Abstract
Growth factors have key roles in liver physiology and pathology, particularly by promoting cell proliferation and growth. Recently, it has been shown that in mouse hepatocytes, epidermal growth factor receptor (EGFR) plays a crucial role in the activation of the xenosensor constitutive androstane receptor (CAR) by the antiepileptic drug phenobarbital. Due to the species selectivity of CAR signaling, here we investigated epidermal growth factor (EGF) role in CAR signaling in primary human hepatocytes. Primary human hepatocytes were incubated with CITCO, a human CAR agonist, or with phenobarbital, an indirect CAR activator, in the presence or absence of EGF. CAR-dependent gene expression modulation and PXR involvement in these responses were assessed upon siRNA-based silencing of the genes that encode CAR and PXR. EGF significantly reduced CAR expression and prevented gene induction by CITCO and, to a lower extent, by phenobarbital. In the absence of EGF, phenobarbital and CITCO modulated the expression of 144 and 111 genes, respectively, in primary human hepatocytes. Among these genes, only 15 were regulated by CITCO and one by phenobarbital in a CAR-dependent manner. Conversely, in the presence of EGF, CITCO and phenobarbital modulated gene expression only in a CAR-independent and PXR-dependent manner. Overall, our findings suggest that in primary human hepatocytes, EGF suppresses specifically CAR signaling mainly through transcriptional regulation and drives the xenobiotic response toward a pregnane X receptor (PXR)-mediated mechanism.
Introduction
Liver is a metabolically active organ that biotransforms a myriad of exogenous and endogenous compounds essentially through the metabolic functions of hepatocytes. These activities are finely regulated by a coordinated network of transcription factors. Among them, two nuclear receptors are known to control the expression of key detoxication enzymes: pregnane X receptor (NR1I2) and constitutive androstane receptor (CAR, NR1I3) (Wang et al., 2012). They may also participate in cancer development and drug resistance (De Mattia et al., 2016). Moreover, increasing evidence indicates that they modulate numerous endogenous signals and crosstalk with signaling pathways involved in the regulation of the homeostasis of bile acids, lipids, hormones, glucose, inflammation, vitamins, and other endobiotics (Wagner et al., 2005; Pascussi et al., 2008; Wada et al., 2009). PXR activation may worsen steatosis, obesity, and insulin resistance (Zhou et al., 2006, 2008; Li et al., 2012). Conversely, CAR activation is often described as beneficial, because it allows maintaining the homeostasis of cholesterol and bile acids and reducing blood glucose and steatosis. However, CAR effect on lipid metabolism was not reproduced in human primary hepatocytes (Lynch et al., 2014).
Considering the central role of CAR and PXR in transforming xenobiotics and regulating cell metabolism, much research has focused on the identification and development of molecules to modulate their activities. However, the study of PXR and CAR function is hindered by several factors. Indeed, the crosstalk between CAR and PXR through recognition of common responsive elements triggers the reciprocal activation of overlapping target genes that coordinate the hepatic response (Xie et al., 2000). For instance, CYP2B6 is a target of human CAR, but can be also efficiently induced by rifampicin (RIF), a specific PXR activator (Goodwin et al., 2001). Similarly CAR can regulate CYP3A4, the prototypical PXR target gene (Maglich et al., 2002). Moreover, CAR and PXR operate as heterodimers with retinoic X receptor (RXR, NR2B1) and share coactivators, such as SRC-1 and PGC-1α, and corepressors, such as SMRT and NCoR (Oladimeji et al., 2016). It is therefore challenging to identify human CAR-specific functions and target genes.
Species-specific direct CAR agonists have been found, such as 6-(4-chlorophenyl)imidazo[2,1-b][1,3]thiazole-5-carbaldehyde O-(3,4-dichlorobenzyl) oxime (CITCO) (Maglich et al., 2003) for human CAR and 1,4-bis-[2-(3,5-dichloropyridyloxy)]benzene, 3,3′,5,5′-tetrachloro-1,4-bis(pyridyloxy) benzene (TCBOBOP) (Tzameli et al., 2000) for mouse CAR. Conversely, phenobarbital (PB) is an indirect CAR activator that does not directly interact with the CAR ligand binding domain and thus does not show species specificity. Pregnenolone 16α-carbonitrile specifically activates mouse PXR, whereas the antibiotic RIF activates human, but not mouse, PXR. Human PXR can also be activated by PB via a receptor binding mechanism, in contrast to the receptor binding-independent mechanism involved in PB-mediated CAR activation (Moore et al., 2000). CAR activation by PB is the key molecular initiating event and promotes hepatocarcinogenesis in rodents. No carcinogenic effect of PB has been reported in humans (Elcombe et al., 2014).
Accumulating evidences indicate that xenobiotic metabolism is repressed during hepatic regeneration and cancer. This stimulated investigations on the involvement of growth factors and cytokines in the downregulation of CYP enzymes because of their important role in liver regeneration and diseases (Berasain and Avila, 2014). For instance, EGF represses CYP2B gene induction by PB in mouse and rat hepatocytes (Kawamura et al., 1999; Koike et al., 2007). Moreover, EGF-mediated CYP2B6 inhibition in human hepatocytes is accompanied by a strong decrease of CAR mRNA expression (Bachleda et al., 2009). The recent finding of a crosstalk between CAR and EGFR (Mutoh et al., 2013) further highlights the connection between cell proliferation/survival and liver detoxification/metabolism. Early and recent studies demonstrated that PB activates CAR transcriptional activity by preventing EGFR phosphorylation (Mutoh et al., 2009; Negishi, 2017). Very recently, it was found that EGF stimulates CAR homodimerization, thus forcing CAR in its inactive form (Shizu et al., 2017). Indeed, CAR activation by CAR ligands (CITCO) and indirect activators (PB) is regulated through homodimer-monomer conversion (Shizu et al., 2017). Moreover, PB competes with EGF for binding to EGFR in human cell lines and mouse hepatocytes (Mutoh et al., 2013).
To date EGF impact on CAR signaling has not been studied in primary human hepatocytes (PHHs). This model presents advantages over human liver cell lines, such as the physiological expression of transcription factors and coregulatory molecules and human CAR localization in the cytoplasm. We investigated EGF effect on CAR transcriptional activity in PHHs, using the human CAR agonist CITCO and the indirect human CAR activator PB. We also downregulated CAR with specific siRNAs to identify specific targets/responses. We found that EGF reduced transcription of CAR target genes, as expected, and also decreased CAR, but not PXR expression. EGF greatly influenced PHH response to direct (CITCO) or indirect (PB) activators and prevented CAR-dependent gene expression. Finally, we demonstrated that in the presence of EGF, CAR target genes are not regulated by PB via CAR, but via PXR.
Materials and Methods
Primary Human Hepatocyte Isolation and Culture.
PHHs were provided by Biopredic International (Rennes, France), or isolated, as described previously (Pichard et al., 2006), from donor organs unsuitable for transplantation or from liver resections performed in adult patients for medical reasons unrelated to our research program. Liver samples were obtained from the Biologic Resource Center of Montpellier University Hospital (CRB-CHUM; http://www.chu-montpellier.fr; Biobank ID: BB-0033-00031), and this study benefitted from the expertise of Dr Jeanne Ramos (hepatogastroenterology sample collection) and Prof Sylvain Lehmann (CRB-CHUM manager). The patients’ clinical characteristics are presented in Table 1. The procedure was approved by the French Ethics Committee and written or oral consent was obtained from the patients or their families.
Liver donor data
PHHs were seeded at confluency (4.105 cells/well) in collagen-coated 24-well dishes and cultured in 5% CO2 humidified atmosphere at 37°C in hepatocyte growth medium (WME medium supplemented with 5 µg/ml insulin, 0.1 µM hydrocortisone, 10 µg/ml transferrin, 250 µg/ml ascorbic acid, 3.75 mg/ml fatty acid-free bovine serum albumin, 2 mM glutamine, penicillin and streptomycin). When indicated, hepatocyte growth medium was supplemented with 10 ng/ml EGF (Peprotech, Neuilly-Sur-Seine, France) (Ismail et al., 1991; Katsura et al., 2002).
siRNA Transfection.
Adherent PHHs were transfected with 10 nM scrambled siRNA (siCT) or ON-TARGET plus SMARTpool siRNAs (pool of four siRNAs) specific for human NR1I3 (the gene encoding CAR) or human NR1I2 (the gene encoding PXR) (Dharmacon, Lafayette, CO) using Lipofectamine RNAiMAX (Life Technologies, Carlsbad, CA) at days 2 and 5 after seeding. At day 5 postseeding, PHHs were incubated with 0.5 mM PB, 1 µM CITCO, 10 µM RIF (Sigma, St. Louis, MO), or vehicle (dimethylsulfoxide) for 24 hours.
RNA Isolation and Quantitative PCR.
RNA was extracted using the TRIzol reagent (Invitrogen/Life Technologies), and then 300 ng of total RNA was reverse transcribed (RT) using the high-capacity cDNA reverse transcription kit (Applied Biosystems, Foster City, CA). Quantitative PCR (qPCR) was performed using the Roche SYBER Green reagent and a LightCycler 480 apparatus (Roche Diagnostic, Meylan, France) with the following program: one step at 95°C for 10 minutes and then 50 cycles of denaturation at 95°C for 10 seconds, annealing at 65°C for 15 seconds, and elongation at 72°C for 15 seconds. The primer sequences are listed in Table 2.
Primer sequences
Microarray Data Acquisition.
Total RNA (500 ng) was amplified and labeled for hybridization on GeneChip HT HG-U133+ PM Array plates. Data were acquired following the manufacturer’s instructions (Affymetrix, High Wycombe, UK). For each condition, five biologic replicates were used. To obtain the intensity value signal, scanned plates were processed using the Affymetrix GCOS 1.4 software, and the Affymetrix CEL files were normalized using the Robust Multichip Average (RMA) implementation of the algorithm. The raw microarray data in .CEL files are available in the Gene Expression Omnibus with accession number GSE68493 at http://www.ncbi.nlm.nih.gov/geo/.
Microarray Data Analysis.
Microarray data were analyzed using the web tool Genomicscape (genomicscape.com) (Kassambara et al., 2015). Briefly, paired significance analysis of microarrays was performed using the 15,000 probe sets with the highest S.D. Significance was calculated using the Wilcoxon test and the list of significantly deregulated genes (fold change, FC, >1.3; false discovery rate <5%) was extracted. Venn diagrams were generated using the Venny tool (http://bioinfogp.cnb.csic.es/tools/venny/index.html). Heat maps of the identified genes of interest were produced using Cluster 3.0 and Treeview. Finally, pathway enrichment was assessed using KEGG [Kyoto Encyclopedia for Genes and Genomes (KEGG)] and the STRING web tool (http://string-db.org).
Protein Analysis.
Total proteins were extracted from PHH cultures using RIPA lysis buffer supplemented with a protease inhibitor cocktail (Santa Cruz Biotechnology, Santa Cruz, CA). Protein concentration was determined using the bicinchoninic acid method, according to the manufacturer’s instructions (Pierce Chemical, Rockford, IL). Proteins (20 μg/sample) were resolved on 10% Precast SDS-polyacrylamide gels (Bio-Rad Laboratories, Marnes la Coquette, France) and transferred to polyvinylidene fluoride membranes (Bio-Rad Laboratories). Membranes were incubated with antibodies against CYP2B6 and GAPDH (sc-67224 and sc-32233; Santa Cruz Biotechnology). Immunocomplexes were detected with a horseradish peroxidase-conjugated mouse secondary antibody (Sigma) followed by enhanced chemiluminescence (Millipore, Molsheim, France). Signals were acquired using a ChemiDoc-XRS+ apparatus (Bio-Rad Laboratories) and quantified with the Image Laboratory software (version 4.1)
DigiWest.
The NuPAGE SDS-PAGE gel system (Life Technologies) was used for protein separation and blotting. Proteins (18 μg per sample) were separated using 4%–12% Bis-Tris gels according to the manufacturer’s instructions. Blotting onto PVDF membranes (Millipore) was performed in standard conditions. For high content Western blot analysis, the DigiWest procedure was performed as described (Treindl et al., 2016).
Statistical Analysis.
Quantitative PCR results were expressed as the mean ± S.E.M. Results were compared using the paired t test. Differences were considered statistically significant when P < 0.05.
Results
EGF Strongly Affects Gene Expression in PHHs.
First, the EGFR pathway activity in PHHs was monitored after long-term incubation (from day 0 to day 5) with 10 ng/ml EGF. Phosphorylation of ERK1/2 (at T202 and Y204), of MEK1/2 (at S217 and S221) and of the immediate downstream kinase RSK1 (p90RSK T573) demonstrated the sustained activation of the EGFR pathway upon incubation with the ligand (Fig. 1A).
Differential gene expression in PHHs following incubation with EGF. (A) DIGIWest analysis of protein expression in PHHs (Liv10, Table 1) following incubation or not with 10 ng/ml EGF for 5 days. (B) Number of upregulated genes and enriched pathways in EGF-treated PHHs [false discovery rate (FDR) = 4.7 × 10−18–1.9 × 10−5]. (C) Number of downregulated genes and enriched pathways in EGF-treated PHHs (FDR = 8.8 × 10−10–0.008). (D) Heat map showing the genes differentially expressed in PHHs (n = 5) following incubation with EGF.
Then, a global transcriptome analysis was performed in PHHs incubated or not with EGF for the same period (day 0 to day 5). Compared with control PHHs (without EGF), 328 genes (187 upregulated and 141 downregulated) were deregulated in EGF-treated cultures. KEGG pathway enrichment analysis revealed the contribution of the differentially expressed genes to specific biologic processes. Genes upregulated following EGF incubation were mostly classified in 14 KEGG pathways mainly associated with cell proliferation and cancer (Komposch and Sibilia, 2015) (Fig. 1B). Conversely, genes downregulated upon incubation with EGF belonged mostly to KEGG pathways related to metabolism. “Drug metabolism cytochrome P450” and “Metabolism of xenobiotics by cytochrome P450” were among the five most deregulated signaling cascades (Fig. 1C). Unsupervised hierarchical clustering of the most significantly deregulated genes (35 up- and 35 downregulated genes) clearly showed that despite the interindividual variability, PHH samples segregated according to whether they had been incubated or not with EGF (Fig. 1D). NR1I3 (asterisk in Fig. 1D) that encodes CAR, the master regulator of CYP2B6 gene induction, was among the genes downregulated upon incubation with EGF (FC = 0.31, q = 0.009).
EGF Reduces the Expression of CAR and CAR-Target Genes.
To determine the role of EGF in CAR expression and signaling in PHHs, cells were incubated with CITCO, a CAR agonist, or with PB, an indirect CAR activator, for 24 hours (from day 2 to day 3 postseeding) in the presence or absence of 10 ng/ml EGF. CITCO was used at a low concentration that does not activate PXR (Maglich et al., 2003). To rule out an indirect effect due to long-term treatment, EGF was added before (day 0) or concomitantly (day 2) with the activators. RT-qPCR analysis confirmed that, compared with untreated cells (UT), EGF specifically and reproducibly decreased CAR mRNA level (Fig. 2A), in the presence or not of CITCO and PB, but had a minor effect on PXR mRNA expression (not shown). Next, as PB can activate both CAR and PXR (Lehmann et al., 1998; Moore et al., 2000), the expression of their prototypical target genes CYP2B6 (Sueyoshi et al., 1999) and CYP3A4 was monitored. In the absence of EGF, PB and CITCO incubation increased CYP2B6 and CYP3A4 mRNA level compared with UT cells. Conversely, incubation with EGF before or concomitantly with the CAR activators strongly and significantly reduced CYP2B6 upregulation by CITCO and PB (Fig. 2B) and of CYP3A4 by CITCO, and to a lesser extent by PB (Fig. 2C). Of note, CYP2B6 basal level (UT samples) was also significantly affected by the presence of EGF in the medium (Fig. 2B).
EGF influences specifically the expression of CAR and its target genes in PHHs. PHHs from two donors (Liv1 and Liv2, Table 1) were incubated with 1 µM CITCO, 0.5 mM PB, or dimethylsulfoxide (DMSO) (UT) for 24 hours in the absence or presence of 10 ng/ml EGF. EGF was added before [day 0 (D0)] or together [day 2 (D2)] with the inducers. The relative expression of CAR (A), CYP2B6 (B), and CYP3A4 (C) mRNA was measured by qPCR and normalized to RPLP0. Results are expressed relative to control (UT) in medium without EGF; *P < 0.05; **P < 0.01, relative to control (no EGF) in the same treatment group; #P < 0.05; ##P < 0.01, relative to control (UT).
These findings suggest that EGF targets rapidly and more specifically CAR signaling in PHHs by modulating its expression and possibly its activity.
EGF Affects CYP2B6 Induction via CAR.
To better understand EGF effect on CAR signaling, CAR expression was downregulated using specific small interfering RNAs (siCAR) and a nontargeting siRNA as a negative control (siCT). After siRNA transfection at day 2 and day 5, CAR mRNA expression was reduced by about 50% independently of incubation with EGF (from day 1 to day 5) and CITCO or PB addition at day 5 for 24 hours (Fig. 3A). In parallel, CAR inhibition reduced CYP2B6 basal expression by twofold (Fig. 3B), as observed in the presence of EGF (Fig. 2B). In the absence of EGF, CYP2B6 mRNA was induced ∼15- and ∼17-fold following CITCO and PB incubation, respectively, in siCT cells. This induction was reduced by more than 50% in CAR-depleted PHHs (Fig. 3B, left). In the presence of EGF, CAR downregulation abolished CYP2B6 mRNA induction mediated by CITCO, whereas it did not significantly affect induction by PB (Fig. 3B, right), although EGF presence greatly affected CYP2B6 mRNA level. Western blot analysis of CYP2B6 expression confirmed these results (Fig. 3C). The different effects observed following incubation of PHHs with CITCO and PB and the impact of CAR downregulation suggest that EGF affects CYP2B6 induction via CAR.
CAR depletion inhibits CYP2B6 induction by PB and CITCO only in the absence of EGF. PHHs from five donors (Liv3–7) were transfected with control (siCT) or CAR-targeting siRNAs (siCAR). Then they were incubated with 0.5 mM PB and 1 µM CITCO or DMSO (UT) for 24 hours, with or without 10 ng/ml EGF (from D0). The relative expression of CAR (A) and CYP2B26 (B) mRNA was measured by qPCR and normalized to RPLP0. Results are expressed relative to UT in siCT-transfected cells in the absence of EGF. (C) CYP2B6 protein level was assessed by Western blotting (representative experiment using PHHs from Liv12). GAPDH protein expression was used as loading control. Results are expressed relative to control (UT) in medium without EGF; ***P < 0.005, relative to control (siCT) in the same treatment group; ###P < 0.005, relative to control (UT).
EGF Differentially Affects the Response to CITCO and PB.
Our previous observation was limited to CYP2B6, the prototypical CAR target gene. Therefore, EGF effect on CAR transcriptional activity in PHHs was assessed by transcriptomic analysis. First, to identify genes affected by EGF, the transcriptomic profiles of siCT-transfected PHHs incubated with CITCO or PB in the presence or absence of EGF were analyzed (Fig. 4, A and B). Independently of EGF presence/absence, more genes were deregulated following incubation with PB than with CITCO (no EGF: 144 vs. 111; with EGF: 109 vs. 58, respectively), in agreement with PB pleiotropic effects (Handschin and Meyer, 2003). Moreover, genes were mostly upregulated (red) in response to PB or CITCO in the presence of EGF, whereas the effect was more heterogeneous (up- and downregulation) in the absence of EGF. Further analysis revealed that only CYP2B6 and CYP2A6 (1.2% of all deregulated genes after CITCO incubation) were significantly deregulated by CITCO in both the presence and absence of EGF (Fig. 4A). However, the FC values were different in the presence and absence of EGF (FC = 2.1 vs. 10.5 for CYP2B6, and FC = 1.4 vs. 14.8 for CYP2A6). On the other hand, upon PB incubation, 45 genes (21.6% of all deregulated genes, including CYP2A6 and CYP2B6) were affected both in the absence and presence of EGF (Fig. 4B), with a weak effect of EGF on their FC (data not shown). Functional characterization demonstrated that most of the 109 genes deregulated by CITCO only in the absence of EGF encoded drug-metabolizing enzymes and were associated with xenobiotic metabolic processes (Table 3). A similar pattern was observed for the 45 PB-deregulated genes that do not depend on EGF (Table 4). Surprisingly, among the genes deregulated by PB (64 genes) or CITCO (56 genes) only in the presence of EGF, many were associated with mRNA processing. Moreover, in the absence of EGF, most of the deregulated genes in response to PB were associated with lipid and steroid metabolism.
EGF differentially affects the response to CITCO and PB. Venn diagrams showing the overlap of genes that are differentially expressed in PHHs from five donors (Liv3–7) upon incubation with 1 µM CITCO (A) or 0.5 mM PB (B) in the absence or presence of 10 ng/ml EGF (from D0). The number of up- and downregulated genes in each experimental condition is indicated in red and green, respectively.
Enriched biologic processes after incubation of PHHs with CITCO
Enriched biologic processes following incubation of PHHs with PB
Finally, comparison of the KEGG pathway enrichment for the genes deregulated in siCT-transfected PHHs incubated with PB (Fig. 5A) or CITCO (Fig. 5B) for 24 hours with or without EGF (relative to UT cells) showed that in the absence of EGF, most of the enriched pathways belonged to metabolism, whatever the treatment. This profile was weakly affected by EGF in cultures incubated with PB. In contrast, in the presence of EGF, CITCO-deregulated genes were no more associated with any enriched KEGG pathway.
KEGG pathway enrichment, relative to untreated cells, of siCT-transfected PHHs from five donors (Liv3–7) incubated with 0.5 mM PB (A) or 1 µM CITCO (B) for 24 hours in the presence or absence of 10 ng/ml EGF (from D0).
Identification of CAR Target Genes.
The role of CAR in gene induction following incubation with CITCO or PB (with or without EGF) was then assessed in PHHs in which CAR was downregulated by siCAR transfection. On the basis of their unresponsiveness to CITCO or PB in siCAR cells compared with siCT cells, a list of CAR target genes was identified (Table 5). Among the 111 genes deregulated by CITCO in the absence of EGF in siCT cells (Fig. 4A), only 15 genes were identified as CAR-target genes. They included previously described CAR targets (CYP2B6, -2A6/7, -2C8/9, 3A4/7/43, EPHX1, ALAS1, and POR) and four new potential up (RDH16, TSKU, CPEB3)- and downregulated (TAGLN) genes (Table 5). Among the 144 genes induced by PB in the absence of EGF in siCT cells (Fig. 4B), most (n = 140) were regulated in a CAR-independent manner. Among the few CAR targets in PB-treated PHHs, CYP2B6 and CYP2A6/7 were also induced by CITCO, whereas STEAP2 was specifically downregulated by PB.
List of genes differentially regulated by CITCO and PB in a CAR-dependent manner
In the presence of EGF, among the 58 genes deregulated following CITCO treatment in siCT cells, only CYP2B6 and CYP2A6 were regulated by CITCO in a CAR-dependent manner. Conversely, upon incubation with PB in the presence of EGF, no gene was regulated in a CAR-dependent manner (Table 5).
These results were validated by qPCR analysis. In siCT cells, CYP2A6 and CYP2A7 (Fig. 6, A and B) were similarly induced by PB and CITCO in the absence of EGF. This induction was strongly reduced following CAR downregulation (siCAR). In the presence of EGF, induction of CYP2A6 and CYP2A7 by CITCO and PB was strongly reduced in siCT cells. CITCO-mediated induction was abolished following CAR downregulation, whereas PB-mediated induction was less affected, as observed for CYP2B6 (Fig. 3B). In the absence of EGF, CYP3A4, CYP3A43, CYP3A7, CYP2C8, CYP2C9, TSKU, EPHX1, and POR (Fig. 6, C–J; Table 5) upregulation by CITCO depended on CAR expression, whereas most of the PB-induced genes were CAR independent, but for CYP2C8 and TSKU. CAR contribution, monitored by the degree of siCAR-mediated inhibition of CITCO-induced response, varied among genes and was positively correlated with EGF inhibitory effect on gene expression. Addition of EGF did not change the FC induction of these genes by PB in controls (siCT) and CAR-depleted cells. Conversely, EGF abolished completely CITCO effect on the induction of these genes (Fig. 6, C–L), as shown for CYP2B6 (Fig. 3B). This pattern was not observed for CPEB3 and STEAP2 (Fig. 6, M and N). These results suggest that EGF deeply affects the drug metabolism response in PHHs.
PHHs from two donors (Liv6 and Liv7, Table 1) were transfected with control (siCT) or CAR-targeting siRNAs (siCAR). They were then incubated with 0.5 mM PB, 1 µM CITCO, or DMSO (UT) for 24 hours, in the presence or not of 10 ng/ml EGF (from D0). The relative expression of CYP2A6 (A), CYP2A7 (B), CYP3A4 (C), CYP3A43 (D), CYP3A7 (E), CYP2C8 (F), CYP2C9 (G), TSKU (H), EPHX1 (I), POR (J), ALAS1 (K), TAGLN (L), STEAP2 (M), and CEBP3 (N) mRNA was measured by qPCR and normalized to RPLP0. Results are expressed relative to not treated siCT-transfected PHHs (UT) in medium without EGF. *P < 0.05; **P < 0.01, relative to control in the same treatment group; ##P < 0.01; ###P < 0.005, relative to control (UT).
Analysis of EGF effect on PHH transcriptome profile highlighted important differences compared with EGF-free cultures. As CAR expression is inhibited by EGF, no effect of CITCO was expected in the presence of EGF. Surprisingly, 58 genes were still upregulated upon incubation of PHHs with CITCO in the presence of EGF (Fig. 4A), and 55 of these genes were deregulated also by PB (Fig. 4B).
Altogether PHH transcriptomic analysis showed that 1) EGF strongly affects the transcriptional activity of the xenosensor CAR and 2) PHH response to PB is much less sensitive to EGF than that to CITCO, suggesting that CAR-independent mechanisms are involved.
In the Presence of EGF, PXR Regulates CAR Target Genes.
It is well known that PXR shares ligand and target genes with CAR in humans (Moore et al., 2000). Recently, Kandel et al. (2016) demonstrated, using human hepatocytes, that CAR and PXR activation by CITCO and RIF, respectively, deregulate the same pathways (including the drug metabolism pathway). Our results showed that EGF negatively affects CAR expression and that a set of genes can still be induced by PB in the presence of EGF in siCAR cells. Therefore, we hypothesized that in the presence of EGF, PXR can take the place of CAR. To test this hypothesis, the effect of CITCO and PB, in the presence or absence of EGF, was assessed in PHHs after siRNA-mediated downregulation of PXR (siPXR). In PHHs transfected with siPXR, a 50% reduction of PXR mRNA was observed, independently of the presence of PB, CITCO, or EGF (Fig. 7A). CAR mRNA level was not affected (Fig. 7B). Different from what observed in siCAR cells, in the absence of EGF, PB- and CITCO-mediated CYP2B6 induction was not inhibited upon PXR downregulation (Fig. 7C). Conversely, in the presence of EGF, PB-mediated CYP2B6 induction was significantly reduced by PXR downregulation, while CITCO-mediated induction was not affected. CYP3A4 mRNA was upregulated by CITCO only in the absence of EGF (Figs. 2B and 7D; Table 5), and its induction by PB depended on PXR, both in the presence or absence of EGF. These data suggest that in the presence of EGF, PB effect on CYP2B6 and CYP3A4 expression could be mediated via PXR rather than CAR. This was confirmed by siRNA-mediated downregulation of both CAR and PXR (Fig. 7E). Indeed, in the presence of EGF, induction of CYP2B6, -3A4, -2C8, -2A6, -2A7, -3A43, and TSKU by PB (Fig. 7F) was strongly reduced when PXR (but not CAR) expression was downregulated. Moreover, this effect was increased in cells in which both CAR and PXR were silenced, particularly for CYP2A6 and CYP2A7.
PHHs from five donors (Liv3–7, Table 1) were transfected with control (siCT), CAR-targeting (siCAR), or PXR-targeting (siPXR) siRNAs. Then they were incubated with 0.5 mM PB, 1 µM CITCO, or DMSO (UT) for 24 hours in the presence or not of 10 ng/ml EGF (from D0). The relative expression of PXR (A), CAR (B), CYP2B6 (C), or CYP3A4 (D) mRNA was measured by qPCR and normalized to RPLP0. Results are expressed relative to not treated siCT-transfected PHHs (UT) in medium without EGF. PHHs from Liv16 were transfected with siCT, siCAR and/or siPXR siRNAs. Then, they were incubated with 0.5 mM PB or DMSO for 24 hours, in the presence of 10 ng/ml EGF (from D0). The relative expression of CAR and PXR (E), CYP2B6, CYP3A4, TSKU, CYP2C8, CYP2A6, CYP2A7, CYP3A43 (F) mRNA was measured by qPCR and normalized to RPLP0 expression level. Results are expressed relative to not treated siCT-transfected PHHs in medium with EGF (white bars). *P < 0.05; **P < 0.01; ***P < 0.001 relative to control in the same treatment group.
Finally, individual analysis of the effect of CITCO or PB (in the absence of EGF) in the different PHH cultures used for gene expression analysis (n = 8) indicated that CITCO-mediated induction of CYP2B6 mRNA expression was variable among samples (Fig. 8A). Moreover, it was significantly lower than the effect mediated by PB in Liv13 and Liv15, and almost no induction could be observed in Liv14 relative to UT cultures. Similarly, the relative CAR mRNA level in the different cultures was more heterogeneous than that of PXR (from 1 to 100 vs. 1 to 10, respectively, not shown). Comparison of CITCO effect and of the PXR/CAR mRNA expression ratio in the eight samples showed that in the absence of EGF, when the PXR/CAR mRNA ratio was low (Fig. 8B) (i.e., high CAR mRNA expression), CITCO-mediated CYP2B6 induction was high (Fig. 8A). Conversely, when the PXR/CAR mRNA ratio was high (i.e., low CAR mRNA expression), CITCO-mediated CYP2B6 induction was low. In the presence of EGF that downregulates CAR expression, CYP2B6 induction was lower after incubation with CITCO than with PB (Fig. 8C), and the PXR/CAR mRNA ratio values were higher (low CAR expression) (Fig. 8D). Moreover, siRNA-mediated CAR or PXR downregulation in two samples that poorly responded to CITCO (Liv13 and Liv15) showed that CYP2B6 upregulation by PB depended on PXR activity (Fig. 8, E and F), both in the presence and absence of EGF.
PHHs from eight donors (Liv3–7 and 13–15, Table 1) were incubated with 0.5 mM PB, 1 µM CITCO, or DMSO (UT) for 24 hours in the presence or not of 10 ng/ml EGF (from D0). The relative expression of CYP2B6 was measured by qPCR and normalized to RPLP0 in the absence (A) or presence (C) of EGF. PXR/CAR mRNA expression ratio in the absence (B) or presence (D) of EGF (expressed as LOG10). PHHs from two donors (Liv13 and Liv15, Table 1) (E and F) were transfected with control (siCT), CAR-targeting (siCAR), or PXR-targeting (siPXR) siRNAs. They were then incubated with 0.5 mM PB, 1 µM CITCO, or DMSO (UT) for 24 hours in the presence or not of 10 ng/ml EGF (from D0). The relative expression of CYP2B6 was measured by qPCR and normalized to RPLP0 in the absence or presence of EGF. Results are expressed relative to not treated siCT-transfected PHHs (UT) in medium without EGF. *P < 0.05; **P < 0.01; ***P < 0.001.
In conclusion, a switch from CAR to PXR dependency in PB-target gene induction is observed when CAR expression is low and in the presence of EGF. This observation could be critical for the development of target therapies.
Discussion
Growth factors play crucial roles in liver physiology, and the involvement of some of them in CAR activity and expression has been widely studied. However, few data are available on EGF role in PHHs. Here, by using a direct and an indirect CAR activator, we show that EGF influences the expression of CAR and CYP2B6, the prototypical CAR target gene, in PHHs. We then found that few genes are strictly CAR dependent in PHHs. Moreover, in this cell model, the indirect CAR activator PB modulates gene expression differentially in the absence or presence of EGF, but mostly in a CAR-independent and PXR-dependent manner. This may influence the design and interpretation of drug investigation studies.
Very few studies have analyzed PHH transcriptome profile following activation of human CAR in vitro. Kandel et al. (2016) performed a genome-wide comparison of the inducible transcriptomes of the nuclear receptors CAR, PXR and PPARα in PHHs (Kandel et al., 2016). Li et al. (2015) studied genome-wide the transcriptome profiles of parental and HepaRG cells where NR1I3 was knocked down (CAR-KO). Surprisingly, we identified only 15 CAR target genes among the 111 genes differentially regulated by CITCO and only four among the 144 genes affected by PB. Conversely, Li et al. (2015) found that in HepaRG cells, the expression of 135 genes and of more than 120 genes was affected by CITCO and by PB, respectively, in a CAR-dependent manner (Li et al., 2015). The difference in the number of deregulated genes could be linked to the cell type (HepaRG cells vs. PHHs). Moreover, CAR silencing by siRNA led to a decrease of only approximately 50% of CAR mRNA expression in PHHs. Therefore, we cannot rule out that the remaining CAR protein is enough to regulate gene expression and that we only identified the major CAR targets. However, the observation that CYP2B6 expression was significantly downregulated in CAR-depleted hepatocytes after treatment with CITCO suggests that the silencing strategy was effective. In addition, PB and CITCO off-target effects have been already reported (Ueda et al., 2002; Li et al., 2015). Therefore, our results are in agreement with PB nonspecific and pleiotropic effects and its possible cross-activation of PXR and also highlight the poorly described CAR-independent CITCO activity.
Most of the identified CAR target genes have been previously described (ALAS1, POR, CYP2A6 -2A7 -2B6 -2C8 -2C9 -3A4 -3A7, EPHX1) (Kandel et al., 2016) and we found only four new genes, of which three were validated by PCR (TSKU, TAGLN, RDH16). CYP2B6, CYP2A6, and CYP2A7 were the only genes modulated in a CAR-dependent manner by both PB and CITCO. CYP2A6 and CYP2A7 are located in close proximity of CYP2B6 on chromosome 19q13.2, and common modulation mechanisms could be involved. Accordingly, several studies found that CYP2A6 (Maglich et al., 2003; Itoh et al., 2006) and CYP2A7 (Kandel et al., 2016) are induced by CITCO in PHHs. TSKU is induced by PB in HepaRG cells and PHHs (Lambert et al., 2009) and is regulated by PXR in osteoblasts (Ichikawa et al., 2006).TAGLN, one of the new genes identified as downregulated by CAR in the present study, encodes transgelin that is mainly expressed in stellate cells and fibroblasts (Petkov et al., 2004). We cannot rule out that PHHs may be contaminated by nonparenchymal cells. Retinol dehydrogenase 16 (RDH16) is involved in retinol synthesis. Its expression increases following PPARα activation in PHHs (Kandel et al., 2016). In silico analysis revealed the presence of putative DR3 and DR4 motifs on its regulatory region, suggesting a possible regulation by CAR (Ebert et al., 2016). Our results demonstrate that in PHHs, only very few genes are strictly CAR dependent, suggesting that complementary mechanisms can promote the expression of xenobiotic-related genes. This could explain in part the difference between humans and mice. Studying CAR transcriptional regulation in rodents, using in vivo and in vitro approaches could help understanding whether our findings are specific to PHHs, or biased by the 2D culture model. Indeed, gene expression (Lauschke et al., 2016b) and energy metabolism (Fu et al., 2013) patterns rapidly change when human/rodent hepatocytes are seeded in culture. Therefore, although PHHs are the most relevant cell model for human studies, data obtained in vitro using this system should be interpreted with caution. During the last decade, cell culture strategies to mimic in vivo situations have been developed. Specifically hepatocytes have been cultured in 3D spheroids (Bachmann et al., 2015; Bell et al., 2016; Vorrink et al., 2017) in the presence or not of stromal cells (Khetani and Bhatia, 2008) and in association or not with medium flux (Vinci et al., 2011; Lin et al., 2015; Lauschke et al., 2016a). These conditions favor the long-term maintenance of cell viability and an in vivo-like hepatic phenotype, as indicated by transcriptomic, proteomic, and metabolomic analyses. This is accompanied by stable expression of phase I and phase II enzymes, phase III transporters, xenoreceptors (CAR and PXR), and transcription factors (HNF4), as well as improved cell-to-cell contacts and bile canalicular structure and activity compared with 2D culture conditions. Therefore, moving from monolayer to 3D cell culture systems can greatly improve nuclear receptor expression and gene-mediated regulation (Vorrink et al., 2017) as well as the in vitro assessment of drug metabolism and toxicity (Lin and Khetani, 2016).
It has been shown that in the liver of EGF2B transgenic mice, the expression of genes predisposing to malignancy is increased, whereas that of enzymes associated with retinoic acid and xenobiotic metabolism is downregulated (Borlak et al., 2005). Interestingly, we obtained similar results in PHHs. KEGG pathway analysis showed that in the presence of EGF, upregulated genes mostly belonged to the cell cycle and carcinogenesis pathways (Komposch and Sibilia, 2015). In addition, in the presence of EGF, pathways associated with drug metabolism-cytochrome P450 were downregulated (Fig. 1C). These pathways are downregulated also in CAR-KO HepaRG cells (Li et al. 2015), suggesting that the EGF-mediated phenotype could be explained, at least in part, by CAR signaling deregulation (Li et al., 2015). This hypothesis is reinforced by the finding that the same KEGG pathways are upregulated in PHHs incubated with CITCO [this study and Kandel et al. (2016)]. This CITCO-mediated effect was almost completely abrogated by the presence of EGF (Fig. 5B).
In our model, we observed that EGF affects similarly CYP2B6 induction by CITCO and PB in short-term culture (day 2) (Fig. 2). In addition, after 6 days of culture in the presence of EGF, CITCO-mediated CYP2B6 induction was lost, in contrast to the induction mediated by PB (Fig. 3). Our transcriptome analysis clearly revealed the strong EGF effect on CAR signaling when using CITCO as activator. In contrast, PB response was almost not affected by the presence of EGF. This is in agreement with PB nonspecific action, but also suggests a secondary compensatory mechanism.
CAR activity regulation involves multiple mechanisms, including the AMPK, insulin, EGFR, and MEK-ERK signaling pathways (Yang and Wang, 2014; Yasujima et al., 2016). Conversely, CAR transcriptional regulation has been poorly studied. The hepatotrophic factor augmenter of liver regeneration reduces P450 enzyme activities in human hepatocytes partially through nuclear factor κB activation and CAR downregulation (Dayoub et al., 2006). In a mouse model in which mouse HGF is overexpressed, CAR but not PXR mRNA expression is reduced (Kakizaki et al., 2007). It would be important to assess whether other growth factors and/or serum have a similar inhibitory effect in PHHs and in HepaRG cells (Aninat et al., 2006). In addition, EGF and serum downregulate CAR, but not PXR and retinoic X receptor expression in HepG2 cells (Osabe et al., 2009) via the SAPK signaling pathway. This may also occur in PHHs.
HNF-4α, PGC-1α, and glucocorticoid receptor are considered to be the main regulators of CAR promoter activity (Pascussi et al., 2003; Ding et al., 2006). Insulin-like growth factor-1 receptor inhibitor (Li et al., 2012) and thyroid hormone (Ooe et al., 2009) also modulate CAR expression. In addition, post-transcriptional regulation by miR-34-a (Lamba et al., 2014) and miR-137 (Chen et al., 2014) leads to NR1I3 downregulation. HNF4α regulates both CAR transcriptional activity (Gonzalez, 2008) and expression (Ding et al., 2006; Kamiyama et al., 2007). However, HNF4α gene expression was not affected by EGF in our model. Direct HNF4α phosphorylation is another possible mechanism that could contribute to EGF effect in hepatocytes (De Boussac et al., 2010; Vető et al., 2017).
Importantly, we found that in the presence of EGF, CAR does not modulate anymore the expression of its target genes (but for CYP2B6 and CYP2A6, to a lower extent). Using a gene silencing strategy, we could show that PB preferentially induces CYP2B6 by selective CAR activation, but can switch to PXR activation when CAR is present in limited amount (in the presence of EGF and when CAR is artificially downregulated). The finding that many genes modulated by PB independently of CAR in the presence of EGF are known PXR target genes (e.g., CYP2B6, CYP3A7, CYP2C8, CYP3A4, EPHX1, CYP3A43, CYP2C9, THRSP, ALAS1, ABCC2, POR, CYP3A5, and ABCB1) further supports the existence of a PXR-mediated compensatory mechanism that might allow overcoming the lack of response to potential threats. This conclusion is consistent with the enhanced role of human PXR in the absence of CAR competition observed in CAR-KO HepaRG cells (Li et al., 2015) and also with the asymmetric regulation of CYP3A4 and CYP2B6 by CAR or PXR previously reported by Faucette et al. (2006). We also observed that induction of drug metabolism enzymes depends on the PXR/CAR ratio. Similarly, a previous report using reporter genes in the mouse showed a crucial role of the PXR/CAR ratio in Cyp2b10 modulation (Ding and Staudinger, 2005). Considering the interindividual variability of CAR and PXR expression, this observation could have a critical impact in drug development.
Altogether, our findings bring new insights into EGF effect on CAR activity in PHHs that might be physiologically relevant and highlight the importance of transcriptional regulation in CAR biology. CAR transcriptional regulation may play an underestimated role in drug metabolism and transport, energy homeostasis, and cell proliferation. Furthermore, unlike the results described in rodents, our work shows that the indirect CAR activator PB modulates gene expression predominantly in a CAR-independent manner, although EGF significantly affects PHH response to this drug. It would be interesting to evaluate this effect using the most recent microfluidic and 3D spheroid culture systems in which nuclear receptor expression is maintained (Lauschke et al., 2016a; Vorrink et al., 2017). Finally, the switch between CAR and PXR for gene modulation in CAR-depleted cells highlights the crosstalk between these xenosensors in the presence of EGF. This crosstalk provides a mechanism for amplifying the body detoxification response to a broad range of chemicals.
In addition to their primary role in drug metabolism, CAR and PXR are also potential pharmacological targets for the treatment of cancers (De Mattia et al., 2016) and metabolic diseases, such as cholestasis, non-alcoholic fatty liver disease, and diabetes (Kakizaki et al., 2007; Gao and Xie, 2012). As PXR and CAR may differentially affect metabolic pathways, it is important to selectively target these receptors. In this context, controlling the microenvironment and culture conditions of the hepatic cells used for investigating CAR activity is crucial.
Acknowledgments
We thank A. Kassambara for bioinformatics advice and E. Vidal and F. Carol for technical support. This work benefitted from the equipment of the transcriptome facility of the University Hospital Centre (CHU) of Montpellier and from the expertise of Véronique Pantesco (v-pantesco@chumontpellier.fr), scientific manager of the facility.
Authorship Contributions
Conducted experiments: de Bousssac, Gondeau, Briolotti, Duret, Treindl, Römer, Templin, Gerbal-Chaloin.
Contributed new reagent and analytic tools: Fabre, Herrero, Ramos, Maurel.
Performed data analysis: de Bousssac, Gerbal-Chaloin, Daujat-Chavanieu.
Wrote or contributed to the writing of the manuscript: de Bousssac, Gerbal-Chaloin, Daujat-Chavanieu.
Footnotes
- Received September 19, 2017.
- Accepted December 14, 2017.
The research leading to these results received funding from the Innovative Medicine Initiative Joint Undertaking (IMI JU) under grant agreement n° 115001 (MARCAR project).
↵1 S.G.-C. and M.D.-C. are coauthors.
Abbreviations
- CAR
- constitutive androstane receptor
- CITCO
- 6-(4-chlorophenyl)imidazo[2,1-b][1,3]thiazole-5-carbaldehyde O-(3,4-dichlorobenzyl) oxime
- CYP
- cytochrome P450
- EGF
- epidermal growth factor
- EGFR
- epidermal growth factor receptor
- FC
- fold change
- KEGG
- Kyoto Encyclopedia for Genes and Genomes
- PB
- phenobarbital
- PCR
- polymerase chain reaction
- PHH
- primary human hepatocytes
- qPCR
- quantitative polymerase chain reaction
- PXR
- pregnane X receptor
- RIF
- rifampicin
- UT
- untreated
- Copyright © 2018 by The American Society for Pharmacology and Experimental Therapeutics
