High-content analysis of constitutive androstane receptor (CAR) translocation identifies mosapride citrate as a CAR agonist that represses gluconeogenesis

Abstract

The constitutive androstane receptor (CAR) plays an important role in hepatic drug metabolism and detoxification but has recently been projected as a potential drug target for metabolic disorders due to its repression of lipogenesis and gluconeogenesis. Thus, identification of physiologically-relevant CAR modulators has garnered significant interest. Here, we adapted the previously characterized human CAR (hCAR) nuclear translocation assay in human primary hepatocytes (HPH) to a high-content format and screened an FDA-approved drug library containing 978 compounds. Comparison of hCAR nuclear translocation results with the Tox21 hCAR luciferase reporter assay database in 643 shared compounds revealed significant overlap between these two assays, with approximately half of hCAR agonists also mediating nuclear translocation. Further validation of these compounds in HPH and/or using published data from literature demonstrated that hCAR translocation exhibits a higher correlation with the induction of hCAR target genes, such as CYP2B6, than the luciferase assay. In addition, some CAR antagonists which repress CYP2B6 mRNA expression in HPH, such as sorafenib, rimonabant, and CINPA1, were found to translocate hCAR to the nucleus of HPH. Notably, both the translocation assay and the luciferase assay identified mosapride citrate (MOS), a gastroprokinetic agent that is known to reduce fasting blood glucose levels in humans, as a novel hCAR activator. Further studies with MOS in HPH uncovered that MOS can repress the expression of gluconeogenic genes and decrease glucose output from hepatocytes, providing a previously unidentified liver-specific mechanism by which MOS modulates blood glucose levels.

1. Introduction

The constitutive androstane receptor (CAR; NR1i3) is a xenobiotic receptor, primarily expressed in the liver, whose canonical role involves the upregulation of detoxification mechanisms such as drug-metabolizing enzymes (DMEs) and transporters to protect the body from foreign chemicals. Upon activation, human CAR (hCAR) induces DMEs such as CYP2B6, its main target gene, and CYP3A4; both enzymes metabolize important clinically-used drugs and are known to participate in drug-drug interactions (DDIs) [1,2]. More recently, the non-canonical roles of xenobiotic receptors, such as the pregnane X receptor (PXR; NR1i2), have been the subjects of intense research, and hCAR is no exception. Studies have found that hCAR plays a significant role in liver physiology, modulating cellular energy homeostasis and cell proliferation [3–5]. This makes hCAR an enticing target for metabolic disorders and has led to a search for novel hCAR activators that may have therapeutic potentials.

In the liver and primary hepatocytes, hCAR is localized in the cytoplasm and upon activation, translocates to the nucleus where it heterodimerizes with the retinoid X receptor (RXR) and enhances the transcription of target genes [6–8]. However, hCAR is localized in the nucleus and constitutively active in most immortalized cell lines, making mechanistic studies of hCAR difficult [9,10]. In addition, hCAR can be activated either directly through ligand binding or indirectly via signaling cascades, adding further complexity to our understanding of hCAR activation [11]. Despite these challenges, many efforts have been made to identify compounds that modulate hCAR activity because of its ability to mediate DDIs and control important physiological processes in the liver.

A number of screening methods have been developed to identify hCAR modulators, employing approaches from computational modeling to cell-based luciferase reporter assays with varying success [12–16]. While computational and in situ approaches have had some success finding hCAR ligands, cell-based assays are the most widely used to identify chemical modulators of this receptor. Modification of the hCAR protein sequence and application of hCAR inverse agonists/antagonists have been successfully used to lower the high basal activity and identify activators of hCAR in cell-based luciferase reporter assays [16–18]. To date, the two largest screens for hCAR modulators have utilized HepG2 cells stably expressing full-length hCAR and a luciferase reporter driven by the CYP2B6 promoter [14,19]. Out of the eight novel hCAR activators chosen for further characterization in a recent study, only four induced CYP2B6 or 3A4 in HPH [19]. Therefore, activation of hCAR in the luciferase assay may not be highly predictive of physiologically-relevant hCAR activation.

One assay that has shown success in predicting hCAR activation is the hCAR nuclear translocation assay in HPH. This assay overexpresses enhanced yellow fluorescent protein (EYFP)-tagged hCAR in HPH and monitors whether hCAR, which is sequestered in the cytoplasm in the quiescent state, undergoes the first step of activation and translocates to the nucleus [20]. In fact, this assay was able to accurately classify the 8 validated compounds from the previous Tox21 study [19]. In the current study, we established a high-throughput hCAR translocation assay and have comprehensively validated the outcomes from screening an FDA-approved drug library containing 978 drugs. Through this approach, we were able to identify mosapride citrate (MOS), a selective 5-Hydroxytryptamine receptor 4 agonist used to increase gastric motility, as a novel activator of hCAR in HPH, which downregulates the expression of glucose 6-phosphatase (G6Pase) and phosphoenolpyruvate carboxykinase (PEPCK). In addition to its role as a gastroprokinetic agent, MOS has been shown to significantly lower fasting glucose levels in human clinical studies [21,22]. Here, we offer evidence that MOS represses gluconeogenic genes and lowers the glucose output of HPH, providing a novel, liver-specific mechanism through which MOS can lower blood sugar levels in humans.

2. Materials and methods

2.1. Chemicals and biological reagents

The FDA Approved Drug Screening Library L1300 (96-well-Z151018–200uL) was obtained from Selleck Chemicals (Houston, TX). Phenobarbital (PB), rifampicin (RIF), MOS, 1-(2-Chlorophenyl)-N-methyl-N-(1-methylpropyl)-3-isoquinolinecarboxamide (PK11195), dibutyryl cyclic-AMP (cAMP), dimethyl sulfoxide (DMSO) and 6-(4-Chlorophenyl)imidazo[2,1-b][1,3]thiazole-5-carbaldehyde-O-(3,4-dichlorobenzyl)oxime (CITCO) were obtained from Millipore Sigma (St. Louis, MO). CAR inhibitor not PXR activator 1 (CINPA1) was obtained from Tocris Bioscience (Minneapolis, MN). PCR primers were synthesized by Integrated DNA Technologies (Coralville, IA). Data from the Tox21 10 K HepG2-CYP2B6-hCAR luciferase assay screen in agonist (AID: 1224892) and antagonist (AID: 1224893) modes were obtained from the Tox21 public database (https://tripod.nih.gov/tox21/assays).

2.2. Culture of HPH

HPH were obtained from BIOIVT (Baltimore, MD) and seeded at 7.5 × 10⁵ cells/well for 12-well, 3.75 × 10⁵ cells/well for 24-well, or 6.0 × 10⁴ cells/well for 96-well collagen-coated plates as described previously [23]. After cell attachment, HPH were cultured in serum-free William’s E Medium (WEM) supplemented with ITS+ (insulin, transferrin, and selenium, 100x; BD Biosciences, Bedford, MA), 0.1 μM dexamethasone (DEX; Millipore Sigma), 100 U/ml penicillin, and 100 μg/ml streptomycin (Pen-Strep; ThermoFisher Scientific, Waltham, MA), and 2 mM L-glutamine (Invitrogen, Carlsbad, CA). Cells used for RNA or protein expression were overlaid with 0.25 mg/mL Matrigel 24 h after seeding (Corning, Corning, NY). Overlaid HPH were treated with compounds in Complete WEM (CYP2B6 and CYP3A4 studies) or WEM with Pen-Strep and L-glutamine with or without 10 μM dibutyryl cAMP and 50 nM DEX (gluconeogenic gene studies) for 24 h or 48 h to isolate RNA or protein, respectively.

2.3. Ad/EYFP-hCAR translocation assay

HPH plated in collagen-coated, black-walled, 96-well plates (Corning Cat. #3603) on Day 0 were infected with Ad/EYFP-hCAR on Day 1 at a concentration of 3 μL/mL of media and incubated for 16 h overnight. On Day 2, the media for each plate was changed and cells were treated with compounds for 8 h. To validate the assay, positive (PB, CITCO, and PK11195) and negative controls (DMSO and RIF) were added to the plate in triplicate, results were analyzed as detailed below, and the z-factor was calculated according to Zhang et al. (1999) [24]. For screening, compounds from the FDA Approved Drug Screening Library L1300 (1 mM in DMSO) were dispensed into each well using the Biomek FX^P Laboratory Automation Workstation with Dual Arm System, Multichannel and Span-8 Pipettors (Beckman Coulter, Indianapolis, IN) for a final DMSO concentration of 1% and a compound concentration of 10 μM. DMSO, PB, CITCO, and PK11195 were dosed in duplicate in column 12 for each plate. After an 8 h incubation, cells were washed with PBS, fixed with 4% paraformaldehyde in PBS for 15 min, stained with 1 µg/mL 4′,6-diamidino-2-phenylindole (DAPI) (Millipore Sigma, St. Louis, MO) for 30 min, and stored at 4 °C protected from light until imaging.

2.4. High-content imaging and analysis

Imaging of plates was completed on a Nikon Eclipse Ti-E inverted microscope (Nikon, Edgewood, NY) equipped with a TI-SH-W Well Plate Holder (Nikon), CFP/YFP/mCherry Filter Cube (Nikon), CFI S Plan Fluor ELWD ADM 20x Objective (Nikon), SpectraX Multi-Spectral Solid-State Excitation Source (Lumencor, Beaverton, OR), and an ORCA-Flash4.0 Camera (Hamamatsu, Hamamatsu City, Japan). A custom acquisition method was developed in the Nikon Elements AR High Content Analysis software package (Version 4.50.00) for plate imaging of DAPI (Ex: 390 nm; Em: 461 nm) and YFP (Ex: 513 nm; Em: 575 nm) channels. Images were acquired in a regular pattern at five points in each well using YFP autofocus with the 20x objective.

A high-content analysis method was established using General Analysis in the Nikon Elements AR High Content Analysis software package. The goal of this analysis method was to calculate the number of cells where hCAR was nuclear localized and normalize to the number of cells expressing EYFP-hCAR to determine percent of hCAR that was translocated. First, an image mask was created for DAPI to define the “Nuclear” region of interest (ROI). Preprocessing utilized the Smooth, Rolling Ball Correction, and Local Contrast functions and only objects ≥ 9 µm in size were allowed while binary processing steps included Filter on Fill Area, Morpho Separate Objects, and Erode functions to accurately portray nuclei. Next, an image mask for YFP was created to define the “hCAR” ROI using Smooth, Rolling Ball Correction, Autocontrast, and Local Contrast functions during pre-processing and Filter on Fill Area and Erode functions during binary processing. A “Cells” ROI image mask was created by dilating the “Nuclear” ROI by 2 µm and then an “Expressing Cells” ROI was created by determining the “Cells” that have “hCAR” to calculate the number of cells expressing hCAR. To calculate the number of cells where hCAR is nuclear localized, a “Combined” ROI was made by including “Nuclear” objects that are ≥20% covered by “hCAR”. Then, a “Nuclear Localized” mask was created by including the “Expressing Cells” having “Combined” and the number of “Nuclear Localized” objects was divided by the number of “Expressing Cells” objects and multiplied by 100 to obtain percent translocation. All values were then corrected by subtracting the average baseline translocation for each plate (1% DMSO wells) and normalized to the PB translocation to obtain normalized hCAR translocation. Each compound was screened in two separate livers and the average normalized hCAR translocation values were obtained.

2.5. Immunofluorescence staining

HPH were treated with either vehicle control (0.1% DMSO) or PB (1 mM) for 8 h then fixed with 4% formaldehyde at room temperature for 15 min. The fixed cells were permeabilized and blocked with PBS containing 0.3% Triton X-100 (Millipore Sigma), 1% normal donkey serum (Millipore Sigma), and 1% bovine serum albumin (BSA, Millipore Sigma) at room temperature for 1 h. Cells were washed 3x with PBS and incubated with anti-CAR antibody (Perseus Proteomics, Tokyo, Japan) diluted 1:500 in PBS containing 1% BSA overnight at 4 °C. Subsequently, cells were washed 3x with PBS and incubated with Rhodamine-conjugated anti-Mouse IgG Secondary Antibody (ThermoFisher, #31660) diluted 1:50 in PBS containing 1% BSA for 1 h at room temperature. After cells were washed 3x with PBS, diluted DAPI was added and incubated in the dark for 2 h at room temperature. The cells were rinsed with PBS 3x before observing under a fluorescence microscope.

2.6. RT-PCR

After 24 h treatment, HPH were washed and TRIzol reagent (ThermoFisher, Rockford, IL) was added for total RNA isolation using a phenol-chloroform extraction. Extracted RNA (1 μg) was reverse transcribed to cDNA using the High Capacity cDNA Archive kit according to manufacturer’s instructions (Applied Biosystems, Foster, CA). To determine gene expression changes, PCR was performed on an ABI StepOnePlus Real-Time PCR system (Applied Biosystems) using Fast SYBR Green Master Mix (Thermofisher, Rockford, IL). Primer sequences for CYP2B6, CYP3A4, G6Pase, PEPCK, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) are: CYP2B6, 5′-AGACGCCTTCAATCCTG ACC-3′ and 5′-CCTTCACCAAGACAAATCCGC-3′; CYP3A4, 5′-GTGGGG CTTTTATGATGGTCA-3′ and 5′-GCCTCAGATTTCTCACCAACACA-3′; G6Pase, 5′- GTGTCCGTGATCGCAGACC-3′ and 5′- GACGAGGTTGAGC CAGTCTC-3′; PEPCK, 5′- GCAAGACGGTTATCGTCACCC-3′ and 5′- GGCATTGAACGCTTTCTCAAAAT-3′; GAPDH, 5′-CCCATCACCATCTTCCAGGAG-3′ and 5′-GTTGTCATGGATGACCTTGGC-3′. Expression values were quantified by the 2^ΔΔCt method, normalizing to GAPDH as the housekeeping gene and determining the fold change compared to control values (n = 3).

2.7. Western blot analysis

HPH were washed 1x with PBS and harvested using cell lysis buffer (Cell Signaling Technology, Danvers, MA) with complete protease inhibitor cocktail (Roche, Mannheim, Germany) before sonication and extraction of soluble protein. Samples were electrophoretically separated on NuPAGE Bis-Tris gels (4–12%) (ThermoFisher) and transferred to polyvinylidine fluoride (PVDF) membranes (ThermoFisher). Subsequently, membranes were blocked with 5% milk in Tris-buffered saline with Tween 20 (TBST) and then incubated with primary antibodies against CYP2B6 (1:500, Abcam), CYP3A4 (1:5000, Millipore Sigma), PGC-1α (sc-517380, 1:500, Santa Cruz Biotechnology, Dallas, TX) or β-actin (1:5000, Millipore Sigma) at 4 °C overnight. Membranes were washed with TBST 3x before incubating with horseradish peroxidase-linked secondary antibodies for 1 h at room temperature and probed with West Pico chemiluminescent substrate (ThermoFisher) and captured with film (CYP2B6 and CYP3A4) or probed with Radiance Q (Azure Biosystems, Dublin, CA) and imaged with the Azure c300 chemiluminescent western blot imaging system.

2.8. Glucose production assay

Overlaid HPH plated in 24-well plates were switched to WEM with only 100 U/ml penicillin, and 100 μg/ml streptomycin and 2 mM L-glutamine added and treated with compounds overnight in triplicate. The next morning, cells were washed 3× with PBS and starved in glucose-free DMEM supplemented with 10 mM HEPES buffer (pH 7.4) with treatments and with/without 10 μM dibutyl cAMP and 50 nM DEX for 2 h. HPH were washed again 3× with PBS and the same treatments as above were added to glucose-free DMEM supplemented with 10 mM HEPES buffer (pH 7.4) and 2 mM sodium pyruvate. Samples of media (100 μL) were taken at 5 h and 24 h to analyze glucose content. The relative amount of glucose in each sample was measured with the Amplex Red Glucose Kit (ThermoFisher) after diluting samples 5–100 times according to the manufacturer’s instructions.

2.9. Statistical analysis

All data are expressed as the mean ± S.D. Statistical comparisons were made using a one- or two-way ANOVA with a Bonferroni post-test. Statistical significance was set at *: p < 0.05, **: p < 0.01. The point-biserial correlation (r_pb) was calculated based on the equation $r_{pb}=\frac{M_1-M_0}{S_n}\sqrt{pq}$ where M₁ is the mean efficacy (agonist assay) or translocation for compounds that induce CYP2B6 and CYP3A4 expression, M₂ is the mean efficacy (agonist assay) or translocation for compounds that either do not induce or repress CYP2B6 and CYP3A4, S_n is the standard deviation across the dataset, p is the proportion of inducers in the dataset, and q is the proportion of non-inducers or repressors in the dataset.

3. Results

3.1. High-throughput EYFP-hCAR translocation assay development and validation

Human CAR translocation, from the cytoplasm to the nucleus of hepatocytes, has been used in numerous studies over the years as a marker of hCAR activation [13,14,20]. Studies have used either nuclear/cytosolic extracts of hCAR protein or assays that employ fluorescently-labeled hCAR protein to study its localization [20,25,26]. These methods are relatively low-throughput, and until now, have only been used to characterize small numbers of hCAR modulators. In this study, we have adapted the adenovirus (Ad)/EYFP-hCAR translocation assay in HPH from its traditional 24-well plate format to a 96-well screening platform, undergo automated high-content microscopy, and images were evaluated by an analysis method to determine the extent of hCAR nuclear translocation (Fig. 1A, B). The first step in adapting this assay to a high-throughput platform was optimization of the experimental time course, HPH plating density, and amount of adenovirus added (Data not shown). From the assay optimization, 8 h compound treatment, 6000 cells per well, and 3 μL of high-titer adenovirus per 1 mL of media were selected in the current study. Once the experimental system was in place, the automated microscopy and analysis methods were developed using several sets of HPH. When EYFP-hCAR is expressed in HPH, it is primarily localized in the cytoplasm; however, EYFP-hCAR also exhibits small amounts of punctate staining in the nuclei under baseline conditions. Therefore, we opted to define hCAR translocation as the percent of EYFP-hCAR expressing cells where more than 50% of the DAPI-stained nucleus overlaps with EYFP-hCAR to ensure that the majority of the nucleus is occupied with EYFP-hCAR to define it as “translocated”. This analysis approach gave consistent results over multiple liver donors and was validated in HPH136 and 139 using known hCAR translocators PB, CITCO, and PK11195 as positive controls, with DMSO and RIF as negative controls. The analysis method was able to successfully differentiate the positive and negative controls, shown by Z-factors > 0.5 for all positive controls in two different liver donors (Fig. 1C). To ensure that EYFP was accurately portraying hCAR translocation in this assay, hCAR was immunostained and imaged in HPH after treatment with DMSO or PB (Fig. 1D). The staining of hCAR was nearly identical to EYFP, validating that the EYFP, a signal of the EYFP-hCAR fusion protein, was accurately reflecting hCAR localization.

Fig. 1. Adapting the Ad/EYFP-hCAR nuclear translocation assay for high-throughput screening.

Inactive hCAR is sequestered in the nucleus and translocates to the nucleus upon activation by 1 mM PB in a representative liver donor, and the analysis method accurately identifies EYFP-hCAR and the DAPI-stained nucleus (A). Workflow for the high-throughput Ad/EYFP-hCAR nuclear translocation assay where HPH are plated in 96-well plates and infected with virus before they are treated with compounds, imaged, and analyzed (B). HPH from liver donors #136 and #139 were treated with negative controls (0.1% DMSO and 10 µM RIF) or positive controls (1 mM PB, 1 µM CITCO, or 10 µM PK11195) in triplicate using 96-well plates for 8 h, cells were fixed and stained with DAPI, five images were taken in each well. hCAR translocation (% of expressing cells with 50% of nucleus overlapping with EYFP-hCAR) was then calculated and Z-factors were obtained for each positive control compared to DMSO (C). HPH from liver donor #144 were treated with 0.1% DMSO or 1 mM PB for 8 h. Immunofluorescence staining with anti-CAR antibody was conducted as detailed in Materials and Methods (D). Outline for comparison of the hCAR nuclear translocation assay with the Tox21 compound library screened with the HepG2-CYP2B6-hCAR luciferase assay (E).

3.2. Screening FDA-approved drugs for hCAR translocation

To identify physiologically-relevant hCAR activators, we screened the FDA-Approved Drug Library containing 978 compounds with each compound at a 10 μM concentration using the validated hCAR translocation assay. The average translocation for each compound was obtained by subtracting basal translocation (control), normalizing to PB translocation, and taking the mean translocation from two separate liver donors. As potent hCAR activators such as phenytoin exhibited translocation ~40% of PB and the compound library was only being screened at a single concentration, compounds that exhibited hCAR translocation over 20% of PB were considered “hits” to ensure the identification of less potent hCAR activators that may be of clinical importance. In addition, we compared our translocation results with results from the Tox21 10 K HepG2-CYP2B6-hCAR luciferase assay screen in agonist (AID: 1224892) and antagonist (AID: 1224893) modes to better understand the similarities and differences between the two systems using the 643 overlapping compounds (Fig. 1E).

In the combined dataset, there were 86 hits in the translocation assay (> 20% PB translocation) compared with 66 agonists and 32 antagonists identified in the hCAR luciferase assays (Fig. 2A). There were 30 agonists and 5 antagonists in the luciferase assays that also translocated hCAR, while 51 translocators were either inconclusive or negative in the hCAR luciferase assays (Fig. 2A). To determine how the assays performed at predicting physiologically-relevant hCAR activators, we completed a literature search focused on the hits in the translocation, agonist, and antagonist assays, for CYP2B6 and/or CYP3A4 induction in HPH or clinical data from human subjects as an indicator of hCAR activation. Overall, we were able to find CYP2B6 or CYP3A4 induction data for 60 compounds (Tables 1–4). To determine any trends in the data, the efficacy for each agonist in the luciferase assay was plotted against the average translocation for each compound (Fig. 2B), and the literature data identifying inducers, repressors, and non-modulators of CYP2B6 and CYP3A4 were overlaid on the assay data (Fig. 2C). While the agonist only and translocator only groups both exhibited several non-modulators and/or repressors, all 18 compounds that activated hCAR in the luciferase assay, translocated hCAR to the nucleus, and had literature data available also induced CYP2B6 and/or CYP3A4, demonstrating the power of combining two separate assays for ligand identification. To determine which assay better predicted CYP2B6 and CYP3A4 induction on its own, the point biserial correlation (r_pb) was calculated, finding that hCAR translocation is more tightly correlated with CYP2B6 and CYP3A4 induction in HPH (r_pb = 0.38) than hCAR activation of the CYP2B6 reporter plasmid in the HepG2-based luciferase assay (r_pb = 0.14).

Fig. 2. Screening of FDA-approved drugs with the Ad/EYFP-hCAR translocation assay compared with the HepG2-CYP2B6-hCAR luciferase assay.

Each compound in the FDA-approved drug library was screened in two separate livers (four donors for the study). Overall, there were 86 hits in the translocation assay, compared with 66 agonists and 32 antagonists identified in the hCAR luciferase assay. Thirty translocators overlapped with the agonist mode, five overlapped with antagonists, and 51 were either inconclusive or inactive in the hCAR luciferase assay (A). Agonist efficacy (% of CITCO) in the hCAR agonist luciferase assay was plotted against average translocation (% of PB) from the hCAR translocation assay and split into agonist only (green triangles), agonist and translocator (yellow circle), and translocator only (grey diamond) groups (B). Literature data for CYP2B6 or CYP3A4 induction was overlaid onto B, with known inducers in red, known repressors in blue, and non-modulators in black (C). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1.

CYP2B6 induction in HPH by selected hits.^*

Compound	Luciferase Assay	Average Translocation	CYP2B6 mRNA Induction (Fold vs. Vehicle Control)
			HPH145	HPH147
Mosapride Citrate	active agonist	68.61%	5.16	13.73
Mitotane (Lysodren)	active agonist	54.23%	6.67	10.30
Loratadine	active agonist	70.01%	3.42	8.24
Aprepitant (MK-0869)	active antagonist	100.06%	2.03	1.55
Rimonabant (SR141716)	active antagonist	76.94%	0.45	0.27
Thioridazine HCl	inactive	56.12%	1.62	3.21
Sorafenib (Nexavar)	inactive	56.06%	0.34	0.23
Felodipine (Plendil)	inactive	53.27%	2.64	4.52
Mycophenolate mofetil (CellCept)	active antagonist	1.01%	0.70	1.26
Phenazopyridine	active agonist	5.00%	0.51	1.14

^*HPH145 and 147 were treated with novel hCAR activators that were hits in either the translocation assay or the luciferase assay at 10 μM for 24 h to determine CYP2B6 mRNA expression.

Table 4.

Summary of literature induction/repression data for hCAR translocators.

Compound	CAS #	Average Translocation	Induces CYP2B6 or 3A4?	Represses CYP2B6 or 3A4?	Reference
Nelfinavir Mesylate	159989–65-8	94.57%	✓		[70]
Pimozide	2062–78-4	75.88%	✕		[56]
Sulconazole Nitrate	61318–91-0	75.70%
Tolvaptan (OPC-41061)	150683–30-0	69.67%
Oxethazaine	126–27-2	63.76%
Econazole nitrate (Spectazole)	24169–02-6	61.64%	✓		[49]
Ritonavir	155213–67-5	57.96%	✓		[44,70,71]
Thioridazine HCl	130–61-0	56.12%	✓		*
Sorafenib (Nexavar)	475207–59-1	56.06%		✓	*
Nicardipine HCl	54527–84-3	54.73%	✓		[13,51]
Pyrithione zinc	13463–41-7	54.39%
Nifedipine (Adalat)	21829–25-4	53.50%	✓		[51,65]
Felodipine (Plendil)	72509–76-3	53.27%	✓		*
Diperodon HCl	537–12-2	50.97%
Azelnidipine	123524–52-7	42.85%
Clomipramine HCl	17321–77-6	34.87%
Tamoxifen Citrate (Nolvadex)	54965–24-1	24.28%	✓		[72]
Trimipramine Maleate	521–78-8	49.72%
Nitrendipine	39562–70-4	47.78%
Dibenzothiophene	132–65-0	44.87%
Altrenogest	850–52-2	44.36%
Darifenacin HBr	133099–07-7	42.80%
Bextra (valdecoxib)	181695–72-7	41.72%
Voriconazole	137234–62-9	41.21%	✓		[49]
Amitriptyline HCl	549–18-8	40.95%	✓		[56]
Benzydamine Hydrochloride	132–69-4	39.06%
Procyclidine HCl	1508–76-5	36.61%
Alverine Citrate	5560–59-8	34.91%
norethindrone	68–22-4	34.72%
Triflupromazine HCl	1098–60-8	29.72%
Clemastine Fumarate	14976–57-9	29.65%
Reboxetine mesylate	98769–84-7	28.72%
Levonorgestrel (Levonelle)	797–63-7	28.66%
Bromhexine HCl	611–75-6	27.57%
Dydrogesterone	152–62-5	26.70%
Pergolide mesylate	66104–23-2	26.61%	✕		[56]
Sertraline HCl	79559–97-0	25.77%	✓		[56]
Orphenadrine citrate (Norflex)	4682–36-4	25.59%
Imipramine HCl	113–52-0	23.65%	✓		[56]
Climbazole	38083–17-9	23.43%	✓		[49]
Drospirenone	67392–87-4	22.89%
Altretamine (Hexalen)	645–05-6	21.35%
Camylofin Chlorhydrate	54–30-8	20.59%
Clofoctol	37693–01-9	34.45%
Loxapine Succinate	27833–64-3	48.32%	✓		[56]
Itraconazole (Sporanox)	84625–61-6	45.44%
Phenytoin sodium (Dilantin)	630–93-3	42.62%	✓		[56]
AMG-073 HCl (Cinacalcet hydrochloride)	364782–34-3	41.73%
Trifluoperazine 2HCl	440–17-5	36.44%
Toremifene Citrate (Fareston, Acapodene)	89778–27-8	30.89%	✓		[73]
Clomifene citrate (Serophene)	50–41-9	28.20%

^*Indicates data from Table 1.

Interestingly, some compounds that were positive in hCAR translocation assay potently repressed CYP2B6 expression in HPH, such as rimonabant and sorafenib (Fig. 3A–D, Table 1). To determine whether it is possible for hCAR antagonists to translocate hCAR to the nucleus, we subjected a known hCAR antagonist, CINPA1, to the hCAR translocation assay. CINPA1 has been shown to bind to hCAR and repress its activity, while its metabolites exhibit either weak antagonism or no activity for hCAR [27,28]. Here, CINPA1 mediates potent nuclear translocation of hCAR in multiple HPH donors, demonstrating that antagonists can also lead to hCAR translocation while repressing its activity in the nucleus (Fig. 3E).

Fig. 3. hCAR antagonists can mediate hCAR nuclear translocation.

Antagonist efficacy (% of PK11195) in the hCAR antagonist luciferase assay was plotted against average translocation (% of PB) in the hCAR translocation assay and split into antagonist only (purple triangle), antagonist and translocator (pink circle), and translocator only (grey diamond) groups (B). Literature data for CYP2B6 or CYP3A4 induction was overlaid onto A, with known inducers in red, known repressors in blue, and non-modulators in black (B). Antagonist and translocator Rimonabant (C), and translocator sorafenib (D) translocate hCAR to the nucleus. Known hCAR antagonist CINPA1 also potently translocates hCAR to the nucleus of HPH in multiple liver donors (E). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.3. MOS induces CYP2B6 and CYP3A4 expression in HPH

One compound we found intriguing in the validation dataset was MOS, which was an agonist in the hCAR luciferase dataset and a hit in the hCAR translocation assay (Table 2). To validate MOS as a hCAR activator, we treated HPH with concentrations of MOS ranging from 100 nM to 10 μM. In HPH obtained from three liver donors MOS exhibited concentration-dependent induction of hCAR target genes CYP2B6 and CYP3A4 at both the mRNA and protein levels, confirming that MOS is an hCAR activator (Fig. 4).

Table 2.

Summary of literature induction data for hCAR agonists.

Compound	CAS #	AC50 (μM)	Efficacy (% of CITCO)	Average Translocation	Induces CYP2B6 or 3A4?	Reference
Clotrimazole (Canesten)	23593–75-1	13.33	20.50%	86.14%	✓	[44]
Artemisinin	63968–64-9	21.13	41.83%	82.90%	✓	[45,46]
Proadifen HCl	62–68-0	11.88	44.01%	79.24%
Butoconazole nitrate	64872–77-1	29.85	46.24%	75.71%
Famprofazone	22881–35-2	13.33	95.22%	72.13%	✓	[47]
Loratadine	79794–75-5	23.71	29.93%	70.01%	✓	*[48]
Tioconazole	65899–73-2	23.71	82.64%	68.89%	✓	[49]
Mosapride citrate	112885–42-4	1.68	46.41%	68.61%	✓	*[39]
Estrone	53–16-7	15.08	39.24%	64.25%	✓	[50]
Isradipine (Dynacirc)	75695–93-1	23.71	38.09%	59.61%	✓	[51]
Mestranol	72–33-3	27.31	42.84%	57.67%
Mitotane (Lysodren)	53–19-0	11.88	52.46%	54.23%	✓	*[52]
Diphenylpyraline HCl	132–18-3	16.79	41.58%	50.06%
Estradiol	50–28-2	10.87	25.50%	48.23%	✓	[50,53,54]
Bergapten	484–20-8	8.63	31.44%	44.79%
Nisoldipine (Sular)	63675–72-9	26.60	30.01%	43.63%
Bisacodyl	603–50-9	9.44	26.02%	43.30%
Phenytoin (Lepitoin)	57–41-0	61.13	17.39%	43.20%	✓	[44,55]
Quetiapine fumarate (Seroquel)	111974–72-2	29.85	21.46%	39.84%
Agomelatine	138112–76-2	23.71	24.10%	39.38%
Malotilate	59937–28-9	13.33	33.75%	38.89%
Ketoconazole	65277–42-1	15.36	30.62%	35.42%	✓	[56]
Lansoprazole	103577–45-3	11.88	48.99%	34.72%	✓	[57–59]
Etomidate	33125–97-2	29.85	29.22%	34.45%
Estradiol Cypionate	313–06-4	26.60	31.01%	28.91%	✓	[50]
Phenoxybenzamine HCl	63–92-3	27.58	118.77%	27.78%	✓	[56]
Bexarotene	153559–49-0	38.57	44.02%	26.33%
Progesterone (Prometrium)	57–83-0	48.26	21.31%	24.68%	✓	[53,56]
Methoxsalen (Oxsoralen)	298–81-7	61.50	38.41%	24.30%	✓	[60]
Terbinafine (Lamisil, Terbinex)	91161–71-6	21.13	22.15%	23.64%	✓	[48,49,61]
Flutamide (Eulexin)	13311–84-7	61.13	37.68%	19.88%
Nabumetone	42924–53-8	11.88	57.19%	14.78%
Albendazole (Albenza)	54965–21-8	6.86	30.88%	9.42%
Bifonazole	60628–96-8	23.71	29.07%	8.82%	✓	[49]
Cinchophen	132–60-5	21.13	22.12%	8.35%
Tretinoin (Aberela)	302–79-4	24.31	103.30%	8.24%
Omeprazole (Prilosec)	73590–58-6	60.39	45.78%	7.51%	✓	[59]
Nitazoxanide (Alinia, Annita)	55981–09-4	0.47	45.82%	7.51%	✕	[19]
Ethoxzolamide	452–35-7	7.50	39.58%	7.23%	✕	[56]
Thiabendazole	148–79-8	7.50	40.98%	6.60%	✓	[62,63]
Phenothrin	26002–80-2	4.86	58.61%	5.50%	✓	[19]
Rosiglitazone maleate	155141–29-0	60.76	20.87%	5.35%	✓	[64,65]
Isotretinoin	4759–48-2	27.39	98.47%	5.32%
Imatinib Mesylate	220127–57-1	13.33	47.50%	5.19%
Phenazopyridine HCl	136–40-3	5.96	79.52%	5.00%	✕	*
Salicylanilide	87–17-2	27.31	103.74%	4.48%
Fenofibrate (Tricor, Trilipix)	49562–28-9	54.48	84.05%	4.36%
Liothyronine Sodium	55–06-1	56.74	52.93%	4.25%
Riluzole (Rilutek)	1744–22-5	11.88	42.06%	2.93%	✕	[56]
Idoxuridine	54–42-2	11.88	30.12%	2.89%
Tolcapone	134308–13-7	33.49	15.84%	2.50%
Resveratrol	501–36-0	19.51	45.06%	2.44%	✕	[66]
Oxfendazole	53716–50-0	6.68	30.67%	2.10%
Febuxostat (Uloric)	144060–53-7	0.07	39.49%	2.07%
Genistein	446–72-0	13.33	55.19%	1.80%
Fenbendazole	43210–67-9	4.33	25.73%	1.75%
Flunixin meglumin	42461–84-7	26.60	27.36%	1.52%
Reserpine	50–55-5	44.50	44.25%	1.38%
Niflumic acid	4394–00-7	29.85	23.45%	1.33%
Leflunomide	75706–12-6	1.06	69.71%	0.79%	✓	[49,67]
Sulindac (Clinoril)	38194–50-2	61.13	58.38%	0.59%
Zileuton	111406–87-2	38.57	21.78%	0.37%
Tolnaftate	2398–96-1	27.31	59.00%	0.23%	✓	[13]
Axitinib	319460–85-0	1.06	105.76%	−0.14%	✕	[19]
Monobenzone (Benoquin)	103–16-2	23.82	50.92%	−0.48%
Daidzein	486–66-8	15.20	41.17%	−1.59%

^*Indicates data from Table 1.

Fig. 4. MOS potently induces CYP2B6 expression over a range of concentrations in HPH.

HPH from liver donors #147 (A), #148 (B), and #149 (C) were treated with 0.1% DMSO, 1 mM PB, 1 µM CITCO, 10 µM RIF, or MOS at 0.1, 0.5, 1, 5, or 10 µM for 24 h or 48 h and harvested for mRNA or protein expression analysis, respectively. Results are expressed as fold over control, mean ± S.D. (n = 3); ^{*, #} P < 0.05, **^{, ##} P < 0.01.

3.4. MOS inhibits the expression of gluconeogenic genes and reduces glucose output from HPH

While multiple mechanisms for the reduction of blood glucose levels by MOS have been proposed, we sought to investigate whether MOS could act on the liver to affect glucose levels. Therefore, we investigated the effect of MOS on the expression of gluconeogenic genes G6Pase and PEPCK with or without stimulation of cAMP in HPH. MOS treatment of cAMP-stimulated HPH at 5 and 10 μM led to the repression of G6Pase and PEPCK expression in two separate livers (Fig. 5). In addition, MOS was able to repress the protein expression of PGC-1α, a key regulator of gluconeogenesis, in HPH151 at 10 μM. Further experiments demonstrated that MOS was able to significantly decrease the glucose output of cAMP-stimulated HPH at both 5 and 24 h time points in a concentration-dependent manner (Fig. 6), indicating that MOS may be able to reduce glucose output from the liver under fasting conditions.

Fig. 5. MOS represses gluconeogenic gene expression in HPH.

HPH from liver donors #150 (A, C) or #151 (B, D) were treated with 0.1% DMSO, 1 µM CITCO, or MOS at 1, 5, or 10 µM in ITS + and dexamethasone-free media with or without 10 µM dibutyl cAMP and 50 nM dexamethasone to induce gluconeogenic gene expression. The mRNA expression of G6Pase (A, B) and PEPCK (C, D) were measured after 24 h treatment; the protein expression of PGC-1α was probed in HPH151 after 48 h under the same experimental conditions (E). Data are expressed as fold over control, mean ± S.D. (n = 3); * P < 0.05; ** P < 0.01.

Fig. 6. MOS decreases glucose output from HPH.

HPH from liver donors #149 (A, C) or #151 (B, D) were treated with 0.1% DMSO, 1 µM CITCO, or MOS at 1, 5, or 10 µM in ITS + and dexamethasone-free media overnight, incubated in glucose-free media with treatments with or without 10 µM dibutyl cAMP and 50 nM dexamethasone for 2 h to deplete intracellular glucose and glycogen, and incubated for 5 h (A, B) or 24 h (C, D) before taking media samples and measuring glucose content. Data represent the mean ± S.D. (n = 3); * P < 0.05; ** P < 0.01.

4. Discussion

In the current study, we have optimized the hCAR translocation assay for high-content screening, evaluated its screening efficacy in comparison to the HepG2-based hCAR luciferase assays, and identified MOS as a novel hCAR activator that represses gluconeogenesis in HPH, a previously unidentified mechanism that could contribute to the clinically observed glucose-lowering effects of MOS.

High-throughput luciferase reporter assays in immortalized cells have been used broadly for identification of nuclear receptor modulators, but have also been associated with difficulties translating data to physiologically-relevant systems. One of the major differences between immortalized cell lines and physiologically-relevant liver cells, such as HPH, is their capacity for drug metabolism. As most compounds are metabolized in the body, this is a major translational issue with HTS assays because a change in the chemical structure of a compound will alter its modulation of xenobiotic receptors. For instance, a metabolite of the hCAR antagonist PK11195, N-desmethyl-PK11195, was identified as an hCAR activator despite only one methyl group being different between the two compounds [29]. Therefore, a metabolism-competent screening system in physiologically-relevant cells would be beneficial in identifying hCAR modulators that may have clinical significance.

While hCAR is constitutively active and localized to the nucleus of immortalized cell lines, it is sequestered in the cytoplasm of primary hepatocytes, and translocates to the nucleus upon activation [8]. The development of Ad/EYFP-hCAR allowed for fluorescent hCAR to be efficiently expressed in HPH and easily visualized with microscopy [20]. Over multiple studies, the hCAR translocation assay in HPH has been able to accurately predict whether a compound would activate hCAR and induce CYP2B6 expression in HPH in a relatively low-throughput fashion. Therefore, we sought to expand this assay into a high-throughput format and determine whether FDA-approved drugs would activate hCAR and induce CYP2B6 expression in HPH. With the advances in high-content microscopy and analysis, we were able to expand this assay to a 96-well plate format, validate an analysis method to characterize hCAR translocation, and identify FDA-approved drugs as novel hCAR modulators. By comparing our hCAR translocation results with the data from the HepG2-CYP2B6-hCAR screen of the Tox21 10 K compound library by the National Center for Advancing Translational Sciences (NCATS), we were able to better understand the utility of both assays [19]. While we can be relatively certain that a compound alters hCAR activity in some way if it can alter hCAR localization, it is harder to fully validate hCAR activity at the functional level. To validate our dataset, we undertook a literature search for modulation of hCAR target genes CYP2B6 and CYP3A4 in HPH by compounds in our combined dataset. While induction of CYP2B6 and CYP3A4 is not a perfect, specific indicator of hCAR activation, as both genes can be influenced by other factors such as PXR activation, CYP2B6 and CYP3A4 are the most widely studied target genes of hCAR. This enables the collection of data for a greater number of compounds since performing hCAR functional studies for each compound would be costly. Combining our literature search with our own CYP2B6 induction data, we found that hCAR translocation was more highly correlated (r_pb = 0.38) with induction of CYP2B6 or CYP3A4 than the luciferase agonist assay (r_pb = 0.14). Clearly, combining data from these two separate hCAR assays can improve their predictive power and increase confidence in the physiological-relevance of each compound tested.

Notably, we found that five out of the 86 compounds that translocated hCAR were also antagonists in the hCAR luciferase assay. While two of the five overlapping compounds were found to induce hCAR target genes in HPH, rimonabant and sorafenib exhibited repression of CYP2B6 expression in HPH. Many studies over the years have posited that hCAR translocation, from the cytoplasm to the nucleus of hepatocytes, is the first, essential step of its activation; given that hCAR must be in the nucleus to regulate the transcription of its target genes, this theory is convincing. While CYP2B6 repression may not be the best indicator of reduced hCAR activity, several lines of evidence suggest that hCAR can be translocated to the nucleus not only by agonists, but by antagonists as well. Two potent antagonists of hCAR, PK11195 and CINPA1, both mediate hCAR nuclear translocation [30] (Fig. 3E). While PK11195 is known to be metabolized into a hCAR activator in HPH, potentially explaining its translocation, CINPA1 and its metabolites do not activate hCAR but mediate hCAR nuclear translocation nonetheless [27–29]. Our data suggests that hCAR translocation is not limited to hCAR agonists and may also occur upon stimulation with antagonists, which is important to note when drawing conclusions from translocation data. In addition, previous studies have shown that mouse CAR antagonist okadaic acid can prevent the dephosphorylation and nuclear translocation of CAR, leading to repression of Cyp2b10 expression, while KN-62 represses Cyp2b10 expression in mouse primary hepatocytes but does not inhibit CAR translocation to the nucleus upon activation [31,32]. Taken together, these studies suggest that there are several types of CAR antagonists that can deactivate CAR at different stages of its activation and is an intriguing area for future study.

One hCAR activator in our validation dataset that piqued our interest was MOS, a 5-Hydroxytryptamine receptor 4 receptor agonist used to treat gastrointestinal ailments such as dyspepsia, gastritis, and irritable bowel syndrome by increasing gastric emptying and gastrointestinal motility [33]. Widely used to treat different forms of gastropathy in Asian countries such as Japan and Korea, MOS has also shown potential to improve glycemic control in subjects who had impaired glucose tolerances or Type 2 diabetes. In several clinical studies, MOS was able to lower fasting glucose levels after a single dose or with repeated dosing over 1–2 weeks [21,22,34–36]. Multiple mechanisms for this decrease in glucose levels have been proposed, including stimulation of insulin release due to increased gastrointestinal motility and improvement of insulin sensitivity [21,35–38]. Nevertheless, the role of the liver in the glucose-lowering capability of MOS is largely unknown. Given that CAR activation in mice leads to a reduction in fasting blood glucose levels by reducing hepatic gluconeogenesis, we sought to further characterize the activation of hCAR by MOS in HPH and investigate whether MOS can reduce gluconeogenic gene expression and glucose output in HPH.

In agreement with a previous report, MOS demonstrated concentration-dependent induction of CYP2B6 and CYP3A4 in HPH [39]. Notably, our data showed for the first time that MOS was able to repress the mRNA expression of G6Pase and PEPCK, two major proteins involved in gluconeogenesis, and to reduce glucose output from HPH. CAR activation is known to repress gluconeogenic gene expression by facilitating the degradation of PGC-1α, a master regulator of gluconeogenesis [40], and we found that MOS also represses PGC-1α protein expression which may contribute the decreased gluconeogenesis observed in HPH treated with MOS. Clinically, the C_max of MOS in humans can reach 700 nM after oral administration of 40 mg MOS [41]. Although this plasma level of MOS is at the lower range of the concentrations used in the current study, it is known that orally-administered drugs can achieve a much higher liver concentration than that of serum [42,43]. Thus, our current in vitro findings may provide a possible liver-specific mechanism by which MOS reduces fasting blood glucose levels (Fig. 7). If this mechanism also occurs in vivo, it would provide a rationale for clinical use of hCAR activators for metabolic disorders.

Fig. 7. Proposed mechanism for MOS-dependent improvement in glycemic control.

While MOS is known to function by increasing gastric motility in the gut and increasing insulin secretion from the pancreas, our study demonstrates that MOS represses gluconeogenesis, providing an additional, liver-specific mechanism by which MOS alters glucose homeostasis.

Although the high-throughput hCAR nuclear translocation assay was robust and able to identify hCAR modulators, there were several limitations to our study. Due to the expense associated with using HPH, we were only able to screen the chemical library at a single concentration (10 μM) in two livers. Therefore, compounds with higher EC₅₀ values may have been missed in the translocation assay while being correctly identified in the luciferase assay, as the maximum compound concentration used in the luciferase screen was 92 μM. We also decided to consider translocation on a per-cell basis and limit translocation to a yes/no; this method provided a robust analysis for detecting true translocators but may have missed compounds which weakly translocate hCAR. Overall, we were able to optimize the Ad/EYFP-hCAR nuclear translocation assay into a high-throughput format, validate the efficacy of the nuclear translocation assay, and identify MOS as a novel hCAR activator that represses gluconeogenesis in HPH, providing a liver-specific mechanism for mosapride-dependent decreases in fasting blood glucose.

Table 3.

Summary of literature induction/repression data for hCAR antagonists.

Compound	CAS #	AC50 (μM)	Efficacy (% of PK11195)	Average Translocation	Induces CYP2B6 or 3A4?	Represses CYP2B6 or 3A4?	Reference
Aprepitant (MK-0869)	23593-75-1	13.33	−87.72%	100.06%	✓		* [65,68,69]
Ethinyl Estradiol	63968-64-9	21.13	−88.13%	80.84%
Rimonabant (SR141716)	62-68-0	11.88	−58.76%	76.94%		✓	*
Diethylstilbestrol (Stilbestrol)	64872-77-1	29.85	−74.19%	75.65%	✓		[16]
Bexarotene	22881-35-2	13.33	−62.01%	26.33%
Adapalene	79794-75-5	23.71	−58.80%	15.59%
Bortezomib (Velcade)	65899-73-2	23.71	−105.84%	15.36%		✓	[14]
Daunorubicin HCl (Daunomycin HCl)	112885-42-4	1.68	−74.75%	11.48%		✓	[14]
Idarubicin HCl	53-16-7	15.08	−120.64%	10.91%
Doxorubicin (Adriamycin)	75695-93-1	23.71	−99.21%	8.77%
Vinorelbine Tartrate	72-33-3	27.31	−117.01%	8.09%
Carmofur	53-19-0	11.88	−85.12%	6.95%
Floxuridine	132-18-3	16.79	−31.85%	6.70%
Zidovudine (Retrovir)	50-28-2	10.87	−39.53%	5.00%
Abitrexate (Methotrexate)	484-20-8	8.63	−77.37%	4.20%	✕		[44]
Clofarabine	63675-72-9	26.60	−60.93%	3.74%
Trifluridine (Viroptic)	603-50-9	9.44	−43.93%	3.45%	✓		[14]
Docetaxel (Taxotere)	57-41-0	61.13	−52.77%	3.08%
Etoposide (VP-16)	111974-72-2	29.85	−79.70%	2.70%
Tenofovir Disoproxil Fumarate	138112-76-2	23.71	−92.48%	2.45%
Pyrimethamine	59937-28-9	13.33	−55.68%	2.39%
Cladribine	65277-42-1	15.36	−75.45%	2.20%
Adrucil (Fluorouracil)	103577-45-3	11.88	−81.01%	1.60%
Doxifluridine	33125-97-2	29.85	−62.82%	1.16%
Mycophenolate mofetil (CellCept)	313-06-4	26.60	−53.41%	1.01%	✕		*
Mycophenolic (Mycophenolate)	63-92-3	27.58	−49.95%	0.80%
Ribavirin (Copegus)	153559-49-0	38.57	−46.51%	0.68%
Fludarabine (Fludara)	57-83-0	48.26	−55.68%	−2.86%
Topotecan HCl	298-81-7	61.50	−107.71%	−2.87%		✓	[14]
Adefovir Dipivoxil (Preveon,	91161-71-6	21.13	−104.76%	−3.00%		✓	[14]
Hepsera)
Ethacridine lactate monohydrate	13311-84-7	61.13	−132.72%	−4.69%
Mitoxantrone Hydrochloride	42924-53-8	11.88	−116.45%	−13.32%

^*Indicates data from Table 1.

Abbreviations:

Ad/EYFP-hCAR

adenovirus-expressing enhanced yellow fluorescent protein-tagged human CAR

CAR

constitutive androstane receptor

CINPA1

CAR inhibitor not PXR activator 1

CITCO

6-(4-Chlorophenyl)imidazo[2,1-b][1,3]thiazole-5-carbaldehyde-O-(3,4-dichlorobenzyl)oxime

cAMP

cyclic adenosine monophosphate

DAPI

4′,6-diamidino-2-phenylindole

DDIs

drug-drug interactions

DEX

dexamethasone

DMEs

drug-metabolizing enzymes

DMSO

dimethyl sulfoxide

G6Pase

glucose-6-phosphatase

HPH

human primary hepatocytes

MOS

mosapride citrate

PB

phenobarbital

PEPCK

phosphoenolpyruvate carboxykinase

PK11195

1-(2-Chlorophenyl)-N-methyl-N-(1-methylpropyl)-3-isoquinolinecarboxamide

PXR

pregnane X receptor

RXR

retinoid x receptor

RIF

rifampicin