AMP-activated Protein Kinase Mediates Phenobarbital Induction of CYP2B Gene Expression in Hepatocytes and a Newly Derived Human Hepatoma Cell Line*

Phenobarbital (PB) administration is known to trig-ger pleiotropic responses, including liver hypertrophy, tumor promotion, and induction of genes encoding drug-metabolizing enzymes. The induction of human CYP2B6 and the rat (CYP2B1) and mouse (Cyp2b10) ho-mologues by PB is mediated by the nuclear receptor constitutive androstane receptor (CAR). The study of CYP2B gene regulation and CAR activity by PB has been difficult due to the lack of a cellular model. In this study, we describe a novel differentiated human hepatoma cell line (WGA), derived from HepG2, which expresses CYP2B6 and CAR. WGA cells represent a powerful system to study the regulation of CYP2B6 gene expression by PB. There is evidence that CAR activity is regulated by phosphorylation and that regulation of some CYP genes depends on the nutritional status of cells. The AMP-activated protein kinase (AMPK) functions as an energy sensor and is activated when cells experience

The cytochrome P450 (CYP) 1 gene family plays a crucial role in the biotransformation of structurally diverse classes of xe-nobiotics including drugs and endogenous compounds such as steroid hormones, vitamins, and fatty acids (1). Preferentially expressed in the liver, members of the CYP1, CYP2, and CYP3 families exhibit a broad substrate specificity and metabolize the majority of administered drugs. Recent studies demonstrated that the CYP2B6 isoform comprises 2-10% of total human liver P450 content. In addition, CYP2B6 has been reported to be involved in the metabolism of almost 25% of current pharmaceuticals (2), and expression of CYP2B6 is strongly induced by structurally diverse compounds, such as phenobarbital (PB), rifampicin, and clotrimazole (3,4). Phenobarbital has been used in the treatment of epilepsy and causes pleiotropic responses in the liver including hypertrophy, tumor promotion, and the induction of genes encoding drug-metabolizing enzymes (5,6). The hepatic expression of many cytochrome P450 genes, including members of the CYP1A, CYP2A, CYP2B, CYP2C, and CYP3A subfamilies, is induced by phenobarbital. The barbiturate induction of human CYP2B6 and murine Cyp2b10 and rat CYP2B1 has been shown to be mediated by constitutive androstane receptor (CAR) (7). CAR is a 49-kDa member of the nuclear receptor family NR1I3 and is expressed preferentially in human liver. It was first shown to activate a DR5-type of retinoid acid response element (␤RARE) in a ligand-independent manner (8,9) and is now considered a constitutively active receptor. CAR binds to and activates the PB response element found in promoters of PB-inducible genes (7). The induction of CYP2B gene expression by PB is controlled by changes in CAR intracellular localization. Upon PB treatment, CAR translocates from the cytoplasm into the nucleus, where it binds and transactivates target genes (10). Translocation is inhibited by addition of okadaic acid, suggesting that phosphorylation controls this process (11,12).
In addition to P450 regulation by exogenous chemicals, there is evidence that nutritional status can be a significant factor in determining cytochrome P450 levels. For example, an enhancement of CYP2B1 gene expression is observed in the liver of streptozotocin-induced diabetic rats, and this diabetic state enhances the PB-induced CYP2B1 gene expression (13). However, PB does not induce CYP2B1 gene expression in obese diabetic Zucker rats (14). Interestingly, the low level of CYP2B1 gene expression observed in these animals was associated with the absence of CAR (15). In addition, CYP2B and CYP3A gene expression is induced by ketone bodies, such as 3-hydroxybutyrate, in primary rat hepatocytes (16). These studies suggest that CYP2B genes can also be regulated by hormones and nutrients and that the energy status of the hepatocyte may be important.
A key sensor of cellular energy change is the AMP-activated protein kinase (AMPK). This kinase is activated when cells experience energy-depleting stresses (17). Recently, AMPK has been implicated in the regulation of glucose uptake in skeletal muscle during physical exercise and also in regulation of hepatic glucose output through the inhibition of phosphoenolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase gene expression (18). Thus, AMPK has been proposed as a target for the development of anti-diabetic drugs (19) and has been shown to be activated by the existing anti-diabetic drugs metformin and rosiglitazone (23,50). In the present study, we identify AMPK as a novel PB-regulated signaling molecule and provide evidence that this protein plays an important role in the regulation of CYP2B gene expression by PB. In addition, we describe a newly derived human hepatoma cell line that will be beneficial in the study of both CAR function and CYP2B gene regulation.

EXPERIMENTAL PROCEDURES
Animals-Animal studies were conducted according to United Kingdom guidelines for care and use of experimental animals. Male Sprague-Dawley rats (200 -300 g) used for isolation of hepatocytes were purchased from Charles River Ltd. (Margate, UK).
Cell Culture-Culture media and antibiotics were obtained from Invitrogen, unless noted otherwise. The human hepatoma cell line HepG2 was obtained from the European Collection of Cell Culture (ECACC 85011430). WGA cells were derived from HepG2 as described under "Results." HepG2 and WGA cells were cultured in a mix of DMEM/DMEM:F12/L15 (50%/25%/25%, respectively) supplemented with 7% NuSerum IV (Becton Dickinson Europe, Le Pont-De-Claix, France), sodium butyrate (400 M), fructose (4 mM), and a mixture of antibiotics and antimycotics (1% of a 100ϫ stock) in a humidified atmosphere at 37°C in the presence of 5% CO 2 .
Rat hepatocytes were prepared by the two-step collagenase perfusion as described previously (25). Cell viability was estimated by trypan blue exclusion, and it was always Ͼ90%. Hepatocytes were cultured in DMEM supplemented with 5% fetal bovine serum (Invitrogen) and 1ϫ ITS (insulin transferin selenium; Becton Dickinson Europe) for 4 h to allow cell attachment. After the attachment period, cells were cultured in DMEM with 5% fetal bovine serum or as detailed under "Results." Isolation of WGA Hepatoma Cells (a Differentiated Variant of HepG2)-The WGA cell population was isolated from HepG2 after metabolic selection as described by Armbruster et al. (26) with some modification. HepG2 cells were first adapted to a new growth medium consisting of Leibovitz's L15 medium (Invitrogen) supplemented with 2% Ultroser G (Biosepra S.A., Cergy-Saint-Christophe, France). Cells were shifted to a L15 glucose-and galactose-free medium (Specialty Media, Phillipsburg, NJ) supplemented with 10 mM fructose and 2% Ultroser G (Biosepra S.A.). This culture medium drives the selection of cells expressing a differentiated hepatocyte phenotype, i.e. the metabolism of fructose. In hepatocytes, fructose is transported across the cell membrane by the facilitative transmembrane transporter, Glut2 (27), and is preferentially metabolized by the aldolase pathway (28). After 3 weeks of culture under these conditions, cells were then placed in an L15 galactose-free and arginine-free medium (Specialty Media) supplemented with glucose (20 mM), glutamine (2 mM), dexamethasone (1 M), and Ultroser G (2%) (Biosepra S.A.). The culture of hepatoma cells in arginine-free medium has also been successfully used to isolate the primary human liver tumor cell line (BC2) that expresses drug-metabolizing enzymes (29). The remaining cell population (WGA) was then expanded in growth medium as detailed under "Cell Culture." RNA Extraction and Reverse Transcription-Cells were washed twice in ice-cold phosphate-buffered saline, and total RNA was extracted using the Qiagen RNeasy mini kit (Qiagen, Crawley, UK) as described by the manufacturer. Total RNA concentration was estimated by A 260 , and the purity of RNA was verified by the A 260 :A 280 ratio. RNA integrity was evaluated by denaturing gel electrophoresis in agarose (1%), formaldehyde (0.66 M) gels. Bands were visualized using ethidium bromide staining.
Total RNA (1 g) was reverse-transcribed using murine leukemia virus reverse transcriptase (PerkinElmer Life Sciences) in the presence of random hexamers in a final volume of 20 l. Reactions were carried out using in a thermal cycler (Peltier PTC225; MJ Research, Inc., Waltham, MA). The cycles were 10 min at room temperature and 15 min at 42°C, and then the reaction was stopped by incubation at 99°C for 5 min.
Real-Time RT-PCR (TaqMan®)-Gene-specific primers and probes (Table I) were designed using Primer Express (PerkinElmer Life Sciences) and synthesized by MWG-Biotech (Ebersberg, Germany). All primers and probes were entered into the NCBI BLAST® program to ensure specificity. The amplification of target genes by real-time PCR was performed from 1 l of cDNA using a Master Mix (PerkinElmer Life Sciences) containing Taq DNA polymerase in a final volume of 20 l. The reaction was performed as suggested by the manufacturer (PerkinElmer Life Sciences). Samples were loaded on a 96-well plate (Micro-Amp Optical; PerkinElmer Life Sciences), and fluorescence intensity was monitored employing an ABI PRISM 7700 sequence detector. All reactions were standardized with 18 S rRNA using TaqMan® rRNA control reagents (PerkinElmer Life Sciences).
Adenoviral Transfection-Primary rat hepatocytes were prepared as described above. After attachment (4 h), medium was replaced by DMEM supplemented with 10% fetal bovine serum. Hepatocytes were infected with adenovirus at the multiplicity of infection (MOI) described in the figure legends for 24 h and then incubated with or without 2.5 mM PB for an additional 24 h. WGA cells were infected at 60 -70% confluence.
Dual Reporter System Assay-WGA cells were seeded at 5 ϫ 10 5 cells/well into black-walled and clear-bottomed 96-well plates (Corning Inc., Corning, NY). Cells were incubated for 24 h at 37°C before transfection. Transfection was performed overnight in growth medium with FuGENE 6 (Roche Applied Science), according to the manufacturer's instructions. Reporter plasmids PBREM/pGL3-tk (700 ng/well) and (NR1) 5 /pGL3-tk (600 ng/well) were co-transfected with pcDNA3-␣1DN expression vector (60 -70 ng/well). All plasmids were co-transfected with pRL-CMV (Renilla luciferase; 1 ng/well). After transfection, fresh growth medium was added with or without P450 inducers for 36 h as stated in the figure legends. Cells were rinsed twice in ice-cold phosphate-buffered saline and lysed in 20 l of passive lysis buffer (Promega) for 15 min at room temperature. Firefly and Renilla luciferase activity was assayed in an Orion microplate luminometer (Berthold Technologies, Bad Wildbad, Germany). Data were analyzed with Simplicity 2.1 software (Berthold Detection System), and results were expressed as the ratio of firefly/Renilla activities. Each determination was carried out in triplicate.
AMP-activated Kinase Activity-Immunoprecipitation kinase assays of AMPK were performed using anti-␣1 or anti-␣2 AMPK antibodies as described previously (31). Results were expressed as a percentage of control for two independent experiments, with each determination carried out in duplicate.

Induction of CYP2B6 Expression in the Human Hepatoma
Cell Line WGA-Studies of the mechanism of CYP2B gene expression have been hampered by the lack of a cell system that exhibits phenobarbital-induced changes in gene expression. In order to address the need for such a system, we derived WGA cells from HepG2 cells by metabolic selection using the ability to grow on fructose as the selection system. The morphology of WGA cells was similar to HepG2, with a dense granulated cytoplasm, clear nucleus, and dense nucleoli. Interestingly, the CYP2B6 mRNA concentration in the WGA cells was higher than that detected in HepG2 cells (Fig. 1A). Simi-larly, WGA cells expressed a Ͼ20-fold higher level of CAR mRNA compared with HepG2 cells (Fig. 1B). Finally, and most importantly, after 24 h in the presence of 2.5 mM PB, CYP2B6 mRNA concentration increased by 4-fold in WGA but not in the parental HepG2 cell line (Fig. 1A). Therefore, WGA cells, unlike HepG2 cells, provide a system to study the regulation of gene expression by PB.
AICAR Mimics the Effect of Phenobarbital on CYP2B6 Gene Expression in Human Hepatoma Cells-In view of the ability to regulate CYP gene expression in vivo in the diabetic state and in other cases of nutrient stress, we investigated whether AMPK may be involved in the regulation of gene expression by barbiturates. AICAR is a cell-permeable compound that is phosphorylated by adenosine kinase in cells to form AICAR, or ZMP (31). This molecule mimics the effect of AMP and activates the AMP-activated kinase (32,33). We therefore incubated WGA hepatoma cells for 24 h in the absence or presence of PB or AICAR. AICAR is almost as potent an inducer of CYP2B6 in WGA cells as PB (3-fold with AICAR versus 4-fold with PB) (Fig. 2A). The addition of PB (2 mM) and AICAR (200 M) together in the culture medium induced CYP2B6 mRNA expression by 5-fold ( Fig. 2A). In contrast, PB and AICAR had much smaller effects on CYP2B6 mRNA expression in HepG2 cells (Fig. 2B).
The Expression of an Active AMPK ␣1 Subunit in WGA Hepatoma Cells Mimics the Phenobarbital Effect on CYP2B6 Gene Expression-ZMP also regulates other AMP-modulated enzymes such as fructose-1,6-bisphosphatase (34) and phosphorylase (35,36), so AICAR is not a completely specific activator of AMPK. Therefore, to further investigate the role of AMPK in the regulation of CYP2B6 expression, we expressed a constitutively active form of AMPK (ad-active-␣1-AMPK) in WGA cells.
Cells were infected with different concentrations (MOI, 1.5-6) of an adenovirus expressing either ad-active-␣1-AMPK or ␤-galactosidase (ad-␤-galactosidase). The expression of the recombinant Myc-tagged AMPK ␣1 subunit was visualized by Western blot (Fig. 3). Recombinant AMPK expression correlated with increasing virus titer. Twenty-four hours after infection, cells were incubated with or without 2 mM PB for an additional 24 h. In the absence of virus, PB induced CYP2B6 expression 4-fold (Fig. 4A). An MOI of 1.5 of the ad-active-␣1-AMPK increased constitutive CYP2B6 expression 5-fold (Fig.  4A). This was not increased further by PB administration (Fig.  4A). In contrast, CYP2B6 expression is unaffected by expression of ␤-galactosidase (Fig. 4A). However, PB or active AMPK expression did not alter the expression of CAR (Fig. 4B).
AICAR and AMPK Induce CYP2B Gene Expression in Primary Rat Hepatocytes-Rat hepatocytes were challenged with PB and AICAR, and the expression of CYP2B1 was quantified by real-time RT-PCR (TaqMan®) (Fig. 5A). The expression of CYP2B is strongly induced (almost 50-fold) after a 24-h culture in 2 mM PB. Importantly, CYP2B expression was also induced (25-fold) in the presence of 200 M AICAR (Fig. 5A).
Infection of primary rat hepatocytes with ad-␤-galactosidase (MOI, ϳ6) did not alter the basal or PB-induced level of CYP2B gene expression (Fig. 5B). However, expression of a dominant negative form of ␣1-AMPK (ad-KD) almost completely abolished the effect of PB on CYP2B gene expression (Fig. 5B).
The NR1 site of the PBREM is the minimal region required for PB induction of the CYP2B6 promoter, and this is the sequence that binds CAR (9). The NR1-luciferase reporter construct was co-transfected with pcDNA3-␣1DN in WGA cells (Fig. 7B). Cells were then cultured in the absence or presence of PB (2 mM) or AICAR (200 M). Addition of PB induced NR1 activity by 4.8-fold compared with control cells. Meanwhile, AICAR induced NR1 activity by Ͼ3-fold (Fig. 7B). The expression of a dominant negative ␣1-AMPK subunit inhibited the PB induction of (NR1) 5 -Luc activity by 70% (Fig. 7B). Moreover, expression of the dominant negative ␣1-AMPK also reduced AICAR induction of (NR1) 5 -Luc activity by 55% (Fig. 7B).
The inhibitory effects of the dominant negative form of ␣1-AMPK on the PB response of the PBREM and NR1 promoters are similar. This suggests that the NR1 site within the PBREM is regulated by AMPK activity.
Phenobarbital Activates AMPK in WGA and H4IIE Hepatoma Cells-WGA cells were cultured for 1 and 6 h in the presence of AICAR (500 M) or PB (2.5 mM). As expected, AICAR activates AMPK activity within 1 h with a 45% increase (control, 100 Ϯ 5%; AICAR, 145 Ϯ 13%). More interestingly, the AMPK activity increased by Ͼ70% in cells cultured with PB for 1 h (PB, 177 Ϯ 40%). In addition, the AMPK activity in cells incubated with PB increased even more after 6 h of culture (PB, 274 Ϯ 52%). Meanwhile, the AMPK activity decreased somewhat after 6 h of culture in the presence of AICAR (AICAR, 112 Ϯ 16%). Analysis of phosphorylation of Thr-172 of AMPK (activating modification) and Ser-79 of acetyl-CoA carboxylase (AMPK substrate) after PB administration confirmed this rather transient regulation of AMPK in WGA cells and rat hepatocytes (Fig. 8). In WGA cells, AICAR activation was maximal at the past dosing, whereas that of phenobarbital occurred at 6 h. In both cases, activation was attenuated at 24 h. In isolated hepatocytes, activation was also observed at 1 and 6 h and again lost at 24 h after treatment. In agreement with these findings, we found that incubation of WGA or rat H4IIE cells with phenobarbital caused a dose-dependent increase in AMPK activity that was accompanied by a decrease in cellular ATP levels (Fig. 9). Therefore, we have observed PB activation of AMPK in three different liver cell systems. DISCUSSION The study of the regulation of both CYP2B gene expression and CAR activity has previously been difficult due to the lack of suitable cellular models. Most studies to date have been performed in primary hepatocytes maintained in a differentiated state by using complex medium or extracellular matrix (37,38). The exogenous expression of CAR in HepG2 hepatoma cells leads to a constitutive nuclear localization of the receptor, despite the absence of any inducers (10). Therefore, this model FIG. 6. AICAR induces P450 CYP2B1/CYP2B2 protein expression in primary rat hepatocyte culture. Primary rat hepatocytes were cultured for 48 h in the absence (Control) or presence of 400 M AICAR, and expression of P450 2B1/2 and CAR proteins was estimated by Western blot. Thirty micrograms of total protein was separated on a 10% polyacrylamide gel, and P450 2B1/2 and CAR proteins were detected by specific polyclonal antibodies. The Western blot shown is representative of two independent experiments. does not allow the study of mechanisms involved in the control of CAR translocation.
In the present study, we have developed a HepG2-derived cell line (WGA) that expresses both CYP2B6 and CAR and, importantly, exhibits an induction of CYP2B6 mRNA expression upon PB treatment. To our knowledge, this is the first report of a human hepatoma cell line expressing CYP2B6 and CAR. The metabolic selection used to obtain WGA cells omits an essential nutrient, thus promoting expression of hepatocytespecific enzymes in order to allow cell growth and division. In our study, HepG2 cells were cultured for 3 weeks in a medium deprived of glucose but supplemented with fructose. The selection on fructose as a carbon source was also successfully used to isolate the well-differentiated murine hepatoma cell line (mhAT3F) that expresses Glut2 (39,40). In the liver, Glut2 expression is restricted to differentiated hepatocytes, and Glut2 expression decreases during the progression of tumorigenicity (41).
The WGA cell line isolated after 3 weeks in fructose also expresses Glut2, and the transcription factor CAR. In addition, a further 4 -5-fold increase of CYP2B6 expression was observed after 24 h of culture in PB (Fig. 5), which is similar to the 6 -7-fold induction of CYP2B6 gene expression observed in primary human hepatocytes (42). Therefore, WGA cells provide a powerful tool to study regulation of CYP2B6 gene expression as well as CAR translocation and function.
Several studies have demonstrated that intracellular phosphorylation events control the PB induction of CYP2B genes (11,(43)(44)(45)(46), although the mechanism involved has not been defined (11,45,46). In addition, both energy status and nutritional environment of the cell can influence PB regulation of CYP2B gene expression. Thus, we examined whether PB could regulate a protein kinase considered to play a key role in cellular response to changes in energy status, namely, AMPactivated protein kinase. We observed increased AMPK activity when WGA cells were cultured for 1 and 6 h with PB. This is the first demonstration that PB regulates this enzyme. The degree of AMPK activation observed with PB is similar to that seen when the hepatocytes were incubated with the "classical" pharmacological activator of AMPK, AICAR. We believe that PB induces AMPK activity by increasing AMP levels (Fig. 9). PB could induce ATP depletion by inducing phosphorylation of glucose (47). Alternatively, in a more probable scenario, AMPK is exquisitely sensitive to any inhibition of the respiratory chain, which causes an increase in the cellular AMP/ATP ratio.
AMPK is not the only protein kinase activated by PB. Primary hepatocytes cultured for 12 h with PB exhibit an increase in p42/44 mitogen-activated protein kinase and protein kinase C activity (48). This is related to the tumor-promoting effect of PB observed in liver, and it is unlikely that activation of protein kinase C affects CYP2B gene expression (45). More interestingly, PB transiently increases phosphorylation of a 34-kDa nuclear protein in primary rat hepatocytes, and this correlates with the appearance of CYP2B1 mRNA (42).
In order to assess whether PB regulation of CYP2B gene expression involves AMPK, we performed two experiments. Firstly, we demonstrated that expression of active AMPK or treatment of cells with AICAR was sufficient to induce CYP2B expression (Figs. 2 and 4). Secondly, we blocked AMPK activity by expressing a dominant negative form of AMPK, and we found that PB could no longer induce CYP2B gene expression (Fig. 5). These data strongly argue that PB activates AMPK in order to induce CYP2B expression and, importantly, that activation of AMPK is sufficient to induce this gene. The inducing effect of AMPK on CYP2B gene expression described here was observed in primary rat hepatocytes as well as in the human hepatoma cell line WGA (Figs. 2 and 5). The absence of additive effects on CYP2B6 expression of PB and ad-␣1-AMPK suggests that a common mechanism is involved. To our knowledge, this is the first evidence that AMPK induces CYP2B gene expression and indeed that AMPK induces gene expression in hepatocytes, although it has been reported that AMPK activation inhibits expression of hepatic genes coding for enzymes of glucose and lipid metabolism. In primary hepatocytes, AMPK activation inhibits glucose-activated gene expression (fatty acid synthase) (49), liver-type pyruvate kinase, and spot 14 (22). Whereas, in the rat hepatoma cell line H4IIE, PEPCK and glucose-6-phosphatase expression was inhibited by AICAR (18). Interestingly, rosiglitazone induces AMPK activity (50) and CYP gene expression (51) and also represses PEPCK gene expression (52). Thus, at least three agents that activate AMPK (PB, AICAR, and rosiglitazone) also regulate these genes.
The induction of CYP gene expression by PB is mediated by CAR, which binds to the PBREM within these gene promoters (7). The PBREM sequence is strongly conserved between mouse, rat, and human CYP2B genes (9). The 51-bp PBREM is characterized by two nuclear receptor sites (NR1 and NR2) separated by a nuclear factor 1 site (NF1) (7). Mutations in NR1 and NR2 revealed that the NR1 site is predominant over the NR2 site and essential for PBREM activity in response to PB treatment (7). The PBREM activation by PB was blocked in WGA cells transfected with a vector expressing a dominant negative form of ␣1-AMPK (pcDNA3-␣1DN) (Fig. 7). It appears that the NR1 site of the PBREM mediates this effect because the ␣1-DN completely abolished the PB response of NR1 (Fig.  7). Thus, PB induction of PBREM activity involves the activation of AMPK and the subsequent regulation of the NR1 site. CAR binds to the NR1 site as a heterodimer with the retinoid X receptor ␣R (7). No changes were observed in CAR expression after AICAR treatment or AMPK overexpression (Figs. 4B and 6). Therefore, AMPK may induce CYP2B promoter activity by promoting CAR nuclear localization or by inducing CAR activity and/or binding to the DNA sequence. We are currently examining whether AMPK activity correlates with changes in CAR localization and whether CAR is directly phosphorylated by AMPK. Recent work has identified CAR as a co-repressor for FOXO1, an insulin-regulated transcription factor, implicating CAR in the regulation of PEPCK transcription (53). Thus, if AMPK does regulate CAR, it is likely to regulate many FOXOregulated genes and may explain many of the insulin-sensitizing effects of the AMPK activator metformin.
In conclusion, for over a decade, PB was used as a prototypical inducer of CYP2B gene expression. The discovery of the importance of CAR in this process was a major step forward in the understanding of such regulation. We now provide evidence that AMPK links PB to CYP2B expression and suggest that AMPK can regulate CAR activity. The use of gene ablation, as well cDNA microarrays, highlights a large spectrum of genes regulated by CAR (53) including PEPCK and carnitine-palmitoyl transferase 1 (CPT1), genes encoding for key enzymes of glucose and lipid metabolism (53). It will be interesting to determine how many of these CAR-regulated genes are also regulated by AMPK.