CAR: Discovery and Development by the Pharmacogenetics Laboratory at NIEHS, NIH

Abstract

Our studies on the molecular mechanisms of phenobarbital induction of hepatic drug metabolism began in 1995 at the Pharmacogenetics Laboratory, National Institute of Environmental Health Sciences, NIH. Almost all pharmacology textbooks at the time described the phenomenon, but the mechanisms remained unknown. Within a few years, we revealed that the nuclear receptor CAR, constitutive active/androstane receptor, is the key molecule in these mechanisms. Since then, CAR research has branched into various directions. CAR is involved in controlling energy metabolism and has been shown to be an essential factor in promoting hepatocellular carcinoma by phenobarbital. It also functionally and physically interacts with other nuclear receptors and factors in cell signaling pathways, controlling many genes and biological responses. Research into CAR phosphorylation has led to new insights into other nuclear receptors such as PXR, RXRα, RORα, ERα, and AR. In this review, we summarize the evolution of CAR research within our group, describing background information and key findings for us to move to new directions. Furthermore, we will illustrate currently unanswered questions, hoping our discussions will inspire scientists with fresh ideas.

Graphical Abstract

1. Introduction

The Pharmacogenetics Laboratory was established at National Institute of Environmental Health (NIEHS), National Institutes of Health (NIH) on July 10, 1983. Soon, it began to investigate phenobarbital (PB) induction of hepatic drug metabolism and determine its induction. First, possible strain differences amongst inbred mice with respect to their responses to PB to induce hepatic CYP2B10, the classic PB target gene, were analyzed. However, despite our hopes, no strain differences were found, ending our effort to investigate the PB induction mechanism.

Our efforts to determine the molecular mechanism of PB induction lay dormant for nearly 10 years until Paavo Honkakoski joined our laboratory and undertook it as his major research aim in 1995. In fact, with the assistance of many hard-working coworkers, this led him to characterize a nuclear receptor. Hereafter, our short history of how CAR was discovered, and how it is activated by PB will be described. Moreover, how CAR, originally characterized as drug activated nuclear receptor, was rediscovered for its roles in regulating physiological functions such as energy metabolism and pathophysiology such as tumor development, will also be discussed. The chronology of our findings on CAR is summarized in Table 1.

Table 1.

Key findings on CAR biology in Pharmacogenetics laboratory at NIEHS

Year	Findings	References
1996 – 2001	PB response DNA elements in Cyp2b10 and Ugt1a1	1, 2, 4, 5, 7
1998 – 1999	Primary function of CAR in Cyp2b10 gene induction by PB	1, 4
1999	CAR nuclear translocation by PB in liver	10
1999 – 2009	CAR phosphorylation at Thr 38 regulates CAR nuclear translocation	10, 12
1999 – 2013	EGF signaling pathway and PP2A in CAR nuclear translocation	10, 11, 16, 21, 23
2002 – 2017	CAR and PXR in liver energy metabolism	9, 31, 35 – 37, 39
2002 – 2023	CAR in PB promoted hepatocellular carcinoma	9, 44, 48 - 52
2005 – 2013	CAR function in liver physiology	60 - 62
2008 – 2014	CAR small molecule modulators	68 -70
2009 – 2020	Conserved phosphorylation sites in nuclear receptor DBD (CAR Thr38 and equivalent residues in other nuclear receptors) and their functions	12, 25, 26, 43, 75 – 77, 87
2015 – 2020	Glucose signal and PXR reciprocal regulation	38 - 40
2016	CAR and diabetes	34
2020 – 2023	Conserved phosphorylation sites in nuclear receptor LBD (PXR Ser350 and equivalent residues in other nuclear receptors) and their functions	40, 41, 88, 89
2016 – 2023	Heterocomplex among nuclear receptors	73, 74

2. Road to CAR

2.1. Prologue

Numerous so-called orphan nuclear receptors were cloned around 1990. This ignited a fever to identify their activator ligands and functions, a movement of reverse biology.

Contrary to this trend, our road began with investigating hepatic functions that induce drug metabolizing enzymes for the sedative drug phenobarbital. The experimental methods used were conventional biochemistry combined with molecular biology techniques.

2.2. Phenobarbital response enhancer module (PBREM)

PBREM is a 51bp DNA sequence first identified in the promoter of a mouse gene to which CAR binds and activates in response to PB (1). When analysis of the promoter began in the early 90’s there was a major experimental obstacle that prevented us from analyzing them: no cell systems which respond to PB had been established. Rat primary hepatocytes were readily used in various experiments at that time. However, mouse counterparts were poorly developed. With Rick Moore, Paavo spent years establishing mouse primary hepatocytes with PB response capability to not only endogenous genes but also to transfected reporter genes as well.

In an initial analysis conducted in 1996, a DNA sequence −1404/−971 exhibited PB response enhancer activity in chloramphenicol acetyl transferase (CAT) based reporter assays(2). Within this region, DNAase footprint assays detected several sensitive regions, one of which contains AGGTCA, which is the half site of a nuclear receptor binding motif. In 1995, a 163 bp sequence (−2318/−2155 bp) located in the rat CYP2B2 promoter was reported to be activated in a CAT reporter assay in response to PB and was subsequently named the PB response element (PBRE)(3). Curiously, the NR half site found in the mouse CYP2B10 promoter is conserved in the PBRE sequence. The 1995 report focused its discussion on NF1 transcription factor, but this half site was not mentioned.

The half site of CYP2B10 promoter responded to PB, but its activity was much lower than the PBRE. Moreover, the DNAase sensitive region was open regardless PB treatment. These facts were of serious concern to us. For this reason, and because the rat’s 163 bp sequence is conserved at a corresponding region as a 132 bp of the CYP2B10 promoter, our research effort was changed from the −1kbp to −2kbp region. PB response activity of 132bp sequence (−2346/−2265) was delineated to a 51bp sequence: TCTGTACTTTCCTGACCTTGGCACAGTGCCACCATCAACTTGCCTGACACC. This sequence is comprised of 2 nuclear receptor binding motifs; DR4 flanked by a NF1 binding site. The 51 bp sequence was named PB response enhancer module, PBREM and the 5’and 3’-DR-4 motifs were named as NR1 and NR2, respectively (Fig. 1) (4). PBREM is conserved in rat and human CYP2b genes as well as in the UDP glucuronosyl transferase UGT1A1 gene (5). In addition, DR-4 motif has been recognized as a core element of CAR binding and its PB response enhancer activity in numerous hepatic genes.

Figure 1. Conserved PBREM sequence.

The 51 bp PBREM sequences of mouse (Cyp2b10, Cyp2b9), rat (CYP2B1), and human (CYP2B6) genes are aligned. Arrows indicate NR1 and NR2 direct repeat motifs. The UGT1A1 gtNR1 response element is also aligned at the top. Conserved bases are shown with (+). Cyp2b10 NR1 and Cyp2b9 NR1* are used for affinity purification of CAR and are indicated by red characters.

2.3. CAR, PB activated nuclear receptor

Having identified PBREM, the search to identify nuclear protein factors which can bind and activate began. To this end, two different experimental strategies were taken. Given the fact that PBREM contains nuclear receptor binding sites and PB induction is relatively liver specific activity, Paavo collected cDNAs encoding liver specific nuclear receptors. These cDNAs were subjected to CAT based reporter assays in mouse primary hepatocytes. Alternatively, Igor Zelko performed DNA affinity chromatography using NR1 oligonucleotides as bait to purify proteins from mouse liver nuclear extracts.

MB67 was one of these nuclear receptors and the cDNA encoding its human counterpart was kindly provided by Dr. David Moore. MB67 was an orphan nuclear receptor that binds to retinoic acid response elements and weakly activates them, constitutively without an activating ligand (6). Fortuitously, MB67 and its mouse cDNA cloned by the Pharmacogenetics Laboratory were soon found to activate PBREM in a PB responsive manner (1). Including PB, 16 known CYP2B10 inducers activated PBREM in mouse primary hepatocytes (7). Due to its intrinsic nature, CAR, Constitutive Active Receptor, replaced MB67 as its name. CAR belongs to the nuclear subfamily 1I, designated NR1I3. Later, the name Constitutive Androstane Receptor was bequeathed to CAR (8), and has been used by peers since.

Purification was strategized by taking the fact that the CYP2B9 gene is not induced by PB (4). Its promoter contains a mutated NR1 within PBREM called NR1* (Fig. 1). Hypothesizing that NR1* does not respond to PB, NR1* oligonucleotides were used as a negative bait to which no proteins bind in response to PB. Nuclear extracts were prepared from non-treated and PB treated mice and chromatographed through NR1 or NR1* conjugated beads. Eluates from each chromatograph were analyzed by SDS polyacrylamide gels and Western blots. Several nuclear proteins were specifically enriched in the eluate from the NR1 but not NR1* in only the PB treated nuclear extracts. CAR and RXRα were found to be two proteins among them. Thus, both the reporter assays with mouse primary hepatocytes and DNA affinity purification provided us with strong evidence that the PB induced nuclear factors activating PBREM is a CAR-RXRα heterodimer. Subsequently CAR KO mice, provided by Dr. Jurgen M. Lehmann, were treated with PB, from which liver RNAs were prepared. cDNA microarray analyses showed that, as expected, the CYP2B10 gene and various other genes encoding for drug metabolizing enzymes were upregulated (9). These findings became historic in the annals of PB actions and opened CAR biology to a new frontier.

3. CAR activation mechanism

Soon after the investigation began, we encountered serious experimental obstacles which had to be resolved for us to pursue our research aim. One obstacle was that in cell-based assays, CAR is constitutively activated. This activity is repressed in primary hepatocytes and liver. A second obstacle is that PB is not a CAR ligand; PB activates CAR indirectly without direct binding. These difficulties provided us with a challenge to develop novel experimental avenues and discoveries.

3.1. Phosphorylation

Rather quickly, it was found that CAR repressed its transcription activity through sequestration in the cytoplasm. In response to PB, CAR translocated from the cytoplasm to the nucleus. Okadaic acid, a protein phosphatase inhibitor repressed CAR nuclear translocation (10). This inhibition suggested that CAR may be phosphorylated. In response to PB, CAR is dephosphorylated for its nuclear translocation. To support this hypothesis, CAR was purified as a complex with protein phosphatase 2A from the liver cytosols of PB treated mice (11). Thus, sequestering phosphorylated CAR in the cytoplasm is the repression mechanism of its constitutive activity. This repression allows PB to derepress / activate CAR in the liver. While work was going smoothly up to this point, we did not expect that it would take us more than 10 tedious years to find a phosphorylated residue and define the regulatory mechanism by which PB elicits dephosphorylation indirectly to activate CAR (Fig.2).

Figure 2. Phosphorylation retains CAR homodimer in the cytoplasm.

CAR forms the homodimer through an interface constituted of three loops between helices 3 and 4, 6 and 7, and 10 and 11. Phosphorylation stabilized this homodimer sequestering CAR in the cytoplasm. In response to PB, the homodimer is dephosphorylated and dissociated into monomers. The DBD undergoes a conformational change and hides the phosphorylation motif through interaction with the LBD

3.1.1. Threonine 38

Threonine 38 (Thr38) is located on the a helix between the 1^st and 2^nd zinc fingers within the DNA binding domain (DBD) of human CAR and constitutes a putative phosphorylation motif of protein kinase C (12). Its corresponding residues were reported to be a motif in various nuclear receptors such as HNF4α and VDR (13–15). Their phosphomimic mutation to aspartic acid altered their transcription activities. Although it was suggested that this motif could be phosphorylated, no experimental evidence for its phosphorylation in in vivo tissues was reported until Thr38 of CAR was shown to be phosphorylated in mouse primary hepatocytes and as well as in human primary hepatocytes (16, 17). To this end, an antibody against a peptide including phosphorylated Thr38 from the CAR molecule was produced. This antibody was used in Western blots to analyze CAR in mouse primary hepatocytes. The obtained results clearly showed that CAR is phosphorylated and dephosphorylated in response to PB. Moreover, this dephosphorylation was inhibited by okadaic acid. By injecting expression plasmids through the tail vein, CAR Thr38Asp mutant was directly expressed in mouse livers (12, 16). Consistent with phosphorylation in primary hepatocytes, the mutant is expressed and retained in the cytoplasm even after PB treatment. With these findings, Thr38 was determined to be phosphorylated residue that regulates CAR activation by PB in the liver.

3.1.2. PB crossover cell growth signal

Retrospectively, there were sporadic reports that PB and cell growth signals such as insulin and epidermal growth factor (EGF) antagonize their signals (18, 19).

Karen I. Hirsch-Ernst’s group demonstrated that EGF represses PB-induced activation of promoter in rat primary hepatocytes (20). This repression occurred at the NR1 within the PBREM of CYP2B1 promoter by preventing a PB induced binding of a nuclear protein. Although this nuclear protein could have been CAR, no experimental support was presented. Therefore, the EGF repression was reexamined in mouse primary hepatocytes; EGF was found to repress PB induced CAR nuclear translocation (21). Moreover, an ERK1 inhibitor(U0126) abrogated the EGF-mediated repression. Thus, EGF transduces its signal through ERK1 to repress the nuclear translocation of CAR. Insulin also repressed CAR nuclear through ERK1. In the opposite direction, CAR spontaneously translocated in the nucleus to activate the CYP2B10 gene in the liver of diabetic mice (22).

3.2. Epilogue

CAR repressed its constitutive activity through sequestration in its phosphorylated form in the cytoplasm. PB antagonizes an EGF signal to elicit dephosphorylation to translocate CAR into the nucleus. With this knowledge in hand, efforts were continued to define the molecular process through which PB activates CAR. Together, with experimental information obtained by many colleagues, Shingo Mutoh, Ryota Shizu and Mack Sobhany carried out key experiments enabling us to connect the dots to finally define the activation process(Fig.3).

Figure 3. PB signaling via EGFR.

EGF binds EGFR activating a signal to phosphorylate ERK1/2 . Phosphorylated /activated ERK1/2 binds the phosphorylated CAR homodimer, preventing it from being dephosphorylated. EGF also activates a signal to SRC kinase that phosphorylates RACK1 at Tyr71. PB directly binds EGFR to repress these ERK1/2 and SRC signals. As a result, the ERK1/2 dissociates from the phosphorylated CAR homodimer. This enables RACK1 and PP2A to bind the homodimer and dephosphorylate Thr38 for CAR monomerization and nuclear translocation. Black arrows indicate the flow of EGF signals. Red arrows and stop arrows show PB signaling.

Isothermal titration was utilized to detect the direct binding of PB to EGF receptor. There are five different sites to which PB binds with an average dissociation constant of 12 μM. Computer dynamic simulations located these binding sites, one of which is the EGF binding site. In addition, PB repressed an EGF induced phosphorylation of Try845 and Try1173 of the EGF receptor in mouse primary hepatocytes. It was concluded that EGF receptor is the initial binding site of PB (23).

This binding converts the cell growth signal to a PB signal. The PB signal is transduced to a phosphorylated CAR homodimer in the cytoplasm. CAR was crystallized in the form of a homodimer (24). The homodimer uses an interface on the opposite side of the CAR molecule from the interface for heterodimerizing with RXRα (Helix 10). Analysis of CAR Thr38Asp in solution showed that it forms the homodimer using the same interface observed in the crystal structure (25). Recombinant CAR Thr38Asp and CAR Thr38Ala proteins were applied on gel filtration columns: the former is eluted as a homodimer, while the latter as a monomer. Therefore, it was concluded that PB signal dephosphorylates the homodimer and converts it to a monomer for nuclear translocation (26).

RACK1, Receptor for activated C kinase, is essential for PP2A to dephosphorylate Thr38 of CAR (23). This was first supported by the finding that CAR is not dephosphorylated in RACK 1 knockdown mouse primary hepatocytes, as well as in vitro dephosphorylation assays. Co-immunoprecipitation assays (CoIP) using CAR and RACK1 proteins ectopically expressed in Huh 7 cells showed that CAR Thr38Asp but not its Ala counterpart, formed the triple complex with endogenous PP2A proteins. Additional CoIP experiments suggested that an unphosphorylated RACK1 directly binds the phosphorylated CAR. As it appeared, phosphorylated CAR, acting as the substrate interacts with PP2A RACK1.

As to the possibility of PB signal transducing to RACK1, RACK1 decreases its phosphorylation levels at Try52 in PB treated mouse livers. In support, Tyr52 phosphorylation was catalyzed by SRC kinase in an in vitro kinase assay. Because SRC is one of the EGF regulated EGFR downstream signals, PB can repress this SRC S-mediated phosphorylation. This allows RACK1 to activate PP2A. Thus, one of the PB signals can be transduced to RACK1 (23). In addition to this SRC signal, PB-ERK1 signal is essential for CAR dephosphorylation and nuclear translocation (21). Where and how the ERK1 signal is transduced remains a question for future investigations. It can be, however, hypothesized that the ERK1 signal may play a role in recruiting PP2A, and RACK1 to CAR. At least these two different PB induced signals through ERK1 and SRC need to be converged on phosphorylated CAR to form the complex for its activation.

Including the years needed to identify CAR as a PB activating nuclear receptor, nearly 20 years had passed while we determined the PB induced signal mechanism by which PB indirectly activates CAR in the liver in 2013. This long-lasting research project constantly provided us with excitement, courage, and satisfaction along with the occasional disappointment, and the joy of accomplishment. Most thankfully besides bread and butter, the project provided the laboratory with the opportunity to work with over 100 young fellows and students and to make many friends worldwide.

4. Energy metabolism

4.1. CAR leads the way to energy

Three years after we showed that CAR is a key factor in PB induction of CYP2B genes, our group obtained pioneering results that implied the diversity of CAR functions. Using an in-house DNA array available at NIEHS at the time, we compared the changes in gene expression 12 hours after intraperitoneal administration of PB in CAR-deficient and wild-type mice (9). Although the number of genes in this DNA array was much smaller than those on currently available arrays, the results were amazingly provocative.

This array study irrefutably showed that CAR is essential for the induction of drug metabolism-related genes such as Cyp2b10 and Cyp3a11. Furthermore, we found multiple genes involved in energy metabolism, including Pepck1 which is involved in gluconeogenesis and Cpt1 in lipid beta oxidation, are only repressed by PB in the wild type mice.

Meanwhile, our group was simultaneously working to identify proteins that bind to CAR. This project was started with the main goal of finding proteins involved in the nuclear translocation of CAR followed by its activation. Among many candidate CAR interacting proteins, most of which we couldn’t publish, we found FOXO1 through yeast two-hybrid screening. It seems this transcription factor may not be involved in CAR nuclear translocation. However, we found a report that shows FOXO1 and PGC1A controlled gluconeogenesis almost concurrently (27). According to this report, FOXO1 co-activates PGC1A for the activation of genes responsible for gluconeogenesis, and this co-activation is inhibited by phosphorylation of FOXO1 by a signal from insulin. Another report that caught our attention showed that CAR binds to PGC1 (28). These findings and reports opened a new research direction for CAR. We hypothesized that gluconeogenesis is one of the physiological pathways regulated by the interaction of FOXO1 with CAR and this mechanism can explain published findings showing phenobarbital treatment decreased blood glucose levels in rodents and humans (29, 30).

Results using hepatoma cell lines expressing CAR constitutively and CAR-deficient mice showed that activated CAR interacts directly with FOXO1. CAR acts as a co-repressor of FOXO1, preventing FOXO1 from coactivating insulin-responsive genes such as PEPCK1 (31). This mechanism explains not only the gene expression changes in our DNA microarray results but is also consistent with results published by other groups in which genes involved in gluconeogenesis are reduced when CAR is activated with PB or TCPOBOP (1,4-Bis[2-(3,5-dichloropyridyloxy)]benzene) (32). Furthermore, a later study showed that activated CAR promotes PGC1 degradation, and the reduced PGC1 did not activate gluconeogenesis-related genes during fasting (33). Taken together, the pivotal role of CAR acting with FOXO1 and PGC1 in gluconeogenesis was established.

Since insulin is known to stimulate phosphorylation of FOXO1 and inactivate it, activation of CAR by PB appears to mimic the action of insulin. Both activated CAR and insulin share FOXO1 as a target and suppress gluconeogenesis. Subsequent studies have revealed the mechanism by which PB inhibits insulin action more directly as an antagonist of insulin binding to its receptor (34). As a result, PB, acting as an insulin receptor antagonist, caused CAR-independent blood glucose level upregulations. After this early stage, a CAR-dependent decrease in blood glucose levels will follow. Conversely, insulin receptor activation by insulin controls CAR activity, weakening the induction of drug metabolizing enzymes such as Cyp2b. PB and insulin interact through the insulin receptor and mutually regulate glucose and drug metabolism. These findings suggest that CAR activation may not be beneficial only for reducing hepatic gluconeogenesis but for suppressing glucose metabolism-related diseases such as type 2 diabetes and fatty liver.

4.2. PXR follows CAR

Our group has revealed that PXR, the nuclear receptor most closely related to CAR, is also involved in hepatic energy metabolism and has elucidated the details of the molecular mechanism (35–37). To clarify the role of PXR in lipid metabolism, we utilized PXRKO mice in fasting conditions (36). PXR activation in these mouse livers with ligands reduced the mRNA levels of Cpt1a (β-oxidation) and Hmgcs2 (ketogenesis) only in wild-type mice. In contrast, Scd1 (lipogenesis) mRNA was elevated in wild-type mice treated with the PXR activating ligand, PCN (Pregnenolone carbonitrile). Consistent with the observed mRNA changes, alterations of hepatic lipids were observed. PCN-treated wild-type mice showed increased and decreased liver triglyceride and serum 3-hydroxybutyrate, respectively. We found PXR directly bound to FOXA2, a transcription factor that activates the Cpt1A and Hmgcs2 genes in the liver during fasting. PXR is directly bound to FOXA2 and suppresses FOXA2-induced activation of the Cpt1a and Hmgcs2 promoters.

PXR is also involved in glucose metabolism in the liver. Mouse PXR has been shown to suppress the transcription of the glucagon-activated G6Pase (glucose-6-phosphatase) gene by directly binding to the transcription factor CREB, which is involved in the regulation of many genes, including those related to energy metabolism. PXR suppresses the transcription of the G6Pase gene by inhibiting CREB binding on the promoter of this gene. When fasting wild-type mice were treated with PCN, the binding of CREB to the G6Pase promoter was reduced compared to untreated mice, while no change was observed in PXRKO mice. Therefore, drug activation of PXR suppresses CREB transcription and downregulates gluconeogenesis (35).

Unlike the case of mice, observational studies have shown that blood glucose levels increase in humans after administration of PXR activators like rifampicin, a tuberculosis drug, or statins, a hyperlipidemic drug. We tried to clarify the reason for this species difference, and in the process, we found functional crosstalk between the nuclear receptor activation and cell signaling pathways. Rifampicin-activated PXR induces the serum/glucocorticoid regulated kinase 2 (SGK2) gene and simultaneously increases the expression of G6Pase to stimulate glucose production. Knockdown of SGK2 significantly attenuated PXR-regulated G6Pase induction and glucose production, indicating that PXR is an essential factor in this pathway (37).

We then revealed that statins utilize PXR and SGK2 to activate PEPCK1 and G6Pase genes, thereby increasing glucose production in human hepatocytes (38). This was the first study to reveal the molecular mechanism of the statin/PXR/SGK2-mediated signaling pathway in hepatic gluconeogenesis. Statin-activated PXR served as a scaffold for protein phosphatase 2C (PP2C) and SGK2, stimulating PP2C to dephosphorylate SGK2 at Thr193. Unphosphorylated SGK2 coactivated PXR-mediated transactivation of gluconeogenic genes in human hepatocytes, thereby promoting gluconeogenesis. This gluconeogenic statin-PXR-SGK2 signal was absent in mice in which hepatic gluconeogenesis is known to be suppressed by statin treatment. These findings revealed the molecular mechanism behind statin-related side effects, such as an increased risk for type 2 diabetes.

Later, we found that low glucose levels regulate VRK1 for phosphorylating PXR Ser350. This phosphorylation enables PXR scaffolding for PP2C to dephosphorylate SGK2, activating the PEPCK1 gene (39). Phosphorylation of mouse PXR Ser347, which is the equivalent of human PXR Ser350, was observed on SULT1E1 gene promoter after fasting and was involved in the regulation of this gene (40). Intriguingly, mice with a PXR Ser347Ala mutant developed hepatic steatosis and hypertriglyceridemia after fasting, suggesting that phosphorylation is a key factor for energy homeostasis in the liver(41). This PXR phosphorylation site in ligand binding domain (LBD) is conserved among many nuclear receptors, and the phosphorylation of these residues was shown to regulate their biological activities as we will discuss in section 7 (42).

4.3. Conserved nuclear receptor phosphorylation in energy metabolism

We observed increased phosphorylation of RXRα T167 during fasting in wild-type mice using a phospho-specific antibody for this residue. Thr 167 in RXRα is equivalent to Thr 38 of human CAR, which is highly conserved among nuclear receptor family members (15). To search for the biological functions of this phosphorylation, we created knock-in mice in which phosphorylation of Thr167 in RXRα does not occur (Rxra^T167A) and found that the phosphorylation contributes to mouse energy metabolism (43). Gene expression changes in the fat and liver of the mutant mice in response to fasting were compared with those in wild type mice. Observed gene expression changes were more significant in wild-type mice than in mutants. In the knock-in mice, fasting regulated gene expression in fat, muscle, and liver was altered, and the integration of their control is disrupted. RXRα +T167A mice showed reduced blood glucose levels after fasting compared to wild-type mice. This correlated with a simultaneous downregulation of lipid metabolism in WAT and was associated with RXRα phosphorylation at Thr 167. Thus, by phosphorylation of Thr 167, RXRα coordinates these three organs to regulate energy metabolism and contribute to blood glucose homeostasis.

5. Hepatocellular carcinoma

5.1. CAR in liver cancer

Because PB is prescribed to humans to treat several conditions, including epilepsy, safety evaluations were performed in animals. Such studies over the past several decades have revealed that PB is involved in hepatocellular tumorigenesis in rodents and has been classified as a non-genotoxic carcinogen. PB is now widely used as a model substance of nongenotoxic carcinogens in the established experimental systems in rodents. A typical experimental system utilizes N-diethylnitrosamine (DEN) as a genotoxic initiator and phenobarbital (PB) as a non-genotoxic promoter. This two-step chemically induced hepatocellular carcinoma (HCC) system has been employed in numerous studies on HCC molecular mechanisms.

Given the long history of nongenotoxic carcinogen research using PB, it was a logical and challenging next target for us to investigate the roles of CAR in this HCC promotion after we identified CAR as the transcription factor for phase I and II drug metabolizing enzyme gene induction by PB. Our study results unambiguously suggest that CAR is responsible for liver carcinogenesis (44). After 32 weeks of treatment with PB in drinking water following DEN initiation, all CAR wild-type mice developed advanced liver tumors, but none of the CAR-deficient mice developed any. The results established that CAR is the indispensable factor for PB HCC promotion. For this study, we used CAR-deficient mice with a background changed from C57 mice, which are resistant to carcinogenesis in this experimental system, to C3H by repeated backcross.

5.2. Gadd45β and CAR in HCC

Shortly after our findings about CAR indispensability in HCC, the expression of growth arrest and DNA damage-inducible 45 beta (Gadd45β), a molecule known to be involved in cellular stress responses, was shown to be strongly induced by TCBOBOP, a CAR ligand, only in wild-type mice but not in CARKO mice (45). Gadd45β is a member of the Gadd45 family, which consists of α, β, and γ, and mediates various protein-protein interactions among protein factors that are induced by stress signals and trigger numerous cellular responses, including DNA repair, cell cycle control, senescence, and apoptosis (46, 47). Therefore, we hypothesized that Gadd45β may play key roles in CAR dependent HCC promotion in rodents.

5.3. Cell signaling pathways and CAR in HCC

Our studies showed that CAR upregulates the expression of Gadd45β, which also acts as a coactivator of CAR-RXRα transcriptional activity (48). Moreover, the CAR-Gadd45β complex suppresses TNFα-induced cell death (49). Gadd45β was highly induced in DEN+PB in both 23- and 32-week-old experimental HCC mice compared to mice treated with DEN alone. Phosphorylation (=activation) of JNK, a downstream factor of TNFα, was decreased by treatment with TCPOBOP in a CAR-dependent manner in primary hepatocytes. Gadd45β increased the formation of a complex between MKK7, a kinase functioning between TNFα and JNK, and CAR. CAR potentiated Gadd45β activity for inhibiting phosphorylation of JNK by MKK7 while CAR and Gadd45β complexes are scaffolding these reactions. An insightful finding at the time was that CAR-TCPOBOP-induced repression in TNFα-induced cell death was not observed in Gadd45β-deficient mice (49).

On the other hand, Gadd45β is known to regulate the p38MAPK pathway. p38MAPK acts as a tumor suppressor, and in liver-specific p38MAPK KO mice, the carcinogenesis initiated by DEN progressed to HCC increased. Hence, we investigated the role of CAR-Gadd45β in the p38MAPK pathway modulation in rodent HCC development. In the phenobarbital-treated wild-type mouse livers, dephosphorylation of p38MAPK occurred, but not in CARKO mice. We found that CAR binds to Gadd45β and inhibits Gadd45β functioning as a scaffold for MAPK kinase 6 (MKK6)(50, 51). Thus, PB-activated CAR suppresses the phosphorylation of p38MAPK. p38 phosphorylation is reduced in PB-treated wild-type mice, but not in Gadd45β KO mice, with or without DEN pretreatment. Phenobarbital-induced hepatocyte proliferation was reduced in the livers of male and female Gadd45β knockout (KO) mice compared with wild-type mice. Thus, these data indicate that the nuclear receptor CAR interacts with Gadd45β to suppress p38 MAPK signaling and induce hepatocyte proliferation in mice. This Gadd45β-regulated male-dominant proliferation may act as a promoting signal for phenobarbital-induced HCC development in future studies.

Next, we investigated Gadd45β gene knockout effects on PB HCC promotion. Compared with wild-type mice, Gadd45β KO mice did not develop HCC in the PB-treated group revealing the indispensable functions of Gadd45β in DEN + PB two-stage HCC promotion in mice. Microarray and qPCR analysis found that two genes, Tgfbr2 and irisin/Fndc5, were up-regulated in PB-treated wild-type mice, but no significant increase was observed in Gadd45β KO mice(52). We focused on these two genes because previous reports have shown that hepatic Irisin/Fndc5 expression is significantly higher in HCC patients and that irisin binds to the TGF-β receptor complex containing the TGFBR2 subunit. Our results disclosed that irisin peptide in cell culture medium increased the proliferation rate of mouse hepatocyte-derived AML12 cells. Microarray analysis revealed that irisin-regulated genes in AML12 cells were significantly associated with genes in the TGFβ pathway. Thus, Gadd45β collaborating with CAR plays a crucial role in mouse HCC development by controlling the irisin/Fndc5 and Tgfbr2 gene expressions(52).

In summary, our results suggest that in the rodent HCC two-stage model system, CAR, with Gadd45β as a cofactor, affects the TNFα and TGFβ pathways and plays a pivotal role in HCC promotion. Secretion factors like Irisin/FNDC5 may modulate TGFβ pathway and kinases including p38MAPK.

This HCC promotion activity of PB is not observed in humans. A recent publication has shown that mouse CAR binds to Yes-associated protein 1 (YAP1) and translocates YAP1 to the nucleus upon activation by PB. However, human CAR cannot bind to YAP1, resulting in the HCC promotion differences between the two species (53). YAP1 has also been shown to functionally interact with the TNFα and TGFβ pathways. Thus, further research is expected to determine how the findings of our group about the Gadd45β, TNFα, and TGFβ pathways are defined in the context of CAR-YAP complex nuclear translocation after mouse CAR activation for HCC development.

A recent report found that the repression of human HCC occurs through CAR inhibiting erythropoietin signaling (54). The erythropoietin signaling pathway also functionally interacts with TNFα, TGFβ, and Gadd45β. Although PB – CAR is not involved in human HCC, this fact does not indicate that the factors and pathways we found in mice are not involved in human HCC. Therefore, it may be interesting to study those factors and pathways in human liver tumorigenesis in the future.

5.4. CAR function in animal physiology and diseases

Over the past three decades, numerous researchers—including former members and collaborators of our group—have significantly contributed to the study of CAR. According to Honkakoski, as of October 2021, a PubMed search using “CAR” or “Nr1i3” retrieved 1,560 publications(55), increasing to 1,710 by July 2025. These studies span a wide range of topics, from structural analyses of CAR to the identification of small-molecule CAR activity modulators and are comprehensively summarized in a number of excellent review articles listed in Honkakoski’s paper.

Research on the physiological activity of CAR and liver pathology has been actively conducted by many groups, including our study of the regulation of expression of the bilirubin metabolic enzyme UGT1A1 by CAR(5, 56–58), studies on steatohepatitis by our collaborators Yamazaki and Kakizaki(59–61), and hyperplasia(62). As summarized in Section 4.1, CAR is critically involved in hepatic energy metabolism— including glucose and lipid pathways—and plays a decisive role in the HCC promoting activity of PB. Numerous studies have since expanded on CAR’s roles in energy, bilirubin, and bile acid metabolism, as well as its involvement in steatohepatitis and hepatocarcinogenesis, with several comprehensive reviews providing overviews of these results(42, 59, 63–66). More recently, CAR’s functions in the intestine have also been increasingly recognized(67). To better harness CAR’s complex physiological functions as therapeutic targets for human health, it will be essential not only to advance current genomics and proteomics approaches but also to elucidate, at the molecular level, the in vivo dynamics of CAR and its associated factors. Experimental systems such as animal models and human-derived organoids will be critical in achieving this goal.

5.5. CAR small-molecule activity modulators and ligands

Research on small-molecule CAR ligands and activity modulators has also been active, including our own studies on diallyl sulfide (DAS), BDE47 and PK11195(68–70). The direction of this research is well summarized in Honkakoski’s review, which classifies ligands and modulators as indirect activators, selective agonists, or inverse agonists, and discusses them in relation to CAR’s structural features(55). The review also describes in detail the strategies used to identify these compounds and highlights species differences and isoform-specific responses to ligand-mediated activation of CAR. These studies of small-molecule activity modulators have yielded important insights for public health, revealing that various endocrine-disrupting chemicals (EDCs) act as CAR ligands and broadly influence its physiological functions(55, 71–73). A recent provocative finding revealed that the potent mouse CAR ligand TCPOBOP induces sustained hepatomegaly in mice even after discontinuation of treatment(74). These findings underscore the need for further research on the impact of EDCs on CAR activity modulation.

6. Road to the future

6.1. Phosphorylation of Thr38 disrupts a local structure of the DBD

As a result, CAR loses its DNA binding capability and alters its protein-protein interactions. These phosphorylation motifs and structural alterations are conserved in the majority of nuclear receptors. The conserved motifs are giving us the platform to extend our findings with CAR to virtually all nuclear receptors. In fact, in addition to CAR, four other nuclear receptors, RXRα, estrogen receptor (ERα), farnesoid X (FXR), and retinoid orphan receptor α (RORα) are phosphorylated at their motifs in mouse tissues and cells in vivo (75). For RXRα and ERα, knock-in (KI) mice bearing the single amino acid mutation to prevent phosphorylation were generated. Analysis of these KI mice confirmed that phosphorylation confers novel physiological functions to ERα and RXRα (43, 76). Thus, investigating how phosphorylation regulates all other nuclear receptors with and without ligands should be essential future research. A more challenging direction of future research is presented by studies of hepatic estrogen sulfotransferase (EST, SULT1E1). To regulate this gene, three nuclear receptors, CAR, RORα, and ERα integrate their functions via phosphorylation at their conserved motifs.

6.2. PB-CAR-RORα

Expression of EST is regulated in the liver of male mice. This expression is constitutively up-regulated in male RORα KO mice, thus indicating that RORα suppresses the transcription of the EST (77). PB induces EST in mouse livers, and this induction is associated with phosphorylation of RORα at the conserved motif Ser100. In cell-based reporter assays, a non-phosphomimic RORα S100A mutant represses a promoter of the EST gene, while the phosphomimic RORα S100D mutant activates the promoter. Evidently, CAR acting as a master regulator, transduced the PB signal to RORα, converting RORα from the non-phosphorylated to phosphorylated form. The phosphorylation, then, converts RORα from a transcriptional repressor to an activator (77).

6.3. PB-CAR-ERα

It began with an unexpected finding that the EST PB induction is severely diminished in the liver of ERα S216A KI mice in which no motif phosphorylation occurs (22). Contrarily, it is normally induced. Given this evidence suggesting that phosphorylated ERα at the conserved motif is required for PB to activate the EST gene, but not the CYP gene, chromatin immunoprecipitation (ChIP) assays were applied to the promoter of the EST gene. ERα becomes phosphorylated in response to PB in wild type but not in ERα KI mice. CAR mediates the PB signal to ERα converting it to phosphorylated form and strengthening its binding to phosphorylated ERα. Then, a CAR-phosphorylated ERα complex activates the EST gene.

6.4. RORα-ERα

Both RORα and ERα are phosphorylated in response to PB to activate the EST gene. This raises the question of which of these two nuclear receptors is phosphorylated first. To answer this question, ChIP assays were performed to examine ERα on the promoter in the liver of RORα KO mice. ERα is detected but is not phosphorylated even after PB treatment. The flow of PB signaling can be the following: PB→CAR→RORα→phosphorylated RORα→ ERα→phosphorylated ERα. This sequential phosphorylation enables these three receptors to form a transcriptionally active triple complex.

Unlike the case in mouse livers, EST has long been known to be highly expressed in human primary hepatocytes. PB represses this expression, which is contrary to the induction in mice (78). Because a human CAR ligand CITCO (6-(4-Chlorophenyl)imidazo[2,1-b]thiazole-5-carbaldehyde-O-(3,4-dichlorobenzyl)oxime) also represses it, CAR should mediate the repression. Results obtained by ChIP assays confirmed both phosphorylated RORα and ERα are on an active promoter of the gene in human primary hepatocytes. Upon PB treatment, the ERα is dephosphorylated but the dephosphorylated ERα remains on the inactive promoter. On the other hand, RORα is dissociated from the promoter altogether (78). Therefore, in human primary hepatocytes, the EST gene is constitutively activated in the same manner as it is activated in diabetic mouse livers. How PB transduces this repression signal remains a subject of future investigation.

Unlike EST gene, human CYP2B is induced by PB, as similarly observed in mouse primary hepatocytes. Because CAR constitutively resides in the nucleus of human primary hepatocytes, there must be a mechanistic reason to repress the human CYP2B gene, which is not working in mouse primary hepatocytes.

6.5. Other nuclear receptors and CAR

A substantial body of prior research has suggested the plausibility that genes regulated by CAR are also influenced through interactions with nuclear receptors. For example, studies using RORα- and RORγ-knockout mice have revealed that many Phase I and Phase II drug-metabolizing genes, known as CAR targets, are in fact regulated by these nuclear receptors(79). Moreover, studies employing CAR ligands have shown that drug-metabolizing gene expression is affected by factors such as sex differences, circadian rhythms, and dietary conditions(80–82). These findings are consistent with our results described in Section 6.2 – 6.4, which propose that CAR regulates gene expression in cooperation with other nuclear receptors, including ERα and RORγ. Moreover, functional interactions between CAR and FXR, HNF4α, LXR, PPARα, and PXR have been reported(42, 63, 83–85). Going forward, it will be important to investigate in more detail how the cooperation between CAR and other nuclear receptors contributes to CAR’s physiological roles. Such studies should extend beyond model genes in cellular systems and incorporate animal models and organoids, along with omics-based analyses. When these findings are supported by detailed molecular-level information, they will serve as essential resources for identifying therapeutic targets.

6.6. Implications in diabetes

In mice, while EST is repressed in males, it is constitutively expressed in diabetogenic db/db mice (86). CAR mediates this expression because CAR KO in diabetic Akita background mice are unable to express EST. ChIP assays detected both phosphorylated RORα and ERα on the active promoter of EST gene in diabetic mouse livers (22).

6.7. In human livers

Levels of EST mRNA were examined with biopsies from human livers. Healthy women kept the mRNA at relatively low levels. Around 50 years of age, women increasingly develop diabetes. Thus, the development of diabetes is associated with the increase of the EST mRNA, likely through estrogen inactivation (78). Based on a common belief that estrogen prevents the development of diabetes the further decrease of estrogen levels may cause an adverse effect on postmenopausal women. It remains, however, a possibility that this lowered active estrogen is beneficial. In either case, EST may be used as a diagnostic marker as well as a therapeutic target. In contrast, men do not show any correlation of EST levels with age or diabetes.

Human livers from biopsies were utilized for ChIP assays to investigate the activation of EST gene. It was found that EST expression increases as levels of phosphorylated ERα on the promoter increase. The same correlation is also observed with phosphorylated RORα. CAR, acting as the master regulator, transduces an activation signal to phosphorylate RORα and ERα on the active promoter. This activation mechanism is conserved in both diabetic mice and humans.

7. Prospective for the future

It is well known that a gene can be regulated through the cooperation of two or three nuclear receptors, a concept known as nuclear receptor crosstalk. For example, a drug activated PXR or CAR cooperates with a liver specific HNF4α to activate CYP genes. Looping of the promoter has been suggested as a regulatory mechanism to restrict nuclear receptors to a transcription start site (40, 87). The detailed mechanism, however, remains unsolved. More critically, no reliable mechanism is presented to understand how nuclear receptors integrate two different signals, such as between xenobiotic and endobiotic signals, between two different xenobiotic signals, or between two different endobiotic signals. For this matter, current findings obtained with studies of the EST gene are shedding light on the experimental basis for future research. If phosphorylation of the conserved motif acts as a communicator to integrate nuclear receptors, the 46 nuclear receptors could build up a nearly unlimited number of information networks.

In addition to the DBD motif, the LBD of 31 out of 46 nuclear receptors conserve a phosphorylation motif (42). Androgen receptor (AR) possesses it at Ser815 and is known to be phosphorylated in mouse as well as human prostate tissues (88). Ingenuity analysis of genes regulated by ectopically expressed AR S815A mutant in PC3 cells shows that phosphorylated AR represses tumor progression in prostates (88, 89). VDR also conserves this LBD motif as well as the DBD motif, and, moreover, VDR is known to repress tumor development. Can it be reasonable to hypothesize that phosphorylation integrates AR and VDR to regulate prostate cancer? Can it help us to develop novel and more effective therapies?

The same Ingenuity analysis showed that phosphorylated AR, but not unphosphorylated AR, advances hepatocellular carcinoma, likely in a male-predominant manner (88). Recent investigations revealed that CAR represses HCC development in humans (53). Can it be reasonable that phosphorylation communicates with AR, regulating the outcome of HCC development? Although these are only two examples, numerous other experimental models must be presented for us to challenge. It is hoped that the concept of conserved phosphorylation as a communication tool for nuclear receptors will be established.

Finally, we close this article by expressing our deepest gratitude to all the members who worked together in Pharmacogenetics; without their courage and hard work, Pharmacogenetics would not have accomplished the research contributions that are described in this article. THANK YOU.