1521-0103/350/1/130–143$25.00 http://dx.doi.org/10.1124/jpet.114.214106THE JOURNAL OF PHARMACOLOGY AND EXPERIMENTAL THERAPEUTICS J Pharmacol Exp Ther 350:130–143, July 2014Copyright ª 2014 by The American Society for Pharmacology and Experimental Therapeutics

Macitentan Does Not Interfere with Hepatic Bile Salt Transport

Alexander Treiber, Päivi Äänismaa, Ruben de Kanter, Stephane Delahaye, Marianne Treher,Patrick Hess, and Patricia SidhartaDepartments of Preclinical Drug Metabolism and Pharmacokinetics (A.T., P.A., R.d.K., S.D.), Toxicology (M.T.), Pharmacology(P.H.), and Clinical Pharmacology (P.S.), Actelion Pharmaceuticals Ltd, Allschwil, Switzerland

Received February 19, 2014; accepted April 23, 2014

ABSTRACTTreatment of pulmonary arterial hypertension with the endo-thelin receptor antagonist bosentan has been associated withtransient increases in liver transaminases. Mechanistically,bosentan inhibits the bile salt export pump (BSEP) leading to anintrahepatic accumulation of cytotoxic bile salts, which eventuallyresults in hepatocellular damage. BSEP inhibition by bosentanis amplified by its accumulation in the liver as bosentan isa substrate of organic anion-transporting polypeptide (OATP)transport proteins. The novel endothelin receptor antagonistmacitentan shows a superior liver safety profile. Introductionof the less acidic sulfamide moiety and increased lipophilicityyield a hepatic disposition profile different from other endothelinreceptor antagonists. Passive diffusion rather than OATP-mediated

uptake is the driving force for macitentan uptake into the liver.Interaction with the sodium taurocholate cotransporting poly-peptide and BSEP transport proteins involved in hepatic bilesalt homeostasis is therefore limited due to the low intrahepa-tic drug concentrations. Evidence for this conclusion is pro-vided by in vitro experiments in drug transporter-expressingcell lines, acute and long-term studies in rats and dogs, ab-sence of plasma bile salt changes in healthy human vol-unteers after multiple dosing, and finally the liver safety profileof macitentan in the completed phase III morbidity/mortalitySERAPHIN (Study with an Endothelin Receptor Antagonist inPulmonary Arterial Hypertension to Improve Clinical Outcome)trial.

IntroductionAbout a decade ago, endothelin receptor antagonists were

introduced as a therapeutic concept for the treatment ofpulmonary arterial hypertension, a debilitating and finallyfatal disease for which no oral treatment option was availablebefore. The dual endothelin receptor antagonist bosentan(Tracleer; Actelion Pharmaceuticals, Allschwil, Switzerland)was approved in 2001 as the first member of this new class,followed by sitaxentan (Thelin; Encysive Pharmaceuticals,Houston, TX) in 2006 and ambrisentan (Letairis/Volibris;Gilead Sciences, Foster City, CA) in 2007. Macitentan(Opsumit, N-[5-(4-bromophenyl)-6-[2-[(5-bromo-2-pyrimidinyl)oxy]ethoxy]-4-pyrimidinyl-N9-propylsulfamide; Actelion Phar-maceuticals) has been developed as a new generation endothe-lin receptor antagonist with sustained receptor binding andimproved receptor potency, pharmacokinetic properties, andliver safety profile (Iglarz et al., 2008; Raja, 2010). Most of theseimprovements result from a modified tissue distribution asmacitentan can freely diffuse into tissues rather than beingdependent on active transport.

Bosentan was approved at doses of 62.5 and 125 mg twicea day, but was initially studied at higher doses for thetreatment of hypertension (Krum et al., 1998) and chronicheart failure (Sütsch et al., 1998). Chronic heart failurepatients treated with 500 mg of bosentan twice a day had an18% incidence of elevated alanine aminotransferase levelsversus 4% on placebo. In a subset of patients concomitantlytreated with the antidiabetic glyburide, 29% had elevatedalanine aminotransferase versus 4% and 0% on either placeboalone or placebo and glyburide, respectively. Changes inliver transaminases were accompanied by dose-dependentincreases in plasma bile salts and alkaline phosphatase.Inhibition of the bile salt export pump (BSEP) by bosentan

and itsmetabolites has been identified as the likelymechanismunderlying the observed changes in plasma transaminases(Fattinger et al., 2001). BSEP is an ATP-dependent transportprotein located at the hepatocanalicular membrane andmediates the rate-limiting step in bile salt secretion from bloodinto bile (Gerloff et al., 1998; Stieger et al., 2000). Bosentan andits metabolites inhibited taurocholate transport in vitro incanalicular rat liver membrane vesicles and in Spodopterafrugiperda (Sf9) cell vesicles overexpressing rat bsep. In rats,plasma bile salts increased in a dose-dependent manner afterdx.doi.org/10.1124/jpet.114.214106.

ABBREVIATIONS: ACT-132577, [5-(4-bromophenyl)-6-(2-(5-bromopyrimidin-2-yloxy)ethoxy)-pyrimidin-4-yl]-sulfamide; AUC, area under theplasma concentration versus time curve; BSEP, bile salt export pump; CHO, Chinese hamster ovary; CLint, intrinsic clearance; CLpo, oralclearance; DMSO, dimethyl sulfoxide; ET, endothelin; HBSS, Hanks’ balanced salt solution; LC-MS/MS, liquid chromatography–tandem massspectrometry; NTCP, sodium taurocholate cotransporting polypeptide; OATP, organic anion-transporting polypeptide; PBPK, physiologically-based pharmacokinetic (modeling); SERAPHIN, Study with an Endothelin Receptor Antagonist in Pulmonary Arterial Hypertension to ImproveClinical Outcome.

130

at ASPE

T Journals on Septem

ber 16, 2018jpet.aspetjournals.org

Dow

nloaded from

intravenous dosing of bosentan (Stieger et al., 2000; Kis et al.,2009).These initial findings in rats were later confirmed with

human BSEP (Mano et al., 2007) and led to the hypothesisthat bosentan treatment initially triggers a disruption of bilesalt homeostasis through dose-dependent blockade of BSEP-mediated bile salt excretion into bile, eventually resulting intheir accumulation in liver cells. As bile salts are cytotoxic athigh concentrations, the observed liver transaminase eleva-tions in man are believed to result from the secondary bile salttoxicity in hepatocytes.The hepatic disposition of bosentan is mediated by organic

anion-transporting polypeptide (OATP) transport (Treiberet al., 2007) followed by extensivemetabolism throughCYP3A4and CYP2C9 and finally excretion of the metabolites into bile(Weber et al., 1999). As a consequence, bosentan pharmacoki-netics are sensitive to concomitant CYP3A4 and/or OATPinhibitors. Although the potent CYP3A4 inhibitor ketoconazoleincreased bosentan in plasma by only about 2-fold (vanGiersbergen et al., 2002), more pronounced elevations wereobserved with the OATP inhibitor rifampicin (van Giersbergenet al., 2007), the human immunodeficiency virus proteaseinhibitor ritonavir/lopinavir (Kaletra; Abbott Laboratories,Abbott Park, IL) (Dingemanse et al., 2010), and cyclosporin A(Binet et al., 2000), the latter two being combined CYP3A4/OATP inhibitors.Conceptually, there are several options for designing drugs

with an improved side-effect profile. On the one hand,improving receptor affinity and pharmacokinetic propertiesmight yield drugs that are effective at lower doses. Thealternative approach is to avoid interactions with targets crit-ically involved in toxicity. Both approaches were combined inthe discovery of macitentan. The present report summarizes theexperimental evidence demonstrating that macitentan does notinteract with hepatic transport proteins critically involved inbile salt trafficking and drug accumulation in the liver.

Materials and MethodsChemicals and Reagents

Macitentan was obtained from Lonza AG (Visp, Switzerland) witha purity of 99.8%. [14C]Radiolabeled macitentan with a specificactivity of 55 mCi/mmol was purchased from GE Healthcare (LittleChalfont, UK). Metabolite ACT-132577 ([5-(4-bromophenyl)-6-(2-(5-bromopyrimidin-2-yloxy)ethoxy)-pyrimidin-4-yl]-sulfamide) wasobtained either from the chemistry department of Actelion Pharma-ceuticals Ltd, or from SynphaBase (Pratteln, Switzerland), with purityin excess of 97%. [14C]ACT-132577 with a specific activity of 56 mCi/mmol was obtained from Quotient Bioresearch (Rushden, North-amptonshire, UK). Both radiolabeled compounds were supplied asacetonitrile solutions with radiochemical purities in excess of 97%.

Bosentan was obtained from the chemistry department of ActelionPharmaceuticals Ltd. Sodium taurocholate was obtained from Sigma-Aldrich (Buchs, Switzerland), and [3H]taurocholic acid with a specificactivity of 4.6–5.0 Ci/mmol was purchased from PerkinElmer (Boston,MA) as a solution in methanol:ethanol (1:3) at a concentration of1 mCi/ml. Estrone-3-sulfate and atorvastatin calcium trihydrate werefrom Sigma-Aldrich. [3H]Estrone-3-sulfate and [3H]atorvastatincalcium with specific activities of 50 Ci/mmol and 10 Ci/mmol, re-spectively, were purchased from American Radiolabeled Chemicals(St. Louis,MO) as solutions in ethanol or ethanol:water (1:1). CyclosporinA was purchased from Fluka (Buchs, Switzerland) and rifampicinfrom Sigma-Aldrich. The liquid scintillation cocktails Filter-Count and

IRGA Safe Plus were purchased from PerkinElmer (Zürich, Switzer-land). Baculovirus-infected Sf9 cell membrane vesicles overexpressinghuman BSEP were obtained from SOLVO Biotechnology (Budapest,Hungary). All media and supplements for Chinese hamster ovary (CHO)and CHO Flp InTM cells were obtained from Invitrogen AG (Basel,Switzerland).

Transport Experiments

Preparation of Stock Solutions. For BSEP and sodium taur-ocholate cotransporting polypeptide (NTCP) inhibition experiments,macitentan and ACT-132577 stock solutions were initially preparedin dimethyl sulfoxide (DMSO) in concentration ranges of 1 mM to 100mM BSEP and 1 mM to 100 mM NTCP, and then diluted with thebuffer used in the transport experiments described herein. Forcellular transport experiments, macitentan and ACT-132577 DMSOstock solutions were prepared in a range from 0.1 mM to 100 mM andagain diluted with transport buffer. Stock solutions of cyclosporin Aand rifampicin were prepared in DMSO in a concentration range from1–50 mM. DMSO was also used to prepare the 10 and 100 mM stocksolutions of atorvastatin, taurocholic acid, and estrone-3-sulfate.

Cell Culture. CHO Flp In cells overexpressing human NTCPwere cultured at passage numbers 5 to 19 on tissue culture dishes of55-cm2 growth area (Sarstedt, Newton, NC) at 37°C in a humidifiedatmosphere containing 5% carbon dioxide. Cells were maintainedin Ham’s F-12 medium supplemented with 10% fetal calf serum,penicillin/streptomycin (100 IU/ml), L-glutamine (1 mM), and hygrom-ycin B (500 mg/ml). For transport experiments, cells from a maximally90% confluent 58-cm2 tissue culture dish were detached with trypsin-EDTA, uniformly resuspended in Ham’s F-12 medium, and seeded ontissue culture dishes (Corning, Tewksbury,MA). The cells were used fortransport experiments 72 to 96 hours later, when they were 80–90%confluent. At 24 hours before starting the transport experiments, cellswere additionally induced by adding 5 mM sodium butyrate (Sigma-Aldrich) to the medium.

CHO cells overexpressing human OATP1B1, OATP1B3, andOATP2B1 and wild-type CHO cells were cultured at passage numbers9 to 60 on tissue culture dishes (Corning) at 37°C in a humidifiedatmosphere containing 5% carbon dioxide. All cell lines weremaintained in Dulbecco’s modified Eagle’s medium containing 1 g/lglucose and supplemented with 10% fetal calf serum, penicillin/streptomycin (100 IU/ml) and L-proline (0.05 mg/ml). The culturemedium for the OATP-expressing CHO cells additionally containedgeneticin (500 mg/ml). For transport experiments, cells from a confluent55-cm2 tissue culture dish were detached with trypsin-EDTA, uni-formly resuspended in Dulbecco’s modified Eagle’s medium, and seededon tissue culture dishes of 8-cm2 growth area. Cells were used fortransport experiments 72 to 96 hours later, when they were 90–100%confluent. At 24 hours before starting the transport experiment, thecells were additionally induced by adding 5 mM sodium butyrate to themedium.

Cryopreserved human hepatocytes (lot SSR; Bioreclamation IVT,Brussels, Belgium) were seeded on collagen-coated 24-well plates(Nunc, Thermo Scientific, Wohlen, Switzerland) at a density of 0.2 �106 viable cells per well. Cells were allowed to attach for about 4 hoursin William’s medium E (Life Technologies Europe B.V., Zug, Switzer-land) supplemented with 10% fetal calf serum, 10mg/ml insulin, and 10mg/ml penicillin/streptomycin before use in uptake experiments.

Transport Experiments with Overexpressing CHO Cells.Transport experiments with CHO Flp In cells expressing NTCP wererun using three 8-cm2 tissue culture dishes for each concentrationinvestigated. After washing the cells three times with 2 ml ofprewarmed (37°C) sodium or choline buffer, we initiated the uptakeexperiment by adding 1 ml of buffer containing either [14C]-labeledmacitentan at various concentrations or 5 mM [3H]-labeled taurocholicacid (appropriately diluted with nonlabeled material). The sodium-containing buffer was composed of 20 mMHEPES (pH 7.4), 116.4 mMNaCl, 1 mM NaH2PO4, 5.3 mM KCl, 0.8 mM MgSO4, and 5.5 mM

Macitentan and Hepatic Bile Salt Transport 131

at ASPE

T Journals on Septem

ber 16, 2018jpet.aspetjournals.org

Dow

nloaded from

D-glucose. The choline-containing buffer had the overall same compo-sition but sodium chloride and NaH2PO4 were replaced with 116.4 mMcholine chloride and 1 mM KH2PO4, respectively. After incubation at37°C for 40 seconds, cellular uptake was stopped by addition of 2 times2 ml of ice-cold choline buffer containing 0.5% bovine serum albumin(Sigma-Aldrich). Bovine serum albumin was included in the washingbuffer to minimize unspecific binding. Cells were washed four timeswith approximately 2 ml of ice-cold choline buffer and then solubilizedby addition of 1ml of 1% (w/v) Triton X-100. After incubation for at least20 minutes, 0.5 ml of the cell lysate was mixed with 3.5 ml ofscintillation cocktail IRGA Safe Plus, and the total radioactivity wasdetermined using a Tri-Carb 2300 TR liquid scintillation analyzer(PackardBioscience, Zürich, Switzerland). Twenty-fivemicroliter–aliquotsof the cell lysates were used to determine the protein content of eachsample. Inhibition experiments were performed by simultaneous ad-dition of 5 mM [3H]taurocholic acid and predefined concentrations ofmacitentan or ACT-132577. Incubations and sample work-up weredone as outlined above. NTCP-mediated transport rates were cal-culated as the difference between sodium and choline buffer.

Transport experiments with OATP-expressing and wild-typeCHO cells were run using three 8-cm2 tissue culture dishes for eachconcentration. After washing the cells three timeswith 2ml of prewarmed(37°C) transport buffer, the uptake experiment was initiated by adding1 ml of buffer containing macitentan or ACT-132577 at concentrations of0.01–100 mM and 0.01–300 mM, respectively. Cellular uptake wasdetermined at 37°C and stopped after 40 seconds by addition of twotimes 2 ml of ice-cold transport buffer containing 0.5% bovine serumalbumin. The latter was included in the washing buffer to minimizenonspecific binding of radioactive compounds. Cells were then rapidlywashed four times with each 2 ml of ice-cold transport buffer andsolubilized by addition of 1 ml of 1% (w/v) Triton X-100. After incubationfor at least 20 minutes, 0.5 ml of the cell lysate was mixed with 5 ml ofscintillation cocktail IRGA Safe Plus and total radioactivity determinedusing a Tri-Carb 2300 TR liquid scintillation analyzer. Twenty-fivemicroliter–aliquots of the cell lysates were used to determine total proteincontent. Before each transport experiment, the time dependence ofcellular uptake was individually determined to optimize experimentalconditions.

The effect of the OATP inhibitors cyclosporin A and rifampicin onthe uptake of 1 mM macitentan was investigated for all three OATPtransporters. The inhibition experiment was started by addition of1 ml of prewarmed transport buffer containing radiolabeled maci-tentan and the inhibitor in a concentration range from 0.05–100 mM.After incubation at 37°C for 20 seconds, cellular uptake wasterminated by addition of 2 times 2 ml of ice-cold transport buffercontaining 0.5% bovine serum albumin. The sample work-up in theseinhibition experiments was performed as previously outlined. Thefinal content of organic solvent in the transport experiments neverexceeded 1%. [3H]Estrone-3-sulfate was used as a positive control.

Transport Experiments with Membrane Vesicles. For trans-port experiments, membrane vesicles expressing human BSEP (50 mgtotal protein) were incubated in the presence and absence of 5 mMATP. Incubations were performed at 37°C for 1 minute or 3minutes intransport buffer containing 10 mM HEPES (pH 7.4), 50 mM sucrose,100 mM KNO3, 10 mM Mg(NO3)2, and 5 mM [3H]taurocholic acid.Taurocholate uptake was stopped by the addition of ice-cold washingbuffer containing 10 mM Tris-HCl (pH 7.4), 50 mM sucrose, and 100mM KCl, followed by collection of membrane vesicles on a cellulosenitrate membrane filter (pore size 0.45 mm) using a rapid filtrationsystem (Millipore, Zug, Switzerland). Before the experiments, filterswere saturated with 1 mM nonlabeled taurocholic acid to minimizenonspecific binding of radiolabeled compound. Retained membraneswere then washed twice with ice-cold buffer and transferred intoscintillation vials. After addition of 3.5 ml of scintillation cocktail, thetotal radioactivity was determined on a Tri-Carb 2300 TR liquidscintillation analyzer. Inhibition experiments were performedby incubating membrane vesicles simultaneously with 5 mM[3H]taurocholic acid and various concentrations of macitentan

ACT-132577. BSEP-mediated transport rates were calculated as thedifference of results obtained in the presence or absence of ATP.

Macitentan Partitioning in Human Hepatocytes. Cellularuptake of macitentan was determined in triplicate with plated,cryopreserved human hepatocytes. After we had removed the William’smedium E, the cells were washed twice with 0.5 ml of prewarmedHanks’ balanced salt solution (HBSS). Medium and washing solutionswere pooled, and the number of unattached cells was counted usinga Vi-CELL counter (Beckman Coulter, Nyon, Switzerland) to estimatethe number of plated hepatocytes in the well. The hepatic uptake ex-periment was started by the addition of 200 ml of prewarmed (37°C)incubation solution containing macitentan in HBSS with 1% DMSO ata final concentration of approximately 100 nM. After 10 minutes’incubation at 37°C on an orbital shaker at 300 rpm, uptake wasterminated by removal of the supernatant followed by washing of thecells with 2 � 0.5 ml of ice-cold phosphate-buffered saline (pH 7.4).Supernatants were fortified with one volume equivalent of acetonitrilecontaining tetradeuterated macitentan as analytic standard. Hepato-cytes were lysed by the addition of 200 ml of a 2:3 mixture of HBSS andacetonitrile containing tetradeuterated macitentan, and incubation atroom temperature for 15 minutes. Calibration samples were preparedand worked up in parallel in a concentration range from 2 to 1000 nMby diluting the macitentan stock solution in DMSO with a 1:1 (v/v)mixture of either acetonitrile and HBSS, or hepatocyte lysate. Allsamples were placed in 96-well plates pending analysis by liquidchromatography–tandem mass spectrometry (LC-MS/MS).

Macitentan Binding in Human Hepatocytes. Macitentanbinding to human hepatocyte homogenate was determined by theuse of rapid equilibrium dialysis and a membrane with a molecularmass cutoff of 8 kDa (Thermo Fisher Scientific, Reinach, Switzerland).Before equilibrium dialysis, human hepatocytes (1 � 106 cells/ml)were metabolically inactivated by an initial incubation at 37°C and800 rpm on a thermomixer for 48 hours, followed by three freeze-thawcycles at room temperature and 220°C, and finally sonication for 10seconds (Vibracell 75043; Bioblock Scientific, Illkirch, France).Macitentan at a final concentration of 0.5 mM was added to thehepatocyte homogenate as a 0.5-mM stock solution in DMSO. Twohundred microliter–aliquots of this mixture were transferred into thedonor compartment of the rapid equilibrium device and dialyzedagainst 350 ml HBSS at 37°C for 4 hours on an orbital shaker in anatmosphere containing 5% CO2. At the end of dialysis, 50-ml aliquotsof the donor compartment were diluted with HBSS, while 50 ml of thereceiver compartment were diluted with 50 ml of blank hepatocytehomogenate to generate samples with the same analytical matrix.Three independent experiments were performed with three replicateseach.

Sample work-up for LC-MS/MS analysis consisted of proteinprecipitation with three volume equivalents of methanol containingtetradeuterated macitentan as analytic standard. After centrifuga-tion at 3220g and 4°C for 20minutes, 5-ml aliquots were transferred ina 96-well plate pending analysis. Calibration samples were preparedand worked up in parallel in a concentration range from 0.5–1000 nMby diluting the macitentan stock solution in DMSO with a 1:1 (v/v)mixture of hepatocyte homogenate and HBSS.

Quantification of Macitentan by LC-MS/MS. The analyticequipment consisted of a Shimadzu HPLC System (Shimadzu,Reinach, Switzerland) connected to an API5000 (AB SCIEX, Concord,ON, Canada). Data acquisition was performed with the Analystsoftware package (version 1.5.1; AB SCIEX). The chromatographicanalysis was achieved on a Phenomenex Luna C8 column (5 mm, 2.0�20 mm i.d.) at room temperature with a flow rate of 0.6 ml/min(Phenomenex, Torrance, CA). Mobile phases consisted of 0.1%aqueous formic acid and acetonitrile. The mass transitions used formacitentan and its tetradeuterated internal standard were 589 to 201and 593 to 205, respectively, both with a scan time of 50 milliseconds.

Determination of Total Protein Content. Total protein con-tent was determined using the Pierce bicinchoninic acid assay (PierceScience, Lausanne, Switzerland) with quantification at a wavelength

132 Treiber et al.

at ASPE

T Journals on Septem

ber 16, 2018jpet.aspetjournals.org

Dow

nloaded from

of 590 nm on a SpectraCount spectrophotometer (Packard Bioscience)according to the supplier’s protocol. Bovine serum albumin was used asa standard. Raw data were analyzed using the PlateReader softwareI-Smart (version 3.0 for Windows; Packard Bioscience).

Data Evaluation. Data from the inhibition experiments wereevaluated by plotting the inhibitor concentration (logarithmic scale)against the BSEP- or NTCP-mediated transport of taurocholic acid. IC50

values were then determined from the plot by nonlinear regression usingEq. 1 with a constraint Bottom . 0 (Giacomini et al., 2010):

y5Top

11 ð xIC50

Þs 1Bottom (1)

in which y is the transport expressed as the percentage of inhibitionrelative to control, x is the inhibitor concentration (mM), s is the slope atthe point of inversion, and Top and Bottom are the maximum andminimum transport rates. For all graphical data evaluations, theGraphPad Prism software package (version 5.0; GraphPad Soft-ware Inc., La Jolla, CA) was used. The fitted parameters arepresented as best-fit parameter and standard error.

Cellular uptakes were normalized for total protein content and areeither expressed as pmol/mg protein or are further normalized forincubation time and expressed as pmol/(mg∙min). The OATP-mediatednet uptake rates are calculated as the difference of OATP-expressingand wild-type CHO cells for each individual concentration and arepresented as mean and S.D. Uptake ratios were calculated from theOATP-expressing and wild-type cells.

The partitioning ratio of macitentan between human hepatocytesand in the incubation medium (Kp) was calculated using Eq. 2:

Kp 5  Ahepatocyte

�Vhepatocyte

cincubation   solution(2)

where Ahepatocyte is the amount of macitentan in hepatocyte lysate,Vhepatocyte is the hepatocyte volume, and cincubation solution is themacitentan concentration in the incubation medium at the end of theexperiment. The hepatocyte volume was estimated from cell diametersmeasured before plating (23 mm). This value is in good agreement withpreviously published data, which are 16.2 mm for human hepatocytes(Mateus et al., 2013) and 24 mm for rat hepatocytes (Treijtel et al., 2005).

The unbound fraction in the hepatocyte homogenate (fu,homogenate)was calculated using Eq. 3:

fu;homogenate 5  cbuffer

chomogenate(3)

where chomogenate and cbuffer are the macitentan concentrations in thedonor and receiver compartments at the end of dialysis. The unboundfraction in human hepatocytes (fu, hepatocyte) was derived from Eq. 4:

fu;hepatocyte 5  1

Dð1�fu;homogenate 2 1Þ1 1(4)

where D is the homogenate dilution factor. The volume of humanhepatocytes was again determined from cell diameters (20 mm) beforehomogenization. The ratio of unbound macitentan concentrationsbetween cells and medium (Kp,uu) as a measure for drug accumulationwas calculated using Eq. 5:

Kp;uu 5  fu;hepatocyte

fu;incubation solution�Kp (5)

where fu, incubation solution is the unbound macitentan concentration ofthe incubation medium. In this equation, fu, incubation solution isassumed to be 1 as the medium does not contain proteins.

Bile Salt Measurements in Animals and Man

Quantification of Bile Salts in Plasma and Serum. Bile saltconcentrations in plasma and serum were determined using an enzymatic

assay based on the reduction ofNAD toNADH,which is subsequently usedto reduce nitrotetrazolium blue to formazan, followed by colorimetricquantification of the latter at a wavelength of 530 nm. For samples fromthe intravenous rat model, bile salts in plasma were quantified using a kitfrom Sigma Diagnostics (St. Louis, MO) and a set of calibration samplesranging from 0–100 mM that was run on the same 96-well plate as theunknown samples. The commercial kit is designed to quantify bile salts inserum but can also be used for plasma (validation data not shown). Bilesalts in sera from the externally performed rat and dog toxicity studies andthe multiple ascending–dose study with macitentan in human healthysubjects (Sidharta et al., 2013) were analyzed in the respective preclinicaland clinical research laboratories.

Animals. For the acute cholestasis model, male Wistar rats, 8–12weeks of age, were delivered from RCC Ltd., Biotechnology andAnimal Breeding Division, Füllinsdorf, Switzerland, and used afteran acclimatization period of at least 7 days. Body weights werebetween 221–345 g at the day of the experiment. All animals werehoused under climate-controlled conditions with a 12-hour light/darkcycle in accordance with the guidelines of the Basel Cantonal Vet-erinary Office (license no. 169). All animals were maintained underidentical conditions and had free access to drinking water and food(batch 3418; Provimi Kliba, Kaiseraugst, Switzerland).

The multiple-dose toxicity studies in Wistar or Sprague-Dawleyrats and Beagle dogs were conducted in certified contract researchorganizations in compliance with principles of Good LaboratoryPractice. All animal experiments adhered to the Principles ofLaboratory Animal Care (National Institutes of Health, 8th edition;http://grants.nih.gov/grants/olaw/Guide-for-the-care-and-use-of-laboratory-animals.pdf).

Bosentan, Macitentan, and ACT-132577 in the Acute RatModel. Macitentan, its metabolite ACT-132577, and bosentan wereformulated as microsuspensions in 7.5% gelatin and intravenouslyadministered via the tail vein at a dose of 25 mg/kg (n 5 6) anda dosing volume of 1 ml/kg. All formulations were prepared freshly onthe day of experiment and stirred well before administration. About0.5 ml of blood were collected into EDTA-containing vials from thesublingual vein before dosing and at 10, 45, and 120 minutes afterdosing. The effect of the gelatin vehicle was investigated in a controlexperiment. Plasma was prepared by centrifugation at approximately4000g and stored frozen at 220°C pending analysis.

Bile Salt Measurements in the Rat and Dog ToxicityStudies. In the multiple-dose oral toxicity studies, male and femaleanimalswere treated once daily withmacitentan either by gavage (rats)or with capsules (dogs). A control group receiving the methylcellulosevehicle or empty capsules was an integral part of all study designs.Doses were selected based on previous dose-range finding studies andpreceding studies of shorter duration. Treatment duration was 4, 13,and 26 weeks in the rat, and 4, 13, and 39 weeks in the dog. Bile saltsin plasma were determined at the end of the toxicity study as part ofthe regular clinical chemistry program.

Bile Salt Measurements in the Multiple Ascending DoseStudy in Man. The multiple ascending dose study with macitentanin healthy human subjects was designed as a double-blind, placebo-controlled, randomized study to investigate the tolerability, safety,pharmacokinetics, and pharmacodynamics of macitentan (Sidhartaet al., 2013). The study followed the principles of the Declaration ofHelsinki and Good Clinical Practice. The protocol and informedconsent form was approved by an independent ethics committee(ethics committee of the Landesärztekammer Baden-Württemberg,Stuttgart, Germany). A total of 32 men received doses of 1, 3, 10, and30 mg of macitentan or placebo for 10 days once daily in fasted state.Each dose was administered sequentially to a group of eight subjects(six on macitentan, two on placebo). The safety evaluation comprisedthe collection of adverse event data, including assessments ofseriousness, severity, relationship to study drug, and outcome. Thesafety assessment comprised laboratory variables, vital signs, 12-leadelectrocardiogram, physical condition, and body weight. Pharmacoki-netic parameters of macitentan and its metabolite ACT-132577 were

Macitentan and Hepatic Bile Salt Transport 133

at ASPE

T Journals on Septem

ber 16, 2018jpet.aspetjournals.org

Dow

nloaded from

assessed for all doses. Bile salts in serum were determined before thedose on day 1 and day 10 of treatment.

Physiologically-Based Pharmacokinetic Modeling

The study used Simcyp Population-Based ADMESimulator (version12; Simcyp Limited, Sheffield, UK), a physiologically-based pharma-cokinetic (PBPK) computer model combined with genetic, physiologic,and demographic variables using Monte Carlo methods and equationsderived from population databases obtained from literature sources.

The physicochemical properties and blood binding of macitentan—molecular weight 588 g/mol, logD 2.9, pKa 6.2, plasma protein binding99.6%, and blood/plasma ratio 0.55—were entered into Simcyp. Thecorresponding values for ACT-132577weremolecular weight 546 g/mol,logD 1.5, pKa 6.1, plasma protein binding 99.5%, and blood/plasmaratio 0.55. Published data were used as source for the pharmacoki-netic, metabolism, and excretion data used for the development of thePBPK model (Bruderer et al., 2012b; Atsmon et al., 2013). Only 4% ofthe oral dose was excreted in feces as unchangedmacitentan after oraldosing of [14C]-labeled macitentan to healthy volunteers, suggestingalmost complete absorption of the dose from the gut. Unchangedmacitentan was not detected in urine.

Oral absorption wasmodeled using a simple first-order model, withthe fraction of the dose absorbed from the gut (fa) set to 1 withoutvariation. The rate constant of absorption (ka), absorption lag time, andQgut, a hybrid term including both villous blood flow and permeabilitythrough the enterocyte membrane (Yang et al., 2007; Pang and Chow,2012), were optimized to fit the observed plasma concentration profileof macitentan. Optimized values were: ka 0.3 h21, lag time 1 hour, andQgut 9.5 l/h. The volume of distribution was calculated based on a fullPBPK model using the tissue partitioning equations of Rodgers et al.(Rodgers et al., 2005; Rodgers and Rowland, 2006, 2007). The predictedvolume of distribution at steady state using tissue volumes for a healthyvolunteer population was 0.36 l/kg for macitentan and 0.22 l/kg forACT-132577. Distribution was assumed to be perfusion-limited for allorgans. The liver to plasma partitioning coefficients for macitentan andACT-132577 were predicted as 0.18 and 0.09, based on physiologic livervolume, intracellular and extracellular water content, neutral andacidic (phospho)lipid content, binding to albumin, and (predicted)binding to lipoproteins.

For the purpose of modeling, it was assumed that macitentanexcretion into feces was the result of biliary excretion rather thanincomplete absorption. Consequently, 4% of the clearance was set tooccur via unchanged excretion into the bile, and the remaining 96%was set to be cleared by hepatic metabolism. Renal clearance was setto zero as no unchanged macitentan was detected in urine. Macitentanblood clearance in man is unknown. However, oral clearance can becalculated using Eq. 6:

CLpo 5dose

AUCpo(6)

The mean plasma area under the concentration versus time curve(AUC) of macitentan was 5759 ng·h/ml at a dose of 10 mg (Atsmonet al., 2013). The oral clearance (CLpo) was calculated as 1.8 l/h. Themetabolic oral clearance (96% of the total oral clearance) of macitentanwas scaled to microliter per minute per milligram protein andmicroliterper minute per million hepatocytes with the assumption, based on invitro and clinical interaction data, that 62% ofmacitentan ismetabolizedto ACT-132577 (Actelion, data on file) using the scaling factors inSimcyp. The intrinsic clearance (CLint) was derived from CLpo using thewell-stirred liver model and Eq. 7:

CLint 5CL

fu;p

B

.P

�12 CL

QH

� (7)

This resulted in a liver microsomal clearance CLint of 0.48 ml/min/pmolCYP3A4 for the formation of ACT-132577. The intrinsic clearance of

1.5 ml/min/million hepatocytes was used to calculate biliary clearancein the PBPK model. The remainder of the metabolic clearance ofmacitentan was not assigned to a specific enzyme.

ACT-132577 has not been dosed intravenously to humans, so thehuman clearance is unknown. However, the clearance of a metabolitecan be calculated using Eq. 8 (Rowland and Tozer, 1989):

AUCACT-132577AUCmacitentan

5 fmCLmacitentan

CLACT-132577(8)

Application of Eq. 8 using the observed AUC values for macitentanand ACT-132577 (corrected for the difference in molecular weight)after a single 10-mg macitentan dose (Atsmon et al., 2013) and the62% fraction of macitentan metabolized to ACT-132577 resulted ina metabolite clearance of 0.32 l/kg.

A coefficient of variation of 30% was assumed for the input pa-rameters single compartment absorption rate constant, lag time, Qgut,and intrinsic clearance. For all other parameters, variation during thesimulations was based on (physiologic) variation of the populationdatabase within Simcyp. The population selected for the trial designwas a healthy volunteer population, male subjects, aged 18–45 years,in fed state. Ten virtual trials of 10 subjects each were run (total size:100) for a single dose of 10 mg and for steady-state simulations witha 30-mg macitentan loading dose followed by 11 daily doses of 10 mg.

ResultsInhibition of taurocholate uptake bymacitentan andmetab-

olite ACT-132577 was investigated using membrane vesiclesexpressing human BSEP at concentrations up to 100 and300 mM, respectively (Fig. 1). IC50 values of macitentan andACT-132577 derived from an analysis of pooled data were18 6 5 mM (n 5 6) and 60 6 14 mM (n 5 4), respectively.Taurocholate was also the substrate in the inhibition experi-ments with human NTCP overexpressed in CHO cells (Fig. 2).Macitentan inhibited taurocholate uptake with a mean IC50

value of 18 6 2 mM (n 5 3), whereas ACT-132577 showed amean IC50 of 14 6 2 mM (n 5 2).We have previously reported on the cellular uptake of

macitentan in OATP1B1- and OATP1B3-overexpressing cellsin the context of the clinical drug-drug interaction studieswith cyclosporin A and rifampicin (Bruderer et al., 2012a).Table 1 summarizes these results of the uptake experimentswith macitentan and ACT-132577 together with new data onOATP2B1. Estrone-3-sulfate was used as a positive control.Uptake ratios between OATP-overexpressing and wild-typecells were calculated. Based on these data, neither maciten-tan nor ACT-132577 are considered substrates for OATP1B1or OATP2B1. Cellular uptake rates of both compounds intoCHO wild-type cells exceeded that of the OATP substrateestrone-3-sulfate by at least 150-fold (at 5 mM), indicating thattheir cellular uptake is mostly driven by high passive diffusion.Figure 3 displays the net uptake rates for macitentan inOATP1B3-overexpressing CHO cells. Net uptake rates con-sistently differed from those in wild-type CHO cells, and thelinear increase up to the highest concentration of 100 mMsuggests that macitentan is likely a substrate of OATP1B3.No saturation in macitentan uptake was observed in thisconcentration range, indicating that the affinity of macitentanfor OATP1B3 transport is rather low. Passive permeation wasagain the major contributor to overall cellular uptake asuptake ratios never exceeded 1.2. The role of OATP in theoverall cellular uptake is therefore considered of little clinical

134 Treiber et al.

at ASPE

T Journals on Septem

ber 16, 2018jpet.aspetjournals.org

Dow

nloaded from

relevance. Based on the results in Table 1, metabolite ACT-132577 is not a substrate for OATP1B3.To verify the above conclusion, macitentan uptake into

OATP-overexpressing cells was additionally determined inthe presence of the known OATP inhibitors cyclosporin A(Shitara et al., 2003) and rifampicin (Vavricka et al., 2002;Hirano et al., 2006). The results are summarized in Table 2.No consistent effect of cyclosporin A was observed over theconcentration range up to 100 mM, that is, at concentrationslargely exceeding its Ki value of 0.2 mM (Shitara et al., 2003).Cyclosporin A inhibition was not evident in OATP1B3 cells,nor was there any consistent effect of rifampicin in eithercell line. Control experiments with both inhibitors using[3H]atorvastatin as an OATP substrate yielded concentration-dependent decreases in net uptake rates for both compounds(data not shown). Overall, these OATP inhibition experimentssupport the conclusion thatmacitentan cellular uptake ismostlydependent on passive diffusion with only a small componentof OATP1B3-mediated uptake.Macitentan uptake into CHO Flp In cells overexpressing

the sodium-dependent taurocholate cotransporting polypep-tide NTCP was investigated in the same concentration rangeas used for OATP transporters. Uptake ratios were deter-mined from experiments in the presence or absence of sodiumusing taurocholic acid at 5 mM as a positive control. Theresults are summarized in Table 3. Uptake ratios were aroundunity over the entire concentration range, indicating thatmacitentan is not a NTCP substrate.

The potential for intracellular accumulation of macitentanhas been determined in human hepatocytes using a previouslypublished method with small modifications (Mateus et al.,2013). Two parallel experiments were performed, in which,first, the partitioning ratio (Kp) of macitentan between hepa-tocytes and culture medium was determined. Mean Kp was724 6 96 indicating a significant partitioning of macitentaninto human hepatocytes. Macitentan binding in hepatocytes(fu,hepatocytes) was derived from binding in hepatocyte homo-genates (fu,homogenate) as determined by rapid equilibriumdialysis. The corresponding values for fu,homogenate andfu,hepatocytes were 0.189 6 0.009 and 0.00099 6 0.00004,respectively. These data suggest that macitentan is highlybound in human hepatocytes and that the free fractiontherein is only around 0.1%. The ratio of unbound macitentanconcentrations in hepatocytes and medium (Kp,uu) as a mea-sure for hepatocellular accumulation was calculated as 0.7 60.1. Macitentan thus does not accumulate in human livercells, as the unbound concentrations in liver cells and incu-bation medium are similar. The significant partitioning ofmacitentan into liver cells is likely the consequence of itselevated lipophilicity and is compensated for by the highbinding to cellular components.The acute effects of macitentan and ACT-132577 on bile salt

homeostasis were tested upon intravenous dosing to the rat.This model was developed by Fattinger et al. (2001) to mecha-nistically rationalize the increased plasma bile salts in clini-cal trials with bosentan. Macitentan and ACT-132577 were

Fig. 1. Inhibition of vesicular taurocholate uptake into membrane vesiclesexpressing human BSEP by (A) macitentan and (B) ACT 132577. Data aremean 6 S.D. of three replicates. Data from one representative experimentare shown. Fig. 2. Inhibition of taurocholate uptake into CHO cells expressing human

NTCP by (A) macitentan and (B) ACT-132577. Data are mean 6 S.D. ofthree replicates. Data from one representative experiment are shown.

Macitentan and Hepatic Bile Salt Transport 135

at ASPE

T Journals on Septem

ber 16, 2018jpet.aspetjournals.org

Dow

nloaded from

individually tested in this model at an intravenous dose of25 mg/kg, with plasma samples taken before and at 10, 45,and 120 minutes after dosing. As individual plasma bile saltconcentrations varied significantly between animals beforedrug administration, the results are also expressed as indi-vidual differences from predose values. Bosentan was in-cluded as a positive control. Results for all three endothelinreceptor antagonists are summarized in Table 4. Bosentanincreased plasma bile salts by 186 13 mM at 10 minutes afterthe dose, which then returned to predose values within 45minutes after dosing. Neither macitentan nor ACT-132577elicited such an increase in plasma bile salts. After 10 minutes,the mean increases were 3.7 6 4.4 mM for macitentan and2.5 6 6.7 mM for ACT-132577, and, thus, were not differentfrom vehicle (0.4 6 3.4 mM).Bile salts in serum were systematically determined in the

oral toxicology program of macitentan in the rat and dog aspart of the clinical chemistry program. Table 5 summarizesthe data collected for all dose groups at the end of therespective study period. In the 4-week rat study, there was noincrease in mean bile salts in the male animals up to thehighest dose of 1500mg/kg and in female rats up to 450mg/kg.Similar to the observations in the intravenous rat model,significant interindividual variability was evident in these ratstudies, most likely resulting from differences in food con-sumption as animals had free access to food over the entirestudy. No difference between dose groups was noted in thebile salt concentrations in the 26-week toxicity study, in whichrats received macitentan doses up to 250 mg/kg.In the 4-week dog study, macitentan doses up to 250 mg/kg

were given. Interanimal variability was significantly lowercomparedwith the rat. Therewas no difference inmean serumbile salts between dose groups in male or female animals.A similar picture was obtained for doses up to 100 mg/kg in

the 13-week study, during which bile salt data were collectedafter 4 weeks of treatment and at study end. No change in bilesalts was observed across dose groups for the entire studyduration. Doses in the 39-week dog study were 5, 30, and 100mg/kg at the start of the study. After 20 weeks, the high dosehad to be reduced to 75 mg/kg. Serum bile salt data werecollected at the end of the study and confirmed the observa-tions from the studies of shorter duration.Changes in serum bile salts were also monitored in the

multiple-ascending dose study with macitentan in whichhealthy volunteers received macitentan doses of 1, 3, 10, and30 mg for a period of 10 days (Sidharta et al., 2013). Each dosegroup consisted of six individuals on active treatment and twoon placebo. The placebo data were pooled from the four activedose groups. Bile salts were collected on the first day beforemacitentan dosing and on day 10 at the end of study. Resultsare shown in Table 6. Bile salt concentrations in serum werein a narrow range from 7–19 mM, and there was no discernibletrend toward increased serum levels at any dose. Inspection ofthe individual data revealed a maximum difference of 5 mMbetween measurements on days 1 and 10, which was observedin a subject receiving placebo. These data confirm the aboveanimal data and show that macitentan treatment in man isnot associated with changes in serum bile salts.A PBPK model of macitentan was developed to allow

comparison of the in vitro transport data to concentrationsof macitentan and ACT-132577 in the portal vein, systemiccirculation, and liver. Table 7 summarizes the observed andpredicted human pharmacokinetic parameters of macitentanand ACT-132577 after single and repeat dosing as derivedfrom the PBPK model. The derived plasma concentration ver-sus time profiles for both compounds after a single oral doseare depicted in Fig. 4, A and B. Projected plasma exposure(AUC) and Cmax data were close to the mean observed data.

TABLE 1Concentration dependence of macitentan and ACT-132577 cellular uptake into OATP-expressing and wild-type CHO cellsUptake rates are expressed as mean 6 S.D. of n = 3.

Concentration Wild-Type CHO OATP1B1 Ratio OATP1B3 Ratio Wild-Type CHO OATP2B1 Ratio

mM pmol/min∙mg pmol/min∙mg pmol/min∙mg

Macitentan0.01 3.3 6 0.5 3.1 6 0.3 0.9 3.9 6 0.7 1.2 3.0 6 0.3 2.9 6 0.1 1.00.05 13 6 1 14 6 0.3 1.0 14 6 0.3 1.0 12 6 1 14 6 0.4 1.20.1 23 6 1 24 6 1 1.1 27 6 1 1.2 24 6 1 26 6 2 1.10.5 120 6 7 126 6 3 1.1 137 6 6 1.1 119 6 1 117 6 5 1.01 228 6 5 236 6 1 1.0 280 6 9 1.2 214 6 2 222 6 6 1.15 1290 6 43 1290 6 67 1.0 1540 6 37 1.2 1050 6 40 1060 6 18 1.0

15 3200 6 108 3300 6 183 1.0 3700 6 144 1.2 3600 6 128 3400 6 136 1.050 10,600 6 306 10,700 6 291 1.0 12,100 6 449 1.1 9600 6 189 8900 6 336 0.9

100 14,800 6 600 14,900 6 426 1.0 17,900 6 479 1.2 20,000 6 755 19,400 6 629 1.05a 3.8 6 0.2 26 6 0.3 6.9 32 6 0.9 8.3 3.8 6 0.2 77 6 1 20.4

ACT-1325770.01 4.0 6 1.0 3.9 6 0.3 1.1 4.5 6 0.9 1.2 2.8 6 0.3 2.6 6 0.2 0.90.05 7.9 6 0.8 8.3 6 0.6 1.1 9.2 6 0.8 1.2 7.1 6 0.5 7.6 6 0.1 1.10.1 12 6 1 13 6 1 1.1 13 6 2 1.1 12 6 1 14 6 1 1.10.5 57 6 3 57 6 4 1.0 64 6 1 1.1 57 6 1 66 6 1 1.21 107 6 5 121 6 8 1.1 145 6 8 1.4 35 6 2 44 6 1 1.35 590 6 22 650 6 26 1.1 720 6 10 1.2 530 6 27 560 6 3 1.1

15 1930 6 93 1920 6 58 1.0 2200 6 80 1.1 1650 6 20 1790 6 16 1.150 6200 6 152 6240 6 33 1.0 6700 6 393 1.1 6800 6 238 6400 6 138 0.9

100 12,000 6 514 12,000 6 377 1.0 13,100 6 542 1.1 12,200 6 564 12,800 6 551 1.1150 19,500 6 383 17,900 6 108 0.9 20,000 6 805 1.0 17,800 6 972 17,200 6 359 1.0300 36,700 6 623 34,900 6 805 0.9 37,000 6 2410 1.0 34,000 6 1084 31,000 6 1648 0.9

5a 3.8 6 0.6 15 6 0.8 4.0 21 6 0.4 5.6 ND ND ND

ND, not determined.aPositive control estrone-3-sulfate.

136 Treiber et al.

at ASPE

T Journals on Septem

ber 16, 2018jpet.aspetjournals.org

Dow

nloaded from

The model predicts a mean Tmax of 7 hours—slightly earlierthan the observed mean of 9 hours but still well within therange of individual values. Similarly, the Tmax for ACT-132577 was predicted at 26 hours versus observed 48 hours.Overall, the PBPK model adequately describes the observedplasma concentration profiles of macitentan and ACT-132577.It was therefore considered suitable to predict portal vein andliver concentrations as well.Figure 4C shows the time course of mean macitentan and

ACT-132577 concentrations in the portal vein, plasma, andliver. Peak concentrations of macitentan in the portal veinand liver after a single 10-mg dose to healthy male subjectswere estimated as 354 nM (208 ng/ml) and 62 nM (36 ng/ml),respectively. The corresponding values for ACT-132577 were420 nM (223 ng/ml) and 39 nM (21 ng/ml). As ACT-132577 isgenerated by hepatic metabolism, the plasma and portal veinconcentrations are identical. The concentration versus timeprofiles for both compounds at steady state are illustrated inFig. 4D. Predicted total peak concentrations for macitentanat steady state were 400 nM (235 ng/ml) in plasma, 419 nM(247 ng/ml) in the portal vein, and 72 nM (43 ng/ml) in theliver. The corresponding values for ACT-132577 were 1722 nM(915 ng/ml) for plasma/portal vein and 159 nM (85 ng/ml) forthe liver.A parameter sensitivity analysis of the PBPK model was

performed to investigate the effect of various degrees ofactive hepatic uptake on the pharmacokinetic parameters of

macitentan, with the liver as a permeability-limited insteadof a perfusion-limited organ. Passive uptake clearance(CLint, passive) into human hepatocytes was estimated fromthe macitentan logD of 2.9 as 5.3 ml/min/106 cells (Ménochetet al., 2012). Four different values for active uptake clearance(CLint, active)—that is, 0, 10, 25, and 50 ml/min/106 cells—werethen arbitrarily assigned to OATP1B1 as a prototypicalhepatic uptake transporter. Ten virtual trials of 10 subjectseach were again run (total size: 100) for a single dose of 10 mgwithout any other changes to the PBPK model. A comparisonbetween the predicted and observed pharmacokinetic param-eters is shown in Table 8. In the permeability-limited PBPKliver model, the best fit between modeled and observedmacitentan AUC0–24 h is achieved with a CLint, active of 25 ml/min/106 cells, for which a ratio of unbound liver and plasmaconcentrations of 1.05 is predicted. The outcome is thenclosest to the perfusion-limited PBPK model, in which theratio of unbound concentrations is per definition set to unity.

Discussion“Dosis sola facit venenum” is the scientific legacy of the

Swiss physician, alchemist, astrologer, and philosopher Para-celsus (1493–1541). About 500 years later, its longer version“all things are poison, and there is nothing without poisonousqualities . . . it is only the dose which make a thing poison” isstill a major paradigm in modern pharmacology in its attemptto find drugs with adequate risk-benefit profiles. Nowadays,the interplay of target affinity and local drug concentrationshas replaced the historical dose term as a measure for drugquantity. Whether a drug reaches its target in sufficient con-centrations is mostly dependent on its physicochemical prop-erties, as drug distribution to the majority of organs andtissues is driven by passive diffusion. Although total drug con-centrations can vary significantly among tissues as a conse-quence of differing binding properties, free drug concentrationsstrive to equilibrate. This general concept needs to be expandedfor organs expressing transport proteins that are capableof maintaining nonequilibrium conditions and that, conse-quently, depending on their transport direction, result in loweror higher free drug concentrations on one side of a membrane.Liver, kidney, brain, and placenta are typical examples of suchorgans, and the discovery and characterization of transporterslike the ATP-dependent efflux pump P-glycoprotein or the

Fig. 3. Net cellular uptake of macitentan in OATP1B3-expressing CHOcells. Data are mean 6 S.D. of three replicates.

TABLE 2Effect of cyclosporin A and rifampicin on the net uptake rate of 1 mMmacitentan in OATP-overexpressingCHO cellsNet uptake rates are expressed as mean 6 S.D. of n = 3.

ConcentrationNet Uptake Rate OATP1B1 Net Uptake Rate OATP1B3 Net Uptake Rate OATP2B1

Cyclosporin A Rifampicin Cyclosporin A Rifampicin Cyclosporin A Rifampicin

mM pmol/min∙mg

0 80 6 25 170 6 43 70 6 21 100 6 33 20 6 22 24 6 260.05 41 6 9 40 6 54 48 6 5 50 6 82 20 6 20 220 6 510.1 50 6 22 70 6 39 40 6 12 40 6 35 17 6 9 220 6 390.5 80 6 24 60 6 8 110 6 17 110 6 22 50 6 15 10 6 231 60 6 27 90 6 61 60 6 19 80 6 57 40 6 24 2 6 475 70 6 19 120 6 25 50 6 24 100 6 19 10 6 14 90 6 50

10 70 6 18 90 6 21 70 6 21 100 6 23 20 6 17 20 6 1430 30 6 37 60 6 32 20 6 33 70 6 23 10 6 65 30 6 2460 50 6 18 230 6 17 40 6 25 70 6 17 20 6 16 10 6 19

100 24 6 1 90 6 19 48 6 8 90 6 22 10 6 23 10 6 24

Macitentan and Hepatic Bile Salt Transport 137

at ASPE

T Journals on Septem

ber 16, 2018jpet.aspetjournals.org

Dow

nloaded from

organic anion-transporting polypeptide family OATP has fun-damentally changed our understanding of drug disposition.Endothelin receptor antagonists have been established as

a therapeutic concept for the treatment of pulmonary arterialhypertension. From a chemical perspective, they requirea negative charge (Wu, 2000; Boss et al., 2002) for theirinteraction with Arg326 of the endothelin receptors (Breuet al., 1995). In the case of sitaxsentan and ambrisentan, thisnegative charge is provided by a carboxylic acid whereasbosentan contains an aromatic sulfonamide. As a conse-quence, the majority of drug molecules are negatively chargedat physiologic pH, which–together with high binding toplasma proteins—limits distribution into tissues. On theother hand, the negative charge makes them candidatesubstrates for anion transporters, and interactions withOATP transport proteins have indeed been described (Katzet al., 2006; Treiber et al., 2007; Spence et al., 2010). Hepaticside effects—mostly manifesting as transient reversibleelevations of liver transaminases in plasma—have also beenreported for some marketed endothelin receptor antagonists(Humbert et al., 2007; Galie et al., 2011). Active drug uptakeby OATP transporters has the potential to increase hepaticdrug concentrations and thus to contribute to liver injury. Thecombination of BSEP inhibition and OATP-mediated accu-mulation in liver cells is the likely mechanism for thecholestatic effect of bosentan (Fattinger et al., 2001; Treiber

et al., 2007). In line with this hypothesis is the observation inpulmonary arterial hypertension patients concomitantly re-ceiving the human immunodeficiency virus protease inhibitorlopinavir/ritonavir, a potent inhibitor of both CYP3A4 metab-olism and OATP transport (Hull et al., 2009; Annaert et al.,2010): bosentan trough plasma concentrations were increasedby 48-fold in this patient population but were not associatedwith a higher frequency of liver injury (Dingemanse et al.,2010), most likely because the hepatic drug burden wasreduced as a consequence of blocked OATP uptake.These points illustrate the necessity to discover novel

endothelin receptor antagonists with an improved safety pro-file. The medicinal chemistry program leading to the discoveryof macitentan and its pharmacologic profile in animal modelswere published previously (Iglarz et al., 2008; Bolli et al., 2012).Key structural changes in macitentan versus bosentan con-stitute the replacement of the sulfonamide by a sulfamidemoiety and the increase in overall compound lipophilicity. Thesulfamide function in macitentan is less acidic than the sulfon-amide in bosentan, as evidenced by the difference in pKa valuesof 6.2 and 5.1 (Iglarz et al., 2008), respectively, thus increasingthe proportion of molecules present in nonionized state atphysiologic pH. The octanol/water partition coefficient logD7.4,a measure of lipophilicity, is 2.9 for macitentan and 1.3 forbosentan. Both factors facilitate membrane permeability andpenetration into tissues. These changes were achieved withoutcompromising target affinity. Macitentan antagonizes thespecific binding of endothelin (ET)-1 to recombinant ETA andETB receptors with IC50 values of 0.5 and 391 nM, respectively(Iglarz et al., 2008). In ex vivo models using rat aorta andtrachea preparations, macitentan behaved as a more dualendothelin receptor with pA2 values of 7.6 and 5.9 for ETA andETB receptors, respectively (Iglarz et al., 2008). Metabolicstability was optimized in vitro and favorably translated inanimals (Bolli et al., 2012) and humans (Sidharta et al., 2011,2013). As a consequence, macitentan was used at daily doses of3 mg and 10 mg in patients in the completed phase III Studywith anEndothelin Receptor Antagonist in PulmonaryArterialHypertension to Improve Clinical Outcome (SERAPHIN) trial(Pulido et al., 2013). Macitentan has an active metabolite,ACT-132577, that forms in a cytochrome P450–catalyzed re-action, circulates in human plasma, and likely contributes tothe overall efficacy.

TABLE 3Concentration dependence of macitentan cellular uptake into NTCP-expressing CHO Flp In cellsUptake rates are expressed as mean 6 S.D. of n = 3.

Concentration NTCP without Na+ NTCP with Na+ Ratio

mM pmol/min∙mg

0.01 3.7 6 0.3 3.4 6 0.2 0.90.05 17 6 1 18 6 1 1.00.1 35 6 1 33 6 2 0.90.5 162 6 4 168 6 2 1.01 310 6 12 324 6 1 1.05 1520 6 98 1500 6 59 1.0

15 4270 6 43 4500 6 240 1.150 13,800 6 124 14,000 6 207 1.0

100 22,500 6 905 23,500 6 314 1.05a 14 6 4 310 6 10 21

aPositive control taurocholic acid.

TABLE 4The effect of bosentan, macitentan, and its metabolite ACT-132577 on plasma bile salts after intravenous administration ata dose of 25 mg/kg

Compound Predose 10 min 45 min 120 min

mM

Plasma bile saltsa

Vehicleb 16 6 5.4 (8.2–19) 16 6 1.6 (7.7–17) 12 6 5.2 (7.7–16) 13 6 3.1 (7.0–21)Bosentan 10 6 0.8 (4.1–29) 29 6 1.2 (5.2–57) 15 6 3.5 (6.3–23) 12 6 6.3 (5.7–20)Macitentan 8.7 6 2.1 (7.2–20) 17 6 4.9 (12–21) 13 6 3.5 (11–25) 18 6 1.1 (15–19)ACT-132577 20 6 1.4 (7.4–41) 21 6 9.4 (13–45) 20 6 1.9 (8.7–26) 23 6 17 (11–35)

D Plasma bile saltsc

Vehicle — 0.4 6 3.4 (24.3–5.5) 22.0 6 5.7 (211–4.0) 20.5 6 5.9 (29.0–7.5)Bosentan — 18 6 13 (23.9–36) 1.1 6 5.1 (26.2–7.2) 21.6 6 3.1 (216–6.1)Macitentan — 3.7 6 4.4 (21.5–10) 1.8 6 4.1 (25.5–5.5) 3.5 6 4.8 (22.5–9.5)ACT-132577 — 2.5 6 6.7 (26.0–11) 24.3 6 11 (221–7.4) 22.0 6 13 (220–14)

aResults are mean 6 S.D. (range) of n = 6.bVehicle: 0.5% methylcellulose in water.cResults are calculated as individual differences from predose values and presented as mean 6 S.D. (range).

138 Treiber et al.

at ASPE

T Journals on Septem

ber 16, 2018jpet.aspetjournals.org

Dow

nloaded from

Beyond improved receptor-binding affinities and pharma-cokinetic properties over bosentan, the structural changes inmacitentan have altered the hepatic disposition and inter-action profile with hepatic transporters. As illustrated in Figs.1 and 2, macitentan and ACT-132577 are inhibitors of the twomajor transport proteins responsible for hepatic bile salttrafficking: NTCP and BSEP. NTCP (SLC10A1) is the majortransporter responsible for moving bile salts from blood intoliver cells (Doring et al., 2012) whereas BSEP mediates therate-limiting step in bile salt secretion from blood into bile(Gerloff et al., 1998; Stieger et al., 2000) (Fig. 5). NTCP wasequally inhibited by macitentan and ACT-132577 with IC50

values in the range of 14–18 mM, whereas BSEPwas inhibitedwith IC50 ranging from 18–60mM. These in vitro data are thus

not different from those of bosentan with reported IC50 valueson BSEP and NTCP of 25–77 mM (Mano et al., 2007; Dawsonet al., 2012; Warner et al., 2012) and 24–30 mM (Leslie et al.,2007), respectively.However, the mechanism of hepatic uptake differs between

bosentan and macitentan. Bosentan uptake into liver cells islargely dependent on OATP transport in animals and hu-mans (Treiber et al., 2004, 2007). In contrast, its increasedlipophilicity allows macitentan to enter cells by passive diffu-sion, as shown in the uptake experimentswithOATP-expressingand wild-type cells (Table 1). The same conclusion was drawnin a recent study on macitentan interactions with other he-patic transport proteins (Weiss et al., 2013). Macitentan isa weak substrate for OATP1B3 (Fig. 3) but not for OATP1B1,OATP2B1, or NTCP. Active transport contributes less than20% to overall cellular uptake. Consequently, macitentanuptake is not vulnerable to OATP inhibition by cyclosporin Aor rifampicin (Table 3). These conclusions were confirmed ina clinical drug-drug interaction study with cyclosporin A, anestablished potent inhibitor of OATP-mediated (Shitara et al.,2003) and NTCP-mediated transport (Mita et al., 2006). Con-comitant cyclosporin A increased macitentan exposure in hu-mans by only 1.1-fold (Bruderer et al., 2012a), compared witha 30-fold increase observed with the proven OATP substratebosentan (Treiber et al., 2007) in the presence of cyclosporin A(Binet et al., 2000).Bosentan elicited a dose-dependent increase in plasma bile

salts in the rat after a single intravenous administration(Fattinger et al., 2001). Macitentan and ACT-132577 wereboth tested in this model alongside with bosentan, all at a

TABLE 5Serum bile salts (in mM) in rat and dog toxicity studies with macitentan

Species Duration Sex Dose

mg/kg

Rat 4 wka 0 50 150 450 1500M 25 6 11 (11–46) 20 6 6.6 (6.6–45) 21 6 8.9 (9.4–34) 25 6 9.1 (12–40) 24 6 12 (10–40)F 20 6 8.1 (12–40) 25 6 8.8 (8.8–45) 27 6 9.9 (8.9–34) 28 6 9.3 (16–39) 37 6 33 (13–128)

26 wkb 0 10 50 250M 32 6 15 (11–57) 17 6 6.6 (9.4–27) 23 6 10 (10–52) 19 6 9.3 (6.5–28.9)F 49 6 27 (19–124) 32 6 16 (13–81) 37 6 31 (11–125) 30 6 16 (11–58)

Dog 4 wkc 0 10 50 250M 8.3 6 1.3 (7.2–10) 7.3 6 0.4 (7.0–7.7) 6.6 6 0.8 (6.0–7.5) 7.7 6 3.3 (6.3–12)d

F 15 6 10 (8.3–33) 15 6 8.2 (8.1–24) 7.8 6 1.6 (6.6–9.6) 9.0 6 3.9 (6.2–16)

13 wk 0 2 5 30 100Me 2.8 6 2.3 (BLQ–7.7)f 1.0 6 0.6 (BLQ–2.3) 1.6 6 0.9 (BLQ–2.7) 1.2 6 0.9 (BLQ–3.1) BLQg

F 1.5 6 0.5 (BLQ–2.0) 3.5 6 3.7 (BLQ–2.3) 1.9 6 0.9 (BLQ–2.3) 1.9 6 0.8 (BLQ–3.4) 1.8 6 1.5 (BLQ–3.9)Mh BLQg BLQg 1.3 6 1.1 (BLQ–3.5) 1.9 6 1.4 (BLQ–4.0) 1.8 6 2.4 (BLQ–7.2)F 3.2 6 3.6 (BLQ–9.7) 1.7 6 1.2 (BLQ–3.1) BLQg BLQg 4.9 6 8.7 (BLQ–6.2)i

39 wk j 0 5 30 100/75k

M 1.4 6 1.9 (0.2–5.9) 2.0 6 3.7 (0.3–11.1) 0.5 6 0.3 (0.1–1.0) 0.7 6 0.8 (BLQ–1.9)l,m

BLQ, below the limit of quantification.an = 10. Data are from day 28 of the 4-week study, all data are mean 6 S.D. (range).bn = 15. Data are from the last day of macitentan treatment.cData are from day 23 of the 4-week study. n = 5 in the control and 250-mg/kg dose groups, and n = 3 in the 10- and 50-mg/kg dose groups.dOne sample missing; mean and S.D. of n = 4.en = 7. Data are from week 4 of the 13-week study.fBLQ: 1.5 mM.gn = 7. Data are from the last treatment week of the 13-week study.hAll animals BLQ.iOne apparent outlier with plasma bile salts of 24 mM at study end, included in the calculation of mean and S.D. but not in range; liver specimens were unremarkable in the

subsequent histologic examination.jn = 8. Data are from the last treatment week of the 39-week study.kThe high dose was reduced from 100 to 75 mg/kg in week 20.ln = 6–7.mBLQ: 0.5 mM.

TABLE 6Serum bile salts in the multiple ascending dose study with macitentan inhealthy human subjectsResults are expressed as mean 6 S.D. (range) of n = 6–8.

DoseBile Saltsa

Day 1 Day 10

mg mM

Placebo 10.7 6 2.4 (7.7–14) 12.2 6 4.0 (8.2–19)1 9.8 6 1.7 (8.2–13) 10.8 6 1.2 (9.7–13)3 9.8 6 1.4 (7.7–11) 11.1 6 1.6 (8.8–13)

10 11.8 6 1.2 (11–14) 14.8 6 2.7 (12–18)30 11.8 6 2.7 (9.5–15) 13.3 6 1.4 (11–15)

aBile salts in serum were determined before dosing on both dosing days. Lowerlevel of quantification: 1 mM.

Macitentan and Hepatic Bile Salt Transport 139

at ASPE

T Journals on Septem

ber 16, 2018jpet.aspetjournals.org

Dow

nloaded from

dose of 25 mg/kg. Results with bosentan reproduced literaturedata, while no effect was observed with macitentan orACT-132577 (Table 4). These data provide evidence in a morephysiologic setting that neither compound interferes with bilesalt homeostasis under conditions that led to an acute cho-lestatic effect with bosentan. Supportive data upon chronic

dosing was provided by the bile salt data from the rat and dogtoxicity studies. After 26 or 39 weeks of treatment withmacitentan at doses up to 250 mg/kg (rat) and 100/75 mg/kg(dog), there was no pattern of elevated plasma bile salts in anytoxicity study nor were there any histologic findings pointingto a cholestatic potential of macitentan (Actelion, data on file).

TABLE 7Simulated and observed pharmacokinetic parameters of macitentan and ACT-132577 after single andmultiple oral doses of 10 mg

Compound PK Parameter Observeda Predictedb

Macitentan SD Cmax (ng/ml) 227 (205–252) 199 (190–209)Tmax (h) 9.0 (5.0–12) 7 (7–7)AUC0–336 h (ng∙h/ml) 5759 (4914–6749) 4966 (4634–5321)

ACT-132577 SD Cmax (ng/ml) 178 (155–204) 226 (217–237)Tmax (h) 48 (24–48) 26 (24–27)AUC0–336 h (ng∙h/ml) 19,749 (17,969–21,705) 18,903 (17,899–19,963)

Macitentan SS Cmax (ng/ml) 371 (290–475) 227 (213–242)Tmax (h) 6.0 (5.0–12) 6 (6–6)AUC0–24 h (ng∙h/ml) 5400 (4107–7099) 3956 (3676–4257)

ACT-132577 SS Cmax (ng/ml) 802 (585–1100) 884 (836–935)Tmax (h) 9.0 (8.0–12) 9 (9–9)AUC0–24 h (ng∙h/ml) 15,541 (11,931–20,244) 20,067 (18,908–21,298)

PK, pharmacokinetic; SD, single dose; SS, steady state.aObserved single-dose data are from Atsmon et al. (2013), and multiple-dose data are from Sidharta et al. (2013).bResults are geometric mean (95% confidence interval).

Fig. 4. Liver, portal vein and systemic plasma concentration-time profiles of macitentan and ACT-132577 after (A–C) single and (D) repeat 10-mg oraldosing. Observed data (A and B) from Atsmon et al., 2013.

140 Treiber et al.

at ASPE

T Journals on Septem

ber 16, 2018jpet.aspetjournals.org

Dow

nloaded from

In contrast, bosentan-treated animals showed elevated bilesalt and alanine aminotransferase levels, which were consid-ered to be consequences of a functional cholestasis (Actelion,data on file).A PBPK model was developed for macitentan and ACT-

132577 with the aim of estimating blood and liver concen-trations. Total mean macitentan concentrations projectedfrom PBPK modeling at steady state in plasma, portal vein,and liver are 400, 419, and 72 nM, respectively. Thecorresponding values for ACT-132577 were 1722 nM forplasma/portal vein and 159 nM for the liver. Correction forplasma protein binding leads to predicted unbound peakmacitentan concentrations of 1.60 nM in plasma, 1.68 nM inthe portal vein, and 0.29 nM in the liver. Free concentrationsfor ACT-132577 in plasma/portal vein and liver were 8.61 and0.80 nM, respectively. These concentrations are significantlybelow those observed in the bile salt transport experimentswith NTCP- and BSEP-expressing cells. The key assumptionof free partitioning of macitentan between blood and liver andequal unbound drug concentrations in both compartments issupported by the experimentally determined Kp,uu of 0.7 inhuman hepatocytes and the sensitivity analysis of the PBPKmodel probing for the impact of various degrees of activehepatic uptake on macitentan pharmacokinetics. In thispermeability-limited PBPK model, the best fit between ob-served and predicted macitentan pharmacokinetics is achievedwhen the ratio of unbound plasma and liver concentrationsapproaches unity (Table 8).Serum bile salts were measured in the multiple ascending

dose study with macitentan in which daily doses up to 30 mgwere given to healthy volunteers for 10 days (Sidharta et al.,2013). As shown in Table 6, there was no difference in bilesalts when the dose groups were compared with placebo. All

individual values were in a narrow range from 7.7–19 mM.Liver transaminases but not bile salts were measured in thelong-term phase III SERAPHIN trial with macitentan (Pulidoet al., 2013). The incidence of liver enzyme elevations morethan 3 times the upper limit of normal was not differentbetween placebo and the 3-mg and 10-mg dose groups. Theincidences of alanine or aspartate aminotransferase eleva-tions more than 3 times above the upper limit of normal were4 and 3% in the 3 mg and 10 mg dose groups, respectively,compared with 4% on placebo (Pulido et al., 2013). Similarly,the incidence of plasma bilirubin elevations was equallydistributed over all dose groups and placebo.In conclusion, there is cumulative evidence demonstrating

the superior liver safety profile of the next generationendothelin receptor antagonist macitentan. Unlike bosentan,macitentan is devoid of a pattern of functional cholestasis inlong-term clinical trials. On a molecular basis, this differenceis the result of discrete changes in chemical structure. Whilemacitentan maintains equal inhibitory potency to bosentanon NTCP and BSEP transport in vitro, replacement of thesulfonamide by a less acidic sulfamide moiety and an in-creased lipophilicity result in a complete change in the hepaticdisposition profile. Bosentan uptake into the liver is an activeprocess, mediated by OATP transport proteins, likely leadingto accumulation in liver cells. In contrast, macitentan par-titions into the liver mostly by passive diffusion. Local drugconcentrations are thus limited by the extensive binding ofmacitentan to plasma proteins to levels that are unlikely toexert an inhibitory potential on hepatic bile salts. Theseconclusions from biochemical and drug disposition data wereconfirmed in an acute rat model and long-term toxicity studiesin the rat and dog. In clinical trials, no change in plasma bilesalts was observed in healthy volunteers upon multiple-dose

TABLE 8Parameter sensitivity analysis of active hepatic uptake on macitentan human pharmacokineticparameters for a single oral dose of 10 mgResults are geometric mean (95% confidence interval).

PBPK Model CLint, active cu, liver/cu, plasmaa AUC0–24 h Cmax

ml/min×106 cells ng*h/ml ng/ml

Perfusion-limited liver — 1 4966 (4634–5321) 199 (190–209)

Permeability-limited liver 0 0.20 10,322 (9828–10,840) 236 (226–247)10 0.55 7557 (7151–7985) 218 (208–229)25 1.05 5395 (5025–5793) 199 (189–209)50 1.85 3661 (3357–3991) 175 (165–185)

Observed — — 5759 (4914–6749) 227 (205–252)

aUnbound macitentan concentrations in liver and plasma at time of peak concentrations in the liver.

Fig. 5. Hepatic transport proteins involved in bile salt trans-port. NTCP as well as OATP1B1, OATP1B3, and OATP2B1 arelocated at the basolateral membrane of hepatocytes andfacilitate the uptake of bile salts in liver cells. BSEP is locatedat the canalicular membrane and promotes the exports of bilesalts into bile canaliculi.

Macitentan and Hepatic Bile Salt Transport 141

at ASPE

T Journals on Septem

ber 16, 2018jpet.aspetjournals.org

Dow

nloaded from

treatment up to 30 mg per day, nor was the incidence of livertransaminase elevations different in macitentan- versusplacebo-treated patients in the long-term morbidity/mortalitySERAPHIN trial. Macitentan at a daily dose of 10 mg istherefore not expected to interfere with hepatic bile salttransport in clinical practice.

Acknowledgments

The authors thank Aude Weigel, Eric Soubieux, Julia Friedrich,and Stephanie Bernhard for dedication and experimental contribu-tions, and Susan Flores and Charlotte Gonzales for help in preparingthe manuscript.

Authorship Contributions

Participated in research design: Treiber, Äänismaa, Delahaye,Treher, Hess, Sidharta.

Conducted experiments: Treiber, Äänismaa, Delahaye, Treher,Hess, Sidharta.

Performed data analysis: Treiber, Äänismaa, Treher, de Kanter,Delahaye, Hess, Sidharta.

Wrote or contributed to the writing of the manuscript: Treiber,Äänismaa, Treher, de Kanter, Hess, Sidharta.

References

Annaert P, Ye ZW, Stieger B, and Augustijns P (2010) Interaction of HIV proteaseinhibitors with OATP1B1, 1B3, and 2B1. Xenobiotica 40:163–176.

Atsmon J, Dingemanse J, Shaikevich D, Volokhov I, and Sidharta PN (2013) In-vestigation of the effects of ketoconazole on the pharmacokinetics of macitentan,a novel dual endothelin receptor antagonist, in healthy subjects. Clin Pharmaco-kinet 52:685–692.

Binet I, Wallnöfer A, Weber C, Jones R, and Thiel G (2000) Renal hemodynamics andpharmacokinetics of bosentan with and without cyclosporine A. Kidney Int 57:224–231.

Bolli MH, Boss C, Binkert C, Buchmann S, Bur D, Hess P, Iglarz M, Meyer S, Rein J,and Rey M, et al. (2012) The discovery of N-[5-(4-bromophenyl)-6-[2-[(5-bromo-2-pyrimidinyl)oxy]ethoxy]-4-pyrimidinyl]-N9-propylsulfamide (Macitentan), an orallyactive, potent dual endothelin receptor antagonist. J Med Chem 55:7849–7861.

Boss C, Bolli M, and Weller T (2002) Endothelin receptor antagonists: structures,synthesis, selectivity and therapeutic applications. Curr Med Chem 9:349–383.

Breu V, Hashido K, Broger C, Miyamoto C, Furuichi Y, Hayes A, Kalina B, LöfflerBM, Ramuz H, and Clozel M (1995) Separable binding sites for the natural agonistendothelin-1 and the non-peptide antagonist bosentan on human endothelin-Areceptors. Eur J Biochem 231:266–270.

Bruderer S, Aänismaa P, Homery MC, Häusler S, Landskroner K, Sidharta PN,Treiber A, and Dingemanse J (2012a) Effect of cyclosporine and rifampin on thepharmacokinetics of macitentan, a tissue-targeting dual endothelin receptor an-tagonist. AAPS J 14:68–78.

Bruderer S, Hopfgartner G, Seiberling M, Wank J, Sidharta PN, Treiber A,and Dingemanse J (2012b) Absorption, distribution, metabolism, and excretion ofmacitentan, a dual endothelin receptor antagonist, in humans. Xenobiotica 42:901–910.

Dawson S, Stahl S, Paul N, Barber J, and Kenna JG (2012) In vitro inhibition of thebile salt export pump correlates with risk of cholestatic drug-induced liver injury inhumans. Drug Metab Dispos 40:130–138.

Dingemanse J, van Giersbergen PL, Patat A, and Nilsson PN (2010) Mutual phar-macokinetic interactions between bosentan and lopinavir/ritonavir in healthyparticipants. Antivir Ther 15:157–163.

Döring B, Lütteke T, Geyer J, and Petzinger E (2012) The SLC10 carrier family:transport functions and molecular structure. Curr Top Membr 70:105–168.

Fattinger K, Funk C, Pantze M, Weber C, Reichen J, Stieger B, and Meier PJ (2001)The endothelin antagonist bosentan inhibits the canalicular bile salt export pump:a potential mechanism for hepatic adverse reactions. Clin Pharmacol Ther 69:223–231.

Galiè N, Hoeper MM, Simon J, Gibbs R, and Simonneau G; Task Force for theDiagnosis and Treatment of Pulmonary Hypertension of the European Society ofCardiology (ESC) and the European Respiratory Society (ERS) (2011) Liver toxicityof sitaxentan in pulmonary arterial hypertension. Eur Heart J 32:386–387.

Gerloff T, Stieger B, Hagenbuch B, Madon J, Landmann L, Roth J, Hofmann AF,and Meier PJ (1998) The sister of P-glycoprotein represents the canalicular bilesalt export pump of mammalian liver. J Biol Chem 273:10046–10050.

Giacomini KM, Huang SM, Tweedie DJ, Benet LZ, Brouwer KL, Chu X, Dahlin A,Evers R, Fischer V, and Hillgren KM, et al.; International Transporter Consortium(2010) Membrane transporters in drug development. Nat Rev Drug Discov 9:215–236.

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

Hull MW, Harris M, Lima V, Guillemi S, Harrigan PR, and Montaner JS (2009)Lopinavir/ritonavir pharmacokinetics in a substitution of high-dose soft-gelatincapsule to tablet formulation. J Clin Pharmacol 49:155–161.

Humbert M, Segal ES, Kiely DG, Carlsen J, Schwierin B, and Hoeper MM (2007)Results of European post-marketing surveillance of bosentan in pulmonary hy-pertension. Eur Respir J 30:338–344.

Iglarz M, Binkert C, Morrison K, Fischli W, Gatfield J, Treiber A, Weller T, Bolli MH,Boss C, and Buchmann S, et al. (2008) Pharmacology of macitentan, an orallyactive tissue-targeting dual endothelin receptor antagonist. J Pharmacol Exp Ther327:736–745.

Katz DA, Carr R, Grimm DR, Xiong H, Holley-Shanks R, Mueller T, Leake B, WangQ, Han L, and Wang PG, et al. (2006) Organic anion transporting polypeptide 1B1activity classified by SLCO1B1 genotype influences atrasentan pharmacokinetics.Clin Pharmacol Ther 79:186–196.

Kis E, Ioja E, Nagy T, Szente L, Herédi-Szabó K, and Krajcsi P (2009) Effect ofmembrane cholesterol on BSEP/Bsep activity: species specificity studies for sub-strates and inhibitors. Drug Metab Dispos 37:1878–1886.

Krum H, Viskoper RJ, Lacourciere Y, Budde M, and Charlon V; Bosentan Hyper-tension Investigators (1998) The effect of an endothelin-receptor antagonist,bosentan, on blood pressure in patients with essential hypertension. N Engl J Med338:784–790.

Leslie EM, Watkins PB, Kim RB, and Brouwer KL (2007) Differential inhibition ofrat and human Na1-dependent taurocholate cotransporting polypeptide (NTCP/SLC10A1)by bosentan: a mechanism for species differences in hepatotoxicity. JPharmacol Exp Ther 321:1170–1178.

Mano Y, Usui T, and Kamimura H (2007) Effects of bosentan, an endothelin receptorantagonist, on bile salt export pump and multidrug resistance-associated protein 2.Biopharm Drug Dispos 28:13–18.

Mateus A, Matsson P, and Artursson P (2013) Rapid measurement of intracellularunbound drug concentrations. Mol Pharm 10:2467–2478.

Ménochet K, Kenworthy KE, Houston JB, and Galetin A (2012) Simultaneous as-sessment of uptake and metabolism in rat hepatocytes: a comprehensive mecha-nistic model. J Pharmacol Exp Ther 341:2–15.

Mita S, Suzuki H, Akita H, Hayashi H, Onuki R, Hofmann AF, and Sugiyama Y(2006) Inhibition of bile acid transport across Na1/taurocholate cotransportingpolypeptide (SLC10A1) and bile salt export pump (ABCB 11)-coexpressing LLC-PK1 cells by cholestasis-inducing drugs. Drug Metab Dispos 34:1575–1581.

Pang KS and Chow EC (2012) Commentary: theoretical predictions of flow effects onintestinal and systemic availability in physiologically based pharmacokinetic in-testine models: the traditional model, segregated flow model, and QGut model.Drug Metab Dispos 40:1869–1877.

Pulido T, Adzerikho I, Channick RN, Delcroix M, Galiè N, Ghofrani HA, Jansa P,Jing ZC, Le Brun FO, and Mehta S, et al.; SERAPHIN Investigators (2013)Macitentan and morbidity and mortality in pulmonary arterial hypertension. NEngl J Med 369:809–818.

Raja SG (2010) Macitentan, a tissue-targeting endothelin receptor antagonist for thepotential oral treatment of pulmonary arterial hypertension and idiopathic pul-monary fibrosis. Curr Opin Investig Drugs 11:1066–1073.

Rodgers T, Leahy D, and Rowland M (2005) Tissue distribution of basic drugs: ac-counting for enantiomeric, compound and regional differences amongst beta-blocking drugs in rat. J Pharm Sci 94:1237–1248.

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

Rodgers T and Rowland M (2007) Mechanistic approaches to volume of distributionpredictions: understanding the processes. Pharm Res 24:918–933.

Rowland M and Tozer N (1989) Clinical Pharmacokinetics: Concepts and Applica-tions, Lea and Febiger, London.

Shitara Y, Itoh T, Sato H, Li AP, and Sugiyama Y (2003) Inhibition of transporter-mediated hepatic uptake as a mechanism for drug-drug interaction between cer-ivastatin and cyclosporin A. J Pharmacol Exp Ther 304:610–616.

Sidharta PN, van Giersbergen PL, and Dingemanse J (2013) Safety, tolerability,pharmacokinetics, and pharmacodynamics of macitentan, an endothelin receptorantagonist, in an ascending multiple-dose study in healthy subjects. J Clin Phar-macol 53:1131–1138.

Sidharta PN, van Giersbergen PL, Halabi A, and Dingemanse J (2011) Macitentan:entry-into-humans study with a new endothelin receptor antagonist. Eur J ClinPharmacol 67:977–984.

Spence R, Mandagere A, Richards DB, Magee MH, Dufton C, and Boinpally R (2010)Potential for pharmacokinetic interactions between ambrisentan and cyclosporine.Clin Pharmacol Ther 88:513–520.

Stieger B, Fattinger K, Madon J, Kullak-Ublick GA, and Meier PJ (2000) Drug- andestrogen-induced cholestasis through inhibition of the hepatocellular bile salt ex-port pump (Bsep) of rat liver. Gastroenterology 118:422–430.

Sütsch G, KiowskiW, Yan XW, Hunziker P, Christen S, StrobelW, Kim JH, RickenbacherP, and Bertel O (1998) Short-term oral endothelin-receptor antagonist therapy inconventionally treated patients with symptomatic severe chronic heart failure.Circulation 98:2262–2268.

Treiber A, Schneiter R, Delahaye S, and Clozel M (2004) Inhibition of organic aniontransporting polypeptide-mediated hepatic uptake is the major determinant in thepharmacokinetic interaction between bosentan and cyclosporin A in the rat. JPharmacol Exp Ther 308:1121–1129.

Treiber A, Schneiter R, Häusler S, and Stieger B (2007) Bosentan is a substrate ofhuman OATP1B1 and OATP1B3: inhibition of hepatic uptake as the commonmechanism of its interactions with cyclosporin A, rifampicin, and sildenafil. DrugMetab Dispos 35:1400–1407.

Treijtel N, van Helvoort H, Barendregt A, Blaauboer BJ, and van Eijkeren JC (2005)The use of sandwich-cultured rat hepatocytes to determine the intrinsic clearanceof compounds with different extraction ratios: 7-ethoxycoumarin and warfarin.Drug Metab Dispos 33:1325–1332.

van Giersbergen PL, Halabi A, and Dingemanse J (2002) Single- and multiple-dosepharmacokinetics of bosentan and its interaction with ketoconazole. Br J ClinPharmacol 53:589–595.

142 Treiber et al.

at ASPE

T Journals on Septem

ber 16, 2018jpet.aspetjournals.org

Dow

nloaded from

van Giersbergen PL, Treiber A, Schneiter R, Dietrich H, and Dingemanse J (2007)Inhibitory and inductive effects of rifampin on the pharmacokinetics of bosentan inhealthy subjects. Clin Pharmacol Ther 81:414–419.

Vavricka SR, Van Montfoort J, Ha HR, Meier PJ, and Fattinger K (2002) Interactionsof rifamycin SV and rifampicin with organic anion uptake systems of human liver.Hepatology 36:164–172.

Warner DJ, Chen H, Cantin LD, Kenna JG, Stahl S, Walker CL, and Noeske T (2012)Mitigating the inhibition of human bile salt export pump by drugs: opportunitiesprovided by physicochemical property modulation, in silico modeling, and struc-tural modification. Drug Metab Dispos 40:2332–2341.

Weber C, Gasser R, and Hopfgartner G (1999) Absorption, excretion, and metabolismof the endothelin receptor antagonist bosentan in healthy male subjects. DrugMetab Dispos 27:810–815.

Weiss J, Theile D, Rüppell MA, Speck T, Spalwisz A, and Haefeli WE (2013) In-teraction profile of macitentan, a new non-selective endothelin-1 receptor antago-nist, in vitro. Eur J Pharmacol 701:168–175.

Wu C (2000) Recent discovery and development of endothelin receptor antagonists.Expt Opin Ther Pat 10:1653–1668 DOI: 10.1517/13543776.10.11.1653.

Yang J, Jamei M, Yeo KR, Tucker GT, and Rostami-Hodjegan A (2007) Prediction ofintestinal first-pass drug metabolism. Curr Drug Metab 8:676–684.

Address correspondence to: Alexander Treiber, Actelion Pharmaceuticals,Ltd, Gewerbestrasse 16, 4123 Allschwil, Switzerland. E-mail: alexander.treiber@actelion.com

Macitentan and Hepatic Bile Salt Transport 143

at ASPE

T Journals on Septem

ber 16, 2018jpet.aspetjournals.org

Dow

nloaded from