Abstract
Intrahepatic cholestasis represents 20%–40% of drug-induced injuries from which a large proportion remains unpredictable. We aimed to investigate mechanisms underlying drug-induced cholestasis and improve its early detection using human HepaRG cells and a set of 12 cholestatic drugs and six noncholestatic drugs. In this study, we analyzed bile canaliculi dynamics, Rho kinase (ROCK)/myosin light chain kinase (MLCK) pathway implication, efflux inhibition of taurocholate [a predominant bile salt export pump (BSEP) substrate], and expression of the major canalicular and basolateral bile acid transporters. We demonstrated that 12 cholestatic drugs classified on the basis of reported clinical findings caused disturbances of both bile canaliculi dynamics, characterized by either dilatation or constriction, and alteration of the ROCK/MLCK signaling pathway, whereas noncholestatic compounds, by contrast, had no effect. Cotreatment with ROCK inhibitor Y-27632 [4-(1-aminoethyl)-N-(4-pyridyl) cyclohexanecarboxamide dihydrochloride] and MLCK activator calmodulin reduced bile canaliculi constriction and dilatation, respectively, confirming the role of these pathways in drug-induced intrahepatic cholestasis. By contrast, inhibition of taurocholate efflux and/or human BSEP overexpressed in membrane vesicles was not observed with all cholestatic drugs; moreover, examples of noncholestatic compounds were reportedly found to inhibit BSEP. Transcripts levels of major bile acid transporters were determined after 24-hour treatment. BSEP, Na+-taurocholate cotransporting polypeptide, and organic anion transporting polypeptide B were downregulated with most cholestatic and some noncholestatic drugs, whereas deregulation of multidrug resistance-associated proteins was more variable, probably mainly reflecting secondary effects. Together, our results show that cholestatic drugs consistently cause an early alteration of bile canaliculi dynamics associated with modulation of ROCK/MLCK and these changes are more specific than efflux inhibition measurements alone as predictive nonclinical markers of drug-induced cholestasis.
Introduction
Many drugs have been reported to induce liver injury in a dose-dependent or dose-independent manner in humans. During the last decades, many drugs have been removed from the market or received a black box warning as a result of their hepatotoxicity and idiosyncrasy. A frequent manifestation of drug-induced liver injury (DILI) is represented by intrahepatic cholestasis, which is characterized by impaired hepatocellular secretion of bile, resulting in accumulation of bile acids (BAs), bilirubin, and cholesterol (Padda et al., 2011). In several population-based studies on DILI, a cholestatic pattern has been found in 20%–40% and a mixed cholestatic and hepatocellular injury pattern was seen in 12%–20% of patients (Lee, 2003), with a large proportion of the drug-induced cholestatic cases reportedly unpredictable.
Early detection of DILI (particularly, cholestasis) remains a challenge for pharmaceutical industries and regulatory agencies. According to the International Transporter Consortium and guidelines on the investigation of drug interactions from the European Medicines Agency (Zamek-Gliszczynski et al., 2012), inhibition of the bile salt export pump (BSEP), which plays a major role in BA canalicular transport (Jansen et al., 1999; Stieger, 2010), should be evaluated during drug development when evidence of cholestatic liver injury has been observed in nonclinical safety studies or in human clinical trials (Kenna, 2014). Indeed, many drugs reported to cause DILI have been identified as efflux transporter inhibitors (Morgan et al., 2010; Stieger, 2010; Dawson et al., 2012; Pedersen et al., 2013), but a significant number of false positives has been found (Pedersen et al., 2013). Moreover, compounds known to interfere with BSEP function are often not associated with significant liver cell injury in preclinical animal models although they have been related to liver damage when administered to humans (Morgan et al., 2010; Dawson et al., 2012), and many cholestatic drugs are not BSEP inhibitors (Pedersen et al., 2013).
Currently, various in vivo and in vitro biologic approaches are used for studying BA transporters. Detection of cholestatic drugs can depend on the choice of method and/or the model and its species origin. Animal models including knockout models, mutant models, and whole tissues can be employed but in vitro systems are more frequently used, based on membrane and cell assays (Brouwer et al., 2013). Inverted vesicles derived from various nontransfected or transfected cell lines are widely used to analyze interactions between substrates and inhibitors with target transporters. However, a number of probe substrates and inhibitors are recommended for evaluation of human ATP-dependent transport into membrane vesicles. Polarized cells are standard models for evaluating drug transport. Primary hepatocytes appear to be the most appropriate cell model to mimic hepatic uptake and efflux functions. However, freshly isolated and cryopreserved hepatocytes are not polarized, and they rapidly lose their functional activities and functional bile canaliculi in standard culture conditions (Guguen-Guillouzo and Guillouzo, 2010). Rat hepatocyte couplets (Graf et al., 1984; Thibault et al.1992) and primary rat and human hepatocytes in a sandwich configuration (Swift et al., 2010; Ellis et al., 2014) have been the most widely used in vitro cell models to analyze hepatic transport processes. Sandwich-cultured human hepatocytes (SCHHs) recapitulate the polarized architecture but they frequently exhibit large interdonor and in vitro time-dependent functional variations. An alternative to SCHHs is the human HepaRG cell line, which expresses features characteristic of mature hepatocytes, exhibits typical functional bile canaliculi and correctly polarized distribution of transport proteins, and produces BAs at levels comparable to those measured in primary human hepatocyte (PHH) cultures (Guguen-Guillouzo and Guillouzo, 2010; Bachour-El Azzi et al., 2015; Sharanek et al., 2015).
The mechanisms underlying drug-induced intrahepatic cholestasis remain poorly understood. Besides the role of hepatobiliary transporter changes, other mechanisms (e.g., altered cell polarity, disruption of cell-to-cell junctions, and cytoskeletal modifications), are believed to participate to the development of intrahepatic cholestasis. In normal hepatocytes, bile canaliculi undergo spontaneous contractions, which are essential for BA efflux (Arias et al., 1993) and alternations in myosin light chain (MLC) subunit 2 phosphorylation and dephosphorylation are required for these contractions (Sharanek et al., 2016). We recently reported that cholestatic drugs can cause in vitro early constriction or dilatation of bile canaliculi associated with deregulation of the Rho kinase (ROCK)/myosin light chain kinase (MLCK) pathway (Sharanek et al., 2016).
In this study, we aimed to determine whether changes in bile canaliculi dynamics and the ROCK/MLCK pathway are more representative targets of cholestatic drugs than efflux transporter inhibition and whether these alterations could better discriminate cholestatic from noncholestatic compounds using human differentiated HepaRG cells. Our results showed that the 12 tested cholestatic drugs could cause bile canaliculi constriction and dilatation associated with different disturbances of the ROCK/MLCK pathway.
Materials and Methods
Reagents.
Cyclosporine A (CsA), chlorpromazine (CPZ), nefazodone (NEF), tolcapone (TOL), diclofenac (DIC), perhexiline (PER), troglitazone (TRO), tacrolimus (TAC) or amiodarone (AMI), acetaminophen (APAP), ximelagatran (XIM), metformin (MET), entacapone (ENT), buspirone (BUS), pioglitazone (PIO), cimetidine (CIM), Y-27632 [4-(1-aminoethyl)-N-(4-pyridyl) cyclohexanecarboxamide dihydrochloride], insulin, and methylthiazoletetrazolium (MTT) were purchased from Sigma (St. Quentin Fallavier, France). Fialuridine (FIA) was supplied by Carbosynth (Compton, UK). Bosentan (BOS) was obtained from Sequoia Research Products (Pangbourne, UK). Calmodulin (CaM) was provided by Merck Chemicals (Fontenay sous Bois, France). [3H]-Taurocholic acid (TCA) was supplied by PerkinElmer (Boston, MA). Specific antibodies against phospho-MLC2 (Ser19) and HSC70 (Heat Shock 70 kDa Protein) were provided by Cell Signaling Technology (Schuttersveld, The Netherlands). Secondary antibodies were obtained from Invitrogen (Saint Aubin, France). GlutaMAX, Hyclone bovine fetal calf serum, penicillin, streptomycin and Williams' E medium were purchased from Thermo Fisher Scientific (Waltham, MA). Hydrocortisone hemisuccinate was purchased from Serb Laboratoires (Paris, France). Other chemicals were of the highest reagent grade.
Cell Cultures and Treatments.
HepaRG cells were seeded at a density of 2.6 × 104 cells/cm2 in Williams’ E medium supplemented with 2 mm GlutaMAX, 100 U/ml penicillin, 100 μg/ml streptomycin, 10% HyClone bovine fetal calf serum, 5 μg/ml insulin, and 50 μM hydrocortisone hemisuccinate. At confluence after a 2-week cell proliferation phase, HepaRG cells were shifted to the same medium supplemented with 1.7% dimethylsulfoxide (DMSO) for 2 additional weeks to obtain confluent differentiated cultures containing equal proportions of hepatocyte-like and progenitors/primitive biliary-like cells (Cerec et al., 2007). These differentiated hepatic cell cultures were used for analytical assays.
PHHs were obtained from Biopredic International (St. Grégoire, France). They were isolated by collagenase perfusion of histologically normal liver fragments from three adult donors undergoing resection for primary and secondary tumors (Guguen-Guillouzo et al., 1982). Donors 1, 2, and 3 were Caucasian men aged 74, 65, and 71 years, respectively. Primary cultures were obtained by seeding 1.5 × 105 hepatocytes/cm2 onto plates in Williams’ E medium supplemented with 10% HyClone bovine fetal calf serum, 100 U/ml penicillin, 100 μg/ml streptomycin, 1 μg/ml insulin, and 2 mM glutamine. The medium was discarded 12 hours after cell seeding, and cultures were maintained thereafter in serum-free medium. PHH cultures were used 3 to 4 days after cell seeding. For drug treatments, differentiated HepaRG cells and PHHs were incubated in a medium containing 2% serum and 1% DMSO. Table 1 provides a list and characteristics of the tested drugs. The drugs were divided into three groups according to their clinical effects: well recognized as cholestatic (group 1), rarely cholestatic (group 2), and noncholestatic (group 3). The compounds were dissolved in DMSO or water and the final concentration of the solvent did not exceed 1%.
Characteristics of the 18 tested drugs
These compounds include the 14 training compounds of the MIP-DILI project plus CPZ, CsA, TAC, and CIM. Compounds were classified into three groups based on their reported clinical cholestatic potential. The list of compounds, their therapeutic use, clinical hepatic effects, and some results obtained in vitro with either vesicles overexpressing human BA transporters or liver cells are displayed.
Cell Viability.
Cytotoxicity of the tested drugs was evaluated using the MTT colorimetric assay. Briefly, cells were seeded in 96- or 24-well plates and exposed to various concentrations of each compound in triplicate for 24 hours. After medium removal, 100 μl serum- and DMSO-free medium containing MTT (0.5 mg/ml) was added to each well and incubated for 2 hours at 37°C. Water-insoluble formazan was dissolved in 100 μl DMSO and absorbance was measured at 550 nm (Aninat et al., 2006). IC20 values (the concentrations causing 20% cytotoxicity) were calculated from the concentration-response curves.
Time-Lapse Cell Imaging.
Phase-contrast images of HepaRG cells and PHHs were captured every 10 minutes using time-lapse, phase-contrast videomicroscopy. The Zeiss Axiovert 200M inverted microscope (Carl Zeiss, Jena, Germany) was equipped with a thermostatic chamber (37°C and 5% CO2) to maintain the cells under normal culture conditions and with an Axiocam MRm camera with a ×10 objective.
Determination of Canalicular Lumen Surfaces.
Canalicular lumen surfaces appeared bright (white) and the hepatocytes/biliary cells were denser (black) using a Zeiss Axiovert 200M phase-contrast microscope. Brightness parameters were adjusted to better distinguish between white and black densities and analysis was performed on at least four images per condition per well. The white canalicular lumen was then quantified using ImageJ software (National Institutes of Health, Bethesda, MD) every 10 minutes for 24 hours. Data obtained during the first 4 hours with all tested compounds are presented and bile canaliculi quantification is expressed as the percent of the control (Supplemental Fig. 1).
MLCK and ROCK Modulation.
MLCK and ROCK implication in bile canaliculi deformations was determined by the use of CaM and Y-27632, two specific modulators. HepaRG cells were treated with CaM, a MLCK- specific activator, and Y-27632, a ROCK-selective inhibitor, in the presence or absence of dilatators and constrictors, respectively (Sharanek et al., 2016). Bile canaliculi alterations after cotreatment were quantified and compared with treatment with the modulator alone as described above.
TCA Clearance.
Cells were first exposed to 43.3 nM [3H]-TCA for 30 minutes to induce its intracellular accumulation and were then washed with standard buffer and incubated with the tested compounds for 2 hours in a standard buffer with Ca2+ and Mg2+. After the incubation time, cells were washed and scraped in 0.5 N NaOH and the remaining radiolabeled substrate was measured through scintillation counting to determine [3H]-TCA clearance. [3H]-TCA clearance was determined based on its accumulation in the cell layers (cells + bile canaliculi) and was calculated relative to the control using the following formula: [3H]-TCA clearance = ([3H]-TCA accumulation in (cells + bile canaliculi)Control × 100)/([3H]-TCA accumulation in cell layersTested compound) (Sharanek et al., 2016).
Real-Time Quantitative Polymerase Chain Reaction Analysis.
Total RNA was extracted from 106 HepaRG cells with the SV Total RNA Isolation System (Promega). RNAs were reverse transcribed into cDNA and real-time quantitative polymerase chain reaction was performed using a SYBR Green mix. Primer pair sequences were as follows: glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (ID 2597; forward: 50-ttcaccaccatggagaaggc-30; reverse: 50-ggcatggactgtggtcatga-30), BSEP (ID 8647; forward: 50-tgatcctgatcaagggaagg-30; reverse: 50-tggttcctgggaaacaattc-30), Na+-taurocholate cotransporting polypeptide (NTCP) (ID 6554; forward: 50-gggacatgaacctcagatt-30; reverse: 50-cgttggatttgaggacgat-30), organic anion transporting polypeptide (OATP)-B (ID 6579; forward: 50-tgattggtatggggctatc-30; reverse: 50- catatcctagggctggtgt-30), multidrug resistance-associated protein (MRP2) (ID 1244; forward: 50-tgagcaagtttgaaacgcacat-30; reverse: 50-agctcttctcctgccgtctct-30), (MRP3) (ID 8714; forward: 50-gtccgcagaatggacttgat-30; reverse: 50-tcaccacttggggatcatt-30), (MRP4) (ID 10257; forward: 50-gctcaggttgcctatgtgct-30; reverse: 50-cggttacaatttcctcctcca-30), albumin (ALB) (ID 213; forward: 50-gctgtcatctcttgtgggctgt-30; reverse: 50-actcatgggagctgctggttc-30), aldolase B (ALDOB) (ID 229; forward: 50-tgcgcccagtacaagaaggacggttg-30; reverse: 50-ctcaagatctcggacggctg-30), CYP3A4 (ID 1576; forward: 50-tcagcctggtgctcctctatctat-30; reverse: 50-tccagggcccacacctctgcct-30), and superoxide dismutase (SOD2) (ID 6648; forward: 50-acaggccttattccactgct-30; reverse: 50-cagcataacgatcgtggttt-30).
Western Blotting Analysis of p-MLC2.
HepaRG cells were treated with the tested compounds for 1, 2, and 3 hours at a concentration causing bile canaliculi alteration, then washed with cold phosphate-buffered saline, and finally resuspended in cell lysis buffer supplemented with protease and phosphatase inhibitors (Roche, Mannheim, Germany). Aliquots containing equivalent total protein content, as determined by the Bradford procedure with bovine serum albumin as the standard, were subjected to SDS/12% PAGE, electrotransferred to Immobilon-P membranes, and incubated overnight with primary antibodies directed against p-MLC2 and HSC70 (Heat Shock 70 kDa Protein). After incubation with a horseradish peroxidase–conjugated anti-mouse/rabbit antibody (Thermo Fisher Scientific, Waltham, MA), membranes were incubated with a chemiluminescence reagent (Millipore, Billerica, MA) and bands were visualized and quantified by densitometry with fusion-CAPT software (Vilber Lourmat, Collégien, France).
Statistical Analysis.
One-way analysis of variance with a multiple-comparisons test (GraphPad Prism software, version 6.00; GraphPad Software Inc., La Jolla, CA) was performed to compare time-dependent MLC2 phosphorylation in different samples. Data were considered significantly different when P < 0.05. The t test was applied to compare values of bile canaliculi quantification between treated and corresponding cotreated cultures (with CaM or Y-27632). Data were considered significantly different when P < 0.001. The t test was also applied to compare values of BA clearance and mRNA levels between treated and corresponding control cultures. Data were considered significantly different when P < 0.05, P < 0.01, and P < 0.001. Each value corresponded to the mean ± S.E.M. of three independent experiments.
Results
Cytotoxicity Determination.
Experiments were performed to first calculate IC20 values for each compound after 24-hour treatment of HepaRG cells using the MTT assay (Table 2). Three concentrations were then tested for further experiments, the highest corresponding to the IC20 value. IC10 or IC20 values are commonly selected for use in early in vitro toxicology studies (Liguori et al., 2008; Pomponio et al., 2015). When IC20 values could not be determined because of absence of toxicity, higher concentrations were tested. Table 2 displays the IC20 values and/or final selected maximum concentrations tested.
Cmax values, IC20 values, and maximum tested concentrations of the 18 tested drugs
Cmax values are from the literature and refer to the maximum serum concentration of the drug (Regenthal et al., 1999; Chu et al., 2007; Xu et al., 2008; Dhillon and Keating, 2009; Keisu and Andersson, 2010; Dawson et al., 2012; Sharanek et al., 2014; Schadt et al., 2015). Cells were incubated for 24 hours with different concentrations of each drug. For cytotoxicity testing, the cells were treated with varying concentrations of each drug for 24 hours and then assayed with the MTT colorimetric test and IC20 values were calculated.
Bile Canaliculi Dynamics.
Bile canaliculi contractions are essential for clearance of BAs; they are characterized by repeated opening and closing processes (Sharanek et al., 2016). We examined bile canaliculi dynamics in living cells by time-lapse imaging during at least 12 hours after the addition of the tested compounds. Depending on the drug, bile canaliculi showed constriction, dilatation, or no change. The five compounds from group 1 caused early alterations of bile canaliculi, usually starting after 1-hour exposure of HepaRG cells (Fig. 1). As previously reported (Sharanek et al., 2016), CPZ (50 µM) and CsA (50 µM) induced constriction, whereas BOS (100 µM) caused dilatation of bile canaliculi. The two other cholestatic compounds from group 1 (NEF and TRO) and four rarely cholestatic compounds from group 2 (DIC, ENT, PER, and TAC) also caused morphologic alterations of bile canaliculi at the same range of concentrations, either constriction with NEF (75 µM), PER (20 µM), and TRO (50 µM) (Supplemental Video 1) or dilatation with ENT (100 µM) (Supplemental Video 2), DIC (200 µM), and TAC (50 µM) (Fig. 1, A and B). Two other compounds from group 2 (MET and CIM) also caused alterations of bile canaliculi dynamics but at much higher although noncytotoxic concentrations (i.e., constriction and dilatation at 3 and 4 mM, respectively) (Fig. 1B).
Effects of cholestatic and noncholestatic drugs on bile canaliculi (BC) dynamics in HepaRG cells. (A and B) Phase-contrast images after 2-hour treatment showing dilatation (red arrows), constriction (green arrows), or no change (yellow arrows). (C) Effects of 10 μM CsA after 3 hours. (D) Effects of TOL after 1 and 6 hours. (E) Effects of negative compounds (XIM and AMI) after 2 hours. (F–I) BC surfaces of HepaRG cells exposed to dilatators (F), constrictors (G), TOL/ENT (H), and control compounds (I) were quantified. Briefly, BC surface quantification was based on brightness parameters that were adjusted to eliminate noncorresponding objects and analysis was performed on at least three different experiments. The white canalicular lumen was then quantified using ImageJ software every 10 minutes for 24 hours. Splices seen on some photographs are due to the software used for time-lapse imaging analysis. Data are expressed relative to untreated cells, arbitrarily set at a value of 100%. Data represent the means ± S.E.M. of three independent experiments. All dilatators showed at least transiently enlarged BC > 125%. Bar, 50 μm.
To note, differential time- or concentration-dependent bile canaliculi alterations were evident with some cholestatic compounds. Thus, among the five constrictors from group 1, only CsA caused dilatation of bile canaliculi at low concentrations (10 µM) (Fig. 1C). Interestingly, one group 2 compound (100 µM TOL) showed dilatation during the first hour followed by constriction of bile canaliculi during the following hours of treatment (Fig. 1D). By contrast, ENT, a dilatator from the same therapeutic family as TOL, caused strong dilatation starting within the first hour without any constriction during the following hours (Fig. 1B). All other tested compounds—namely, XIM, AMI, APAP, FIA, BUS, and PIO (group 3)—did not alter bile canaliculi dynamics at noncytotoxic concentrations (Fig. 1E; Supplemental Video 3).
Morphologic canalicular alterations induced by the tested compounds were quantified by measuring the canalicular surface using ImageJ software every 10 minutes for 24 hours. Data obtained during the first 3 hours with all tested compounds are displayed in Fig. 1, F–I. When observed, several altered bile canaliculi were detected in each selected culture area as with each cholestatic drug. Interestingly, the results were comparable to those previously obtained using the canalicular distribution of zona occludens-1 protein to measure bile canaliculi areas (Sharanek et al., 2016); for example, BOS caused a 1.5-fold increase in bile canaliculi surfaces using either method.
Interestingly, constrictions appeared to be irreversible, whereas dilatations were only transient, as seen with TOL. Noticeably, AMI showed an unspecific increase in bile canaliculi surface quantification starting after 1 hour due to the appearance of intracytoplasmic vesicles corresponding to the induction of phospholipidosis (Fig. 1E). Drug effects on bile canaliculi structures are summarized in Supplemental Table 1.
ROCK and MLCK Alterations with Cholestatic Drugs.
MLC2 (the regulatory subunit of myosin 2, which is predominantly distributed in the pericanalicular region of hepatocytes) is known as a target of ROCK; its phosphorylation/dephosphorylation reflects the dynamic rhythm of bile canaliculi and was found to be impaired by CPZ, CsA, and BOS (Sharanek et al., 2016). On the basis of our previous study, we analyzed the 18 compounds at three critical time points (1, 2, and 3 hours) by Western blotting. However, we first verified with six compounds that similar alterations of MLC2 phosphorylation occurred after 30 minutes whatever the tested compound (Supplemental Fig. 2). Western blot analysis of MLC2 showed an increased phosphorylation tendency between 1 and 3 hours with constrictors (CPZ, CsA, PER, NEF, TRO, and MET) and inversely dephosphorylation tendency with dilatators (TAC, BOS, CIM, and DIC) (Fig. 2A). Depending on the compound, either a constrictor or a dilatator, large individual quantitative differences were observed in their effects on bile canaliculi measurements of dilatation and constriction (Fig. 2B). The potent constrictors (PER, TRO, and NEF) showed a much higher rate of phosphorylation than MET after 3-hour treatment (Fig. 2, A and B). Interestingly, ENT showed a decrease in MLC2 phosphorylation during bile canaliculi dilatation, whereas TOL did not show any significant change, likely due to rapid morphologic changes of bile canaliculi from dilatation to constriction after 1-hour treatment (Fig. 2B). All noncholestatic drugs did not significantly affect MLC2 phosphorylation/dephosphorylation (Fig. 2, C and D).
Alterations of MLC2 phosphorylation/dephosphorylation by cholestatic drugs in HepaRG cells. (A) Representative Western blots of p-MLC2 compared with HSC70 (Heat Shock 70 kDa Protein), a loading control, using anti-S19 phospho-MLC2 and anti-HSC70 antibodies at various time points (1, 2, and 3 hours) in cells treated with group 1 and group 2 cholestatic drugs. (B and D) Graphical representation of MLC2 phosphorylation/dephosphorylation quantified using fusion-CAPT software. (C) Western blots of p-MLC2 obtained with group 3 molecules. Data are expressed in arbitrary units (A.U.) and represent means ± S.E.M. of three independent experiments. One-hour treatment was arbitrarily set at a fixed value of 2.5 for all of the drugs. *P < 0.05.
To further analyze alterations of bile canaliculi dynamics by cholestatic compounds, involvement of ROCK and MLCK was investigated using Y-27632, a ROCK-specific inhibitor, and CaM, a MLCK-specific activator. Y-27632 (10 µM) caused bile canaliculi dilatation, whereas CaM (5 µM) alone caused no bile canaliculi alteration (Fig. 3A). Alteration of bile canaliculi size after cotreatment with a drug and CaM or Y-27632 was quantified and compared with the drug treatment alone. As expected, cotreatment with CaM reduced bile canaliculi dilatation induced by the dilatators (BOS, DIC, CIM, and TAC) (Fig. 3, B and F). By contrast, cotreatment with Y-27632 partly counteracted the early phase of constriction induced by the constrictors (TRO, NEF, PER, and MET) (Fig. 3, C and D). TOL showed a lower dilatation during the first hour when cotreated with CaM and less constriction after 1 hour when cotreated with Y-27632, whereas ENT showed a much lower dilatation in the presence of CaM (Fig. 3E). Interestingly, cotreatment with BOS and Y-27632 showed an increase in the extent of dilatation compared with BOS or Y-27632 alone (Fig. 3G).
Alterations of the ROCK/MLCK pathway by cholestatic drugs in HepaRG cells. (A) Bile canaliculi (BC) surface quantification after Y-27632 (10 µM) and CaM (5 µM) treatments for 4 hours. (B) Phase-contrast images of cells treated with 200 µM DIC with or without the MLCK activator CaM (5 µM) for 2 hours. (C) Phase-contrast images of cells treated with 75 µM NEF with or without the ROCK inhibitor Y-27632 (10 µM) for 2 hours. (D–F) BC surface quantification with constrictors as described above (D), TOL and ENT (E), and dilatators (F). (G) Phase-contrast images of cells treated with 100 µM BOS with or without Y-27632 (10 µM) or with Y-27632 (10 µM) alone for 2 hours. Data are expressed relative to untreated cells, arbitrarily set at a value of 100%. Data represent the means ± S.E.M. of three independent experiments. #P < 0.001. Bar, 50 μm.
TCA Clearance.
The influence of all of the tested drugs on the efflux of [3H]-TCA, a predominant BSEP substrate, was assessed in HepaRG cells maintained in a standard buffer. Cells were first incubated with [3H]-TCA for 30 minutes and then treated for 2 hours with all drugs at three different concentrations, the highest causing bile canaliculi alterations. [3H]-TCA clearance, reflecting BA accumulation in HepaRG hepatocytes and thus the cholestatic potential, was determined by quantification of the intracellular accumulation of radiolabeled TCA. The results are displayed in Fig. 4. Two categories of compounds could be distinguished. The first was composed of eight compounds that caused a dose-dependent decrease in TCA clearance after 2-hour treatment and included the five group 1 cholestatic drugs (NEF, TRO, BOS, CPZ, and CsA) as well as three group 2 drugs (PER, TOL, and TAC). The second category was composed of the other 10 drugs that did not inhibit TCA efflux at the concentrations used and included four group 2 drugs (DIC, ENT, MET, and CIM) and the six drugs that are classified as noncholestatic in vivo (XIM, AMI, APAP, FIA, BUS, and PIO). Noticeably, DIC caused a concentration-dependent increase in TCA clearance, reaching 4-fold at 200 µM.
Effects of tested drugs on BA clearance activity in HepaRG cells. (A–C) [3H]-TCA clearance in HepaRG cells treated with group 1 (A), group 2 (B), and group 3 (C) drugs for 2 hours. [3H]-TCA clearance was determined based on its accumulation in cell layers (Ctlr indicates control). Data are expressed relative to the levels found in untreated cells, arbitrarily set at a value of 100%. Data represent the means ± S.E.M. of three independent experiments. *P<0.05; **P<0.01; ***P<0.001 (compared with untreated cells).
Modulation of BA Transporter Expression.
Genes encoding the major BA transporters can also be deregulated by cholestatic drugs (Pauli-Magnus and Meier, 2006) and the deregulation of their mRNA expression is usually linked to their activity (Bramow et al., 2001). Six major genes encoding either efflux (BSEP, MRP2, MRP3, and MRP4) or uptake (NTCP and OATP-B) transporters were analyzed by real-time quantitative polymerase chain reaction after a 24-hour treatment with three concentrations of each tested drug. Different patterns, depending on the compound and the tested concentration, were evident and are shown in Table 3 and Supplemental Figs. 3–8. Both BSEP and NTCP were strongly repressed by all group 1 cholestatic compounds. In addition, OATP-B was repressed by CsA, BOS, and NEF, whereas MRP2, MRP3, and MRP4 were overexpressed only by CsA. Most compounds in group 2 (TOL, DIC, PER, MET, CIM, and TAC) also repressed BSEP, NTCP, and OATP-B. Moreover, MRP2, MRP3, and MRP4 were upregulated by TAC. In addition, MRP2 was upregulated by DIC and MRP3 was downregulated by PER. Noticeably, none of the tested transporters was significantly deregulated by ENT. Some BA transporters were also deregulated by noncholestatic compounds. Thus, BUS downregulated BSEP and NTCP, whereas PIO and AMI inhibited BSEP and NTCP expression. In addition, transcripts of four other genes (ALB, ALDOB, CYP3A4, and SOD2) were also measured (Supplemental Table 2). A slight or significant decrease in expression of the liver-specific genes (ALB and ALDOB) was also observed with most drugs, which likely reflected cell adaptation or some cytotoxicity after a 24-hour treatment and not a direct drug-induced cholestatic effect. Induction of CYP3A4 mRNAs with CPZ and their inhibition by CsA, TAC, and BOS were in agreement with previous reports. SOD2 mRNA expression was increased with CsA and CPZ and was not altered with control compounds.
Effects of tested drugs on transcript levels of different hepatic transporters.
HepaRG cells were exposed to the three groups of compounds at the indicated concentration for 24 hours. mRNA levels were measured by real-time quantitative polymerase chain reaction analysis. All results are expressed relative to the levels found in control cells, arbitrarily set at a value of 1.
Comparative Analysis of Bile Canaliculi Dynamics in PHHs.
Hepatocytes from three different donors were analyzed by phase-contrast cell imaging. Similar results as those described with HepaRG hepatocytes were obtained within the first hours of drug exposure (i.e., bile canaliculi constriction with CPZ, CsA, NEF, and TRO; Fig. 5A) as well as with TOL after 2 hours, and bile canaliculi dilatation with TOL (between 1 and 2 hours), ENT, DIC, and BOS (Fig. 5B); however, these effects were not observed in cells from all donors with DIC, ENT, and TAC. Moreover, the rarely cholestatic drug PER that caused bile canaliculi constriction in HepaRG cells did not alter bile canaliculi structures in any of the three PHH populations (Fig. 5C). As expected, the noncholestatic drug FIA was ineffective. The time course of appearance of bile canaliculi alterations was approximately the same for all of the cholestatic compounds but the extent of dilatation appeared to be reduced compared with that observed in HepaRG cells.
Disruption of bile canaliculi (BC) rhythmic movements in 4-day PHH cultures. (A and B) Phase-contrast images after 2-hour treatment showing untreated cells (control) and cells treated with constrictors (50 µM CPZ, 50 µM CsA, 75 µM NEF, and 100 µM TRO), with 100 µM TOL, or with dilatators (100 µM ENT, 200 µM DIC, and 100 µM BOS). (C) Results obtained with hepatocytes from the three donors treated with different drugs. nc indicates no change. Bar, 50 μm.
Discussion
Both bile canaliculi dynamics and ROCK/MLCK activities can be impaired by cholestatic drugs (Sharanek et al., 2016). In this study, we demonstrated by using a set of 18 cholestatic and noncholestatic compounds that the 12 drugs that were classified cholestatic on the basis of reported clinical findings caused disturbances in both bile canaliculi dynamics and the ROCK/MLCK pathway. By contrast, noncholestatic compounds exhibited no effect on these pathways. Comparison of these data with TCA efflux showed that some cholestatic drugs were ineffective or enhanced TCA clearance, whereas some noncholestatic drugs exerted an inhibitory effect.
Drug-induced impairment of bile canaliculi dynamics can result in either early constriction or dilatation of the canalicular lumen. These morphologic changes were observed with the two groups of cholestatic drugs (i.e., well recognized as either causing clinical cholestasis or being responsible for rare cases of clinical cholestasis) (Table 1). Bile canaliculi constriction was irreversible, representing a terminal step leading to cell death; however, dilatation could be reversible and did not impede (at least within 24 hours) cell survival. Interestingly, the two types of bile canaliculi alterations could be observed with some drugs, showing either time or concentration dependence. TOL first caused dilatation and later constriction, whereas CsA caused dilatation at low concentrations and constriction at high concentrations. However, other bile canaliculi constrictors (CPZ, NEF, PER, and TRO) did not show clear evidence of dilatation prior to constriction or differences in the effects on bile canaliculi dynamics at lower concentrations or shorter times. Since a cholestatic liver usually develops after chronic drug administration in patients and is characterized by dilatated bile canaliculi (Imanari et al., 1981; Watanabe et al., 1991; Chung et al., 2002), experiments are ongoing to determine whether dilatation of bile canaliculi could occur after prolonged repeated treatments with constricting drugs at lower concentrations. No alteration in bile canaliculi was observed with the six noncholestatic drugs. Noticeably, additional noncholestatic drugs have been tested, including ambrisentan (a member of the same endothelin receptor antagonist family as BOS) and ibuprofen, and were found to be similarly ineffective on the dilatation and constriction of bile canaliculi (data not shown).
Noteworthy, bile canaliculi constriction and dilatation was associated with different alterations of the ROCK/MLCK pathway. Activation of the ROCK pathway and inhibition of the MLCK pathway associated with increased phosphorylation and dephosphorylation of MLC2 were observed with constrictors and dilatators, respectively (Fig. 6). Importantly, differential quantitative effects were observed, supporting the conclusion that cholestatic drugs can act by different mechanisms, involving different targets on the ROCK/MLCK pathway (Sharanek et al., 2016). The involvement of MLC2 in the occurrence of bile canaliculi constriction or dilatation was also demonstrated by using Y-27632, a ROCK-specific inhibitor, and CaM, a MLCK-specific activator, which caused inhibition of bile canaliculi constriction and dilatation respectively, giving support to the conclusion that disturbances of the ROCK/MLCK pathway preceded alterations of bile canaliculi dynamics. Interestingly, BOS combined with Y-27632 had an additive effect on bile canaliculi dilatation, showing that both drugs acted via distinct MLCK and ROCK enzymatic targets (Fig. 3G).
Schematic representation of the molecular targets of cholestatic drugs. Dilatators and TOL (within the first hour) inhibit Ca2+/CaM-dependent MLCK leading to dilatation. Constrictors and TOL (3 h) activate ROCK activity and maintain abnormal high MLC2 phosphorylation, thereby leading to bile canaliculi (BC) constriction (Actin - Myosin interaction). Y-27632 inhibits ROCK activity and causes MLC2 dephosphorylation, thereby leading to BC dilatation.
Drug inhibition studies of BSEP have mostly been performed using membrane vesicles overexpressing BSEP (Morgan et al., 2010). Although PHHs represent a more relevant physiologic in vitro model, they have been only rarely used (Kostrubsky et al., 2006; Swift et al., 2010; Pedersen et al., 2013). A large proportion of cholestatic drugs have been identified as BSEP inhibitors (Morgan et al., 2010; Stieger, 2010; Dawson et al., 2012; Pedersen et al., 2013), but a significant number of false positives has been found, using either vesicle assays or SCHHs (Pedersen et al., 2013). Using the TCA clearance assay, we also found that not all cholestatic drugs are inhibitors of TCA clearance. Indeed, compared with deformation of bile canaliculi associated with disruption of the ROCK/MLCK pathway induced by the 12 cholestatic drugs and not by the six noncholestatic drugs, alteration of TCA clearance was observed only with four of seven drugs known to rarely cause cholestasis in patients (DIC, ENT, MET, and CIM). Noticeably, if DIC and ENT inhibit BSEP using transfected vesicles, MET and CIM are not classified or only weakly shown to be inhibitors of BSEP (Morgan et al., 2010). Moreover, the potent cholestatic drug CPZ was not detected as a BSEP inhibitor using transfected vesicles (Pedersen et al., 2013), whereas it appeared as a bile canaliculi constrictor and an inhibitor of TCA efflux in this study. This discrepancy could be attributed to CPZ-induced generation of reactive oxygen species that cannot be observed with transfected vesicles (Anthérieu et al., 2013). Two additional major differences were evident between transfected cells and TCA clearance assays: we observed inhibition of TCA clearance and bile canaliculi deformation with TOL and no change with PIO, whereas opposite data were reported with transfected vesicles (Morgan et al., 2010; Pedersen et al., 2013). Noticeably, PIO has been classified as either rarely cholestatic (May et al., 2002) or noncholestatic (Pedersen et al., 2013) and as a potent BSEP inhibitor (Morgan et al., 2010; Dawson et al., 2012) or not (Kaimal et al., 2009; Pedersen et al., 2013) using transfected vesicles. However, in the study by Pedersen et al. (2013), PIO at 50 µM did not significantly inhibit TCA efflux in SCHHs. Our results were in agreement with this finding; at 100 µM, PIO had no effect on TCA efflux or bile canaliculi dynamics. Together, these data clearly emphasize the importance of both cell model-based systems and assays to estimate BSEP inhibition. In our experimental whole-cell system, TCA efflux could not be considered as reflecting only BSEP activity. Indeed, TCA clearance was reduced by CsA to approximately 20% of control values. This strong inhibition could be due not only to an inhibition of canalicular efflux but also to an inhibition of basolateral efflux. By contrast, a dose-dependent increase in TCA efflux was obtained with DIC, likely reflecting an enhanced basolateral efflux and/or a reduced influx activity. Moreover, with some drugs (whether cholestatic or not, such as BUS), only a slight increase in TCA efflux was observed at low concentrations, whereas efflux levels at higher concentrations were comparable to those measured in control HepaRG cell cultures, confirming previously reported concentration-dependent effects of drugs on TCA efflux using SCHHs (Kostrubsky et al., 2006). With the two compounds MET and CIM, high noncytotoxic concentrations were required to observe changes in bile canaliculi dynamics, whereas TCA efflux levels were comparable to those measured in controls but significantly decreased with low drug concentrations. By contrast, ENT did not inhibit TCA efflux at any of the concentrations tested. Together, these data support the conclusion that not all cholestatic drugs induce accumulation of TCA in HepaRG cells and that its clearance can be modulated by different mechanisms other than direct alteration of BSEP activity.
Most studies have focused on interactions between drugs and BSEP without considering perturbation of expression of this transporter. Recently, one group (Garzel et al., 2014) analyzed 30 drugs for their ability to perturb BSEP expression in PHH cultures after 24-hour treatment and classified them into three categories according to their capacity to strongly (>60%), moderately (20%–60%), or not repress the transporter. In this study, we analyzed four of these compounds. Similarly, we found a strong repression of BSEP expression with TRO and BOS, which were tested at comparable concentrations, but we demonstrated a strong repression with NEF and CsA, which were tested at higher concentrations. As shown here, the effects of cholestatic and noncholestatic drugs on BA transporter expression depended on the drug concentration and the transporter tested (Supplemental Fig. 3A). To note, most cholestatic drugs repressed BSEP, NTCP, and OATP-B at concentrations that caused bile canaliculi deformations, suggesting that in vitro experimental conditions (drug concentrations, in vitro models, etc.) are critical for obtaining correct results. The three BA transporters were also repressed by some noncholestatic drugs. Interestingly, it has been found that downregulation of transporters also occurs at the mRNA level in the human cholestatic liver, as observed for BSEP, NTCP, and OATP (Zollner et al., 2001). More variable results were obtained with MRPs (Table 3). The most important changes were induction of MRP2–MRP4 with CsA and TAC and of MRP2 with DIC. MRP3 upregulation has been described as a compensatory mechanism to BA accumulation in the cholestatic rat liver (Donner and Keppler 2001). Since transcript levels were measured 24 hours after drug addition, their changes more likely reflected secondary effects rather than direct and specific drug effects.
Interestingly, different responses were obtained with drugs of the same family as observed with TOL and ENT (as described above) as well as CsA and TAC, which caused bile canaliculi constriction and dilatation, respectively. TAC has been reported to cause rare cases of cholestasis in infants. Accordingly, TAC induced dilatation at only 50 μM in this study. Since TAC is used at 10- to 100-fold lower therapeutic doses than CsA, the absence of hepatotoxic and cholestatic effects at concentrations up to 25 μm was in full agreement with reports of its clinical safety (Mihatsch et al., 1998). Accordingly, considering the Cmax values measured in patients (1.15 and 0.09 µM for CsA and TAC, respectively), cholestatic concentrations are expected to be reached normally only in patients treated with CsA. However, considering the 12 tested cholestatic drugs, no direct correlation could be established between Cmax values (Table 2) and their in vitro concentrations inducing bile canaliculi deformation, confirming data reported with BSEP inhibition assays on a large set of compounds (Dawson et al., 2012; Köck et al., 2014).
Bile canaliculi deformation was also investigated in PHH cultures from three donors treated with some of the tested compounds. If bile canaliculi constriction or dilatation was induced by all cholestatic drugs except PER, donor–donor differences were observed. Only four of eight tested cholestatic compounds were found to cause alterations or no changes of bile canaliculi structures in the three hepatocyte populations. Among the factors that could explain donor differences were variations in drug metabolism and detoxification capacity and/or BA transporter levels. Nevertheless, these data with PHHs confirm those obtained with HepaRG cells and support the suitability of this model as a reproducible, easy-to-use liver cell model for screening and mechanistic studies for the identification of drug-induced cholestasis.
In summary, early bile canaliculi deformations resulting in constriction or dilatation of the canalicular lumen associated with alterations of the ROCK/MLCK pathway were observed with all tested cholestatic drugs, including those that are not BSEP inhibitors (Table 4). Together, our results favor the conclusion that alterations of bile canaliculi dynamics with the involvement of the ROCK/MLCK pathway are more specific predictive markers than TCA clearance and direct BSEP inhibition to screen the cholestatic potential of new chemical entities.
Summary of the results on the 18 tested drugs
Bile canaliculi deformation (i.e., constriction or dilatation), BSEP inhibition (vesicles), [3H]-TCA clearance (control = 100%), BSEP mRNA expression (control = 1), MLC2 phosphorylation state, and MLCK/ROCK implication are summarized. Human BSEP inhibition levels in transfected vesicles are taken from the literature and expressed as high (++), medium (+), and low (−) or absent.
Acknowledgments
The authors thank R. Le Guevel (ImPACcell Platform, BIOSIT, Université Rennes 1, Rennes, France) for imaging analysis.
Authorship Contributions
Participated in research design: Burbank, Burban, Sharanek, Guillouzo.
Conducted experiments: Burbank.
Performed data analysis: Burbank, Weaver, Guguen-Guillouzo, Guillouzo.
Wrote or contributed to the writing of the manuscript: Burbank, Weaver, Guillouzo.
Footnotes
- Received May 4, 2016.
- Accepted August 11, 2016.
This research was supported by the European Union Mechanism-Based Integrated Systems for the Prediction of Drug Induced Liver Injury (MIP-DILI) project [Contract MIP-DILI-115336]. The MIP-DILI project has received support from the Innovative Medicines Initiative Joint Undertaking, resources of which are composed of financial contributions from the European Union’s Seventh Framework Programme (FP7/2007–2013) and in-kind contributions from members of the European Federation of Pharmaceutical Industries and Associations. M.G.B. was financially supported by a CIFRE (Convention Industrielle de Formation par la Recherche) PhD contract with Servier. A.B. and A.S. were supported by the MIP-DILI project.
↵
This article has supplemental material available at dmd.aspetjournals.org.
Abbreviations
- ALB
- albumin
- ALDOB
- aldolase B
- AMI
- amiodarone
- APAP
- acetaminophen
- BA
- bile acid
- BOS
- bosentan
- BSEP
- bile salt export pump
- BUS
- buspirone
- CaM
- calmodulin
- CIM
- cimetidine
- CPZ
- chlorpromazine
- CsA
- cyclosporine A
- DIC
- diclofenac
- DILI
- drug-induced liver injury
- DMSO
- dimethylsulfoxide
- ENT
- entacapone
- FIA
- fialuridine
- HSC 70
- Heat Shock 70kDa Protein
- MET
- metformin
- MLC
- myosin light chain
- MLCK
- myosin light chain kinase
- MRP
- multidrug resistance-associated protein
- MTT
- methylthiazoletetrazolium
- NEF
- nefazodone
- NTCP
- Na+-taurocholate cotransporting polypeptide
- OATP
- organic anion transporting polypeptide
- PER
- perhexiline
- PHH
- primary human hepatocyte
- PIO
- pioglitazone
- ROCK
- Rho kinase
- SCHH
- sandwich-cultured human hepatocyte
- SOD
- superoxide dismutase
- TAC
- tacrolimus
- TCA
- taurocholic acid
- TOL
- tolcapone
- TRO
- troglitazone
- XIM
- ximelagatran
- Y-27632
- 4-(1-aminoethyl)-N-(4-pyridyl) cyclohexanecarboxamide dihydrochloride
- Copyright © 2016 by The American Society for Pharmacology and Experimental Therapeutics