1521-009X/45/3/246–259$25.00 http://dx.doi.org/10.1124/dmd.116.074120DRUG METABOLISM AND DISPOSITION Drug Metab Dispos 45:246–259, March 2017Copyright ª 2017 by The American Society for Pharmacology and Experimental Therapeutics

Specific Inhibition of the Distribution of Lobeglitazone to the Liver byAtorvastatin in Rats: Evidence for a Rat Organic Anion Transporting

Polypeptide 1B2–Mediated Interaction in Hepatic Transport s

Chang-Soon Yim, Yoo-Seong Jeong, Song-Yi Lee, Wonji Pyeon, Heon-Min Ryu, Jong-Hwa Lee,Kyeong-Ryoon Lee, Han-Joo Maeng, and Suk-Jae Chung

College of Pharmacy and Research Institute of Pharmaceutical Sciences, Seoul National University, Gwanak-gu, Seoul, Republic ofKorea (C.-S.Y., Y.-S.J., S.-Y.L., W.P., H.-M.R., S.-J.C.); Korea Institute of Toxicology, Yuseong-gu, Daejeon, Republic of Korea (J.-H.L.);Life Science Research Center, Daewoong Pharmaceutical Company Ltd., Yongin-si, Gyeonggi-do, Republic of Korea (K.-R.L.); and

College of Pharmacy, Gachon University, Yeonsu-gu, Incheon, Republic of Korea (H.-J.M.)

Received November 1, 2016; accepted January 5, 2017

ABSTRACT

Cytochrome P450 enzymes and human organic anion transportingpolypeptide (OATP) 1B1 are reported to be involved in the pharmaco-kinetics of lobeglitazone (LB), a new peroxisome proliferator–activatedreceptor g agonist. Atorvastatin (ATV), a substrate for CYP3A andhumanOATP1B1, is likely to be coadministeredwith LB in patients withthe metabolic syndrome. We report herein on a study of potentialinteractions between LB and ATV in rats. When LB was administeredintravenously with ATV, the systemic clearance and volume of distri-bution at steady state for LB remained unchanged (2.67 6 0.63 ml/minper kg and 289 6 20 ml/kg, respectively), compared with that of LBwithout ATV (2.34 6 0.37 ml/min per kg and 271 6 20 ml/kg, re-spectively). Although the tissue-to-plasma partition coefficient (Kp) of

LB was not affected by ATV in most major tissues, the liver Kp for LBwas decreased by ATV coadministration. Steady-state liver Kp valuesfor three levels of LB were significantly decreased as a result of ATVcoadministration. LB uptake was inhibited by ATV in rat OATP1B2-overexpressing Madin–Darby canine kidney cells and in isolated rathepatocytes in vitro. After incorporating the kinetic parameters for thein vitro studies into a physiologically based pharmacokinetics model,the characteristics of LB distribution to the liver were consistent withthe findingsof the in vivostudy. It thusappears that thedistributionof LBto the liver ismediated by the hepatic uptake of transporters such as ratOATP1B2, and carrier-mediated transport is involved in the liver-specific drug–drug interaction between LB and ATV in vivo.

Introduction

Lobeglitazone (LB), an agonist for peroxisome proliferator–activatedreceptor g (PPARg), is approved in Korea for the treatment of type2 diabetes mellitus (T2DM). Previous studies (Sauerberg et al., 2003;Kim et al., 2004; Lee et al., 2005) indicate that LB has a therapeuticadvantage over rosiglitazone and pioglitazone (i.e., clinically approve-d/used PPARg agonists with similar structural motifs) in terms ofpotency for the PPARg receptor. As a result, an improvement in insulinsensitivity is expected even at a lower dosage (e.g., 0.5 mg/d): the lower

dosage of LB is likely to reduce the incidence of adverse effects(Kim et al., 2014).We previously demonstrated that LB was primarily metabolized by

major cytochrome P450 enzymes (namely, CYP1A2, CYP2Cs, andCYP3A4) (Lee et al., 2015a,b) and interacts with human organic aniontransporting polypeptide (hOATP) 1B1, with an IC50 value of 2.44 mM(Lee et al., 2015b). This suggests that drug–drug interactions (DDIs) arepossible for LB in drug metabolism as well as in its distribution totissues. Clinical pharmacokinetics studies of LB with other prescriptiondrugs such as warfarin or amlodipine were recently conducted in Korea,and no significant differences in the systemic pharmacokinetics of LBwere reported after coadministration (Jung et al., 2015; Kim et al., 2015).These studies were designed to investigate potential changes in thesystemic pharmacokinetics of LB when these drugs were coadministered,considering the fact that they are known to share some of the same met-abolic pathways of LB (i.e., CYP1A2, CYP2C9, and CYP3A4 withwarfarin; CYP3A4 with amlodipine) (Guengerich et al., 1991; Kaminskyand Zhang, 1997). The lack of appreciable differences in pharmacokineticsindicates that a DDI mediated by drug metabolism is less likely for LB.

This research was supported by the National Research Foundation of Korea,funded by the Korean government [Grant 2009-0083533].

A portion of this work was previously presented as a poster Interaction betweenLobeglitazone and Atorvastatin in Rats: Potential Involvement of Transporter-Mediated Interaction in the Hepatic Uptake. Yim CS, Jeong YS, Lee SY, Ryu HM,Lee W, Dae-Duk K, Chung SJ: (2016) 11th International Society for the Study ofXenobiotics Meeting; 2016 Jun 12–16; Busan, Republic of Korea.

dx.doi.org/10.1124/dmd.116.074120.s This article has supplemental material available at dmd.aspetjournals.org.

ABBREVIATIONS: ACN, acetonitrile; ANOVA, analysis of variance; ATV, atorvastatin; AUC, area under the concentration-time curve; B/P, blood-to-plasma concentration ratio; BCA, bicinchoninic acid; BSA, bovine serum albumin; CL, systemic clearance; DDI, drug–drug interaction; DMSO,dimethylsulfoxide; DPBS, Dulbecco’s phosphate-buffered saline; FRT, flippase recognition target; fu, fraction unbound; hOATP, human organicanion transporting polypeptide; J, rate of cellular uptake; KHB, Krebs-Henseleit buffer; LB, lobeglitazone; LC, liquid chromatography; MDCK,Madin–Darby canine kidney; MRT, mean residence time; MS/MS, tandem mass spectrometry; OATP, organic anion transporting polypeptide;PBPK, physiologically based pharmacokinetic; PPARg, peroxisome proliferator–activated receptor g; PS, unbound passive diffusionalclearance; rOATP, rat organic anion transporting polypeptide; SD, Sprague-Dawley; T2DM, type 2 diabetes mellitus.

246

http://dmd.aspetjournals.org/content/suppl/2017/01/09/dmd.116.074120.DC1Supplemental material to this article can be found at:

at ASPE

T Journals on O

ctober 5, 2020dm

d.aspetjournals.orgD

ownloaded from

In addition to DDIs caused by complications in drug metabolism,based on the literature, it is clear that DDIs are possible as a result ofinteractions at the drug transporter level. For example, it was reportedthat cyclosporine inhibits the transport of rosuvastatin to the liver,probably mediated by hOATP1B1, thereby increasing the plasma levelof rosuvastatin by 7.1-fold (Simonson et al., 2004). Interestingly,however, evidence for interactions caused by drug transporters may beless obvious and would be difficult to detect in some cases. For example,prior studies reported that concentrations of metformin in the liver andkidney are elevated after coadministration of cimetidine or pyrimeth-amine, which is probably caused by the inhibition of mouse multidrugand toxin extrusion 1 protein expressed in the liver and kidney of mice(Ito et al., 2012; Shingaki et al., 2015) without any apparent indicationof DDI in the systemic pharmacokinetics. It was also noted that theconcentration of ciprofloxacin in the liver was decreased after co-administration of 24-nor-ursodeoxycholic acid by inducing the expres-sion of mouse multidrug resistance-associated protein 4 in the basolateralside of the mouse liver (Wanek et al., 2016), whereas the systemicpharmacokinetics remained unchanged. These results indicate thatconventional methods for detecting DDIs are not adequate for someinteractions involving transporters, and the possibility of DDI needs tobe examined in the target tissue. For LB, however, the possibility oftransporter-mediated DDIs is not known.Atorvastatin (ATV), one of the most frequently prescribed statins

(Hsyu et al., 2001; Laufs et al., 2016), is reported to be metabolized byCYP3A4 (Park et al., 2008) and is transported by hOATP1B1 andhOATP1B3 (Knauer et al., 2010). It is now well established thathyperlipidemia can occur simultaneously with T2DM (Iglay et al.,2016); thus, a combined treatment involving statins and hypoglycemicagents (e.g., LB) may become necessary clinically (Gu et al., 2010). It isnoteworthy that ATV and LB share similar pathways for drug metab-olism (CYP3A4) and distribution to tissues (hOATP1B1). Unfortu-nately, however, the possibility of DDIs between ATV and LB does notappear to have been examined to date.The objective of this study, therefore, was to determine whether DDIs

between LB and ATV occur in rats. We were particularly interested inthe possibility of DDI at the level of metabolism by CYP3A and thecarrier-mediated transport via rat organic anion transporting polypeptide(rOATP) 1B2 (i.e., ortholog of hOATP1B1 and hOATP1B3) (Hagenbuchand Meier, 2004) in the pharmacokinetics of LB in the absence orpresence of ATV coadministration. The findings indicate that although noapparent interaction was found between LB and ATV in their systemicpharmacokinetics, there was a significant interaction for the hepaticdistribution of LB, most likely at the level of the hepatic distribution viarOap1b2, after coadministration of ATV.

Materials and Methods

Chemicals and Reagents

LB (98.5% purity) was kindly provided by Chong Kun Dang Pharmaceuticals(Seoul, Korea). ATV (98% purity) was purchased from Tokyo Chemical Industry(Tokyo, Japan). Glipizide (96% purity), bicinchoninic acid (BCA), bovine serumalbumin (BSA), Hanks’ balanced salt solution, sodium bicarbonate, HEPES, anddimethylsulfoxide (DMSO) were obtained from Sigma-Aldrich (St. Louis, MO).[3H]-Inulin (194 mCi/g) and [3H]-estradiol-17b-D-glucuronide (specific activity,41.4 Ci/mmol) (both from American Radiolabeled Chemicals, St. Louis, MO)were also used in this study. SDS and polyethylene glycol 400 were purchasedfrom Georgia Chemical (Norcross, GA) and Duksan Pure Chemicals (Ansan,Korea), respectively. Dulbecco’smodified Eagle’smedium, nonessential amino acidsolution, Dulbecco’s phosphate-buffered saline (DPBS), penicillin/streptomycin,and fetal bovine serum were obtained from Welgene (Daegu, Korea). High-performance liquid chromatography (LC)– grade acetonitrile (ACN) and formicacid were purchased from Fisher Scientific (Pittsburgh, PA) and Fluka (Cambridge,

MA), respectively. An ammonium formate solution (5 M; Agilent Technologies,Santa Clara, CA) was also used. Pooledmale rat liver microsomes and an NADPH-regenerating system solution were purchased from Corning Gentest (Woburn,MA). Zoletil 50 (tiletamine-HCl/zolazepam-HCl) was purchased from VirbacLaboratories (Carros, France) and Rompun (xylazine-HCl) was from Bayer Corp.(Shawnee Mission, KS). All other chemicals were of reagent grade or greater andwere used without further purification.

Animals

Male Sprague-Dawley (SD) rats (body weight, 240–270 g; Orient Bio Inc.,Seongnam, Korea) were used in all in vivo/hepatocyte isolation studies.Experimental protocols involving the animals used in this study were reviewedand approved by the Seoul National University Institutional Animal Care and UseCommittee, according to theNational Institutes of Health Principles of LaboratoryAnimal Care (publication number 85-23, revised in 1985).

Intravenous Bolus Administration Study

Identification of Systemic DDIs in Rats. Overnight fasted male SD rats wereanesthetized by an intramuscular administration of 50 mg/kg tiletamine-HCl/zolazepam-HCl (Zoletil 50) and 10 mg/kg xylazine-HCl (Rompun). Afterwe confirmed the induction of anesthesia, the femoral artery (for collecting bloodsamples) and vein (for administering and supplementing body fluids) werecatheterized with polyethylene tubing (PE 50; Clay Adams, Parsippany, NJ) filledwith heparinized saline (20 U/ml; for arterial cannulae) and normal saline (forvenous cannulae), respectively. Upon recovery from the anesthesia, four types ofdosing solutions (i.e., 0.5 mg/ml LB, 0.5 mg/ml LB with 2.5 mg/ml ATV,2.5 mg/ml ATV, or 2.5 mg/ml ATV with 0.5 mg/ml LB) were preparedimmediately prior to administration and given intravenously to rats by a bolusinjection. The vehicle for the dosing solutions consisted of DMSO/polyethyleneglycol 400/saline [0.5:4:5.5 (v/v/v)] and the injection volume was 2 ml/kg. Bloodsamples (150 ml for each sample) were collected from the arterial catheter at 0, 5,15, 30, 60, 120, 240, 360, and 480minutes (for the study of LB pharmacokinetics)or at 0, 5, 15, 30, 60, and 120 minutes (for the study of ATV pharmacokinetics)after administration. Immediately after blood collection, a volume of saline identicalto the volume of the blood sample was given to the animal to compensate for fluidloss. The plasma fraction was separated from blood samples by centrifugation(16,100g for 5minutes at 4�C) and stored at –80�Cuntil it was used in a LC–tandemmass spectrometry (MS/MS) assay for LB and ATV (see below).

Intravenous Bolus Administration Study

Identification of DDIs in Rat Tissues. Overnight fasted male SD rats wereanesthetized and catheterized as described above. Upon recovery from anesthesia,four types of dosing solutions (i.e., 0.5 mg/ml LB, 0.5 mg/ml LB with 2.5 mg/mlATV, 2.5 mg/ml ATV, or 2.5 mg/ml ATV with 0.5 mg/ml LB) were similarlyprepared and administered intravenously to rats by a bolus injection. The animalswere euthanized at 30, 60, 120, or 240 minutes after administration, and tissue(i.e., lungs, adipose, brain, heart, kidneys, muscle, liver, spleen, and gut) sampleswere collected. In this study, samples of trunk blood were also collected. Aftertissue isolation, the samples were washed four times with ice-cold DPBS andweighed. Twice the volume of the tissue weight of DPBS was then added forhomogenization (Ultra Turrax homogenizer; IKA, Staufen, Germany). Themixture was then stored at280�C until used in determining the concentrationof LB and ATV.

When it was necessary to determine the volume of trapped blood in tissuesamples, rats were injected intravenously with a solution of inulin containing atrace amount of [3H]-inulin in normal saline (injection volume, 1 ml/kg). Bloodsamples were collected at 0.5 minutes, and the animals were immediatelyeuthanized by cervical dislocation and tissue samples were collected. Bloodsamples were then centrifuged at 10,770g for 10 minutes and the supernatant wascollected. For tissue samples, approximately 100 mg of the tissue sample wasweighed and digested with 2 ml Soluene-350 (PerkinElmer Life and AnalyticalSciences, Waltham, MA). Hydrogen peroxide (200 ml; 35%; Junsei Chemical,Tokyo, Japan) was then added and the resulting sample was incubated overnightat room temperature to prevent quenching (Saito et al., 2001). Aliquots of plasma(50 ml) or tissue sample were mixed with Ultima Gold (PerkinElmer), and theradioactivity of the samplewas determined by liquid scintillation counting (Tri-Carb

DDIs between Lobeglitazone and Atorvastatin in Rats 247

at ASPE

T Journals on O

ctober 5, 2020dm

d.aspetjournals.orgD

ownloaded from

3110TR Liquid Scintillation Analyzer; PerkinElmer). Considering the possibilitythat inulin might not be distributed to erythrocytes (Gaudino and Levitt, 1949), theestimated trapped plasma volume was subsequently divided by 0.55 (i.e., plasma toblood volume ratio). In all subsequent studies in which a tissue concentration ofdrugs was reported, the concentration found in the sample was corrected bysubtracting the portion in the trapped blood volume.

Intravenous Infusion study of LB with and without ATV in Rats

When it was necessary to further determine the distribution of LB to the liver, aconstant LB concentration in the plasma was achieved by intravenous infusionand the liver concentration was determined in the absence and presence of ATVcoadministration. Thus, overnight fastedmale SD rats, weighing 240–270 g, wereanesthetized and catheterized, as described above. After the rats recovered fromanesthesia, LBwas intravenously infused at a rate of 50, 250, or 1000 ng/min withor without ATV solution (10 mg/10 ml per min of DMSO/polyethylene glycol400/saline [0.5:2:7.5 (v/v/v)]) using an infusion pump (Harvard Apparatus,Holliston, MA). Blood samples (150 ml) were collected from the catheterconnected to the artery at 30, 60, 120, 240, 300, and 360minutes. Upon collectingthe last blood sample, the animals were euthanized and a liver sample wascollected. The plasma and liver samples were processed for analysis as describedabove.

Determination of Blood-to-Plasma Concentration Ratio and Plasma ProteinBinding of LB

LB was added to fresh blank blood at final concentrations of 0.1, 1, or 10 mM,and the mixture was incubated at 37�C in a water bath for 60 minutes. Afterincubation, themixturewas centrifuged (16,100g) and the plasmawas collected asthe supernatant: the concentration of LB was then determined by LC-MS/MSanalysis to calculate the blood-to-plasma concentration ratio (B/P).

In this study, the extent of plasma protein binding was also determined for LBby the rapid equilibrium dialysis method, according to the manufacturer’srecommended protocol (Thermo Fisher Scientific, Waltham, MA). Briefly, theplate of the dialysis device was rinsed with 20% ethanol for 10 minutes. LB [0.1,0.3, 1, 3, or 10 mM as the final concentration for the rat plasma; 0.5, 5, or 50 mMfor rat liver microsomes (see below); 0.5, 1, 2, 5, 10, or 20mM for 5mMBSA (seebelow) in the transport medium (9.7 g/l Hanks’ balanced salt solution, 2.38 g/lHEPES, and 0.35 g/l sodium bicarbonate, pH adjusted to 7.4)] and ATV [finalconcentration: 0.1, 1, or 10 mM for 5 mM BSA in the transport medium] wereadded to the medium [i.e., the plasma, the microsomal incubation medium, or thetransport medium]. Aliquots of the mixture (200 ml) and protein-free solution(350 ml) were placed into the sample chamber and the buffer chamber,respectively, and the device was sealed/incubated on a shaker (150 rpm) at37�C for 4 hours. After the incubation was complete, an aliquot (50 ml) wascollected from each side of the chamber. To ensure that the matrix of the samplesmatched, a 50-ml aliquot of blank medium was then added to the protein-freesolution sample, and an equal volume of protein-free solution was also added tothe medium. The resulting sample was then analyzed for LB and ATV todetermine the unbound fraction (fu). In this study, the recovery was between 70%and 100% in all experiments.

Stability of LB in Rat Liver Microsomes

Themetabolic stability of LBwith or without ATV in rat liver microsomes wasdetermined in this study. The reaction mixture (total volume of 1 ml) consisted ofan NADPH-regenerating solution containing 1.3 mM NADP+, 3.3 mM glucose-6-phosphate, 0.4 U/ml glucose-6-phosphate dehydrogenase, and 3.3 mMMgCl2,as recommended by the manufacturer’s protocol. Depending on the study, anaqueous solution (100 mM potassium phosphate buffer, pH 7.4) containing LB(final concentration in the reaction mixture: 1, 2, 5, 10, 20, or 50 mM) withoutATV, or LB (the final concentration: 5 mM)with ATV (the final concentration: 1,5, 10, 20, 50, 100, or 200 mM) was also added. The concentration of organicsolvent (i.e., DMSO) in the incubation mixture was maintained at less than 1% tolimit its influence on microsomal enzymes. After preincubation (37�C for10 minutes), the reaction was initiated by the addition of rat liver microsomes(50 ml; the final microsomal protein concentration was 0.5 mg/ml). Preliminaryexperiments were carried out to determine the incubation time for the given LBconcentrations for a linear rate of LB disappearance (data not shown). Sampleswere collected (50 ml) at various predetermined times and then an ice-cold ACN

solution (i.e., for the termination of the reaction) containing glipizide (100 ng/ml,internal standard) was added. The LB concentration in the mixture wasdetermined by the LC-MS/MS assay.

rOATP1B2 Cloning

The protein coding region of rOATP1B2 (GenBank accession numberNM_031650.3) was cloned from a rat total liver mRNA library (Takara Shuzo,Kyoto, Japan). The RNA (1 mg) was reverse transcribed using the PrimeScriptfirst-strand cDNA synthesis kit (Takara) by incubating the mixture at 65�C for5 minutes, cooling to 4�C followed by adding RNase inhibitor and RTase into thereaction mixture, and then the following reaction was performed by incubating at,42�C for 60 minutes, heating to 95�C for 5 minutes, and cooling to 4�C. Specificprimers for cloning of the rOATP1B2 coding region were 59-GCTAGCAGT-GATTGCAGACGTTCCCA-39 (sense strand; the NheI site is underlined) and 59-AAGCTTGTCCATCCTTGCCCCATTCT-39 (antisense strand; the HindIII siteis underlined). The polymerase chain reaction was performed using ExTaq DNApolymerase (Takara) using the following settings: 30 cycles of denaturation at94�C for 30 seconds, a primer-annealing step at 63�C for 30 seconds, andextension at 72�C for 3 minutes. The amplicon was cloned into pcDNA5/FRT(flippase recognition target) vector (Invitrogen, Carlsbad, CA), and the identity ofthe insert confirmed by sequencing. The plasmid containing the wild type form ofrOATP1B2 was selected and transfected into Madin–Darby canine kidney(MDCK) II/FRT cells using the FuGENE transfection reagent. The transfectedcells were incubated in the culture medium containing 0.1 mg/ml hygromycin B(Invitrogen) for several weeks for selection. Expression of rOATP1B2 wasconfirmed by the reverse transcription polymerase chain reaction in the trans-fected cells. Functional expression was determined by comparing the uptake ofradiolabeled estradiol-17b-glucuronide (a standard substrate of the transporter;Cattori et al., 2000) in the transfected cells with that in mock-transfectedMDCKII/FRT cells (Supplemental Fig. 1). In this study, at least a 60-fold higheruptake was found in rOATP1B2 transfected cells with estradiol-17b-glucuronide.Since themeasurements in this study involved uptake across the apical membrane,this observation indicates that the transporter is expressed ubiquitously in theplasma membrane, rather than localized in the basolateral membrane (i.e.,endogenous site of expression), in the transfected cells. Protein concentrationswere determined by a BCA assay (Smith et al., 1985) and used for the correctionof the transport function.

Uptake of LB in rOATP1B2-Transfected MDCKII/FRT Cells

To determine the kinetic characteristics of rOATP1B2 transport, MDCKII/FRTcells expressing the anion transporter were used. Thus, rOATP1B2 cells weregrown in Dulbecco’s modified Eagle’s mediumwith 10% (v/v) fetal bovine serum,1% (v/v) nonessential amino acid solution, 100 U/ml penicillin/streptomycin, and10 mM HEPES. All cells were kept at 37�C with 5% CO2 and 95% relativehumidity. Cells were seeded in poly(L)-ornithine (Sigma-Aldrich)–coated 24-well(2.0 cm2) cell-culture plates (Corning, Corning, NY) in supplemented Dulbecco’smodified Eagle’s medium at a density of 0.5� 106 cells/well and were then grownin a humidified atmosphere of 5%CO2 at 37�C for 2 days. Preliminary studies werecarried out with LB to determine the appropriate reaction time in rOATP1B2 cellsas well as mock cells. In this study, the uptake of LB was proportionally increasedwith the incubation time up to 12 minutes (Supplemental Fig. 2); thus, a 10-minuteincubation time was used in subsequent studies involving LB and rOATP1B2-expressing cells.

For measurement of LB transport in rOATP1B2 cells, the cells were firstwashed twice with prewarmed DPBS and preincubated with transport mediumcontaining 2% (final concentration) DMSO (i.e., solubilizing agent for LB). Froma pilot experiment, it was found that DMSO up to 2% in the transport medium hadno appreciable effect on the function of the transporter (Supplemental Fig. 3),consistent with other literature findings (Da Violante et al., 2002; Taub et al.,2002). LB was prepared in DMSO and diluted with 5 mM BSA in transportmedium (final DMSO concentration of 2%). We found that the addition of BSAwas necessary to suppress the nonspecific adsorption of the drugs on the ex-perimental apparatus and/or cell surface (Takeuchi et al., 2011). The uptake of LBwas measured in six different concentrations ranging from 0.5 to 20mM (i.e., 0.5,1, 2, 5, 10, or 20 mM) in rOATP1B2 and in mock cells. After preincubation for10 minutes at 37�C, the transport medium was removed and replaced withprewarmed medium containing LB in the presence of and absence of ATV

248 Yim et al.

at ASPE

T Journals on O

ctober 5, 2020dm

d.aspetjournals.orgD

ownloaded from

(200 ml). Upon completion of the incubation, the aqueous medium was firstaspirated and the cells were washed twice with ice-cold DPBS containing 0.1%BSA (400 ml) and once with ice-cold DPBS (400 ml). The final rinsing solutionwas then aspirated and 0.1% (w/v) SDS added to the cell suspension for lysis. Apreliminary experiment indicated that the addition of BSA up to 3% in the rinsingsolution had no effect on the protein concentration in the cell lysate. In addition,the effect of limiting the nonspecific binding on the experimental apparatus and/oron cell surface was not statistically different for BSA concentrations ranging from0.1% to 10% in the rinsing solution (Supplemental Fig. 4). The resulting mixturewas agitated for 30minutes, and anACNsolution containing glipizide (100 ng/ml,internal standard) was added. After vortexing for 5 minutes and centrifugation for5 minutes (16,100g), an aliquot (50 ml) of the supernatant was collected and theconcentration of LB in the supernatant was determined by LC-MS/MS assay (seebelow). The transport rate was normalized by the protein concentration in thesample as determined by a BCA assay. When necessary, the number of cells wasestimated using the predetermined relationship [y (mg protein/ml) = 0.304� 106 × x(cells/ml) + 0.0335, r2 = 0.994] between the protein concentration (inmg protein/ml)and the number of cells per milliliter (in 106 cells/ml).

When the inhibitory effect of ATV on the transport function of LB inrOATP1B2 cells was studied, the uptake of LB (5mM)was determined in the cellsin the presence of various concentrations of ATV (i.e., 0, 0.01, 0.1, 1, 10, or30 mM). The transport medium containing DMSO (2%, final concentration) wasprepared and BSA (5 mM, final concentration for minimizing nonspecificbinding) was added. After a 10-minute incubation at 37�C, the uptake wasterminated by aspiration of the transport medium. The cells were then rinsed twicewith ice-cold DPBS containing 0.1%BSA (400ml) and once with ice-cold DPBS(400 ml). Cells were lysed for 30 minutes in 0.1% (w/v) SDS and the mixture wasagitated. AnACN solution containing glipizide (100 ng/ml, internal standard) wasthen added to the mixture. After vortexing for 5 minutes and centrifugation for5 minutes (16,100g), an aliquot (50 ml) of the supernatant was collected and theconcentration of LB was determined by an LC-MS/MS assay. The data werenormalized by the protein concentration in the sample as determined by BCAassay.Preliminary studies were carried out in an attempt to determinewhether the openingof the tight junction is required for transport in the case of rOATP1B2-expressingcells. Therefore, the cells were pretreatedwith EDTA according to a literature report(Wang et al., 2016) with minor modifications. Briefly, a Ca2+/Mg2+-free transportmedium containing 500 mMEDTAwas prepared and rOATP1B2-expressing cellswere pretreated for 2 hours. The transport experiment was then performed asdescribed above. Although the uptake of LB was slightly increased in both mockand rOATP1B2-expressing cells with comparable Km and Ki values, this findingsuggests that the kinetic properties of carrier-mediated transport are not significantlyaltered for LB by EDTA pretreatment. Therefore, we chose to exclude the EDTApretreatment step before the transport experiment, to prevent the formation of anypossible artifacts resulting from the EDTA pretreatment.

Uptake of LB in Isolated Rat Hepatocytes

Rat hepatocytes were isolated using the two-step collagenase perfusionmethodwith minor modifications (Kotani et al., 2011). Briefly, under anesthesia, acannula was inserted into the portal vein followed by perfusing with a Ca2+/Mg2+-free buffer at the flow rate of 20 ml/min for 5 minutes. Upon the initiation of theperfusion, the inferior vena cavawas dissected to allow the perfusate to exit.Whennecessary, the perfusion medium was switched to a buffer containing 50 mMCaCl2 and 0.5 mg/ml collagenase (Sigma-Aldrich) at a flow rate of 20 ml/min for10 minutes. Hepatocytes from the digested liver were dispersed in fresh perfusionbuffer and separated from the connective tissue by filtration through a sterile50-mesh (the size of the sieve opening of 280 mm). Rat hepatocytes wereseparated from nonparenchymal cells in the crude hepatocyte fraction as a pelletby centrifugation at 50g for 5 minutes. The cells were then resuspended in Krebs-Henseleit buffer (KHB) and centrifuged again at 50g for 5 minutes. In this study,hepatocytes having a viability of greater than 80%, as determined by a trypan blueexclusion assay, were used in subsequent experiments. The resulting cell pelletwas resuspended in Williams’ media E containing 10% (v/v) fetal bovine serum,1% (v/v) penicillin/streptomycin, and 0.01% (v/v) insulin/transferrin/selenium(i.e., 10 mg/ml insulin, 5.5 mg/ml transferrin, and 5 mg/ml selenium) premix(Sigma-Aldrich) at pH 7.4. The hepatocytes were plated in 24-well collagen I(Sigma-Aldrich)–coated plates at a density of 5 � 105 viable cells per well in0.5 ml supplemented Williams’ media E. The plate was equilibrated from 5 to

6 hours at 37�C in an atmosphere containing 5% CO2 to allow the cells to adhereto the collagen-coated surface.

In this study, hepatocyte uptake was measured for six concentrations of LB(i.e., 0.5, 1, 2, 5, 10, or 20 mM). After approximately 5 to 6 hours of cell plating,the medium was aspirated and the attached cells were rinsed twice withprewarmed DPBS. LB was dissolved in KHB containing DMSO (5%, finalconcentration) and BSA (5 mM, final concentration). In a preliminary study, theaddition of DMSO up to a final concentration of 10% did not appear to have anappreciable impact on transport function (Supplemental Fig. 5), consistent withprevious observations (Da Violante et al., 2002). The transport reaction wasinitiated by the addition of a KHB solution containing LB (200 ml) on the top ofthe plate. According to our study (Supplemental Fig. 6), the accumulation of LB inthe hepatocytes was proportional to the incubation time up to 2 minutes and thisincubation time was used in subsequent studies. Cellular uptake was terminatedby aspiration of the transport medium; cells were rinsed twice with ice-cold DPBScontaining 0.1% BSA (400 ml), followed by an additional washing with ice-coldDPBS (400 ml). The hepatocytes were lysed for 30 minutes by the addition of0.1% (w/v) SDS and agitation. An ACN solution containing glipizide (100 ng/ml,internal standard) was then added to the lysate and the concentration of LBdetermined by a LC-MS/MS assay. When necessary, the transport rate wasnormalized by the amount of protein in the sample as determined by a BCA assay.The number of hepatocytes was also estimated using the predetermined relation-ship [y (mg protein/ml) = 0.600 � 106 × x (hepatocytes/ml) 2 0.007, r2 = 0.998]between the protein concentration (in mg protein/ml) and the number ofhepatocytes per milliliter (in 106 hepatocytes/ml).

To determine the inhibitory effect of ATV on the transport of LB in rat isolatedhepatocytes, LB (5 mM) uptake was measured in the presence of variousconcentrations of ATV (i.e., 0, 0.001, 0.01, 0.1, 1, 10, 30, or 100 mM). Thus, themedium was aspirated, followed by rinsing twice with prewarmed DPBS. TheDPBS was then aspirated, and the cells were preincubated for 60 minutes at 37�Cwith KHB containing various concentrations of ATV (i.e., 0, 0.001, 0.01, 0.1, 1,10, 30, or 100 mM) and BSA (5 mM, final concentration). After preincubation for60 minutes (Amundsen et al., 2010; Shitara et al., 2013), the medium wasswitched to KHB containing BSA (5 mM, final concentration) and 5 mM LBalong with various concentrations (i.e., 0, 0.001, 0.01, 0.1, 1, 10, 30, or 100 mM)of ATV. After incubation at 37�C for 2 minutes, uptake was terminated byaspirating the medium, followed by rinsing twice with ice-cold DPBS containing0.1%BSA (400ml) and once with ice-cold DPBS (400ml). The hepatocytes werelysed for 30 minutes in 0.1% (w/v) SDS and the mixture was agitated. An ACNsolution containing glipizide (100 ng/ml, internal standard) was then added to thelysate. The concentration of LB was determined by an LC-MS/MS assay. Whennecessary, the transport rate was normalized by amount of protein in the sample asdetermined by a BCA assay.

LC-MS/MS Assay for LB and ATV

The concentration of LB and ATV in rat plasma samples, tissue homogenates,or cell lysates was determined using an LC-MS/MS assay method as previouslydescribed (Lee et al., 2009; Kim et al., 2012) with a minor modification. Briefly,an aliquot (50 ml) of the sample was vortex mixed with an ACN solutioncontaining glipizide (100 ng/ml, internal standard) followed by centrifugation(16,100g for 5 minutes at 4�C). An aliquot (5 ml) of the supernatant was directlyinjected onto the LC-MS/MS system. In this study, the LC-MS/MS system wasequipped with a Waters e2695 high-performance LC system (Milford, MA) andAPI 3200 QTRAP mass spectrometer (Applied Biosystems, Foster City, CA).The mobile phase, consisting of (A) 0.1% formic acid in ACN and (B) 10 mMammonium formate in purified water, was delivered at the flow rate of 0.3 ml/minusing a gradient elution involving 30% of A (0 minutes), from 30% to 80% of A(0–0.5 minutes), 80% of A (0.5–1.5 minutes), from 80% to 30% of A (1.5–2minutes), and 30% of A (2–5minutes). Chromatographic separation was carriedout on a reversed-phase high-performance LC column (Eclipse XDB-C18,3.5 mm, 2.1 � 100 mm; Agilent Technologies) at 25�C while the temperature inthe autosampler was maintained at 4�C during the analysis. Samples were ionizedusing a turbo ion spray interface in the positive ionization mode and monitored atthe following Q1/Q3 transitions (mass/charge ratio): 481.3/258.2 for LB,559.3/440.4 for ATV, and 445.8/320.9 for glipizide. The common source/gasconditions for LB, ATV, and glipizide were as follows. The pressure of the curtaingas, ion spray voltage, source temperature, ion source gas 1, and ion source gas2 was 25 psi, 4500 V, 550�C, 60 psi, and 60 psi, respectively. The declustering

DDIs between Lobeglitazone and Atorvastatin in Rats 249

at ASPE

T Journals on O

ctober 5, 2020dm

d.aspetjournals.orgD

ownloaded from

potentials for LB, ATV, and glipizide were 77.9, 45.0, and 47.5 V, respectively.The entrance potentials were 7.4, 4.0, and 4.0 V, collision energies were 47, 27,and 17 V, and the collision cell exit potentials were 4.0, 6.0, and 8.0 V,respectively. The detector response was linear in the concentration rangeexamined (i.e., 5–5000 ng/ml) for both LB and ATV in the rat plasma sampleswith interday/intraday precisions of less than 10% and an accuracy within 8% ofthe theoretical value, indicative of a valid assay for LB and ATV. In addition toplasma samples, the calibration curves for LB and ATV in the rat tissuehomogenates were linear (i.e., 0.992# r2# 0.999 for LB and 0.995# r2# 0.999for ATV, respectively) in the concentration range studied. The concentration ofLB in cell lysates was also determined with the linear calibration curves (i.e.,0.998# r2 # 0.999 in cell lysates from mock and rOATP1B2 cells, and 0.998#r2 # 0.999 in cell lysates from isolated rat hepatocytes).

Data Analysis

In Vitro Kinetic Analysis. When necessary, the in vitro metabolic kineticparameters (e.g., Km, Vmax, and Ki) were determined by fitting the data to eqs. 1and 2 (Segel, 1975):

V ¼ Vmax × ½S�Km þ ½S� ð1Þ

V ¼ Vmax × ½S�Km ×

�1þ ½I�

Ki

�þ ½S�

ð2Þ

where V was the rate of metabolic reaction, Vmax is the maximal rate ofmetabolism, Km is the Michaelis–Menten constant, Ki is the inhibition constant,[S] is the LB concentration as a substrate, and [I] is the ATV concentration as aninhibitor, respectively.

In addition, the cellular transport rate (J) was estimated from the amount of LBuptake in cell lysates, divided by the incubation time, and expressed in eqs. 3 and4.

In the case where only passive transport was present,

J ¼ PS × ½S� ð3Þ

In the case of parallel passive and carrier-mediated transports,

J ¼ Jmax × ½S�Km þ ½S� þ PS × ½S� ð4Þ

where PS is the unbound clearance via passive diffusion and [S], Km, and Jmax arethe unbound concentration of LB, the unbound Michaelis-Menten constant, andthe maximal rate of the carrier-mediated transport, respectively. Kineticparameters (e.g., PS, Km, and Jmax) were estimated by the simultaneous fittingof the in vitro data to eqs. 3 and 4 using nonlinear regression analysis. Themeasured value for uptake in mock cells may contain some statistical variabilityand the correction may not be adequate in some cases (e.g., resulting in a negativetransport value after the correction). As indicated in the literature (DeLean et al.,1978), a simultaneous fitting approach may result in more reliable values inestimating vitro kinetic parameters. Therefore, we chose to use this method ofkinetic analysis.

When it was necessary to analyze the kinetics of the concentration-dependentinhibition of LB transport by ATV, eq. 5 was fitted to the in vitro data to estimatethe half maximal inhibitory concentration (IC50). In addition, the Ki value wasestimated using eq. 6 (Cheng and Prusoff, 1973):

R ¼ Rmax 2 ðRmax 2R0Þ ×� ½I�½I� þ IC50

�ð5Þ

Ki ¼ IC50�1þ ½S�

Km

� ð6Þ

where R, Rmax, R0, [I], [S], and Km are the ratio of the amount of LB transport tothe value for the control group treated without ATV, the maximum value of R, theminimumvalue of R, the ATV concentration, the unbound concentration of LB asa substrate, and the unbound Michaelis–Menten constant, respectively. It has beenreported that the adoption of Hill’s slope factor in the Cheng–Prusoff equation may

introduce additional errors (Lazareno and Birdsall, 1993). Therefore, the IC50 wasestimated without usingHill’s slope tominimize additional errors (eq. 5).WinNonlinProfessional 5.0.1 software (Pharsight Corporation, Mountain View, CA) was usedfor the nonlinear regression analysis using the equations described above.

Noncompartmental Pharmacokinetic Analysis. Standard noncompartmen-tal analyses were carried out using WinNonlin to calculate the pharmacokineticparameters (Gibaldi and Perrier, 1982), including the area under the plasmaconcentration–time curve (AUC) from time zero to infinity (AUCinf), eliminationclearance (CL), terminal half-life, mean residence time, and apparent volume ofdistribution at steady state (Vss).

The tissue-to-plasma concentration ratio of LB (i.e., Kp) was calculated usingeq. 7 (Sawada et al., 1985) after the administration of an intravenous bolus of LBand/or ATV to rats:

Kp ¼ AUCinf;tissue=AUCinf;plasma ð7Þ

where AUCinf,tissue represents the area under the drug concentration–time curve intissue and AUCinf,plasma represents the area under the drug concentration–timecurve in plasma after the intravenous bolus administration. For the calculation ofthe tissue-to-plasma partition coefficient (Kp), the contribution of trapped blood inthe collected tissue was corrected using the volume of blood trapped in tissuesamples.

Physiologically Based Pharmacokinetic Modeling and Simulation. In thisstudy, a physiologically based pharmacokinetic (PBPK) model (see the Appendix)was constructed assuming the simultaneous intravenous bolus administration forLB and ATV. The PBPK model consisted of nine tissues corresponding to thedifferent tissues of the body (lung, adipose, brain, heart, kidneys, muscle, liver,spleen, and gut) and the tissues were assumed to be connected by the circulatorysystem (arterial and venous side). Physiologic and anatomic variables, required inthe PBPK calculation, were obtained from the literature (Brown et al., 1997) [i.e.,essentially the default values found in Simcyp software (Jamei et al., 2009), version15, release 1; Simcyp Limited, Sheffield, UK]. In this calculation, the rate ofLB/ATV distribution to tissues, including the liver, was assumed to be perfusionrate limited and the standard mass balance differential equations were used withminor modifications (Lee et al., 2011) (see the Appendix).

After the adequacy of the constructed PBPK model (for single intravenousbolus administration of LB and ATV, respectively) (Supplemental Fig. 7) wasconfirmed with the Simcyp animal simulator (version 15, release 1), a simulationwas carried out with Berkeley Madonna software (version 8.3.18; University ofCalifornia, Berkeley, CA) since a DDI module for rats was not available in theSimcyp software. In this study, the fourth order of the Runge–Kutta method wasused for the numerical integration.

Statistical Analysis. When it was necessary to compare the mean valuesbetween/among groups, the unpaired t test or one-way analysis of variance(ANOVA), followed by the Tukey post hoc test, was used. In this study, data areexpressed as means 6 S.D. P , 0.05 was denoted as statistically significant.

Results

DDI between LB and ATV in Systemic and Tissue Pharmacokineticsin Rats

Themean plasma concentration-time profiles are shown in Fig. 1A foran intravenous bolus administration of 1 mg/kg LB to rats with or withoutsimultaneous administration of 5 mg/kg ATV. The LB concentrationprofile in the case of ATV coadministration was comparable to thatwithout ATV (Fig. 1A). The pharmacokinetic parameters, as estimatedfrom a standard moment analysis, are summarized in Table 1. The CL ofLBwith or without ATVwas 2.676 0.63 and 2.346 0.37ml/min per kg,respectively, and these values were not significantly different from eachother. The Vss for LB with and without ATV was 289 6 20 and 271 620 ml/kg, respectively. Furthermore, the secondary pharmacokineticparameters for LB (e.g., AUCinf, terminal half-life, and mean residencetime) were not affected by the coadministration of ATV. Similar to thecase of LB, ATV concentrations in rat plasma with or without LBco-administration were not changed, as shown in Fig. 1B and the summaryof pharmacokinetic parameters for ATV (Table 1). Collectively, these

250 Yim et al.

at ASPE

T Journals on O

ctober 5, 2020dm

d.aspetjournals.orgD

ownloaded from

observations suggest that there is no appreciable interaction in systemicpharmacokinetics when LB and ATV are simultaneously given in rats.The mean tissue concentration-time profiles for LB and ATV are

depicted in Fig. 2 for an intravenous bolus administration to rats at theLB dose (i.e., 1 mg/kg) with or without ATV coadministration (i.e.,5 mg/kg). The tissue-to-plasma partition coefficient (Kp) found in thisstudy, along with the volume of blood trapped in tissues for LB andATV, is listed in Table 2. When LB was administered intravenouslywithout ATV, the liver appeared to be the primary organ for distribution,consistent with our previous findings (Lee et al., 2015b). The estimatedKp,liver value for LB was 2.34 in the absence of ATV coadministration,whereas the value was reduced to 1.55 in the presence of ATVcoadministration. For the case of the distribution of LB to tissues otherthan the liver, ATV coadministration did not appear to significantly affectthe distribution of LB. For the case of ATVwithout LB coadministration,the liver was one of the primary organs for this distribution (i.e., Kp,liver =97.8 and Kp,gut = 125). However, the coadministration of LB reduced theKp,liver and Kp,gut values for ATV to 65.4 and 64.1, respectively. Theseobservations suggest that DDIs might be a possibility at the level of drugdistribution when LB and ATV are simultaneously administered.

Dose Dependence of the Liver-to-Plasma Concentration Ratio ofLB in Rats

In this study, different levels of LB in steady state were attained, andthe concentrations of LB in the plasma/liverwere determined, in an attempt toexamine the dependence ofKp,liver on LB dosage andATV coadministration.

The mean plasma concentration–time profiles are shown in Fig. 3A forintravenous infusions of LB to rats at rates of 50, 250, or 1000 ng/minwithand without simultaneous infusion of ATV at a rate of 10 mg/min. For allinfusion rates, the concentrations of LB in the plasma for 4, 5, and 6 hoursof infusionwere not statistically different among the three time points for agiven rate, confirming that a steady state had been reached by 4 hours afterthe infusion. It was also noted that ATV coadministration did not result ina change in LB concentration in plasma at a 6-hour infusion.The LB concentration was determined in the liver at 6 hours: the liver-

to-plasma concentration ratio for LBwithout ATV coadministration was3.14 6 0.46 (at 50 ng/min), 1.55 6 0.52 (at 250 ng/min), and 0.959 60.259 (at 1000 ng/min) (i.e., P, 0.001 by one-wayANOVA). In addition,the liver-to-plasma concentration ratio for LB with ATV coadministrationwas 1.38 6 0.39 (at 50 ng/min), 0.701 6 0.013 (at 250 ng/min), and0.714 6 0.184 (at 1000 ng/min), indicating that the coadministration ofATV significantly decreases the liver-to-plasma concentration ratio for LB(Fig. 3B). For the 1000-ng/min LB infusion condition, the liver-to-plasmaconcentration ratio for LB was not further decreased by ATV, probablybecause the transport of LB would be fully saturated by the highconcentration of LB. Collectively, these observations indicate that thedistribution of LB to the liver ismediated by saturable transport process(es)and that transport to the liver is inhibited by ATV in rats.

B/P and Plasma Protein Binding

The B/P ratios of LB were 0.6316 0.023, 0.6116 0.022, and 0.66760.077 (i.e., no statistical difference, by one-way ANOVA) at concentrations

Fig. 1. (A) Temporal profiles for the plasma concentration of LB after an intravenous bolus administration of 1 mg/kg LB alone (filled circles) and 1 mg/kg LB with 5 mg/kgATV (open circles) to rats. (B) Temporal profiles for the plasma concentration of ATV after an intravenous bolus administration of 5 mg/kg ATV alone (filled circles) and5 mg/kg ATV with 1 mg/kg LB (open circles) to rats. Data are expressed as means 6 S.D. of quadruplicate runs in (A) and pentaplicate runs in (B).

TABLE 1

Pharmacokinetic parameters for LB and ATV after an intravenous bolus administration of LB (1 mg/kg dose) with andwithout coadministration of 5 mg/kg ATV to rats, and after intravenous bolus administration of 5 mg/kg ATV with and

without 1 mg/kg LB to rats, respectively

Data are expressed as means 6 S.D. of quadruplicate runs for LB and pentaplicate runs for ATV.

ParameterLB ATV

LB LB Plus ATV ATV ATV Plus LB

AUCinf (mg · min/ml) 422 6 55 382 6 85 57.5 6 26.6 52.1 6 16.4Terminal half-life (min) 99.1 6 4.7 86.4 6 10.8 22.3 6 0.9 22.3 6 1.8Mean residence time (min) 117 6 18 111 6 19 16.2 6 1.9 16.3 6 1.4CL (ml/min per kg) 2.34 6 0.37 2.67 6 0.63 102 6 41 103 6 33Vss (ml/kg) 271 6 20 289 6 20 1610 6 600 1670 6 537

DDIs between Lobeglitazone and Atorvastatin in Rats 251

at ASPE

T Journals on O

ctober 5, 2020dm

d.aspetjournals.orgD

ownloaded from

of 0.1, 1, and 10 mM, respectively. Considering the lack of concentrationdependence in the distribution to blood cells, the average B/P for LB (i.e.,0.636) for the three concentrations was assumed to be representative and thevalue was used in subsequent calculations.The fraction of unbound LB in the rat plasma (fu,plasma) was also

independent in the LB concentration range from 0.3 to 10 mM. Therepresentative (i.e., averaged) fu,plasma value of LBwas then calculated as0.00267 6 0.00036. The fraction of unbound LB in the microsomalincubation medium (fu,MIC) was independent in the concentration rangefrom 0.5 to 50 mM, and the representative fu,MIC value of LB wasestimated to be 0.4796 0.063. However, the unbound fraction of LB at5 mM BSA in the transport medium (fu,BSA) increased with increasingLB concentration in the range from 0.5 to 20 mM (i.e., approximately

0.111–0.447). For the case of fu,BSA of ATV, the value was independentin the concentration range from 1 to 100 mM, and the representativevalue for fu,BSA of ATVwas estimated to be 0.5886 0.045. These valueswere used in subsequent calculations.

Stability of LB in the Presence and Absence of ATV in Rat LiverMicrosomes

The concentration-metabolic reaction rate profile for LB in rat livermicrosomes is shown in Fig. 4A. The Eadie–Hofstee transformation ofthe data is also presented as an inset in Fig. 4A. In general, the metabolicrate decreased proportionally with the increasingV/[LB] value, suggestingthat a saturable process is involved in the metabolic reaction. As a result,

Fig. 2. (A) Temporal profiles for plasma and tissue concentration of LB after an intravenous bolus administration of 1 mg/kg LB alone (open circles) and 1 mg/kg LB with5 mg/kg ATV (closed circles) to rats. (B) Temporal profiles for the plasma and tissue concentrations of ATV after intravenous bolus administration of 5 mg/kg ATV alone(filled triangles) and 5 mg/kg ATV with 1 mg/kg LB (open triangles) to rats. Data are expressed as means 6 S.D. of triplicate runs.

TABLE 2

Tissue-to-plasma partition coefficient (Kp) for LB and ATV in rats and the volume of blood trapped in rat tissues

Kp values were obtained from triplicate runs. Volumes of trapped blood in the tissue are expressed as means 6 S.D. of triplicate runs.

TissueLB ATV

Volume of Trapped Blood in TissuesLB LB Plus ATV ATV ATV Plus LB

ml/g tissueLungs 0.240 0.290 1.34 1.96 0.283 6 0.035Adipose 0.216 0.216 0.158 0.335 0.0309 6 0.0063Brain 0.0252 0.0296 0.0349 0.0270 0.0176 6 0.0014Heart 0.343 0.410 0.400 0.396 0.139 6 0.004Kidneys 0.406 0.478 2.51 3.28 0.209 6 0.083Muscle 0.109 0.170 0.480 0.586 0.0387 6 0.0082Liver 2.34 1.55 97.8 65.4 0.105 6 0.026Spleen 0.165 0.170 0.884 0.461 0.0430 6 0.0209Gut 0.234 0.196 125 64.1 0.107 6 0.017

Kp,tissue 5 (AUCinf,tissue/AUCinf,plasma).

252 Yim et al.

at ASPE

T Journals on O

ctober 5, 2020dm

d.aspetjournals.orgD

ownloaded from

assuming that simple Michaelis–Menten kinetics is involved as shown ineq. 1, a nonlinear regression analysis indicates that theKm,MIC andVmax,MIC

values are 4.37 6 0.90 mM and 996 6 110 nmol/min per milligram ofprotein, respectively. The inhibitory effect of ATV on the metabolicreaction rate of LB in rat liver microsomes was also determined (Fig. 4B)in this study. Based on a nonlinear regression analysis using eq. 2, theKi,MIC

value was estimated to be 27.06 5.7 mM. Using the kinetic characteristicsobtained in this study, the estimated extraction ratio was calculated to be0.03 for LB. Therefore, we assumed that a correction of Kp,liver for LB wasnot necessary and this value is reported without correction.

Uptake of LB in rOATP1B2-Transfected MDCK Cells andIsolated Rat Hepatocytes

The unbound concentration-uptake rate profile for LB in mock cellsand rOATP1B2 cells is shown in Fig. 5A. Since both LB and ATV are

drugs that bind highly to plasma proteins (i.e., fu,plasma values of 0.00267and 0.0567, respectively), it is possible that they might also bindnonspecifically to the surfaces of the experimental apparatus, therebycausing experimental artifacts. In this study, we chose to include BSA inthe incubation medium and washing buffer to prevent/minimize non-specific binding to the experimental apparatus and/or cell surface(Takeuchi et al., 2011). As a result, the fu,BSA values of LB weremultiplied by the total LB concentration to estimate the concentration ofunbound LB in the transport medium. Using this experimental designand a nonlinear regression analysis, Km,OATP and Jmax,OATP values were2.06 6 0.45 mM and 2.42 6 1.27 pmol/min per 106 cells, respectively(eqs. 3 and 4).The profile for the unbound concentration-uptake rate for LB in rat

hepatocytes is shown in Fig. 5B. Similar to the case of transfectedcells, the PS, Km,hepa, and Jmax,hepa values were estimated to be 21.066.0 ml/min per 106 cells, 16.36 10.0 mM, and 6376 384 pmol/min per

Fig. 3. (A) Temporal profiles for the plasma concentration of LB after an intravenous infusion at dosing rates of 50 ng/min, 250 ng/min, and 1000 ng/min (shown as filledcircles, triangles, and diamonds for LB alone and as open circles, triangles, and diamonds for LB with 10 mg/min ATV, respectively). (B) Liver-to-plasma concentration ratioof LB at 6 hours after an intravenous infusion at the dosing rates of 50, 250, and 1000 ng/min to rats. Asterisks in the black circles indicate statistical differences from thelowest LB dose (i.e., by one-way ANOVA, followed by the Tukey post hoc test), and asterisks in the white circles indicate statistical differences from the control (i.e., theabsence of ATV) by the unpaired t test (*P , 0.05; **P , 0.01; ***P , 0.001). Data are expressed as means 6 S.D. of quadruplicate runs.

Fig. 4. (A) The concentration-metabolic reaction rate curve for LB in an incubation with rat liver microsomes. The solid line was generated by the best-fit parametersobtained from the nonlinear regression analysis based on Michaelis–Menten kinetics and indicates the linearly regressed line of the data (eq. 2). The inset represents anEadie–Hofstee plot of the concentration-metabolic reaction rate curve. (B) The inhibitory effect of ATV on the metabolic reaction rate of LB (5 mM) for an incubation withrat liver microsomes. The solid line was generated by the best-fit parameters obtained from the nonlinear regression analysis based on eq. 3. Each symbol represents themean 6 S.D. of triplicate runs.

DDIs between Lobeglitazone and Atorvastatin in Rats 253

at ASPE

T Journals on O

ctober 5, 2020dm

d.aspetjournals.orgD

ownloaded from

106 cells, respectively (eqs. 3 and 4), based on a nonlinear regressionanalysis. Based on the inhibitory effect of ATV on the transport rate of LBin rOATP1B2 cells (Fig. 5C) and rat hepatocytes (Fig. 5D), a nonlinearregression analysis indicated that the IC50,OATP and IC50,hepa values were0.473 6 0.278 and 0.984 6 0.812 mM, respectively. Based on therelationship given by Cheng and Prusoff (1973), the Ki,OATP and Ki,hepa

values were estimated to be 0.2966 0.174 and 0.8756 0.739mMby eq.6. Based on nonlinear regression analysis of virtual data (SupplementalTable 1) and eqs. 5 and 6, we found that the assigned and calculated Ki (inmM) were 0.875 and 0.833 6 0.520 (Supplemental Table 2). Theseobservations suggest that the variability found in the current study doesnot significantly affect the outcome of the estimation of Ki values.

Simulation of LB Pharmacokinetics after a Single IntravenousDose and DDIs Using a PBPK Model

In this study, we attempted to use a PBPKmodel to predict the plasmaand tissue kinetics of LB at a dose of 1 mg/kg (Fig. 2A) using primarilyin vitro experimental results (Table 3). Our calculations indicate that theprofiles for the plasma and liver concentration could be reasonablypredicted when LB was given to rats intravenously without ATVcoadministration (Fig. 6). Accordingly, the ratios of AUCplasma to the

AUCliver (i.e., Kp,liver) were calculated to be 2.186 0.09 (Table 4) fromin vitro data (i.e., Km,MIC, Vmax,MIC, PS, Km,hepa, and Jmax,hepa), which isreasonably close to the experimentally determined value (i.e., 2.34;Table 4).Assuming that the current PBPK model for LB is predictive, the

PBPK model for LB may also be used to mimic in vivo DDIs betweenLB and ATV by using the metabolic and transport kinetic parameters forLB obtained from in vitro studies. The observed and simulated LBconcentration profiles for the plasma and liver are presented in Fig. 6,and their pharmacokinetic parameters are summarized in Table 4. Forthe case of the Kp,liver value of LB with ATV, the predicted value (1.71)was similar to that based on an experimental determination (1.55) whenthe Ki,OATP value was used in the calculation.

Discussion

The possibility of a DDI occurring between LB and ATV wasexamined, considering the likelihood of the concurrent development ofhyperlipidemia/T2DM (Ervin, 2009) and overlapping metabolic (CYP3A)/transport (rOATP1B2) pathways for the two drugs. No apparent DDI wasfound (Fig. 1; Table 1) between the two drugs in their respective systemicpharmacokinetics, even though interactions between LB and ATV were

Fig. 5. (A) The unbound concentration-uptake rate curve for LB in mock-MDCK cells (filled circles) and rOATP1b2-expressing MDCK cells (open circles). (B) Unboundconcentration-uptake rate curve for LB in rat hepatocytes at 4�C (filled circles) and 37�C (open circles). The solid lines were generated by the best-fit parameters obtainedfrom the simultaneous nonlinear regression analysis based on eqs. 4 and 5, respectively. (C and D) The inhibitory effect of ATV on the uptake of LB in rOATP1b2-expressing MDCK cells (C) and rat hepatocytes (D). The solid line was generated by the best-fit parameters obtained from the nonlinear regression analysis based on eq. 6.Asterisks indicate statistical differences (**P, 0.01; ***P, 0.001) from the control group (i.e., without ATV) by one-way ANOVA, followed by the Dunnett post hoc test.Each symbol represents the mean 6 S.D. of triplicate runs in (A) to (C) and octuplicate runs in (D).

254 Yim et al.

at ASPE

T Journals on O

ctober 5, 2020dm

d.aspetjournals.orgD

ownloaded from

both detected in vitro at the metabolic level (Fig. 4B) and at transport (Fig.5, C and D). Despite its high lipophilicity, the volume of distribution waslimited for LB (i.e., 2716 20 ml/kg), likely because it extensively bindsto plasma proteins. In addition, the reduction in Kp,liver for LB (i.e., from2.34 to 1.55) by ATV did not appear to be significant in the total Vss,considering the fact that the contribution of the reduced distribution of LBto the liver would be approximately 27ml/kg, which is relatively small forthe Vss of 271 ml/kg. In other clinical DDI studies involving LB withmetformin (i.e., T2DM agent), LB with warfarin (i.e., an anticoagulantwith a narrow therapeutic index), or LB with amlodipine (i.e., an

antihypertensive agent) (Shin et al., 2012; Jung et al., 2015; Kim et al.,2015), no apparent changes in the systemic pharmacokinetics was found.In contrast, however, the liver concentration of LB was significantlydecreased by the coadministration of ATV (Fig. 2; Table 2). Aconcentration-dependent decrease was readily apparent for theKp,liver ofLB for different levels of steady-state LB concentrations (Fig. 3) and theextent of liver distribution was determined. These findings are the firstindication of this “liver-specific” DDI between LB and ATV in rats.Other tissue-specific DDIs have also been noted. A decrease in theconcentration of metformin in the liver and kidney was reported, without

TABLE 3

Summary of kinetic parameters for LB and ATV used in PBPK calculation

Parameter LB ATV Reference/Comment

Molecular weight (g/mol) 480.5 558.6 DrugBank (http://www.drugbank.ca)Log P 4.3 5.7 ChemAxon (https://www.chemaxon.com)Compound type Zwitterions Monoprotic acidpKa

Acid 7.61 4.3 ChemAxon (https://www.chemaxon.com)Base 3.96

fu,plasma 0.00267 0.0567 Determineda; Watanabe et al. (2010)b

B/P 0.636 1.2 Determineda; Watanabe et al. (2010)b

DistributionKp 0.0252–2.34 0.0270–97.8 Determined

Distribution to liverPS (ml/min per 106 cells) 21.0 6 6.0 DeterminedJmax,hepa (pmol/min per 106 cells) 637 6 384 DeterminedKm,hepa (mM) 16.3 6 10.0 Determined

EliminationVmax,MIC (pmol/min per mg protein) 996 DeterminedKm,MIC (mM) 4.37 Determinedfu,MIC 0.479 0.557 Determineda; Watanabe et al. (2010)b

CLu,int (ml/min) 1530 Calculated using the CL value of ATV fromintravenous bolus study with LB coadministration

InhibitionKi,MIC (mM) 27.0 DeterminedKi,hepa (mM) 0.875 6 0.739 DeterminedKi,OATP (mM) 0.296 6 0.174 Determinedfu,BSA 0.111–0.447 0.588 Determined

aFor LB.bFor ATV.

Fig. 6. (A) Observed and simulated plasma concentration–time profiles for LB after intravenous bolus administration of 1 mg/kg LB with or without 5 mg/kg ATV to rats.Black circles, solid line, white circles, and the dashed line represent the observed (LB alone), simulated (LB alone), observed (LB with ATV), and simulated (LB with ATV,with the Ki,OATP value) plasma concentrations of LB, respectively. (B) Observed and simulated liver concentration–time profiles for LB after intravenous bolusadministration of 1 mg/kg LB with or without 5 mg/kg ATV to rats. Black circles, solid line, white circles, and the dashed line represent the observed (LB alone), simulated(LB alone), observed (LB with ATV), and simulated (LB with ATV, with the Ki,OATP value) liver concentration of LB, respectively. Symbols represent the mean 6 S.D. oftriplicate runs. The input parameters for LB and ATV are summarized in Table 3.

DDIs between Lobeglitazone and Atorvastatin in Rats 255

at ASPE

T Journals on O

ctober 5, 2020dm

d.aspetjournals.orgD

ownloaded from

any apparent change in the systemic pharmacokinetics, for thecoadministration of cimetidine or pyrimethamine: this tissue-specific change was mediated by the inhibition of mouse multidrugand toxin extrusion 1 protein (Ito et al., 2012; Shingaki et al., 2015).In addition, an increase in the donepezil concentration in theheart and brain was reported, without any appreciable change insystemic pharmacokinetics, after cilostazol coadministration, prob-ably by inhibition of the breast cancer resistance protein (Takeuchiet al., 2016)The metabolic rate of LB was also decreased by ATV (Fig. 4), although

the Ki,MIC values for ATV (i.e., 15.0 mM, unbound concentration) asestimated from invitro datawere relatively high. Since the estimated unboundliver concentration of ATV in vivo would be close to 71.8 nM (calculatedfrom the observed Cmax,liver of ATV when ATV was administered withLB in the tissue study by multiplying fu,plasma,ATV/Kp,liver,ATV), themaximum concentration of free ATV in the liver (i.e., 1.27 mM) wouldstill be approximately 12-fold lower than the calculatedKi,MIC value, evenif all ATV in the plasma is present in the unbound form. Taken together,the hepatic metabolism of LB is not likely to be significantly inhibited bythe ATV dose administered in this study. In addition, the reducedmetabolic activity by ATV from in vitro studies did not appear to beconsistent with in vivo observations in which a lower hepatic concentra-tion of LB by ATV coadministration was found. Therefore, we did notpursue the mechanism responsible for the metabolic interaction betweenLB and ATV in vitro any further.It is now well established that rOATP1B2 is primarily expressed

in the liver in rats and ATV is one of its representative substrates(Knauer et al., 2010). These findings indicate that the transport ofLB in rOATP1B2-expressing cells had a Km,OATP of 2.06 6 0.45mM with a Ki,OATP by ATV of 0.2966 0.174 mM (Fig. 5, A and C).The uptake of LB was saturable and inhibited by ATV, with Ki,hepa

values of 0.875 6 0.739 mM in the study involving isolated rathepatocytes. Collectively, our observations suggest that the distributionof LB to the liver is mediated by hepatic uptake transporter(s), such asrOATP1B2, and the liver distribution of LB is inhibited by thecoadministration of ATV, a rOATP1B2 substrate. As corroborativeevidence for the involvement of rOATP1B2 in liver-specific DDIs, wefound that the Kp,liver of ATV declined from 97.8 to 65.4 by thecoadministration of LB (Table 2).Based on in vitro studies, LB appeared to be transported to the

liver via passive diffusion and carrier-mediated transport(s).Assuming that a similar process occurs in vivo, evaluating theKp,liver,pass of LB, representing the liver-to-plasma concentrationratio for LB by passive diffusion only, was necessary. We used themethod of Rodgers and Rowland (2006) for this (i.e., 0.794 forLB). Interestingly, the current value was comparable to the Kp,liver

value measured from the intravenous infusion study at the highest

dose of LB with ATV (i.e., 0.714 6 0.184; Fig. 3B). The ATVplasma concentration was approximately 4.8 mM and the estimatedKi,OATP value by ATV was approximately 0.3 mM in rOATP1B2-expressing cells. Therefore, carrier-mediated transport by rOATP1B2appears to be fully inhibited under the above conditions. In subsequentcalculations, the computational estimate was assumed to be adequate andwas used.It has been reported that “albumin-mediated uptake mechanism(s)” in

hepatocytes may occur for highly bound drugs such as LB and ATV(Poulin et al., 2016). The calculated fu,plasma,surf value for ATV wasapproximately 5-fold higher than fu,plasma for ATV (i.e., fu,plasma,surf andfu,plasma for ATV were 0.280 and 0.0567, respectively), indicating thatthe potency of ATV for inhibiting the hepatic uptake of LB might beunderestimated by as much as 5-fold if the fu,plasma of ATV is usedinstead of the fu,plasma,surf of ATV (Poulin and Haddad, 2015). In thisstudy, the simulation using fu,plasma,surf, instead of fu,plasma, yieldedimproved predictions (Supplemental Fig. 8).The theoretical Kp,liver assuming LB without the coadministration

of ATV was calculated to be 2.18 6 0.09, comparable to the observedKp,liver value (i.e., 2.34). In addition, the estimated Kp,liver assuming thatLB was coadministered with ATV was 1.716 0.13, close to 1.55 undera similar in vivo situation. Taken together, the simulatedKp,liver valuesfor LB with or without ATV were in reasonable agreement with theobserved data. It is noteworthy that the kinetic calculation wasprimarily based on parameters obtained from in vitro studies usingrOATP1B2-expressing cells. Therefore, the liver-specific DDI be-tween LB and ATV may be due to the interaction of rOATP1B2between the two drugs. Since rOATP1B2 is known to be functionallysimilar to hOATP1B1 (Hagenbuch and Meier, 2003), a similarinteraction (e.g., the lack of a DDI in systemic exposure while theliver concentration for LB reduced by ATV) may occur between LBand ATV in humans. This aspect of a DDI involving LB and ATVwarrants further investigation.Although LB was reported to be a PPARg agonist, it possessed a

strong affinity for PPARa (i.e., EC50 = 0.02 mM) (Lee et al.,2007). It was reported that fenofibrate, a PPARa agonist, mayimprove insulin sensitivity, probably by interactions between thedrug and its receptor, as shown in an animal model of diabetes andobesity (Koh et al., 2003). Interestingly, the effective dose of LBwas significantly lower than that for rosiglitazone and pioglita-zone [i.e., (in effective dose) LB, 0.5 mg/d; rosiglitazone, 2–8 mg/d; pioglitazone, 15–45 mg/d] (Balfour and Plosker, 1999;Hauner, 2002; Park et al., 2014). Therefore, it is possible that thehigher potency of LB may be partly mediated by the interaction ofLB with PPARa. Assuming this, the liver may be regarded as anadditional site of action for LB, since the receptor is highly expressedin the liver (Braissant et al., 1996). Considering that the coadminis-tration of LB and ATV led to a liver-specific decrease in thedistribution of the two drugs, efficacy could be affected for the twodrugs. In particular, since the liver is likely to be closely linked to thepharmacological activity of the drugs (i.e., the primary site of actionfor ATV and potentially the secondary site for LB), additional studiesmay be necessary to confirm the impact of the liver-specific DDI onthe pharmacological activity.In conclusion, ATV coadministration led to a significant decrease in

the liver concentration of LB with no appreciable change in its plasmaconcentration in rats. Metabolic interactions between LB and ATVappeared to cause only minor kinetic changes and are not likely toaccount for the liver-specific decrease in LB levels in vivo. LB was asubstrate for rOATP1B2 and carrier-mediated transport was inhibited byATV, even at low concentrations. The transport variables obtained fromin vitro studies and the PBPK model assuming a carrier-mediated

TABLE 4

Observed and simulated pharmacokinetic parameters for LB for an intravenousbolus administration of 1 mg/kg LB alone and 1 mg/kg LB with 5 mg/kg ATV

to rats

Simulated data are expressed as means6 S.D. of three virtual rats from in vitro experiments intriplicate runs.

ParameterObserved Simulated

LB LB Plus ATV LB LB Plus ATVa

AUCinf,plasma (mg · min/ml) 435 509 493 6 0.08 492 6 29AUCinf,liver (mg × min/ml) 1020 789 1070 6 46 842 6 68Kp,liver 2.34 1.55 2.18 6 0.09 1.71 6 0.13

aValues were determined using Ki,OATP values. Kp,liver values were 1.83 6 0.22 using Ki,hepa

values.

256 Yim et al.

at ASPE

T Journals on O

ctober 5, 2020dm

d.aspetjournals.orgD

ownloaded from

transport process in distribution to the liver were adequate for mimickingthe in vivo pharmacokinetics of LB with or without ATV coadminis-tration. Considering that LB and ATVmight be used in combination, theliver-specific DDI caused by the combination found in this study may betherapeutically relevant.

Appendix

In this study, the pharmacokinetics of LB was assumed to beadequately explained by taking nine typical tissues into consideration.Thus, the distribution of LB to the lung compartment may be expressedas shown in eq. A1:

VLU ×dCLU

dt¼ QCO ×

�CVB 2

CLU × B=PKp;LU

�ðA1Þ

where VLU is the volume of lung, CLU is the drug concentration inthe lung, QCO is the cardiac output, CVB is the venous bloodconcentration of drug, B/P is the blood-to-plasma concentrationratio, and Kp,LU is the tissue-to-plasma partition coefficient of thelung.In the venous blood compartment (eq. A2):

VVB ×dCVB

dt¼ QAD ×

CLU × B=PKp;AD

þ QBR ×CBR × B=P

Kp;BR

þ QHE ×CHE × B=PKp;HE

þ QKI ×CKI × B=PKp;KI

þ QMU ×CMU × B=PKp;MU

þ QLI ×CLI × B=P

Kp;LI

þ QSP ×CSP × B=PKp;SP

þ QGU ×CGU × B=P

Kp;GU

þ QRE ×CRE × B=P

Kp;RE2QCO × CVB þ Dose rate ðA2Þ

where VVB is the volume of the venous blood; QAD, QBR, QHE, QKI,QMU, QLI, QSP, QGU, and QRE are the blood flows to the adipose, brain,heart, kidney, muscle, liver, spleen, gut, and the rest of body,respectively; CAD, CBR, CHE, CKI, CMU, CLI, CSP, CGU, and CRE arethe drug concentration of the adipose, brain, heart, kidney, muscle, liver,spleen, gut, and the rest of body, respectively; and Kp,AD, Kp,BR, Kp,HE,Kp,KI, Kp,MU, Kp,LI, Kp,SP, Kp,GU, and Kp,RE are the tissue-to-plasmapartition coefficient of the adipose, brain, heart, kidney, muscle, liver,spleen, gut, and the rest of the body, respectively. Since the kinetics ofthe rest of the body were negligible for the pharmacokinetics of both LBand ATV (i.e., Vss calculated by the Øie–Tozer equation for the ninemajor tissues and plasma accounted for 89.4% and 167% of Vss based onthe moment analysis of the plasma concentration profiles of LB andATV, respectively) (Øie and Tozer, 1979), the Kp,RE value wasarbitrarily set to 0.001. Dose rate refers the dosing rate of LB and/orATV.In the arterial blood compartment of LB and ATV (eq. A3):

VAB ×dCAB

dt¼ QCO ×

�CLU × B=P

Kp;LU2CAB

�ðA3Þ

where VAB is the arterial blood volume and CAB and CAP are thedrug concentration in arterial blood and plasma, respectively. Inour previous report (Lee et al., 2015b), the excretion of LB to the

bile was found to be negligible (i.e., 0.31%–0.37% of total dose ofLB). Therefore, we assumed that the elimination of LB and ATV ismainly by hepatic metabolism (Lennernäs, 2003; Lee et al.,2015b).In the liver compartment of LB and ATV (eq. A4):

VLI ×dCLI

dt¼ ðQLI 2QSP 2QGUÞ × CAB þ QSP ×

CSP × B=PKp;SP

þ QGU ×CGU × B=P

Kp;GU2QLI ×

CLI × B=PKp;LI

2CLu;int × fu;plasma × CLI

Kp;LI

ðA4Þ

Metabolism of LB without ATV (eq. A5):

CLu;int ¼ Vmax;MIC

Km;MIC × fu;MIC þ fu;plasma × CLI

Kp;LI

× MPPGL × LW ðA5Þ

Metabolism of LB with ATV (eq. A6):

CLu;int ¼ Vmax;MIC

Km;MIC × fu;MIC ×�1þ

fu;plasma;ATV × CLI;ATVKp;LI;ATV

Ki;MIC × fu;MIC;ATV

�þ fu;plasma × CLI

Kp;LI

× MPPGL × LW

ðA6Þ

Distribution to the liver of LB without ATV (eq. A7):

Kp;LI ¼

�PSþ Jmax;hepa × HPGL × LW

Km;hepaþfu;plasma;surf × CAP

�× Kp;LI;pass

PSðA7Þ

Distribution to the liver of LB with ATV (eq. A8):

Kp;LI ¼

�PSþ Jmax;hepa × HPGL × LW

Km;hepa ×

�1þfu;plasma;surf;ATV × CAP;ATV

Ki;hepa × fu;BSA;ATV

�þfu;plasma;surf × CAP

�× Kp;LI;pass

PSðA8Þ

In the liver compartment of ATV (eq. A9):

VLI ×dCLI

dt¼ ðQLI 2QSP 2QGUÞ × CAB þ QSP ×

CSP × B=PKp;SP

þ QGU ×CGU × B=PKp;GU

2QLI ×CLI × B=PKp;LI

ðA9Þ

where VLI is the liver volume; CLu,int is the unbound intrinsicclearance; Vmax,MIC is the maximal rate in the microsomal incubation;Km,MIC is the Michaelis–Menten constant in the microsomal incubation;and LW, MPPGL, and HPGL were the liver weight (9 g), microsomalprotein per gram of liver (46 mg/g liver), and hepatocellularity per gramof liver (108 � 106 hepatocytes/g liver). For scaling the kineticparameters from in vitro experiments to an in vivo situation, the Simcypdefault values were used.Kp,LI,pass is the tissue-to-plasma partition coefficient of the liver with

when drug transport into and out of tissue is mediated only by symmetricpassive diffusion (i.e., theoretically the ratio of the free fraction in plasmato that in liver; fu,plasma,surf/fu,liver), where fu,plasma,surf is the unboundfraction of drug at cell surface in plasma (Poulin et al., 2016),respectively (eq. A10). Kp,liver,pass was calculated using mechanistic insilico prediction methods (Rodgers and Rowland, 2006) within Simcyp,

DDIs between Lobeglitazone and Atorvastatin in Rats 257

at ASPE

T Journals on O

ctober 5, 2020dm

d.aspetjournals.orgD

ownloaded from

Authorship ContributionsParticipated in research design: Yim, Chung.Conducted experiments: Yim, Jeong, S.-Y. Lee, Pyeon, Ryu.Contributed new reagents or analytic tools: J.-H. Lee.Performed data analysis: Yim, Jeong, Maeng, Chung.Wrote or contributed to the writing of the manuscript: Yim, Jeong, J.-H. Lee,

K.-R. Lee, Maeng, Chung.

References

Amundsen R, Christensen H, Zabihyan B, and Asberg A (2010) Cyclosporine A, but not tacro-limus, shows relevant inhibition of organic anion-transporting protein 1B1-mediated transport ofatorvastatin. Drug Metab Dispos 38:1499–1504.

Balfour JAB and Plosker GL (1999) Rosiglitazone. Drugs 57:921–930, discussion 931–932.Braissant O, Foufelle F, Scotto C, Dauça M, and Wahli W (1996) Differential expression ofperoxisome proliferator-activated receptors (PPARs): tissue distribution of PPAR-alpha, -beta,and -gamma in the adult rat. Endocrinology 137:354–366.

Brown RP, Delp MD, Lindstedt SL, Rhomberg LR, and Beliles RP (1997) Physiological parametervalues for physiologically based pharmacokinetic models. Toxicol Ind Health 13:407–484.

Cattori V, Hagenbuch B, Hagenbuch N, Stieger B, Ha R, Winterhalter KE, and Meier PJ (2000)Identification of organic anion transporting polypeptide 4 (Oatp4) as a major full-length isoformof the liver-specific transporter-1 (rlst-1) in rat liver. FEBS Lett 474:242–245.

Cheng Y and Prusoff WH (1973) Relationship between the inhibition constant (K1) and theconcentration of inhibitor which causes 50 per cent inhibition (I50) of an enzymatic reaction.Biochem Pharmacol 22:3099–3108.

Da Violante G, Zerrouk N, Richard I, Provot G, Chaumeil JC, and Arnaud P (2002) Evaluation ofthe cytotoxicity effect of dimethyl sulfoxide (DMSO) on Caco2/TC7 colon tumor cell cultures.Biol Pharm Bull 25:1600–1603.

DeLean A, Munson PJ, and Rodbard D (1978) Simultaneous analysis of families of sigmoidalcurves: application to bioassay, radioligand assay, and physiological dose-response curves. Am JPhysiol 235:E97–E102.

Ervin RB (2009) Prevalence of metabolic syndrome among adults 20 years of age and over, by sex,age, race and ethnicity, and body mass index: United States, 2003-2006. Natl Health Stat Rep 13:1–7.

Gaudino M and Levitt MF (1949) Inulin space as a measure of extracellular fluid. Am J Physiol157:387–393.

Gibaldi M and Perrier D (1982) Pharmacokinetics, Marcel Dekker, New York.Gu Q, Dillon CF, and Burt VL (2010) Prescription drug use continues to increase: U.S. prescriptiondrug data for 2007-2008. NCHS Data Brief 42:1–8.

Guengerich FP, Brian WR, Iwasaki M, Sari MA, Bäärnhielm C, and Berntsson P (1991) Oxidationof dihydropyridine calcium channel blockers and analogues by human liver cytochrome P-450IIIA4. J Med Chem 34:1838–1844.

Hagenbuch B and Meier PJ (2003) The superfamily of organic anion transporting polypeptides.Biochim Biophys Acta 1609:1–18.

Hagenbuch B and Meier PJ (2004) Organic anion transporting polypeptides of the OATP/ SLC21family: phylogenetic classification as OATP/ SLCO superfamily, new nomenclature andmolecular/functional properties. Pflugers Arch 447:653–665.

Hauner H (2002) The mode of action of thiazolidinediones. Diabetes Metab Res Rev 18 (Suppl 2):S10–S15.

Hsyu PH, Schultz-Smith MD, Lillibridge JH, Lewis RH, and Kerr BM (2001) Pharmacokineticinteractions between nelfinavir and 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitorsatorvastatin and simvastatin. Antimicrob Agents Chemother 45:3445–3450.

Iglay K, Hannachi H, Joseph Howie P, Xu J, Li X, Engel SS, Moore LM, and Rajpathak S (2016)Prevalence and co-prevalence of comorbidities among patients with type 2 diabetes mellitus.Curr Med Res Opin 32:1243–1252.

Ito S, Kusuhara H, Yokochi M, Toyoshima J, Inoue K, Yuasa H, and Sugiyama Y (2012)Competitive inhibition of the luminal efflux by multidrug and toxin extrusions, but notbasolateral uptake by organic cation transporter 2, is the likely mechanism underlying thepharmacokinetic drug-drug interactions caused by cimetidine in the kidney. J Pharmacol ExpTher 340:393–403.

Jamei M, Marciniak S, Feng K, Barnett A, Tucker G, and Rostami-Hodjegan A (2009) The Simcyppopulation-based ADME simulator. Expert Opin Drug Metab Toxicol 5:211–223.

Jung JA, Lee SY, Kim TE, Kim JR, Kim C, Huh W, and Ko JW (2015) Lack of the effect oflobeglitazone, a peroxisome proliferator-activated receptor-g agonist, on the pharmacokineticsand pharmacodynamics of warfarin. Drug Des Devel Ther 9:737–743.

Kaminsky LS and Zhang ZY (1997) Human P450 metabolism of warfarin. Pharmacol Ther 73:67–74.

Kim B, Shin HS, Kim JR, Lim KS, Yoon SH, Yu KS, Shin SG, Jang IJ, and Cho JY (2012)Quantitative and qualitative analysis of CKD-501, lobeglitazone, in human plasma and urineusing LC–MS/MS and its application to a pharmacokinetic study. Chromatographia 75:671–677.

Kim BY, Ahn JB, Lee HW, Kang SK, Lee JH, Shin JS, Ahn SK, Hong CI, and Yoon SS (2004)Synthesis and biological activity of novel substituted pyridines and purines containing 2,4-thiazolidinedione. Eur J Med Chem 39:433–447.

Kim CO, Sil Oh E, Kim C, and Park MS (2015) Pharmacokinetic interaction between amlodipineand lobeglitazone, a novel peroxisome proliferator-activated receptor-g agonist, in healthysubjects. Clin Ther 37:1999–2006.e1.

Kim SG, Kim DM, Woo JT, Jang HC, Chung CH, Ko KS, Park JH, Park YS, Kim SJ, and Choi DS(2014) Efficacy and safety of lobeglitazone monotherapy in patients with type 2 diabetes mellitusover 24-weeks: a multicenter, randomized, double-blind, parallel-group, placebo controlled trial.PLoS One 9:e92843.

Knauer MJ, Urquhart BL, Meyer zu Schwabedissen HE, Schwarz UI, Lemke CJ, Leake BF, KimRB, and Tirona RG (2010) Human skeletal muscle drug transporters determine local exposureand toxicity of statins. Circ Res 106:297–306.

Koh EH, Kim MS, Park JY, Kim HS, Youn JY, Park HS, Youn JH, and Lee KU (2003) Perox-isome proliferator-activated receptor (PPAR)-a activation prevents diabetes in OLETF rats:comparison with PPAR-g activation. Diabetes 52:2331–2337.

Kotani N, Maeda K, Watanabe T, Hiramatsu M, Gong LK, Bi YA, Takezawa T, Kusuhara H,and Sugiyama Y (2011) Culture period-dependent changes in the uptake of transporter substratesin sandwich-cultured rat and human hepatocytes. Drug Metab Dispos 39:1503–1510.

Laufs U, Karmann B, and Pittrow D (2016) Atorvastatin treatment and LDL cholesterol targetattainment in patients at very high cardiovascular risk. Clin Res Cardiol 105:783–790.

Lazareno S and Birdsall NJ (1993) Estimation of competitive antagonist affinity from functionalinhibition curves using the Gaddum, Schild and Cheng-Prusoff equations. Br J Pharmacol 109:1110–1119.

Lee HW, Ahn JB, Kang SK, Ahn SK, and Ha DC (2007) Process development and scale-up ofPPAR a/g dual agonist lobeglitazone sulfate (CKD-501). Org Process Res Dev 11:190–199.

Lee HW, Kim BY, Ahn JB, Kang SK, Lee JH, Shin JS, Ahn SK, Lee SJ, and Yoon SS (2005)Molecular design, synthesis, and hypoglycemic and hypolipidemic activities of novel pyrimidinederivatives having thiazolidinedione. Eur J Med Chem 40:862–874.

Lee JH, Ahn SH, Maeng HJ, Lee W, Kim DD, and Chung SJ (2015a) The identification oflobeglitazone metabolites in rat liver microsomes and the kinetics of the in vivo formation of themajor metabolite M1 in rats. J Pharm Biomed Anal 115:375–382.

Lee JH, Noh CK, Yim CS, Jeong YS, Ahn SH, Lee W, Kim DD, and Chung SJ (2015b)Kinetics of the absorption, distribution, metabolism, and excretion of lobeglitazone, a novelactivator of peroxisome proliferator-activated receptor gamma in rats. J Pharm Sci 104:3049–3059.

Lee JH, Woo YA, Hwang IC, Kim CY, Kim DD, Shim CK, and Chung SJ (2009) Quantification ofCKD-501, lobeglitazone, in rat plasma using a liquid-chromatography/tandem mass spectrom-etry method and its applications to pharmacokinetic studies. J Pharm Biomed Anal 50:872–877.

Lee KR, Chae YJ, Maeng HJ, Lee J, Kim DD, Chong S, Shim CK, and Chung SJ (2011)Physiologically based pharmacokinetic modeling of SNU-0039, an anti-Alzheimer’s agent, inrats. J Pharmacokinet Pharmacodyn 38:637–651.

Lennernäs H (2003) Clinical pharmacokinetics of atorvastatin. Clin Pharmacokinet 42:1141–1160.Øie S and Tozer TN (1979) Effect of altered plasma protein binding on apparent volume ofdistribution. J Pharm Sci 68:1203–1205.

Park JE, Kim KB, Bae SK, Moon BS, Liu KH, and Shin JG (2008) Contribution of cytochromeP450 3A4 and 3A5 to the metabolism of atorvastatin. Xenobiotica 38:1240–1251.

Park MK, Kim TE, Kim J, Kim C, Yoon SH, Cho JY, Jang IJ, Yu KS, and Lim KS (2014)Tolerability and pharmacokinetics of lobeglitazone, a novel peroxisome proliferator-activatedreceptor-g agonist, after a single oral administration in healthy female subjects. Clin DrugInvestig 34:467–474.

Poulin P (2015) A paradigm shift in pharmacokinetic-pharmacodynamic (PKPD) modeling: rule ofthumb for estimating free drug level in tissue compared with plasma to guide drug design.J Pharm Sci 104:2359–2368.

Poulin P, Burczynski FJ, and Haddad S (2016) The role of extracellular binding proteins in thecellular uptake of drugs: impact on quantitative in vitro-to-in vivo extrapolations of toxicity andefficacy in physiologically based pharmacokinetic-pharmacodynamic research. J Pharm Sci 105:497–508.

Poulin P and Haddad S (2015) Albumin and uptake of drugs in cells: additional validation exercisesof a recently published equation that quantifies the albumin-facilitated uptake mechanism(s) inphysiologically based pharmacokinetic and pharmacodynamic modeling research. J Pharm Sci104:4448–4458.

Poulin P, Kenny JR, Hop CECA, and Haddad S (2012) In vitro-in vivo extrapolation of clearance:modeling hepatic metabolic clearance of highly bound drugs and comparative assessment withexisting calculation methods. J Pharm Sci 101:838–851.

Rodgers T and Rowland M (2006) Physiologically based pharmacokinetic modelling 2: predictingthe tissue distribution of acids, very weak bases, neutrals and zwitterions. J Pharm Sci 95:1238–1257.

Saito T, Zhang ZJ, Ohtsubo T, Noda I, Shibamori Y, Yamamoto T, and Saito H (2001) Homo-zygous disruption of the mdrla P-glycoprotein gene affects blood-nerve barrier function in miceadministered with neurotoxic drugs. Acta Otolaryngol 121:735–742.

and fu,plasma,surf was calculated using eq. A10 (Poulin et al., 2012, 2016;Yun and Edginton, 2013):

fu;plasma;surf ¼PLR × funionized;plasma

funionized;cells

1þ ðPLR2 1Þ × funionized;plasmafunionized;cells

ðA10Þ

where PLR is the plasma to tissue concentration ratio of extracellularbinding proteins and the value for the liver was 13.3 (Poulin et al., 2012).The funionized,plasma and funionized,cells are the fraction of uninonized drugsin plasma and cells. The values of funionized,plasma and funionized,cells werecalculated using the pKa value(s) of each drug, and the physiologicalpH values on both sides of the membrane (e.g., 7.1 for liver and 7.4for plasma) based on Hendersone–Hasselbalch equations (Yun andEdginton, 2013; Poulin, 2015).In other tissue compartments:

VTissue ×dCtissue

dt¼ Qtissue ×

�CAB 2

Ctissue × B=PKp;tissue

�ðA11Þ

where Vtissue, Ctissue, and Kp,tissue are the tissue volume, the drugconcentration in the tissue, and the tissue-to-plasma partition coefficient,respectively.

258 Yim et al.

at ASPE

T Journals on O

ctober 5, 2020dm

d.aspetjournals.orgD

ownloaded from

Sauerberg P, Bury PS, Mogensen JP, Deussen HJ, Pettersson I, Fleckner J, Nehlin J, FrederiksenKS, Albrektsen T, Din N, et al. (2003) Large dimeric ligands with favorable pharmacokineticproperties and peroxisome proliferator-activated receptor agonist activity in vitro and in vivo.J Med Chem 46:4883–4894.

Sawada Y, Hanano M, Sugiyama Y, and Iga T (1985) Prediction of the disposition of nine weaklyacidic and six weakly basic drugs in humans from pharmacokinetic parameters in rats.J Pharmacokinet Biopharm 13:477–492.

Segel IH (1975) Enzyme Kinetics, Wiley, New York.Shin D, Kim TE, Yoon SH, Cho JY, Shin SG, Jang IJ, and Yu KS (2012) Assessment of thepharmacokinetics of co-administered metformin and lobeglitazone, a thiazolidinedione anti-hyperglycemic agent, in healthy subjects. Curr Med Res Opin 28:1213–1220.

Shingaki T, Hume WE, Takashima T, Katayama Y, Okauchi T, Hayashinaka E, Wada Y, Cui Y,Kusuhara H, Sugiyama Y, et al. (2015) Quantitative evaluation of mMate1 function based onminimally invasive measurement of tissue concentration using PET with [(11)C]metformin inmouse. Pharm Res 32:2538–2547.

Shitara Y, Takeuchi K, and Horie T (2013) Long-lasting inhibitory effects of saquinavir andritonavir on OATP1B1-mediated uptake. J Pharm Sci 102:3427–3435.

Simonson SG, Raza A, Martin PD, Mitchell PD, Jarcho JA, Brown CDA, Windass AS,and Schneck DW (2004) Rosuvastatin pharmacokinetics in heart transplant recipients admin-istered an antirejection regimen including cyclosporine. Clin Pharmacol Ther 76:167–177.

Smith PK, Krohn RI, Hermanson GT, Mallia AK, Gartner FH, Provenzano MD, Fujimoto EK,Goeke NM, Olson BJ, and Klenk DC (1985) Measurement of protein using bicinchoninic acid.Anal Biochem 150:76–85.

Takeuchi K, Sugiura T, Umeda S, Matsubara K, Horikawa M, Nakamichi N, Silver DL, IshiwataN, and Kato Y (2011) Pharmacokinetics and hepatic uptake of eltrombopag, a novel platelet-increasing agent. Drug Metab Dispos 39:1088–1096.

Takeuchi R, Shinozaki K, Nakanishi T, and Tamai I (2016) Local drug-drug interaction ofdonepezil with cilostazol at breast cancer resistance protein (ABCG2) increases drug accumu-lation in heart. Drug Metab Dispos 44:68–74.

Taub ME, Kristensen L, and Frokjaer S (2002) Optimized conditions for MDCK permeability andturbidimetric solubility studies using compounds representative of BCS classes I-IV. Eur J PharmSci 15:331–340.

Wanek T, Halilbasic E, Visentin M, Mairinger S, Römermann K, Stieger B, Kuntner C, Müller M,Langer O, and Trauner M (2016) Influence of 24-nor-ursodeoxycholic acid on hepatic dispo-sition of [(18)F]ciprofloxacin, a positron emission tomography study in mice. J Pharm Sci 105:106–112.

Wang X, Wang N, Yuan L, Li N, Wang J, and Yang X (2016) Exploring tight junction alterationusing double fluorescent probe combination of lanthanide complex with gold nanoclusters. SciRep 6:32218.

Watanabe T, Kusuhara H, Maeda K, Kanamaru H, Saito Y, Hu Z, and Sugiyama Y (2010)Investigation of the rate-determining process in the hepatic elimination of HMG-CoA reductaseinhibitors in rats and humans. Drug Metab Dispos 38:215–222.

Yun YE and Edginton AN (2013) Correlation-based prediction of tissue-to-plasma partition co-efficients using readily available input parameters. Xenobiotica 43:839–852.

Address correspondence to: Dr. Suk-Jae Chung, College of Pharmacy andResearch Institute of Pharmaceutical Sciences, Seoul National University,1 Gwanak-ro, Gwanak-gu, Seoul 08826, Republic of Korea. E-mail: sukjae@snu.ac.kr

DDIs between Lobeglitazone and Atorvastatin in Rats 259

at ASPE

T Journals on O

ctober 5, 2020dm

d.aspetjournals.orgD

ownloaded from