Abstract
HepaRG cells possess the unique property to differentiate in vitro and to express various functions of mature hepatocytes, including the major cytochromes P450 (P450s). In the present study, we carefully analyzed mRNA expression and activity of the major P450s and their responsiveness to three prototypical inducers, phenobarbital, rifampicin, and omeprazole, in differentiated HepaRG cell cultures over a 4-week period after low and high seeding. Only minor differences were observed in P450 activities when measured by two cocktails of probe substrates, probably related to the choice and/or concentration of substrates. Similar results were obtained from the two cell seeding conditions. Expression and activities of several P450s were dimethyl sulfoxide-dependent. However, basal P450 expression and activities as well as their responsiveness to the prototypical inducers were well maintained over the 4-week period, and a good correlation was observed between transcript levels and corresponding activities. Thus, CYP1A2, CYP2B6, and CYP3A4 were found to accurately respond to their respective prototypical inducers, i.e., omeprazole, phenobarbital, and rifampicin. Likewise, basal expression of several phase II enzymes, transporters, and nuclear receptors, and response to inducers were also well preserved. More genes were found to be induced in HepaRG cells than in primary human hepatocytes, and no marked variation was noticed between the different passages. Taken together, these data support the conclusion that HepaRG cells represent a promising surrogate to primary human hepatocytes for xenobiotic metabolism and toxicity studies.
Drug-induced hepatotoxicity represents a major clinical problem accounting for approximately 50% of all cases of acute liver failure, and it is a major cause of attrition in drug development. Estimation of cytochrome P450 (P450) induction and inhibition and prediction of drug-drug interactions are an important consideration for the development of novel therapeutic agents (Park and Miller, 1996; Pelkonen et al., 2008). Predicting the ability of a drug to modulate P450 expression at an early stage of its discovery and development should reduce the risk of failure in the clinic and, more importantly, permit the identification of alternative noninducing/noninhibiting chemical structures. Over the past decades, various in vitro and/or ex vivo models have been developed to investigate drug metabolism. In vitro liver cell systems represent a good experimental approach to screen potential hepatotoxic compounds and to investigate mechanisms by which chemicals induce liver lesions (Guillouzo, 1998). Primary human hepatocytes are considered to be the most pertinent model for in vitro testing of the induction/inhibition potential of drug candidates (Guillouzo, 1998; Gómez-Lechón et al., 2004; Hewitt et al., 2007). However, their use is limited by scarce availability of donor organs, large interdonor functional variability, and early phenotypic changes in vitro. Most hepatocyte cell lines, mainly originated from tumors, have indefinite proliferative capacity, but they are considered inappropriate for prediction of hepatotoxicity in preclinical drug development due to the low levels, if any, of major P450 enzymes and several transporters. An exception is represented by the HepaRG cell line derived from a human liver carcinoma (Gripon et al., 2002). When seeded at low density (LD), HepaRG cells transdifferentiate into bipotent hepatic progenitors and actively divide before acquiring typical morphological and functional characteristics of adult human hepatocytes in primary culture under appropriate culture conditions (Cerec et al., 2007). Indeed, after 2 weeks at confluence in the presence of 2% dimethyl sulfoxide (DMSO) and corticosteroids, they appear as hepatocyte-like colonies surrounded by biliary-like cells. Transdifferentiation of differentiated HepaRG cells can be avoided by seeding at high density (HD) (Aninat et al., 2006; Cerec et al., 2007).
Differentiated HepaRG cells express various liver functions, including P450s, phase II enzymes, transporters, and nuclear receptors at levels comparable with those found in primary hepatocytes and are responsive to prototypical inducers, suggesting that they could represent a surrogate to the latter in drug metabolism and toxicity studies (Aninat et al., 2006; Le Vee et al., 2006; Guillouzo et al., 2007; Josse et al., 2008; Kanebratt and Andersson, 2008b; Turpeinen et al., 2009). Moreover, some evidence has been provided that LD-seeded HepaRG cells can retain relatively stable expression and activities of P450s for several weeks at confluence (Josse et al., 2008; Kanebratt and Andersson, 2008a). However, long-term maintenance of P450 activities and responsiveness to inducers has not been fully characterized, especially in HD-seeded cultures. The present study was undertaken within the European Community Framework Programme LIINTOP project to compare transcript and activity levels of the major P450s and their responsiveness to the prototypical inducers, phenobarbital (PB), rifampicin (RIF), and omeprazole (OME), over a 4-week period in differentiated HepaRG cells. For this purpose, P450 activities were simultaneously estimated using two different cocktail substrate settings. Such P450-substrate cocktails are increasingly used for the determination of basal and induced P450 activities (Zhou et al., 2004). Indeed, the use of a variety of substrate combinations, numbers of substrates, or analytical methods has revealed no significant differences between the results obtained from individual or cocktail analyses (Dierks et al., 2001; Kim et al., 2005; Lahoz et al., 2008a). In addition, expression of various other genes related to xenobiotic metabolism and transport was also measured for comparison purposes. We show that basal mRNA and activity levels of major P450s, expression of several phase II enzymes and transporters, as well as inducibility by RIF, PB, and OME were well maintained in differentiated HepaRG cells during the 4-week testing period, whether the cells were seeded at LD or HD.
Materials and Methods
Chemicals.
DMSO, RIF, PB, OME, and insulin were purchased from Sigma-Aldrich (St. Quentin Fallavier, France). Williams' E medium was obtained from Eurobio Laboratories (Les Ulis, France). Fetal calf serum (FCS) was supplied by Perbio (Brebières, France). Penicillin streptomycin carboxydichlorofluorescein di-acetate (CDFDA) and BODIPY FL vinblastine, a fluorescent analog of the anticancer drug vinblastine, were from Invitrogen (Cergy Pontoise, France). Hydrocortisone hemisuccinate was obtained from Upjohn Pharmacia (Guyancourt, France). Substrates and metabolite standards used for the Unidad Mixta Fundación Hospital La Fe-Advancell (UAM) cocktail were as follows: 4′-hydroxydiclofenac, 6-hydroxychlorzoxazone, hydroxybufuralol, midazolam, 1′-hydroxymidazolam, phenacetin, bufuralol, mephenytoin, hydroxymephenytoin, and acetaminophen (supplied by BD Biosciences Discovery Labware, Bedford, MA). Substrates used for the Novamass (NM) cocktail were as follows: bupropion, midazolam, and OME were purchased from Sequoia Research Products (Pangbourne, UK); testosterone was obtained from Fluka (Buchs, Swizerland); and phenacetin was obtained from ICN Biomedicals (Costa Mesa, CA). Metabolite standards O-desmethyldextromethorphan, 6-hydroxychlorzoxazone, desethylamodiaquine, hydroxytolbutamide, and 6β-hydroxytestosterone were purchased from BD Biosciences Discovery Labware. All other chemicals were obtained from Sigma-Aldrich (St. Louis, MO).
Cell Cultures.
HepaRG cells were cultured at a density of either 2.6 × 104 cells/cm2 (LD) or 0.45 × 106 differentiated cells/cm2 (HD) as described previously (Gripon et al., 2002; Aninat et al., 2006). LD-seeded HepaRG cells were first incubated in the Williams' E medium supplemented with 10% FCS, 100 units/ml penicillin, 100 μg/ml streptomycin, 5 μg/ml insulin, 2 mM glutamine, and 5 × 10−5 M hydrocortisone hemisuccinate for 2 weeks. Maximal liver-specific activities were attained after two additional weeks in the same medium added with 2% DMSO (Gripon et al., 2002). HD-seeded HepaRG cells were directly incubated in the DMSO-containing medium; they did not transdifferentiate and retained their morphological and functional characteristics. The medium of both LD and HD cultures was renewed every 2 or 3 days.
Human hepatocytes from five adult donors undergoing resection for primary and secondary tumors were obtained by perfusion of histologically normal liver fragments (Guguen-Guillouzo et al., 1982). In brief, hepatocytes were seeded at a density of 190,000 cells/cm2 in 24-well plates in Williams' E medium supplemented with 10% FCS, 100 units/ml penicillin, 100 μg/ml streptomycin, 1 μg/ml insulin, 2 mM glutamine, and 1 μg/ml bovine serum albumin. Cells were allowed to attach for 12 h, at which time medium was replaced with the same medium deprived of FCS and supplemented with 10−7 M dexamethasone. The medium was renewed every day.
Induction Treatments.
For induction studies in HepaRG cells, experimental conditions were based on the observations that P450 induction by DMSO was lost 24 h after its depletion, and that no marked changes occurred in P450 transcript levels after two additional days (Aninat et al., 2006). HepaRG cells (from passages 14, 17, and 20) were seeded either at LD and cultured for up to 56 days or at HD and cultured for up to 28 days (Fig. 1). Three days before inducer addition, the medium was changed to a DMSO-free medium containing 2% FCS, and after that period the cells were exposed to the prototypical inducers (10 μM RIF, 1 mM PB, or 50 μM OME) or their vehicle, i.e., DMSO (final concentration of 0.5%), for RIF and OME and water for PB, for 48 h (mRNA quantification) or 72 h (measurements of P450 activities). Human hepatocytes were exposed to the prototypical inducers or their vehicle 1 to 3 days after seeding. Treatments also lasted either 48 h (mRNA quantification) or 72 h (measurements of P450 activities).
Isolation of RNA and Reverse Transcriptase-Quantitative Polymerase Chain Reaction Analysis.
Total RNA was extracted from 106 HepaRG cells or 106 human hepatocytes with the SV total RNA isolation system (Promega, Madison, WI), which directly included a DNase treatment step. RNAs were reverse-transcribed into cDNA by using a High-Capacity cDNA Archive kit (Applied Biosystems, Foster City, CA). Real-time quantitative polymerase chain reaction (PCR) was performed by the fluorescent dye SYBR Green methodology using the SYBR Green PCR Master Mix (Applied Biosystems) and the ABI Prism 7000 (Applied Biosystems). Table 1 (supplemental data) shows primer pairs for each transcript chosen with Primer 3 (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi), except for CYP2A6, CYP2C8, CYP2C19, and CYP2D6, which were provided by Biopredic International (Rennes, France) (Girault et al., 2005). The amplification curves were read with the ABI Prism 7000 SDS software using the comparative cycle threshold method. The relative quantification of the steady-state mRNA levels was calculated after normalization of the total amount of cDNA tested using 18S RNA as a reference. Furthermore, a dissociation curve was performed after the PCR analysis to verify the specificity of the amplification. Results were expressed as percentage of mRNA levels measured in freshly isolated hepatocytes (FIH; arbitrarily set at 100%) for basal expression studies, or as percentage of mRNA levels measured in the corresponding control samples exposed to the vehicle (arbitrarily set at 100%) for induction studies.
Evaluation of P450 Activities.
Two substrate cocktails were used: the UAM cocktail and the NM cocktail.
The UAM cocktail.
Activity assays used direct incubation of cell monolayers with a cocktail of eight substrates as described previously (Lahoz et al., 2008b). The substrate mixture stock solutions were prepared in DMSO and then methodically diluted in the incubation medium to obtain the following final concentrations: 10 μM phenacetin (CYP1A2), 5 μM coumarin (CYP2A6), 10 μM bupropion (CYP2B6), 10 μM diclofenac (CYP2C9), 50 μM mephenytoin (CYP2C19), 10 μM bufuralol (CYP2D6), 50 μM chlorzoxazone (CYP2E1), and 5 μM midazolam (CYP3A4). The final concentration of DMSO during incubation was 0.5% (v/v). Incubations with the cocktail of substrates lasted 2 h. Metabolism of all compounds was linear during this incubation period. At the end, aliquots of media (300 μl) were collected and stored at −80°C until delivery and analysis in UAM laboratories. Samples were subsequently extracted twice with ethyl acetate 1:1 (v/v), and dextromethorphan was used as recovery standard. The organic phase was transferred to a clean tube, evaporated under vacuum, dissolved in 100 μl HEPES (5% acetonitrile), and submitted for analysis. Metabolites formed and released into the culture medium were quantified by high-performance liquid chromatography (HPLC) tandem mass spectrometry. This system comprised a Micromass Quattro Micro (Waters, Milford, MA) triple quadrupole mass spectrometer in electrospray ionization mode, interfaced with an Alliance 2795 HPLC (Waters Chromatography). Chromatography was performed at 35°C, and an aliquot (20 μl) was injected into a Teknokroma C18 column (100 mm × 2.1 mm, 3-μm particle size) at a flow rate of 0.4 ml/min. The mobile phase was 0.1% formic acid in acetonitrile (A) and 0.1% formic acid in water (B). The proportion of acetonitrile was increased linearly from 0 to 90% in 6 min, and then the column was allowed to re-equilibrate at the initial conditions for 4 min. The column eluent was directed without splitting to an electrospray ionization interface, operating at 320°C and using nitrogen as cone gas (50 l/h). The tandem mass spectrometry experiments were carried out with a triple quadrupole analyzer operating in multiple reaction monitoring mode. Enzymatic activities were expressed as picomole of metabolites formed per hour per milligram of total protein.
The NM cocktail.
Activity assays were performed by direct incubation of cell monolayers with a cocktail of 10 substrates as described previously (Turpeinen et al., 2005; Tolonen et al., 2007). The substrate mixture stock solutions were prepared in DMSO and diluted in incubation media to obtain the following final concentrations: 5 μM melatonin (CYP1A2), 2 μM coumarin (CYP2A6), 2 μM bupropion (CYP2B6), 5 μM amodiaquine (CYP2C8), 8 μM tolbutamide (CYP2C9), 5 μM OME (CYP2C19), 1 μM dextromethorphan (CYP2D6), 10 μM chlorzoxazone (CYP2E1), 1 μM midazolam (CYP3A4), and 5 μM testosterone (CYP3A4). The final concentration of DMSO during incubation was 0.5% (v/v). After a 4-h incubation with the cocktail of substrates, aliquots (300 μl) were collected and stored at −80°C until delivery and analysis. Frozen samples were thawed at room temperature and centrifuged for 10 min at 13,200g, then 20 μl of the supernatant was injected into the liquid chromatography tandem mass spectrometry. A Waters 2695 Alliance HPLC system (Waters) was used together with a Waters Sunfire RP18 column (2.1 × 100 mm column with 5-μm particle size) and a Luna C18 precolumn (2.0 × 4.0 mm) (Phenomenex, Torrance, CA) at 30°C. The HPLC eluents were aqueous 1% formic acid + 10 mM ammonium acetate, pH 2.4 (A) and methanol (B). The gradient elution with 5 to 50 to 80% B was applied in 0 to 1.0 to 4.0 min, followed by 0.5-min isocratic elution with 80% B and column equilibration, resulting in a total time of 8 min/injection. The eluent flow rate was 0.5 ml/min. The flow was split postcolumn with an Accurate Post-Column Stream Splitter (LC Packings, Amsterdam, The Netherlands) with the mass spectrometry ion source-to-waste ratio of 1:3. Data were acquired using a Micromass Quattro Micro triple quadrupole mass spectrometer (Altrincham, Cheshire, UK), equipped with a Z-spray electrospray ion source. Multiple reaction monitoring mode using a polarity switching between positive and negative ion mode was applied. The capillary voltage used was 4200 V and extraction cone voltage 2 V for all compounds. The dissolution temperature and source temperature were 280°C and 150°C, respectively. Nitrogen was used as a drying gas with a flow rate of 700 l/h and as a nebulizer gas with a full flow rate. The collision cell argon pressure was set to 3.8 × 10−3 mbar. The dwell times for monitoring each reaction were 100 msec, and the delay between positive and negative polarities during hydroxychlorzoxazone detection (2.7–3.8 min) was 300 msec. Enzyme activities were expressed as picomole of metabolites formed per hour per milligram of total protein.
Efflux Transport Assays.
Analysis of efflux transport was performed by using two fluorescent substrates: CDFDA, a substrate of multidrug resistance-associated protein (MRP) 2 and BODIPY FL vinblastine, a substrate of bile salt export pump (BSEP) and multidrug resistance protein (MDR) 1. After washing with the uptake buffer (136 mM NaCl, 5.3 mM KCl, 1.1 mM KH2PO4, 0.8 mM MgSO4, 1.8 mM CaCl2, 11 mM d-glucose, 10 mM HEPES, pH7.4), HepaRG cells were incubated for 30 min at 37°C with fluorescent substrates (3 μM CDFDA or 2 μM vinblastine) then washed with chilled phosphate-buffered saline and observed under fluorescence microscopy.
Statistical Analysis.
Each value corresponded to the mean ± S.D. of three independent experiments. The Kruskal-Wallis nonparametric test was used to compare mRNA levels and P450 activities between different time points. The Mann-Whitney U test was applied to compare mRNA levels and P450 activities between inducer-treated cultures and corresponding control samples. Data were considered significantly different when p < 0.05.
Results
Basal mRNA Levels of Phase I and II Metabolizing Enzymes, Membrane Transporters, and Nuclear Receptors.
Basal mRNA levels of 20 genes were analyzed by reverse transcriptase-quantitative PCR (RT-qPCR) in FIH and 1- and 3-day-cultured primary human hepatocytes and LD- and HD-seeded HepaRG cells at different times for up to 4 weeks after differentiation. These genes were as follows: 10 P450s (CYP1A1, CYP1A2, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1, and CYP3A4), two phase II enzymes [glutathione transferase A1/A2 (GSTA1/A2) and UDP-glucuronosyl transferase 1A1 (UGT1A1)], five membrane transporters [breast cancer resistance protein (BCRP), BSEP, MDR1, MRP2, and Na+-dependent taurocholic cotransporting polypeptide (NTCP)], and three nuclear receptors [aryl hydrocarbon receptor (AhR), constitutive androstane receptor (CAR), and pregnane X receptor (PXR)]. All expression values were compared with those measured in 1-day human hepatocyte cultures set at 100%. Differential changes were observed in basal gene expression in primary human hepatocytes (Fig. 2) as a function of time in culture; they could be classified into five groups characterized by either a decrease followed by an increase at day 3 (CYP2C8, CYP2C9, CYP3A4, UGT1A1, CAR, NTCP), a continuous and strong decrease (CYP1A2, CYPB6, CYP2E1, BSEP), a decrease followed by a relative stability (CYP1A1, CYP2C19, GSTA1/2, BCRP), a continuous increase (MDR1, MRP2), or a relative stability (CYP2A6, CYP2D6, AhR, PXR). Compared with 1-day primary hepatocyte cultures, all the genes tested were expressed in DMSO-exposed HepaRG cultures; the levels of transcripts were either markedly lower (CYP1A2, CYP2C8, CYP2D6, CYP2E1, BSEP, NTCP), at the 30 to 50% level (CYP1A1, CYP2A6, CYP2B6, PXR), or comparable or even higher (CYP2C9, CYP2C19, CYP3A4, GSTA1/2, UGT1A1, MDR1, MRP2, AhR, CAR) (Fig. 3, A and C). However, when HepaRG cells were shifted to a DMSO-free medium for 72 h before mRNA quantification, some major differences were found (Fig. 3, B and D). Transcripts of CYP1A1, CYP2A6, CYP2B6, CYP3A4, and UGT1A1 were decreased, whereas those of CYP1A2, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1, GSTA1-A2, AhR, CAR, and transporters were only slightly altered. Moreover, most genes were much less expressed in both proliferative and poorly differentiated HepaRG cells cultured for 5 or 15 days, respectively, in the absence of DMSO after LD seeding; the only exception was CYP1A1, which was highly expressed (Fig. 3B). All data showed a long-term, relatively stable gene expression of all tested drug-metabolizing enzymes and transporters in differentiated HepaRG cells from both LD- and HD-seeded cultures maintained at confluence for up to 4 weeks.
Basal Activities of Various P450s.
The UAM and NM cocktails were used for the determination of P450 activities in HepaRG cells and primary human hepatocytes. In addition to the use of several different substrates, the major differences between the two assay procedures were the use of lower substrate concentrations and longer incubation times in the NM assay. Concentrations of the four common substrates were 5-fold (bupropion, chlorzoxazone, midazolam) and 2.5-fold (coumarin) higher in the UAM (Fig. 4) than in the NM (Fig. 5) assay. Lower values were frequently obtained with the NM cocktail. Compared with the values measured in freshly isolated hepatocyte suspensions, only CYP2A6, CYP2C9, and CYP2E1 activities were markedly decreased after 24 h of culture using the UAM probe cocktail (Fig. 4A). Our results showed important interdonor variability between human hepatocyte populations for some activities, especially for CYP2C19 (from 0.98 to 10.04 pmol/h/mg protein; n = 5) and CYP2B6 (from 1.04 to 31.62 pmol/h/mg protein; n = 5).
In DMSO-exposed HepaRG cultures (Fig. 4B), P450 activities were found to be either lower (CYP1A2, CYP2A6, CYP2C9, CYP2E1) or relatively close (CYP2B6, CYP2C19, CYP2D6, CYP3A4) to those of 1-day primary hepatocyte cultures. In HepaRG cells maintained in DMSO-free medium (Figs. 4C and 5B), most P450 activities were lower, especially CYP2B6 and CYP3A4. As expected, P450 activities were all quite low in 5- and 15-day LD-seeded cultures, corresponding to proliferating and poorly differentiated HepaRG cells, respectively.
All activities were relatively well maintained over the 28-day period of confluence. Using the NM cocktail, P450 activity values were either comparable with (CYP1A2, CYP2B6, CYP2C19, CYP2D6) or lower than (CYP2C9, CYP3A4) those estimated with the UAM one. Furthermore, low CYP2A6 and CYP2E1 activities were found with the UAM cocktail, whereas they were undetectable with the NM cocktail. However, other P450 activities were found to be relatively well maintained by using the NM cocktail, and CYP2C8 activity was demonstrated (amodiaquine deethylation).
Functional Activities of Membrane Transporters.
To verify whether canalicular efflux transporters remained functional in hepatocyte-like HepaRG cells after several weeks at confluence, transport of specific fluorescent substrates was analyzed on day 56 after LD seeding. First, no morphological modification of bile canalicular structures was observed when HepaRG cells were cultured in DMSO-free medium for 5 days (Fig. 6A). Moreover, accumulation of CDFDA, a fluorescent substrate of MRP2, was restricted to bile canaliculi of HepaRG hepatocytes (Fig. 6B). In addition, specific accumulation of BODIPY FL vinblastine, a fluorescent substrate of BSEP/MDR1, was also evidenced in canalicular spaces of HepaRG cells. Similar observations were made in HD-seeded HepaRG cell cultures (data not shown).
Induction of P450s by Prototypical Inducers at mRNA and Activity Levels.
To further estimate the functional capacity of HepaRG cells maintained at confluence for several weeks, their responsiveness to three prototypical inducers (RIF, PB, and OME) was assessed over a 4-week period. At appropriate time points, HepaRG cultures were shifted to a DMSO-free medium supplemented with 2% FCS for 72 h and then treated with the prototypical inducers for 48 and 72 h for mRNA (Fig. 7) and activity (Fig. 8) measurements, respectively. Comparable results were obtained in LD- and HD-seeded HepaRG cultures, and the levels of P450 induction were relatively consistent for 4 weeks in differentiated HepaRG cells in both culture conditions (nonsignificant variations, Kruskal-Wallis test).
The main results are summarized in Table 1. As expected CYP1A2, CYP2B6, and CYP3A4 transcripts and activities were strongly increased by their prototypical inducers. Thus, CYP1A2 mRNA and activity were induced by OME (Figs. 7C and 8C), and CYP2B6 and CYP3A4 mRNA and corresponding activities were induced by both PB and RIF (Figs. 7, A and B, and 8, A and B). In addition, various other P450s were modulated by the three inducers (Supplemental Tables 2, 3, and 4). CYP1A1 and CYP1A2 transcripts were augmented in both HepaRG cells and human hepatocytes by OME. CYP1A1 expression was strongly increased by OME (450–850-fold) and also weakly increased by PB (3–6-fold) in HepaRG cells. CYP1A2 activity was induced by the three compounds in HepaRG cells. CYP1A1 activity was not specifically measured by the probe substrates. Expression and activity of CYP2B6 and CYP3A4 were also induced by the three compounds in both LD- and HD-seeded HepaRG cells. However, induction of CYP3A4 was demonstrated only by 3-hydroxylation of omeprazole (NM) and 1-hydroxylation of midazolam (UAM). No induction of 1-hydroxylation of midazolam or 6β-hydroxylation of testosterone was evidenced with the NM cocktail.
Expression of CYP2A6 was induced by PB in HepaRG cells, whereas it was increased by the three prototypical inducers in human hepatocytes. On the other hand, CYP2E1 expression was decreased by all three inducers in both human hepatocytes and HepaRG cells. However, CYP2A6 and CYP2E1 activities were not detected with the cocktails in DMSO-free cultured HepaRG cells whatever the inducer.
RIF and OME decreased CYP2D6 expression in HepaRG cells but not in primary hepatocyte cultures. CYP2D6 activity was decreased by the three inducers using the NM cocktail (with dextromethorphan as a substrate), but this activity was increased by RIF and PB in both LD- and HD-seeded HepaRG cells by using the UAM cocktail (with bufuralol as a substrate). However, these variations of activities were low compared with the -fold inductions of other P450s.
Induction of CYP2C (CYP2C8, CYP2C9, and CYP2C19) expression by PB and more weakly by RIF was observed in HepaRG cells, whereas only RIF induced CYP2C8 and CYP2C19 expression in human hepatocytes. CYP2C9 activity was weakly increased by OME as well as CYP2C8 by RIF and PB in HepaRG cells using the NM cocktail. Conflicting results were found for CYP2C19 activity with the two cocktails: a strong decrease of activity by OME and an increase by RIF were observed with the UAM cocktail (mephenytoin as a substrate), whereas this activity was strongly induced by PB and RIF and more weakly by OME with the NM cocktail (5-hydroxylation of omeprazole as the model reaction).
It is noteworthy that induction rates were much lower in human hepatocytes than in HepaRG cells (Fig. 7). CYP1A1, CYP2C8, CYP2C9, CYP2C19, and CYP3A4 were found to be induced by PB in HepaRG cells but not in human hepatocytes (Supplemental Table 2).
Induction of Phase II Enzymes, Transporters, and Nuclear Receptors Expression.
Expression of phase II enzymes, transporters, and nuclear receptors was investigated in both HepaRG cells and primary human hepatocytes in response to prototypical inducers (Supplemental Table 5). Overall effects were relatively well maintained in differentiated HepaRG cells whether they were seeded at LD or HD over the 4-week period. GSTA1/2 and UGT1A1 expression was induced by the three inducers in LD- and HD-seeded HepaRG cells and only by RIF and OME in primary human hepatocytes. Likewise, expression of efflux transporters MDR1 and MRP2 was increased by the three inducers in HepaRG cell cultures and only by RIF and OME in primary hepatocyte cultures. An induction of BCRP was observed after PB and OME exposure in HepaRG cells and only after OME treatment in human hepatocytes. In addition, an inhibition of the uptake-transporter NTCP expression by OME and a decreased expression of the efflux transporter BSEP by RIF were evidenced in both differentiated HepaRG cells and human hepatocytes. Some responses of nuclear receptors to inducers were also observed. A RIF-dependent decrease and a PB-dependent induction of CAR expression were observed in HepaRG cells.
Discussion
Because human hepatocytes are a precious, limited resource and show an extensive interdonor variability in response to inducers, alternatives are required for use in drug discovery. Previous studies have shown that human hepatoma HepaRG cells express most of the major drug-metabolizing enzymes when they are differentiated in vitro. The present results confirm and extend these observations; they show that expression and activities of the major P450s as well as their responsiveness to prototypical inducers were well maintained over a 4-week period in differentiated HepaRG cells obtained from either LD- or HD-seeded cultures. Only random and limited differences were noticed between these two seeding conditions. Contrary to LD seeding that requires 1 month before HepaRG cells can be used as fully differentiated (Cerec et al., 2007), HD seeding prevents transdifferentiation (Aninat et al., 2006; Cerec et al., 2007) and is associated with only transient and limited decrease in certain liver-specific functions. Consequently, HD-seeded cultures are of potentially great value for high volume, convenient use of the HepaRG cell line.
An extensive analysis of P450 activities was performed using substrate cocktails. This approach requires fewer sample numbers because the same culture is incubated with all the substrates. Most results obtained with the UAM and NM cocktails used in the present study were found to be comparable. However, some differences were observed, which could be due to differences in the experimental conditions and/or the choice of probe substrates. Indeed, UAM and NM cocktails were composed of 8 and 10 substrates, respectively (Turpeinen et al., 2005; Lahoz et al., 2008b), and only 4 substrates were in common: bupropion for CYP2B6, coumarin for CYP2A6, chlorzoxazone for CYP2E1, and midazolam for CYP3A4. P450 activities were higher with the UAM cocktail, and detectable CYP2A6 and CYP2E1 activities were demonstrated only in the presence of DMSO with this cocktail. This effect could be explained by the fact that substrate concentrations in the UAM cocktail are close to or higher than their Km values, whereas lower substrate concentrations are used in the NM cocktail that has been primarily designed for drug-drug interactions studies. No evidence of any kind of interaction or interference between the substrates has been observed in the latter (Turpeinen et al., 2005). Because testosterone is rapidly metabolized in HepaRG cells (Turpeinen et al., 2009), it is likely that the absence of detectable 6ß-OH-testosterone metabolites with the NM cocktail was due to use of an inadequate concentration of testosterone and the conjugation of all the metabolites formed during the incubation period. Two major differences were observed after OME exposure: CYP1A2 induction was much higher using the UAM cocktail and CYP2C19 activity was increased using the NM cocktail and decreased using the UAM cocktail. In both cases, such effects could be attributed to the choice of different substrates. In the case of CYP1A2, phenacetin and melatonin were used in UAM and NM assays, respectively. Melatonin has been proven to be a less selective substrate (FDA-Guide, 2001) than phenacetin for CYP1A2 induction studies. Regarding CYP2C19, UAM and NM cocktails contained mephenytoin and omeprazole as a substrate, respectively. CYP2C19 inhibition has been described to occur when OME was used at high concentrations (Ko et al., 1997). Because OME is a substrate for CYP2C19, its potential intracellular accumulation during induction studies could inhibit mephenytoin metabolism by CYP2C19.
In support of previous observations (Aninat et al., 2006) using the two substrate cocktails, we showed that several P450s that were strongly increased in the presence of DMSO recovered their basal levels after DMSO withdrawal. Activities of CYP3A4 and CYP2B6 were the most sensitive to DMSO, and those of CYP2A6 and CYP2E1 became undetectable in cells cultured for 3 days in a DMSO-free medium, indicating that the DMSO-free conditions were not suitable for their expression. The cause of the loss of CYP2E1 activity was unclear. Its inhibition by insulin has been reported in a rat hepatoma cell line (Moncion et al., 2002). However, medium depletion or lower concentrations of insulin did not prevent CYP2E1 activity loss in HepaRG cells (data not shown).
A good correlation was observed between transcripts and corresponding activities. However, -fold inductions of mRNA levels were much higher, especially for CYP1A2 (800-fold for transcripts versus 30-fold for activity with the UAM cocktail), CYP2B6 (50-fold for transcripts versus 30-fold for activity with the UAM cocktail), and CYP3A4 (80-fold for transcripts versus 10-fold for activity with the UAM cocktail). As previously emphasized (Aninat et al., 2006), maintenance of HepaRG cells in conditions in which basal P450 activities were low resulted in strong increased levels after treatment with prototypical inducers. -Fold inductions of P450 transcripts and activities were much higher than in primary hepatocytes, especially for CYP3A4, demonstrating that HepaRG cells were still highly responsive when treatments with inducers were started 3 days after DMSO withdrawal.
Whereas basal expression and activity of P450s and responsiveness to prototypical inducers are variable in human hepatocyte populations (Morel et al., 1990; Madan et al., 2003), they were always found at comparable levels in differentiated HepaRG cells when comparing different passages. Several genes, including P450s, which have been found to be induced in vitro in only a fraction of human hepatocyte populations (Goyak et al., 2008), were responsive in differentiated HepaRG cells. Accordingly, important interdonor variation was found for basal CYP2B6 and CYP2C19 activities in primary human hepatocytes, and expression of CYP1A1, CYP2C8, CYP2C9, CYP2C19, and CYP3A4 was induced by PB in only HepaRG cells (Table 1). More recently, induction of CYP1A1/2 activities in HepaRG cells, but not in primary human hepatocytes, was also reported after PB treatment (Turpeinen et al., 2009). A major difference between HepaRG cells and primary hepatocytes was the continuous expression of CYP1A1 in the former; this activity could be related to their transformed state rather than to the presence of biliary cells. RIF is another well established inducer, and our data (Table 1) reproduced well the previous findings in primary hepatocytes (Morel et al., 1990; Lahoz et al., 2008b) and HepaRG cells (Aninat et al., 2006; Kanebratt and Andersson, 2008b). Therefore, HepaRG cells appear to be representative of a large fraction of human hepatocyte populations.
Phase II enzymes and plasma transporters are also targets of microsomal enzyme inducers. However only limited studies still exist on the regulation of UGTs in human hepatocytes (Soars et al., 2004; Nishimura et al., 2008). We show that in addition to the major P450s, several phase II enzymes and transporters were modulated by PB, RIF, and/or OME. The three inducers increased both UGT1A1 and GSTA1/A2 transcripts in differentiated HepaRG cells; this result is in agreement with previous observations in primary hepatocytes. It has been reported that PB is an inducer of GSTA1/2 (Morel et al., 1993) and that OME and RIF are inducers of UGT1A1 (Nishimura et al., 2008). Note that the -fold inductions of these two genes with the three inducers were much lower than those observed with P450s, ranging mostly between 1.5- and 5-fold, with the exception of UGT1A1 induction by PB and OME that reached 7 to 10-fold (Table 1). The low mean increase of UGT1A1 (1.5-fold) and the absence of modulation of GSTA1/A2 in 3-day hepatocytes exposed to PB in the present study can be explained by interdonor variability in response to this compound (Morel et al., 1993; Soars et al., 2004).
Various transporters were also analyzed at the transcriptional level and found to be modulated in HepaRG cells (Table 1). MDR1 and MRP2 were increased by the three inducers, and BCRP was increased by OME and PB. NTCP was inhibited mainly by OME, and BSEP was inhibited by RIF and slightly by OME. Our results fully agree with previous reports showing induction of MDR1 and MRP2 by PB in HepaRG cells (Le Vee et al., 2006; Lambert et al., 2009). Our results replicate data previously obtained with human hepatocytes, demonstrating consistent and wide interdonor variability in response to identical inducers (Jigorel et al., 2006).
In summary, although HepaRG cells are originated from only one donor and are the product of in vivo transformation, they exhibit a drug metabolism capacity, including responsiveness to chemical modulators, which largely reflects that observed with the majority of human hepatocyte populations in primary culture and, in addition, offer several unique advantages including the following: 1) the data are reproducible during several passages; 2) the functional activities are well maintained for several weeks at confluence; and 3) the levels of activities can be modulated by selecting appropriate culture conditions, especially the composition of the culture medium. Therefore, we conclude that HepaRG cells not only represent a promising surrogate to primary human hepatocytes for investigating induction of xenobiotic-metabolizing enzymes and transporters and drug-drug interactions, but they also represent a unique metabolically competent cell model for in vitro chronic toxicity studies.
Acknowledgments.
We thank Dr. David Steen for critical reading of the manuscript.
Footnotes
This work was supported by the European Community [Contracts LIINTOP-STREP-037499 and Predict-IV-202222 (to A.G.)]; and the Ministerio Ciencia e Inovación/Instituto de Salud Carlos III for a Miguel Server [Contract CP08/00125] (to A.L.).
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
doi:10.1124/dmd.109.030197.
↵ The online version of this article (available at http://dmd.aspetjournals.org) contains supplemental material.
-
- P450
- cytochrome P450
- LD
- low density
- DMSO
- dimethyl sulfoxide
- HD
- high density
- PB
- phenobarbital
- RIF
- rifampicin
- OME
- omeprazole
- FCS
- fetal calf serum
- CDFDA
- carboxydichlorofluorescein di-acetate
- UAM
- Unidad Mixta Fundación Hospital La Fe-Advancell
- NM
- Novamass
- PCR
- polymerase chain reaction
- FIH
- freshly isolated hepatocytes
- HPLC
- high-performance liquid chromatography
- MRP
- multidrug resistance-associated protein
- BSEP
- bile salt export pump
- MDR
- multidrug resistance protein
- RT-qPCR
- reverse transcriptase-quantitative PCR
- GSTA1/A2
- glutathione transferase A1/A2
- UGT1A1
- UDP-glucuronosyl transferase 1A1
- BCRP
- breast cancer resistance protein
- NTCP
- Na+-dependent taurocholic cotransporting polypeptide
- AhR
- aryl hydrocarbon receptor
- CAR
- constitutive androstane receptor
- PXR
- pregnane X receptor.
- Received September 12, 2009.
- Accepted December 17, 2009.
- Copyright © 2010 by The American Society for Pharmacology and Experimental Therapeutics