The Orphan Nuclear Receptor Constitutive Active/Androstane Receptor Is Essential for Liver Tumor Promotion by Phenobarbital in Mice

Hepatocellular carcinoma (HCC) comprises 90% of liver tumors occurring worldwide in humans, with more than half a million new cases reported every year. Diagnosis of this tumor, once restricted mostly to Asia, is now rising in Europe as well as North America (1). Environmental factors, such as viral hepatitis and chemical exposure, are causally implicated in HCC occurrence and development of HCC is a multistage process, involving initiation, promotion and progression (2). The molecular-based mechanism(s) underlying each stage of HCC development remains enigmatic. In rodents, HCC can be produced by initiating with genotoxic carcinogens (e.g., nitrosamines) and subsequent promotion by nongenotoxic agents such as phenobarbital (PB), 1,1,1-trichloro-2,2-bis(p-chlorophenyl)ethane, and 17α-ethynylestradiol (3, 4, 5). Such rodent models have been used to provide insight into understanding HCC development and, in particular, the liver tumor promotion that occurs without gene mutation.

Development of eosinophilic foci and adenoma in rodents is characteristic of liver tumor promotion by agents such as PB (6). The mutation of the β-catenin gene, which occurs frequently in human HCC, has been observed frequently in the PB-promoted eosinophilic foci (7, 8). It was shown that in mice lacking the c-Jun gene, early development of liver tumors by PB was reduced (9). In addition, knocking out genes such as Connexin32 and N-acetylglucosaminyltransferase III has also been reported to reduce the PB-induced tumor promotion in these mice (10, 11). Despite the apparent correlation of tumor promotion with these genes, it has not been shown that they are direct targets of PB action. Typified by the prototypic PB, nongenotoxic carcinogens generally induce hepatic xenobiotic-metabolizing enzymes such as cytochrome P450 (CYP) and certain transferases. Moreover, enzyme induction is tightly correlated with HCC development (12), although an endogenous factor directly connecting them has not yet been identified. The nuclear receptor constitutive active/androstane receptor (CAR) was first identified as a PB-activated transcription factor that regulates the genes encoding those enzymes (13, 14, 15). We have now used Car^−/− mice to investigate the role of CAR in the development of HCC and present the experimental evidence indicating an essential role for this receptor in liver tumor promotion by PB.

The original background strain of Car^+/+ and Car^−/− mice used in our previous work (16) was changed to a liver tumor-susceptible mouse strain by repeated backcrossing to C3H/HeNCrlBR mice (Charles River Laboratories, Wilmington, MA) until microsatellite analysis showed that mice contained 98% of the C3H markers. Mice were housed in a specific pathogen-free rodent facility, and experiments were done in accordance with National Institutes of Health guidelines for the care and use of laboratory animals. The genotypes of offspring were determined by analyzing the mutant allele using polymerase chain reaction (PCR) with genomic DNA.

Tumor initiation and promotion were carried out according to a standard protocol (17). A total of 269 male mice were given a single intraperitoneal injection of diethylnitrosamine [DEN (90 mg/kg)] at 5 weeks of age and divided into four groups (group 1, 70 Car^+/+ mice; group 2, 71 Car^+/+ mice; group 3, 62 Car^−/− mice; and group 4, 66 Car^−/− mice). The mice in groups 2 and 4 were chronically treated with PB (500 ppm) in drinking water at 7 weeks of age until they were sacrificed or moribund. Seventy-two and 79 mice were randomly selected and underwent necropsy at 30 and 39 weeks of age, respectively. The remaining 118 mice were kept for further investigations including an analysis of survival. Sera were collected, and the livers were weighed and saved for analysis. Macroscopic liver lesions were visually counted. Liver sections were stained with hematoxylin and eosin and evaluated blindly for hepatocellular proliferative lesions according to established criteria (18). Bromodeoxyuridine (BrdUrd) was administered in the drinking water (200 ppm) for 3 days before necropsy. Immunohistochemical staining was performed on 10% formalin-fixed, paraffin-embedded sections. The mouse monoclonal anti-BrdUrd and anti-β-catenin antibodies (Transduction Laboratories, Lexington, KY) were used, respectively. Immunoreactivity was visualized with the Vectastain Elite ABC kit (Vector Laboratories, Burlingame, CA) and 3,3′-diaminobenzidine (Dako, Carpinteria, CA). Slides were counterstained with hematoxylin.

Mice received Purina Pico Chow #5058 (Ralston Purina Co., St. Louis, MO) and water ad libitum. Food consumption was measured over 7 days and calculated on a gram of feed consumed/body weight basis and showed no statistically significant difference between Car genotypes or treatments. In addition, no difference in water consumption was observed between Car^+/+ and Car^−/− mice, but water consumption was slightly greater in PB-treated groups.

TRIzol reagent (Invitrogen, Carlsbad, CA) was used to isolate total RNA from the livers of mice treated with PB for 23 weeks. RNA was reverse transcribed by Superscript II (Invitrogen) to synthesize cDNA. Real-time PCR was carried out with an ABI Prism 7700 (Applied Biosystems, Foster City, CA), using appropriate TaqMan probe and primers 5′-6FAM-CATCACGGCCAATGTTATCTGCTCCA-TAMRA-3′, 5′-ACCCCACGTTCCTCTTCCA-3′, and 5′-CAGCAGGCGCAAGAACTGA-3′ for CYP2B10. Assay-on-Demand Gene Expression Products (Applied Biosystems) were used as probe for NADPH-CYP reductase (Mm00435876-ml) and CYP3A11 (Mm00731567-ml). CYP2B10, CYP3A11, and reductase mRNAs were quantified by normalizing them against glyceraldehyde-3-phosphate dehydrogenase (TaqMan rodent glyceraldehyde-3-phosphate dehydrogenase control reagent; Applied Biosystems).

C57BL/6 mice are resistant to the development of liver tumors, and tumor occurrence is less than one-tenth that of C3H/He mice (17). To begin our tumor study, the C57BL/6X129 background of the original Car^−/− mice (16) was first changed to C3H/He by repeated backcrosses. The Car^−/− C3H/He mice were viable, and newborns appeared normal and healthy. Groups of Car^+/+ and Car^−/− mice received injection with DEN to initiate liver tumors, and half of the mice in each group were subsequently treated with PB in drinking water. Twenty mice from each group were subjected to a pilot necropsy to examine induction of hepatic genes and assess tumor development after 23 weeks of PB treatment. First, real-time PCR was performed to evaluate the expression of known CAR target genes, Cyp2b10, Cyp3a11 and NADPH-CYP reductase (CYP Red; Fig. 1,A). CYP2B10 mRNA was strongly induced in all DEN+PB-treated Car^+/+ mice (>100-fold induction compared with that in the DEN-treated Car^+/+ mice). No induction of the mRNA was observed in any of the DEN+PB-treated Car^−/− mice. Western blots also showed the absence of CYP2B10 protein in the liver of those Car^−/− mice (data not shown). CYP reductase transfers electrons from NADPH to CYP and is an essential component of the CYP-mediated monooxygenase system. CYP reductase mRNA was also induced only in the Car^+/+ mice. In addition, PB treatment preferentially increased CYP3A11 mRNA in the Car^+/+ mice; the average inductions were 3.0- and 1.25-fold in the Car^+/+ and Car^−/− mice, respectively. The results showed clearly that PB, even with chronic treatment, does not induce these genes in the absence of CAR. Proliferative liver lesions had already developed in the DEN+PB-treated Car^+/+ mice by 23 weeks of PB treatment (Table 1). The DEN+PB-treated Car^+/+ mice exhibited a 16-fold higher multiplicity of proliferative liver lesions than the DEN+PB-treated Car^−/− mice. Liver weight as well as serum alanine aminotransferase (ALT) and alkaline phosphatase (ALP), indicative of liver damage during tumor development, also began to increase only in DEN+PB-treated Car^+/+ mice (Table 1). However, in the Car^−/− mice, only small putative preneoplastic foci of cellular alteration were found and no histologically diagnosed liver neoplasia was observed. At 32 weeks of PB treatment, all DEN+PB-treated Car^+/+, but none of the Car^−/− mice, developed advanced liver tumors (Fig. 1,B). As expected, liver weight and ALT and ALP were profoundly elevated in the DEN+PB-treated Car^+/+ mice at this time point, whereas such elevations were observed neither in the other treated mice (Table 1) nor in the respective untreated mice (data not shown). Thorough visual analysis of the livers revealed large lesions and high multiplicity in all DEN+PB-treated Car^+/+ mice (Table 1). In sharp contrast, the DEN+PB-treated Car^−/− mice had approximately 15-fold fewer liver lesions that were on average less than 2 mm. The phenotypes of the lesions in these Car^−/− mice were similar to those in the DEN-treated Car^+/+ and Car^−/− mice. Histologic analysis revealed eosinophilic adenoma, increased BrdUrd staining, and nuclear localization of β-catenin in the DEN+PB-treated Car^+/+ mice (Fig. 1,C). The nuclear localization of β-catenin is a suggested characteristic of eosinophilic liver lesions (7). The eosinophilic adenomas were found in 19 of 20 DEN+PB-treated Car^+/+ mice, but in none of the DEN+PB-treated Car^−/− mice. With respect to the hepatocellular adenoma and carcinoma (Table 2), all DEN+PB-treated Car^+/+ mice, but none of the DEN+PB-treated Car^−/− mice, developed either or both of them; this difference was statistically significant (P < 0.001) by Fisher’s exact test. These results strikingly indicated a lack of liver tumor promotion by PB in the Car^−/− mice, which was further substantiated by the survival of the mice. The DEN+PB-treated Car^+/+ mice started to die at 39 weeks, and all of them died by 52 weeks of PB treatment (Fig. 2). All of the dead mice bore multiple and large liver tumors.

In the absence of administration of specific promoter, DEN-initiated liver neoplasia is assumed to occur based on endogenous promotion and typically results in the development of basophilic and other noneosinophilic tumor phenotypes. Noneosinophilic adenomas were observed in the DEN-treated Car^+/+ (4 of 20) and Car^−/− (3 of 20) and DEN+PB-treated Car^+/+ (8 of 20) mice (see “Other adenoma” in Table 2), with no significant differences in incidence. However, the DEN+PB-treated Car^−/− mice did not develop adenoma of any phenotype, which was statistically different from that observed in the DEN+PB-treated Car^+/+ mice (P < 0.001). Thus, PB treatment appeared to inhibit DEN-initiated tumor formation when complemented with the Car^−/− genotype. A tumor-preventive effect of PB was reported previously in which DEN was injected into preweaned B6C3F1 mice, and the subsequent PB treatment decreased the number of liver tumors (19). Whether the tumor-inhibitory effect of PB observed with the preweaned mice is mechanistically similar to that observed with the Car^−/− mice remains an interesting question for future research.

Our present study provided direct evidence that CAR activation is essential in liver tumor promotion by PB. A recent DNA microarray analysis of hepatic gene expression in Car^−/− mice showed that CAR up- and down-regulates more than 70 genes. In addition to the coordinated up-regulation of drug-metabolizing enzymes, CAR down-regulated genes that encode gluconeogenic enzymes, those involved in fatty acid oxidation, and proteins such as angiogenin and fibronectin (16). Nongenotoxic carcinogens such as PB are known to stimulate cell proliferation and suppress apoptosis, leading to tumor development. Because the signaling mechanisms that regulate energy metabolism are interconnected with those controlling cell proliferation and apoptosis, the role of CAR might inadvertently extend to alter the fate of initiated cells. Our preliminary examination found that the expression of the oncogenes c-Jun and c-Myc was down-regulated and that expression of the Mdm2 gene was slightly elevated in the Car^−/− mice, yet those changes displayed no obvious correlation with PB promotion (data not shown). The microarray study also revealed that more than 70 PB-modulated genes are not regulated by CAR (16). Whereas the possibility cannot be ruled out that, in addition to CAR, another PB-regulated factor or factors may equally be required for the development of PB-promoted liver neoplasia, identifying the downstream event of CAR activation that actually leads to liver tumor development is an urgent issue of our current investigations.

In conclusion, our present study has revealed that PB promoted liver tumors only in Car^+/+ mice, indicating that the nuclear receptor CAR mediates the liver tumor promotion. Once CAR-mediated pathways of tumor promotion are defined at the molecular level, the receptor can be useful as a drug target for HCC prevention.

We thank members of Integrated Laboratory Systems Inc. for technical assistance and members of M. Negishi’s laboratory for helpful discussion.