and ibuprofen (Goldstein, 2001).Prediction of CYP2B6 induction and inactivation is another lesser

studied area in preclinical drug research. Several substrates, inhibitors,and inducers of CYP2B6 have been identified making it an importantenzyme involved in DDI. Recently Zamek-Gliszczynski et al. (2014)reported that suppression of CYP2B6 by an NCE could be a potentialpathway for DDI. Physiologically based pharmacokinetic (PBPK)models used for predicting these interactions rely on accurate turnoverrate of CYP2B6 mRNA and protein.As the need for more predictive in vitro hepatocyte models is

increasing, there is now a strong demand for technologies that maintaintheir differentiated in vivo-like phenotype and for a longer period oftime. Several models have become available in recent years, however,the main issues are the requirement of large cell numbers, lowthroughput, and expensive equipment, which render these systems

Amanda Moore is a former employee of Hepregen Corporation and owns equityin that company.

dx.doi.org/10.1124/dmd.115.067173.s This article has supplemental material available at dmd.aspetjournals.org.

ABBREVIATIONS: 2D, Two-dimensional; AUC, area under the plasma-concentration time curve; CAR, constitutive androstane receptor; cDNA,complementary DNA; Cmax, maximum plasma concentration; DDI, drug-drug interaction; DMSO, dimethyl sulfoxide; NCE, new chemical entity;P450, cytochrome P450; PBPK, physiologically based pharmacokinetics; PXR, pregnane X receptor.

250

http://dmd.aspetjournals.org/content/suppl/2015/12/09/dmd.115.067173.DC1Supplemental material to this article can be found at:

at ASPE

T Journals on February 4, 2020

dmd.aspetjournals.org

Dow

nloaded from

unattractive for the pharmaceutical industry (Bachmann et al., 2015).The HepatoPac coculturemodel consists of primary hepatocytes that aremicropatterned to form a discrete microarchitecture or “hepatocyteislands” that are surrounded by supporting fibroblasts resulting in long-term viability and metabolic function for several weeks (Khetani andBhatia, 2008). Many groups have been able to successfully use theHepatoPac model for metabolite profiling and improved in vitro–invivo correlation (IVIVC) for low clearance compounds and livertoxicity studies (Wang et al., 2010; Chan et al., 2013; Trask et al.,2014). The HepatoPac model has also been successfully used forthe mechanistic understanding of complex drug-drug interactions(Ramsden et al., 2014).Whereas earlier studies by Khetani and Bhatia (2008) indicated that

transcription factors, such as the pregnane X receptor (PXR) andconstitutive androstane receptor (CAR), together with coactivators andcorepressors are maintained during the extended culture time, nostudies have been reported in which a thorough evaluation has beenconducted to understand the advantages and limitations of theHepatoPac model compared with the 2D-induction model. Theobjectives of this study were to: 1) compare differences in geneexpression levels of major metabolic enzymes and transporters in thesame batch of hepatocytes when cultured in the 2D conformation(monoculture) to the micropatterned cocultured system (HepatoPac); 2)compare induction of major (CYP3A4, CYP2B6, and CYP2C)metabolic enzymes between the two systems; and 3) determine thedegradation rate (kdeg) of CYP3A4 and CYP2B6 in HepatoPac.

Materials and Methods

Human HepatoPac cultures, proprietary hepatocyte maintenance, and in-duction application media were purchased from Hepregen Corporation (Med-ford, MA). The HepatoPac and the monocultures were prepared usingcryopreserved human hepatocytes (Table 1) purchased from Life Technologies/Thermo Fisher (Grand Island, NY) or BioreclamationIVT (Baltimore, MD).Williams’ E medium, Cryopreserved hepatocyte recovery media, HepatocyteMaintenance Supplement Pack, Hepatocyte Plating Supplement Pack werepurchased from Life Technologies. Rifampin, phenytoin, phenobarbital anddiclofenac sodium salt were purchased from Sigma-Aldrich (St. Louis, MO).Midazolam, 19OH-midazolam, 49OH-diclofenac, bupropion, hydroxybupropionand isotopically labeled internal standards ([13C6]hydroxydiclofenac, [

13C3]19-hydroxymidazolam,D6 hydroxybupropion) were purchased fromBDBiosciences/Gentest (Woburn, MA). MagMAX total RNA recovery bead kits were purchasedfrom Life Technologies. Applied Biosystems reverse transcription–polymerasechain reaction (PCR) and TaqMan real-time PCR assays were purchased fromThermoFisher Scientific (Waltham, MA).

Induction in HepatoPac Cultures. Induction-certified cryopreserved hepa-tocytes (Table 1) purchased from commercial vendors were used to prepare theHepatoPac cultures by Hepregen Corporation, using previously describedmethodology (Khetani and Bhatia, 2008). Hepatocytes were seeded at a densityof 32,000 cells/well with 3T3-J2 murine embryonic fibroblasts in 24-well plates.HepatoPac cultures were maintained in proprietary media at Hepregen Cor-poration for 9 days post seeding prior to initiating experiments. For inductionexperiments, proprietary serum-free application medium supplied by HepregenCorporation was used. The plates were cultured in an incubator with 10% CO2

and 95% relative humidity at 37�C. Cells were treated with 0.1% dimethylsulfoxide (DMSO), rifampin, phenytoin, or phenobarbital for 72 hours. Drug-containing media was changed every 24 hours for three days. Concentration ofsolvent (DMSO) in the culture media did not exceed 0.1%. All experiments wereperformed in triplicate.

Induction in Hepatocyte Monocultures. Monocultures were prepared inparallel to the HepatoPac cultures, using the same lots of cryopreservedhepatocytes, in 96-well format. Cells were plated in collagen-coated plates at adensity of 0.8 million viable cells per milliliter. Cells were maintained inWilliams’ E medium containing Hepatocyte Maintenance Supplement Pack and10% bovine serum for 6–8 hours before being changed to medium devoid ofserum. Twenty-four hours after plating, the cells were treated with variousconcentrations of rifampin (0.01–10 mM), phenytoin (1–250 mM), or pheno-barbital (10–1000 mM) for 72 hours. Drug-containing media was changed every24 hours for 3 days. Concentration of solvent (DMSO) in the culture media didnot exceed 0.1%. Cultures were maintained in an incubator with 5% CO2 and95% humidity. All experiments were performed in triplicate.

P450 Activity Assay. At the end of induction period induction meda fromboth cultures was removed and the cells were rinsed with either serum-freeinduction medium or Hanks’ balanced salt solution (HBSS). Cultures were thenincubated with 15 mM midazolam, 40 mM diclofenac, or with 100 mMbupropion, all of which were dissolved in DMSO and then spiked into serum-free medium or HBSS, for 30 minutes. Total concentration of solvent in theactivity assay did not exceed 0.5%. Fifty microliters of supernatant wastransferred to plates containing 100 ml of acetonitrile containing [13C6]hydroxydiclofenac, [13C3]19-hydroxymidazolam, or D6-hydroxybupropion asinternal standards. Samples were stored at –20�C until further analysis.

Nonspecific Binding of Rifampin. Rifampin (1mM)was incubated with 3T3murine embryonic fibroblast cells (n = 3) at 37�C in the media used forHepatoPac-induction studies. The media was collected at 0, 1, 2, 4, 8, and 24hours, and rifampin concentration was measured in these samples. Rifampin(1 mM) was also incubated with plated monocultured hepatocytes (n = 3) inthe induction media at 4�C. Media samples were collected at 0, 1, 2, 4, 8, and24 hours, and rifampin concentration was measured in these samples usingliquid chromatography–mass spectrometry.

Liquid Chromatography–Tandem Mass Spectrometry Analysis ofSamples. Five-microliter aliquots of samples were injected onto a Unisol C18column (2.1 � 30 mm, 5 mm) (Agela Technologies, Newark, DE) using a CTCAnalytics PAL (LEAP) autosampler (CTC Analytics AG, Zwingen, Switzer-land), and Agilent 1100 Series Binary Pumps (Agilent Technologies, Palo Alto,CA). The analytes were eluted using a gradient method with a flow rate from0.8 to 1.8 ml/min. Mobile phase A was 10 mM ammonium acetate in water(pH 4.0) and mobile phase B was acetonitrile/methanol (50/50, v/v). The analytes,19-hydroxymidazolam, 49-hydroxydiclofenac, hydroxybupropion, and rifampin,were analyzed using an API 5500 mass spectrometer (AB Sciex, Foster City,CA) in electrospray-ionization multiple-reaction-monitoring mode using thefollowing mass transitions 342→203, 310→266, 256.2→167.2, 823.6→791.4,respectively. Isotope-labeled internal standard transitions, 345→206, 316→272,and 262.2→167.2, were used for 19-hydroxymidazolam13C3, 49-hydroxy-diclofenac 13C6, and D6 hydroxybupropion, respectively. The peak area ratioof analyte/internal standard was used for calculations. Data were analyzed inGalileo LIMS system version 3.3 (Thermo Electric Corporation, Philadelphia,PA). The lower limit of quantitation (LLOQ) ranged from 0.0005 to 0.05 mMand the upper limit of quantitation (ULOQ) ranged from0.2 to 2mM, depending onanalytes tested, instrument sensitivity, linear dynamic range of instrument, andanalyte concentration range of samples. All analyte concentrations of testedsamples are within linear range of specific testing. The back-calculated concen-trations of the calibration standards are within 630% of the nominal values.

Relative Expression of mRNA. Following P450 activity measurements, themedia was removed and the cells were stored in RNAlater at –20�C until RNAisolation. Total RNA was isolated from cells using the MagMAX bead-basedsystem according to the manufacturers’ protocol (Life Technologies). The RNAconcentration and quality of each sample was measured using a NanoDrop 2000spectrophotometer. A two-step reverse transcriptase (RT)- PCR reaction wasconducted by reverse transcribing 50 ng of total RNA to cDNA using TaqManReverse Transcription Reagents, according to the TaqMan Universal PCRMaster Mix protocol. PCR reactions were then prepared by adding an aliquot ofcomplementary DNA (cDNA) (2 ml) to a reaction mixture containing the

TABLE 1

Demographic information of human hepatocyte donors

Donor ID Sex Age (yr) RaceCause

of DeathMedical History

Hu1624 F 72 Caucasian AspirinBPB F 42 Caucasian Head traumaNON F 35 Caucasian SIGSWa Alcohol, tobacco,

IV drug abuse

aSIGSW, self-inflicted gun-shot wound

Hepatocyte Coculture System to Assess P450 Induction 251

at ASPE

T Journals on February 4, 2020

dmd.aspetjournals.org

Dow

nloaded from

TaqMan Fast Universal PCRMaster Mix solution, primers, and probes for P450enzymes. GAPDH was measured as a housekeeping gene and all data werenormalized to the expression of GAPDH. Fold change in mRNA expression oversolvent control was expressed using the DDCt method.

Basal Expression of Drug Disposition Genes in HepatoPac andMonocultures. Basal expression of drug metabolism genes was determinedfrom the cDNA of DMSO-treated hepatocytes from both culture models.Microfluidic cards with assays for desired genes and endogenous controls werecustom made at Life Technologies. The cards were loaded with a mixture of 50ml of diluted cDNA and 50 ml of fast real-time PCR master mix. The cards wererun according to manufacturer’s (Life Technologies) instructions on ViiA 7 real-time PCR instrument with an array block. The relative difference in geneexpression between HepatoPac cultures and the monocultures was determinedusing comparativeDDCt method. GAPDHwas used as the housekeeping gene inall gene expression studies. The DCt values from the monocultures and theHepatoPac cultures were compared using a Student’s t test to determine if theexpression levels in HepatoPac are significantly different from the monocultures.

Determination of kdeg and t1/2 of CYP3A4 and CYP2B6. HepatoPaccultures for three lots of hepatocytes (n = 3) were treated for 72 hours with either0.1% DMSO or 10 mM rifampin to achieve maximal induction of message andactivity. Following the 72 hours-induction period, the recovery of P450 activityandmRNA expression were measured at 0, 24, 48, 72, 96, and 144 hours. Duringthis recovery period, the hepatocytes were maintained in 10% serum-containingmedium. Fold change in activity and mRNA was calculated.

Data Analysis and Fitting. To determine the EC50 and Emax values, the datafrom concentration-response curves were fitted to a three-parameter sigmoid(Hill) model, according to the following equation:

E ¼ Emax*C g

EC50 g þ C gð1Þ

The baseline value of induction (E0) was fixed to 1. Data fitting was doneusing Graphpad Prism 5.0 (La Jolla, CA).

To determine kdeg of the CYP3A4 and CYP2B6 protein (as measured byfunctional activity) and mRNA, fold change was first converted to the natural logand then plotted against time (hr). The data were fitted to a linear regressionmodel and the slope (k) of the loss of enzyme activity or mRNA was used tocalculate the t1/2 using eq. (2). This t1/2 represents the degradation rate of theprotein and message.

t1=2 ¼ 0:693k ðslopeÞ ð2Þ

Predicting Plasma Exposure of Rifampin. Pharmacokinetics of rifampinwas simulated using the minimal PBPK model implemented in the Simcypsimulator (version 13; Simcyp Ltd, Sheffield, UK). The default pharmacokineticparameters of rifampin in the Simcyp simulator used to predict its plasmaexposure are presented in Supplemental Table 1. Plasma concentration profile ofrifampin given at the oral dose of 5, 10, 15, 75, or 600 mg for 7 days wassimulated. The predicted maximum plasma concentration (Cmax) at steady statewas corrected using the protein binding of rifampin (fu = 0.15, data source:Simcyp). The observed midazolam area under the plasma-concentration timecurve (AUC) change reported in Kharasch et al. (2011) and the simulatedunbound Cmax of rifampin was then fitted to a nonlinear regression model usingGraphPad Prism software (v6.0).

Results

Basal Expression of Drug Metabolism Genes in the HepatoPacModel Compared with Monocultures. Expression of 76 drug-disposition genes and 4 housekeeping genes was determined invehicle-treated cells (72 hours) from both monoculture and theHepatoPac coculture models. The HepatoPac cultures were about 12-days-old and the monocultures were 4-days-old at the time of thisanalysis. For the HepatoPac studies we used a 24-well format that has30,000 hepatocytes /well as opposed to the 96-well format that has 5000hepatocytes/well. The 24-well format ensures sufficient yield of RNAto perform gene expression studies.

TABLE2

Emax

andEC50forCYP3A

4activ

ityandmRNA

aftertreatm

entwith

rifampin,

phenobarbital,andphenytoinfor72

hoursin

either

HepatoP

acor

monoculturesin

serum-freemedia.

NON

BPB

Hu1

624

HepatoP

acMon

oculture

HepatoP

acMon

oculture

HepatoP

acMon

oculture

Emax

EC50

Emax

EC50

Emax

EC50

Emax

EC50

Emax

EC50

Emax

EC50

mM

mM

mM

mM

mM

mM

mRNA

Rifam

pin

13.1

(0.33)

0.1(0.053)

24.4

(0.6)

1.1(0.05)

16.5

(1.36)

0.1(0.16)

34.0

a1.0a

21.1

(1.14)

0.1(0.11)

12.4

(0.28)

0.2(0.06)

Phenobarbital

15.2

(0.61)

97.2

(0.045)

18.8

(2.91)

163.3(0.21)

12.7

(0.44)

112.3(0.04)

18.2

(2.38)

358.2(0.11)

23.3

(2.02)

85.7

(0.09)

9.0(0.54)

101.5(0.097)

Phenytoin

14.7

(4.2)

70.7

(0.27)

25.8

(2.6)

55.0

(0.093)

14.5

(0.99)

56.9

(0.062)

11.5

(1.56)

42.6

(0.18)

17.6

(2.03)

31.9

(0.098)

6.3(0.27)

12.0

(0.096

)Activity

Rifam

pin

6.3(0.2)

0.03

(0.1)

17.1

(1.0)

0.6(0.1)

4.2(0.3)

0.1(0.2)

ND

ND

12.3

(0.4)

0.1(0.1)

7.6(1.5)

0.2(0.6)

Phenobarbital

6.7(0.1)

99.3

(0.05)

15.6

(1.3)

191.1(0.1)

7.4(0.7)

110.4(0.1)

6.3(1.7)

298.2(0.2)

13.7

(1.5)

96.9

(0.1)

6.1(0.6)

163.0(0.1)

Phenytoin

8.0(0.3)

32.0

(0.09)

12.0

(1.0)

42.3

(0.1)

9.0(1.0)

48.1

(0.1)

13.2

(15.4)

147.0(0.5)

9.2(0.5)

37.2

(0.08)

6.2(2.6)

126.0(0.4)

ND,no

tdeterm

ined.

aDataarefrom

48-hourincubatio

nwith

rifampin.

CYP3A

4activ

itywas

measuredusingmidazolam

asthesubstrateandthemRNA

change

was

measuredusingreal-tim

ePCRwith

TaqMan

probes.Datashow

naremean(S.E.)of

threedeterm

inations.

252 Dixit et al.

at ASPE

T Journals on February 4, 2020

dmd.aspetjournals.org

Dow

nloaded from

Comparative statistics using a t test were used to compare the data,and a P value less than 0.05 was considered significant. The P valuesare included in the Supplemental Table 2. Significant differences wereobserved between the HepatoPac model compared with monoculturesof hepatocytes. As shown in Fig. 1A, expression levels of CYP3A4 inthe HepatoPac model were 2- to 200-fold higher than that in themonocultures. Human donor lot BPB had the largest disparity in theexpression of CYP3A4 between the two culture models. This could beattributed to the low basal expression of CYP3A4 in this donor in themonoculture model of hepatocyte induction. The expression ofCYP2C9 and CYP2C19 was only 2- to 3-fold higher in the HepatoPacmodel in all lots; however, the expression of CYP2C8 was 4- to 22-foldhigher in BPB and NON lots. Basal expression of CYP2D6, CYP2E1,UGT1A1, and SULT1B1 was significantly higher in the HepatoPaccultures than in the hepatocyte monoculture. The expression ofarylacetamide deacetylase (AADAC), an enzyme responsible for themetabolism of rifampin (Nakajima et al., 2011), was also significantlyhigher in the HepatoPac cultures compared with the monocultures.HepatoPac cultures also had higher expression of BSEP, MDR3,

OCT1, OATP1A2, and OATP1B3 than that in the monocultured cellsas shown in Fig. 1B. Significant interdonor differences were observedin the expression of some uptake transporters. OATP1B3 was 17- and11-fold higher in NON and BPB lots, respectively. However, in thedonor lot Hu1624, expression of OATP1B3 was only about 3-foldgreater in the HepatoPac cultures. OCT1 expression in lots BPB andNON was .2-fold higher in HepatoPac cultures compared with thehepatocyte monocultures. MDR1 expression levels were.2-fold lower

in the HepatoPac model compared with the 2D model in lots 1624 andBPB.Expression of nuclear receptors, including those implicated in

inductive response by xenobiotics such as PXR, AHR (arylhydrocarbonreceptor), GR (glucocorticoid receptor), and PPAR (peroxisome pro-liferative receptor), were not substantially different in the HepatoPacmodel (Supplemental Fig. 1). Expression of CAR was higher in theHepatoPac model in lot BPB but was similar to the monocultures for theother two lots. Levels of NCOR1 were higher in all three lots of thecocultured system compared with the 2D model.The expression of several drug metabolism genes was lower in the

HepatoPac model than in the monoculture models (Fig. 1A). Genetargets GSTA2 aswell as certain sulfotransferase geneswere lower in theHepatoPac model. Efflux transporters such as BCRP,MRP4, andMRP6were lower in the HepatoPac model in all three lots studied (Fig. 1B).Basal Activity of CYP3A4 and CYP2B6 in HepatoPac Cultures

following 72-Hour Incubation. The activity of CYP3A4 and CYP2B6in HepatoPac was determined using midazolam and bupropion,respectively, as probe substrates, following treatment with 0.1%DMSOin serum-free media for 72 hours. After the treatment period, CYP3A4activity was monitored over 8 days in culture in serum-containingmaintenance media (Fig. 2A). Immediately after switching the culturesto serum-containing medium, the CYP3A4 activity increased about 4-fold from 0–96 hours post-treatment. The formation of hydroxymida-zolam did not show appreciable change over 96–192 hours of culture inserum-containing media. All three lots followed the same pattern ofchange in CYP3A4 activity. Change in the basal activity of CYP2B6

Fig. 1. cDNA from vehicle-treated cells from both monocultures and HepatoPac cultures of hepatocytes were used to determine the basal expression of drug dispositiongenes. Gene expression in HepatoPac cultures was compared with that of monocultured hepatocytes, and the data were expressed as relative change over monocultured cells.(A) Relative expression of Phase I and Phase II enzymes. (B) Relative expression of uptake and efflux transporters in three lots of hepatocytes. Data are shown as mean (S.D.)of three determinations.

Fig. 2. Change in basal activity of CYP3A4and CYP2B6 in HepatoPac cultures measuredeither with midazolam or bupropion as thesubstrate in three lots of hepatocytes. (A) and(B) show the change in CYP3A4 and CYP2B6activity, respectively, following a 72-hourtreatment with 0.1% DMSO in serum-freemedia. The x-axis represents time after a 72-hour treatment with 0.1% DMSO. Data areshown as mean (S.D.) of three independentdeterminations.

Hepatocyte Coculture System to Assess P450 Induction 253

at ASPE

T Journals on February 4, 2020

dmd.aspetjournals.org

Dow

nloaded from

following an induction assay in serum-free media was also monitored(Fig. 2B). The formation of hydroxybupropion remained consistentover the 7-day recovery period indicating that the HepatoPac culturesmaintain metabolic activity for at least 2 weeks after receipt.Comparison of CYP3A4 Induction in HepatoPac Cultures and

Hepatocyte Monocultures by Prototypical Inducers. BothHepatoPacand hepatocyte monocultures were treated with the prototypical inducersrifampin (0.001–10mM), phenytoin (1–200mM), and phenobarbital (10–1000 mM) for 72 hours and fold change in P450 activity and geneexpression were determined. The dose response curves for activity andmRNA from the HepatoPac (Figs. 3 and 4) and from the hepatocytemonocultures (Figs. 5 and 6)were fitted to the sigmoidal Emaxmodel. TheEC50 for rifampin-mediated induction of CYP3A4 mRNA in theHepatoPac cultures was 0.09 mM in the three lots, and the EC50 inmonocultures for CYP3A4 gene expression ranged from 0.24 to 1.2 mMin the three lots tested (Table 2). The EC50 values for phenytoin andphenobarbital-mediated induction of CYP3A4 gene expression rangedfrom 30 to 75mMand 85 to 112mM, respectively, inHepatoPac cultures.Overall, the Emax and EC50 for induction of CYP3A4 activity was inagreement with the mRNA data.

CYP2C9 and CYP2C8 Induction in HepatoPac and Monocul-tures of Hepatocytes. Preliminary studies (data not shown) showedthat a 48-hour incubation time was insufficient to achieve significantinduction of CYP2C9 mRNA. Hence we decided to employ a longer(72-hour) incubation period to assess the induction of all genesincluding CYP2C9. We compared the induction of CYP2C9 andCYP2C8 in the HepatoPac cultures to that in the monoculturesfollowing treatment with prototypical inducers. HepatoPac culturesshowed a robust increase in CYP2C9 expression and activity followinga 72-hour treatment with rifampin and phenobarbital (Figs. 3 and 4).Rifampin-mediated induction of CYP2C9 in HepatoPac showed anEmax of 3- to 4.5-fold for mRNA and 3- to 3.5-fold for activity (Table 3).The EC50 of rifampin in lot NONwas about 10-fold lower in HepatoPacversus the monoculture. Phenobarbital caused about a 3-fold change inmRNA of CYP2C9 and about 6- to 7-fold increase in activity ofCYP2C9. Rifampin and phenobarbital caused only a modest change inCYP2C9 expression and activity in monocultures of hepatocytes. Thefold response in CYP2C9 by rifampin and phenobarbital in thehepatocyte monocultures was often below the threshold and couldnot be fit to a sigmoidal Emax model (5 and 6). Phenytoin did not cause

Fig. 3. Dose-response curves for CYP3A4, CYP2B6, CYP2C8, and CYP2C9 mRNA in HepatoPac following treatment with rifampin, phenobarbital, and phenytoin for 72hours in three lots of hepatocytes. Graphs in rows A, B, C, and D show dose-response curves for CYP3A4, CYP2B6, CYP2C8, and CYP2C9, respectively. Data shown aremeans (S.D.) of three determinations in three lots of human hepatocytes.

254 Dixit et al.

at ASPE

T Journals on February 4, 2020

dmd.aspetjournals.org

Dow

nloaded from

an increase in CYP2C9 mRNA or activity in both the HepatoPac andthe hepatocytemonocultures. The Emax for induction of CYP2C8mRNAby rifampin and phenobarbital was 2- to 3-fold higher in the HepatoPacthan in monocultures of hepatocytes in lots BPB and NON (Table 4).

The Emax in lot Hu1624 was higher for rifampin but similar forphenobarbital in the HepatoPac model than in monocultures. Phenytoindid not show a robust inductive response for CYP2C8 in Hu1624 andBPB lots in the hepatocyte monocultures. Phenytoin did cause

Fig. 4. Dose-response curves for CYP3A4 (row A) and CYP2C9 (row B) activity in HepatoPac following a 72-hour treatment with rifampin, phenobarbital, and phenytoin inthree lots of hepatocytes. Data shown are means (S.D.) of three independent determinations.

Fig. 5. Dose-response curves for CYP3A4, CYP2B6, and CYP2C8 mRNA in three lots of hepatocytes. Monocultures of hepatocytes were treated with rifampin,phenobarbital, and phenytoin for 72 hours. Rows A, B, and C show dose-response curves for CYP3A4, CYP2B6, and CYP2C8, respectively. Data shown are means (S.D.)of three independent determinations.

Hepatocyte Coculture System to Assess P450 Induction 255

at ASPE

T Journals on February 4, 2020

dmd.aspetjournals.org

Dow

nloaded from

induction of CYP2C8 gene in monocultures of lot NON; however, thedose-response curve was not robust, and hence could not be fit to asigmoidal Emax model.Induction of CYP2B6 Expression in HepatoPac and Hepatocyte

Monocultures. Induction of CYP2B6 expression was evaluated in bothculture models. In both models, all compounds tested showedrobust CYP2B6 expression in a concentration-dependent manner(Figs. 3 and 5). Induction of CYP2B6 was either similar to or about2-fold greater in HepatoPac cultures than in the monoculturemodel. Phenobarbital and phenytoin showed higher Emax than didrifampin in the HepatoPac cultures; however, the Emax values weresimilar for all three compounds in the monocultures (Table 5).Degradation kinetics of CYP3A4 and CYP2B6 protein. In this

experiment, HepatoPac cultures were first treated with either 0.1% DMSOor 10 mM rifampin in serum-free–induction media for 72 hours to induceCYP3A4 and CYP2B6 protein and mRNA. The cultures were maintainedin the serum-containingmedia after the 72 hours of drug treatment to avoid

the loss of activity that occurs when treatment in serum-free media isprolonged. The fold change in activity and message were measured atseveral time points following the 72-hour inductive period. At 0 hours aftertreatment with rifampin, CYP3A4 activity was induced 6- to 10-fold andthe mRNAwas induced about 7- to 11-fold in three lots. CYP2B6 activityand mRNA were also induced in two hepatocyte lots following treatmentwith 10 mM rifampin. The loss of activity and mRNA was plotted againsttime and the data were fitted to a linear regressionmodel (Fig. 7 and 8). Theslope of the line was then used to calculate the half-life (t1/2). The t1/2 ofCYP3A4 protein was determined to be 43–56 hours and that for CYP3A4mRNA was 21–31 hours (Table 6). The degradation t1/2 for CYP2B6protein was 70 hours and that for mRNA was 36–40 hours in the two lots(Table 7). These data show that mRNAhas a shorter t1/2 than the protein forboth P450s. The t1/2 values for both functional protein and mRNA showedminimal interdonor variability. The expression of CYP2D6 in rifampin-treated and vehicle-treated cells was determined as a negative control in thekdeg experiments. The DCT values of CYP2D6 in all treatments remained

Fig. 6. Dose-response curves for change in CYP3A4 (row A) and CYP2C9 (row B) activity in hepatocyte monocultures following a 72-hour treatment with rifampin,phenobarbital, and phenytoin in three lots of hepatocytes. Data shown are means (S.D.) of three independent determinations.

TABLE 3

Emax and EC50 for CYP2C9 activity and mRNA after treatment with rifampin, phenobarbital, and phenytoin for 72 hours in either HepatoPac or monoculturesin serum-free media.

NON BPB Hu1624

HepatoPac Monoculture HepatoPac Monoculture HepatoPac Monoculture

Emax EC50 Emax EC50 Emax EC50 Emax EC50 Emax EC50 Emax EC50

mM mM mM mM mM mM

mRNARifampin 4.57 (0.11) 0.13 (0.08) 3.37 (0.37) 1.56 (0.25) 4.13 (0.22) 0.99 (0.1) ND ND 3.54 (0.10) 0.23 (0.09) ND NDPhenobarbital 4.0 (0.25) 254 (0.09) 3.4 (0.33) 389.2 (0.12) 3.3 (0.55) 112.8 (0.28) ND ND ND ND ND NDPhenytoin ND ND ND ND ND ND ND ND ND ND ND ND

ActivityRifampin 3.5 (0.1) 0.1 (0.1) 3.7 (0.2) 1.2 (0.1) 3.3 (0.3) 1.2 (0.2) ND ND 3.1 (0.1) 0.2 (0.1) ND NDPhenobarbital 5.1 (0.6) 417.5 (0.1) 3.7 ND 7.2 (1.6) 848.9 (0.2) ND ND 5.8 (0.8) 449.4 (0.2) ND NDPhenytoin ND ND ND ND ND ND ND ND ND ND ND ND

ND, curves could not be fit to a sigmoidal Emax model.CYP2C9 activity was measured using diclofenac as the substrate and the mRNA change was measured using real-time PCR with TaqMan probes. Data shown are mean (S.E.) of three

determinations.

256 Dixit et al.

at ASPE

T Journals on February 4, 2020

dmd.aspetjournals.org

Dow

nloaded from

consistent over the recovery phase of the experiment (7 days). The foldchange in CYP2D6 in rifampin-treated cells compared with the vehicle-treated cells was close to unity at all the time points in the study. These dataindicate that the change in the expression of inducible genes such asCYP3A4and CYP2B6 in the HepatoPac system is not an artifact of the in vitromodel.In Vivo EC50 for Rifampin. PBPK modeling using Simcyp was

used to predict the free steady-state Cmax of rifampin after oral dosing.The simulated Cmax of rifampin at the 600-mg dose is similar to reportedclinical Cmax of rifampin at 600 mg (Garg et al., 2013). The change inmidazolam AUC after treatment with varying oral doses of rifampinwas reported in Kharasch et al. (2011). The correlation between themidazolam AUC change and the predicted free Cmax of rifampin showsthat the in vivo EC50 for rifampin is 0.05 mM (Fig. 9).Nonspecific Binding of Rifampin. The concentration of rifampin

was measured in the media after incubation with murine fibroblasts.Figure 10 shows the % change in rifampin over the incubation period.Rifampin was also incubated in induction media with monoculturedhuman hepatocytes at 4�C to prevent compound loss from metabolism.In both systems the change in the concentration of rifampin was lessthan 25% from the nominal, indicating minimal nonspecific binding ofthe compound to either cell type.

Discussion

In the current work we report a comparative analysis of inductiveresponse between the HepatoPac model and monocultured models aswell as the relative expression of several metabolic enzymes, trans-porters, and transcription factors in the two culture models using thesame donors of cryopreserved hepatocytes.The magnitude and the potency of inductive response in the

HepatoPac cultures are in agreement with those observed in themonocultures, albeit with some exceptions. The HepatoPac modelappears to be more sensitive to rifampin, as was observed by the shift inthe potency of rifampin in HepatoPac versus monocultures. In lots

NON and BPB, but not in Hu1624, the EC50 of rifampin for CYP3A4was about 10-fold lower in HepatoPac, thus displaying an equal orgreater sensitivity to the PXR agonist. Data from our laboratory andfrom others (Zhang et al., 2014) have shown that the EC50 of rifampin inmonocultured or in sandwich-cultured hepatocytes is in the range of0.5–1 mM. Rifampin showed minimal nonspecific binding to media orto cells in the monoculture or in the coculture format (Fig. 10),indicating that the shift in EC50 cannot be attributed to change in the freefraction. Clinical data shows that rifampin can cause significant DDIwith midazolam at doses as low as 5 mg/day given for 5 days (Kharaschet al., 2011). We used the rifampin model in the Simcyp simulator topredict free plasma concentrations of rifampin at the doses used in thestudy. The fCmax was correlated with the decrease in the midazolamAUC using a sigmoidal Emax model (Fig. 9). The correlation shows thatthe apparent clinical EC50 of rifampin could be as low as 0.05 mM.These analyses suggest that HepatoPac may be a more sensitive modelthan monocultured hepatocytes for estimating the potency of rifampin.Future studies that include incubation time longer than the traditional48–72 hours could be useful to determine the inductive potential ofweaker inducers. However, it is essential to understand any changes inthe metabolic activity after prolonged exposure to serum-free media.The potency of rifampin could be influenced by the differences in the

activity and the expression of transporters in the two culture models.Rifampin is a substrate as well as an inhibitor of OATP1B1 (Hiranoet al., 2006; Niemi et al., 2006). It was shown by Tirona et al. (2003)that PXR activation occurs at lower rifampin concentrations inOATP1B1-expressing cells compared with those that are OATP1B1-naïve. We found that the expression of OATP1B1 in the HepatoPacmodel is not significantly different from that in monocultures.However, the expression of OATP1B3, which is involved in rifampindisposition (Yamaguchi et al., 2011), was higher in HepatoPac cultures.Studies have shown that hepatocytes cultured for 7 days showed higherOATP1B3 activity (Zhu et al., 2014). Interestingly, expression ofOATP1B3 was only higher in lots BPB and NON and not in Hu1624.

TABLE 4

Emax and EC50 for CYP2C8 mRNA after treatment with rifampin, phenobarbital, and phenytoin for 72 hours in either HepatoPac or monocultures in serum-free media.

NON BPB Hu1624

HepatoPac Monoculture HepatoPac Monoculture HepatoPac Monoculture

Emax EC50 Emax EC50 Emax EC50 Emax EC50 Emax EC50 Emax EC50

mM mM mM mM mM mM

Rifampin 10 (0.18) 0.12 (0.03) 6.0 (0.17) 0.5 (0.07) 5.0 (0.41) 0.2 (0.18) ND ND 9.2 (0.43) 0.2 (0.08) 3.8 (0.11) 0.2 (0.1)Phenobarbital 11.85 (0.64) 219.8 (0.061) 6.8 (0.36) 215.6 (0.075) 6.6 (0.59) 159.7 (0.11) 3.4 (0.15) 80.7 (0.09) 7.2 (0.37) 97.7 (0.06) 6.6 (0.95) 178.3 (0.21)Phenytoin 10.5 (0.97) 69.1 (0.11) ND ND 5.8 (0.44) 17.4 (0.16) 3.7 (0.75) 117.7 (0.23) 6.2 (0.42) 16.1 (0.14) ND ND

ND, curves could not be fit to a sigmoidal Emax model.mRNA change was measured using real-time PCR with TaqMan probes. Data shown are mean (S.E.) of three determinations.

TABLE 5

Emax and EC50 for CYP2B6 mRNA after treatment with rifampin, phenobarbital, and phenytoin for 72 hours in either HepatoPac or monocultures in serum-free media.

NON BPB Hu1624

HepatoPac Monoculture HepatoPac Monoculture HepatoPac Monoculture

Emax EC50 Emax EC50 Emax EC50 Emax EC50 Emax EC50 Emax EC50

mM mM mM mM mM mM

Rifampin 10.7 (0.67) 0.4 (0.09) 6.7 (0.44) 0.8 (0.15) 7.0 (0.28) 0.3 (0.07) ND ND 7.3 (0.29) 0.3 (0.07) 5.3 (0.11) 0.5 (0.047)Phenobarbital 13.1 (1.25) 251.4 (0.1) 7.9 (0.51) 131.5 (0.1) 16.9 (4.8) 221.4 (0.33) 8.7 (1.3) 217.2 (0.13) 14.9 (1.2) 143.1 (0.1) 6.5 (0.24) 91.4 (0.065)Phenytoin 18.3 (9.6) 66.7 (0.83) ND ND 13.3 (1.03) 13.3 (0.12) 6.0 (0.39) 8.5 (0.162) 9.7 (0.46) 5.0 (0.09) 6.7 (0.21) 7.5 (0.07)

ND, curves could not be fit to a sigmoidal Emax model.mRNA change was measured using real-time PCR with TaqMan probes. Data shown are mean (S.E.) of three determinations.

Hepatocyte Coculture System to Assess P450 Induction 257

at ASPE

T Journals on February 4, 2020

dmd.aspetjournals.org

Dow

nloaded from

This corresponds to the observation that the EC50 of rifampin wasshifted in BPB and NON but not in Hu1624. Further investigations withclinical inducers that are substrates of uptake transporters are currentlybeing conducted in our laboratory.The HepatoPac model also showed induction of a CAR-responsive

gene, CYP2B6. All three inducers showed a robust increase in the geneexpression of CYP2B6 in both culture models. The potency of rifampinin lot NONwas about 2-fold lower for this gene in HepatoPac. CYP2B6is preferentially regulated by hCAR (Faucette et al., 2006), whereasrifampin preferentially activates PXR (Faucette et al., 2007). The lackof a more significant shift in potency for CYP2B6 induction by rifampincould result from its activation of only one regulatory pathway.Determining an Emax and EC50 for CYP2C9 in hepatocyte-induction

assays is challenging since the magnitude of response is often close to orbelow the threshold (Yajima et al., 2014). We attempted to use theHepatoPac model to derive a dose-response relationship for CYP2C9with prototypical inducers. The HepatoPac cultures show a higher and

consistent response for both of the CYP2C genes compared with themonocultures. In the HepatoPac model we observed that a 3.5-foldchange in mRNA and up to a 7-fold change in activity withphenobarbital, a dual PXR/CAR transactivator (Sahi et al., 2009).Phenobarbital showed the highest inductive response for CYP2C9expression and activity in the HepatoPac model, perhaps owing to itsdual activation of PXR and CAR. Phenytoin, a preferential hCARtransactivator (Küblbeck et al., 2011) failed to significantly increaseCYP2C9 mRNA or activity in either culture model. A previous reportby Sahi et al. (2009) has also shown that phenytoin does not induceCYP2C9. Although the dynamic range of the inductive response is notas large as that for other P450 enzymes, these data suggest thatHepatoPac can be a useful tool to assess CYP2C9 induction forcompounds with complex DDIs, such as ritonavir. Ritonavir showsclinically significant induction of warfarin clearance (Knoell et al., 1998);however, this induction of CYP2C9 is not adequately captured in vitro(Dixit et al., 2007).

Fig. 7. Kinetics of degradation of CYP3A4 protein and mRNA in three lots of human hepatocytes. HepatoPac cultures were treated with 10 mM rifampin for 72 hours, andthe recovery of CYP3A4 activity and mRNA to baseline was measured. All assays were run triplicate. Open circles (s) represent CYP3A4 activity and filled circles (d)represent CYP3A4 mRNA. Data shown are means (S.D.) of three determinations. The data were fit to linear regression, and slope was used to calculate half-life of CYP3A4protein and mRNA. Solid lines show the regression fit.

TABLE 6

kdeg and t1/2 of CYP3A4 protein and mRNA in three lots of human hepatocytes after induction with 10 mM rifampin for 72 hours. Fold change inCYP3A4 activity or mRNA over DMSO control was monitored for 7days. Data were fitted to linear regression and the slope of the line was

converted to the half-life. Data are shown as the mean (S.E.) of three independent determinations.

NON Hu1624 BPB

mRNA Activity mRNA Activity mRNA Activity

kdeg (1/hr) 0.0261 (0.00124) 0.0144 (0.0015) 0.0323 (0.0014) 0.0162 (0.00123) 0.0223 (0.0011) 0.0123 (0.00153)t1/2 (hr) 26.6 49 21.5 43 31.1 56R2 0.99 0.97 1 0.95 0.99 0.94

258 Dixit et al.

at ASPE

T Journals on February 4, 2020

dmd.aspetjournals.org

Dow

nloaded from

CYP2C8 induction has been quantified by several researchers, butthe inductive response was inconsistent in different donors. In theHepatoPac cultures, CYP2C8 was inducible in all the lots and by allthe inducers used in the current work. However, in the monoculturesCYP2C8 gene was inducible by phenytoin in only one out of the threelots tested. CYP2C8 is regulated by multiple receptors including bothPXR and CAR (Chen and Goldstein, 2009). Hence the reasons behindthe lack of induction ofCYP2C8 by phenytoin inmonocultures are unclear.Accurate estimation of degradation rates of P450 enzymes is

essential for improved prediction of induction- and inhibition-baseddrug interactions.We harnessed the longevity of the HepatoPac culturesso as to allow measurement of the depletion kinetics of both mRNAand protein function following an induction period. Maintaining thecultures in serum-containing media during the recovery phase allowedus to retain the basal activity (Fig. 2) and follow the recovery of enzymeback to baseline. Having used this technique, we report the average

degradation t1/2 of the CYP3A4 mRNA to be 26 hours and that ofprotein to be 49 hours. To understand if the half-life of PXR- and CAR-regulated P450s could be different, we also investigated the t1/2 ofCYP2B6. The half-life of CYP2B6 mRNA (38 hours) appears to belonger than that of CYP3A4. The depletion kinetics of CYP2B6 proteinas measured by functional activity was;70 hours, which is longer thanthat for CYP3A4, indicating differential kinetics. The reason behindthese differences in half-lives of different P450s is currently unclear.The Simcyp PBPK model uses 32 hours as the t1/2 of CYP2B6 in theliver, which was measured using immunoreactive CYP2B6 in liverslices from a single donor (Renwick et al., 2000). The turnover ratereported here could prove useful for development of better PBPKmodels aimed at predicting DDI attributable to induction, inactivation,or suppression of CYP2B6.Several clinical studies have attempted to establish a kdeg for

CYP3A4. A clinical study with rifampin as the inducer studied the

TABLE 7

kdeg and t1/2 of CYP2B6 mRNA and functional protein in three lots of human hepatocytes after induction with 10 mMrifampin for 72 hours. Fold change in CYP2B6 activity or mRNA over DMSO control was monitored for 7days.

NON Hu1624 BPB

mRNA Activity mRNA Activity mRNA Activity

kdeg (1/hr) 0.0167 (0.0026) 0.0099 (0.001) 0.01809 (0.0028) ND 0.0191 (0.0021) 0.0103 (0.001)t1/2 (hr) 41 70 38 ND 36 67R2 0.851 0.896 0.859 ND 0.922 0.937

ND, not determined.Data were fitted to linear regression and the slope of the line was converted to the t1/2. Data are shown as the mean (S.E.) of three

independent determinations.

Fig. 8. Degradation kinetics of CYP2B6 mRNA and protein in three lots of human hepatocytes. HepatoPac cultures were treated with 10 mM rifampin for 72 hours and foldchange in CYP2B6 activity and mRNA were measured over a recovery period of 7 days. All assays were run in triplicate. The open circles (s) represent CYP2B6 activityand filled circles (d) represent CYP2B6 mRNA. Data shown are means (S.D.) of three determinations. The data were fit to linear regression and the resulting slope was usedto calculate t1/2 and kdeg. Solid lines show the regression fit.

Hepatocyte Coculture System to Assess P450 Induction 259

at ASPE

T Journals on February 4, 2020

dmd.aspetjournals.org

Dow

nloaded from

depletion kinetics of CYP3A4 using midazolam as a probe for CYP3A4(Reitman et al., 2011). The half-life of CYP3A4 protein from this studywas suggested to be 7.7 days. This is longer than that reported in our invitro study and can be explained by insufficient time points in theclinical study. However, studies with other inducers, such as St. JohnsWort (Imai et al., 2008), carbamazepine (Magnusson et al., 2008), andmethadone (Yang et al., 2008), have reported half-lives more akin tothat observed in the current study. The change in urinary 6b-hydroxycortisol-to-cortisol ratio has also been used to define the t1/2and was reported to be 3 days (Yang et al., 2008). Recently, Ramsdenet al. (2015) used the HepatoPac cultures to determine the kdeg ofCYP3A4 protein using small-interfering RNA and interleukin 6knockdown of the gene. This study reported the half-life of CYP3A4protein in the range of 22–39 hours, which is similar to that reportedhere. However, in the study by Ramsden et al., the t1/2 of CYP3A4mRNA was not obtained by direct measurement in all the donors andwas estimated by modeling the enzyme recovery data. Since thedegradation rate of protein or mRNA is considered a system parameterfor PBPK modeling, accurate measurement of this parameter isessential for DDI predictions.Finally, this report shows the utility of hepatocyte cocultures for

determining the induction potential of several drug-metabolizingP450s. Long-term coculture models maintain the functionality of drugtransporters and metabolic enzymes, which are important determinantsof intracellular drug concentrations. In spite of these advantages,challenges such as low throughput and high cost makes the HepatoPac

model less desirable than monoculture models for routine assessment ofDDI. Overall, we demonstrate that this model can be used to performextended drug treatments that could include an induction/inactivationand recovery periods to evaluate complex DDI.

Authorship ContributionsParticipated in research design: Dixit, Hariparsad, Moore.Conducted experiments: Dixit, Moore, Tsao.Performed data analysis: Dixit, Moore.Wrote or contributed to the writing of the manuscript: Dixit, Hariparsad,

Moore.

References

Bachmann A, Moll M, Gottwald E, Nies C, Zantl R, Wagner H, Burkhardt B, Sánchez JM,Ladurner R, Thasler W, Damm G and Nussler A (2015) 3D Cultivation Techniques for PrimaryHuman Hepatocytes. Microarrays 4:64–83.

Berthiaume F, Moghe PV, Toner M, and Yarmush ML (1996) Effect of extracellular matrixtopology on cell structure, function, and physiological responsiveness: hepatocytes cultured ina sandwich configuration. FASEB J 10:1471–1484.

Buxboim A, Ivanovska IL, and Discher DE (2010) Matrix elasticity, cytoskeletal forces andphysics of the nucleus: how deeply do cells ‘feel’ outside and in? J Cell Sci 123:297–308.

Chan TS, Yu H, Moore A, Khetani SR, Tweedie D, and Tweedie D (2013) Meeting the challengeof predicting hepatic clearance of compounds slowly metabolized by cytochrome P450 using anovel hepatocyte model, HepatoPac. Drug Metab Dispos 41:2024–2032.

Chen Y and Goldstein JA (2009) The transcriptional regulation of the human CYP2C genes. CurrDrug Metab 10:567–578.

Dixit V, Hariparsad N, Li F, Desai P, Thummel KE, and Unadkat JD (2007) Cytochrome P450enzymes and transporters induced by anti-human immunodeficiency virus protease inhibitors inhuman hepatocytes: implications for predicting clinical drug interactions. Drug Metab Dispos35:1853–1859.

Einolf HJ, Chen L, Fahmi OA, Gibson CR, Obach RS, Shebley M, Silva J, Sinz MW, UnadkatJD, and Zhang L, et al. (2014) Evaluation of various static and dynamic modeling methods topredict clinical CYP3A induction using in vitro CYP3A4 mRNA induction data. Clin Phar-macol Ther 95:179–188.

Fahmi OA, Kish M, Boldt S, and Obach RS (2010) Cytochrome P450 3A4 mRNA is a morereliable marker than CYP3A4 activity for detecting pregnane X receptor-activated induction ofdrug-metabolizing enzymes. Drug Metab Dispos 38:1605–1611.

Faucette SR, Sueyoshi T, Smith CM, Negishi M, Lecluyse EL, and Wang H (2006) Differentialregulation of hepatic CYP2B6 and CYP3A4 genes by constitutive androstane receptor but notpregnane X receptor. J Pharmacol Exp Ther 317:1200–1209.

Faucette SR, Zhang TC, Moore R, Sueyoshi T, Omiecinski CJ, LeCluyse EL, Negishi M,and Wang H (2007) Relative activation of human pregnane X receptor versus constitutiveandrostane receptor defines distinct classes of CYP2B6 and CYP3A4 inducers. J PharmacolExp Ther 320:72–80.

Garg V, Chandorkar G, Yang Y, Adda N, McNair L, Alves K, Smith F, and van Heeswijk RP(2013) The effect of CYP3A inhibitors and inducers on the pharmacokinetics of telaprevir inhealthy volunteers. Br J Clin Pharmacol 75:431–439.

Goldstein JA (2001) Clinical relevance of genetic polymorphisms in the human CYP2C sub-family. Br J Clin Pharmacol 52:349–355.

Gómez-Lechón MJ, Jover R, Donato T, Ponsoda X, Rodriguez C, Stenzel KG, Klocke R, Paul D,Guillén I, and Bort R, et al. (1998) Long-term expression of differentiated functions in he-patocytes cultured in three-dimensional collagen matrix. J Cell Physiol 177:553–562.

Hirano M, Maeda K, Shitara Y, and Sugiyama Y (2006) Drug-drug interaction between pit-avastatin and various drugs via OATP1B1. Drug Metab Dispos 34:1229–1236.

Imai H, Kotegawa T, Tsutsumi K, Morimoto T, Eshima N, Nakano S, and Ohashi K (2008) Therecovery time-course of CYP3A after induction by St John’s wort administration. Br J ClinPharmacol 65:701–707.

Inoue K, Inazawa J, Suzuki Y, Shimada T, Yamazaki H, Guengerich FP, and Abe T (1994)Fluorescence in situ hybridization analysis of chromosomal localization of three human cy-tochrome P450 2C genes (CYP2C8, 2C9, and 2C10) at 10q24.1. Jpn J Hum Genet 39:337–343.

Kharasch ED, Francis A, London A, Frey K, Kim T, and Blood J (2011) Sensitivity of in-travenous and oral alfentanil and pupillary miosis as minimal and noninvasive probes forhepatic and first-pass CYP3A induction. Clin Pharmacol Ther 90:100–108.

Khetani SR and Bhatia SN (2008) Microscale culture of human liver cells for drug development.Nat Biotechnol 26:120–126.

Knoell KR, Young TM, and Cousins ES (1998) Potential interaction involving warfarin andritonavir. Ann Pharmacother 32:1299–1302.

Küblbeck J, Jyrkkärinne J, Molnár F, Kuningas T, Patel J, Windshügel B, Nevalainen T, LaitinenT, Sippl W, and Poso A, et al. (2011) New in vitro tools to study human constitutive androstanereceptor (CAR) biology: discovery and comparison of human CAR inverse agonists. MolPharm 8:2424–2433.

Magnusson MO, Dahl ML, Cederberg J, Karlsson MO, and Sandström R (2008) Pharmacody-namics of carbamazepine-mediated induction of CYP3A4, CYP1A2, and Pgp as assessed byprobe substrates midazolam, caffeine, and digoxin. Clin Pharmacol Ther 84:52–62.

Nakajima A, Fukami T, Kobayashi Y, Watanabe A, Nakajima M, and Yokoi T (2011) Humanarylacetamide deacetylase is responsible for deacetylation of rifamycins: rifampicin, rifabutin,and rifapentine. Biochem Pharmacol 82:1747–1756.

Niemi M, Kivistö KT, Diczfalusy U, Bodin K, Bertilsson L, Fromm MF, and Eichelbaum M(2006) Effect of SLCO1B1 polymorphism on induction of CYP3A4 by rifampicin. Pharma-cogenet Genomics 16:565–568.

Ramsden D, Tweedie DJ, St George R, Chen LZ, and Li Y (2014) Generating an in vitro-in vivocorrelation for metabolism and liver enrichment of a hepatitis C virus drug, faldaprevir, using arat hepatocyte model (HepatoPac). Drug Metab Dispos 42:407–414.

Fig. 9. Correlation between the predicted clinical Cmax of rifampin and the decreasein midazolam exposure. The clinical exposure was simulated using the Simcypdefault model and parameters for rifampin. Unbound Cmax and the change inmidazolam AUC were fitted to the sigmoidal Emax curve using GraphPad Prism. Theapparent potency (EC50) of rifampin using these data are predicted to be 0.05 mM.

Fig. 10. Nonspecific binding of rifampin (1 mM) to murine fibroblasts or humanhepatocytes. Rifampin was incubated with fibroblasts or hepatocytes in inductionmedia at 37�C or 4�C, respectively. The change in concentration of rifampin ispresented as % change from nominal. Data are shown as a mean 6 S.D. (n = 3).

260 Dixit et al.

at ASPE

T Journals on February 4, 2020

dmd.aspetjournals.org

Dow

nloaded from

Ramsden D, Zhou J, and Tweedie DJ (2015) Determination of a Degradation Constant forCYP3A4 by Direct Suppression of mRNA in a Novel Human Hepatocyte Model, HepatoPac.Drug Metab Dispos 43:1307–1315.

Reitman ML, Chu X, Cai X, Yabut J, Venkatasubramanian R, Zajic S, Stone JA, Ding Y, WitterR, and Gibson C, et al. (2011) Rifampin’s acute inhibitory and chronic inductive drug inter-actions: experimental and model-based approaches to drug-drug interaction trial design. ClinPharmacol Ther 89:234–242.

Renwick AB, Watts PS, Edwards RJ, Barton PT, Guyonnet I, Price RJ, Tredger JM, Pelkonen O,Boobis AR, and Lake BG (2000) Differential maintenance of cytochrome P450 enzymes incultured precision-cut human liver slices. Drug Metab Dispos 28:1202–1209.

Sahi J, Shord SS, Lindley C, Ferguson S, and LeCluyse EL (2009) Regulation of cytochromeP450 2C9 expression in primary cultures of human hepatocytes. J Biochem Mol Toxicol 23:43–58.

Tirona RG, Leake BF, Wolkoff AW, and Kim RB (2003) Human organic anion transportingpolypeptide-C (SLC21A6) is a major determinant of rifampin-mediated pregnane X receptoractivation. J Pharmacol Exp Ther 304:223–228.

Trask OJ, Jr, Moore A, and LeCluyse EL (2014) A micropatterned hepatocyte coculture model forassessment of liver toxicity using high-content imaging analysis. Assay Drug Dev Technol 12:16–27.

Wang WW, Khetani SR, Krzyzewski S, Duignan DB, and Obach RS (2010) Assessment of amicropatterned hepatocyte coculture system to generate major human excretory and circulatingdrug metabolites. Drug Metab Dispos 38:1900–1905.

Yajima K, Uno Y, Murayama N, Uehara S, Shimizu M, Nakamura C, Iwasaki K, Utoh M,and Yamazaki H (2014) Evaluation of 23 lots of commercially available cryopreserved he-patocytes for induction assays of human cytochromes P450. Drug Metab Dispos 42:867–871.

Yamaguchi H, Takeuchi T, Okada M, Kobayashi M, Unno M, Abe T, Goto J, Hishinuma T,Shimada M, and Mano N (2011) Screening of antibiotics that interact with organic anion-transporting polypeptides 1B1 and 1B3 using fluorescent probes. Biol Pharm Bull 34:389–395.

Yang J, Liao M, Shou M, Jamei M, Yeo KR, Tucker GT, and Rostami-Hodjegan A (2008)Cytochrome p450 turnover: regulation of synthesis and degradation, methods for de-termining rates, and implications for the prediction of drug interactions. Curr Drug Metab9:384–394.

Zamek-Gliszczynski MJ, Mohutsky MA, Rehmel JL, and Ke AB (2014) Investigational small-molecule drug selectively suppresses constitutive CYP2B6 activity at the gene transcriptionlevel: physiologically based pharmacokinetic model assessment of clinical drug interactionrisk. Drug Metab Dispos 42:1008–1015.

Zhang JG, Ho T, Callendrello AL, Clark RJ, Santone EA, Kinsman S, Xiao D, Fox LG, EinolfHJ, and Stresser DM (2014) Evaluation of calibration curve-based approaches to predictclinical inducers and noninducers of CYP3A4 with plated human hepatocytes. Drug MetabDispos 42:1379–1391.

Zhu Q, Xia H, Xia CQ, Yang Q, Doshi U, Li AP, and Liao M (2014) Culture duration-, donor-,and medium-dependent changes in OATP1B3-mediated telmisartan uptake in human hepato-cytes. Drug Metab Lett 7:117–125.

Address correspondence to: Dr. Niresh Hariparsad, 4019E Vertex 1, 50 NorthernDrive, Boston, MA 02210. E-mail: niresh_hariparsad@vrtx.com

Hepatocyte Coculture System to Assess P450 Induction 261

at ASPE

T Journals on February 4, 2020

dmd.aspetjournals.org

Dow

nloaded from