Cytochromes P450 (P450) mediate the oxidative metabolism of numerous structurally diverse exogenous and endogenous compounds. CYP2A6, which is predominantly expressed in the liver, plays a major role in the metabolism of nicotine, coumarin, and clinically used drugs such as tegafur, losigamone, letrozole, and valproic acid (Nakajima et al., 2002). CYP2A6 can also activate nitrosamines such as 4-methylnitrosoamino-1-(3-pyridyl)-1-butanone and N-nitrosodiethylamine (Tiano et al., 1994). Large interindividual variability exists in the enzymatic activity and protein level of CYP2A6 in human livers (Shimada et al., 1994). This is principally due to genetic factors since some polymorphic alleles affect the enzymatic activity or transcriptional activity (http://www.cypalleles.ki.se/cyp2a6.htm). However, in our recent study, large interindividual variability in enzymatic activity was still observed within the homozygotes of the wild type of the CYP2A6 gene (Nakajima et al., 2006). An association of the enzymatic activity and protein level of CYP2A6 with the CYP2A6 mRNA levels in individual liver samples indicates that the variability occurs at a pretranslational level (Yamano et al., 1990). The large interindividual variability may be a consequence of exposure to compounds that are capable of inducing CYP2A6.

Many P450 isoforms are inducible by xenobiotics. The induction serves as a common cellular defense mechanism, usually leading to increased detoxification of xenobiotics but sometimes forming more toxic and carcinogenic metabolites. CYP2A6 is known to be induced by several drugs such as rifampicin and phenobarbital (Donato et al., 2000; Pichard-Garcia et al., 2000; Rae et al., 2001). However, the molecular mechanism of the induction has never been clarified. Rifampicin and phenobarbital are well known as inducers of human CYP3A4, CYP2B6, and CYP2C (Pascussi et al., 2003). Nuclear receptors of the pregnane X receptor (PXR; NR1I3) and constitutive androstane receptor (CAR; NR1I2) are activated by these inducers and translocate into the nucleus to dimerize with retinoid X receptor (RXR)-α (NR2B1). The heterodimer of PXR/RXRα or CAR/RXRα binds to the enhancer region of the target genes and recruits nuclear receptor coactivators, such as steroid receptor coactivator-1 or glucocorticoid receptor-interacting protein-1, to form a multiprotein complex that leads to the initiation of gene transcription (Kliewer et al., 1998; Lehmann et al., 1998). In the present study, we examined whether PXR and/or CAR are responsible for the induction of CYP2A6. Recently, it has been reported that the induction of CYP3A4 (Tirona et al., 2003) and CYP2C9 (Chen et al., 2005) via PXR and CAR is synergistically activated by hepatocyte nuclear factor (HNF)-4α (NR2A1). Peroxisome proliferator-activated receptor-γ coactivator (PGC-1α) is a versatile coactivator for numerous nuclear receptors including PXR and CAR (Shiraki et al., 2003; Bhalla et al., 2004). Therefore, the role of HNF-4α and PGC-1α as a coactivator in CYP2A6 regulation was also investigated.

Materials and Methods

Materials. Rifampicin, phenobarbital sodium, and clotrimazole were purchased from Wako Pure Chemical Industries (Osaka, Japan). 6-(4-Chlorophenyl)imidazo[2,1-b][1,3]thiazole-5-carbaldehyde O-(3,4-dichlorobenzyl)oxime (CITCO) was purchased from Biomol Research Laboratories (Plymouth Meeting, PA). Antibodies to RXRα, RXR, and PGC-1α and normal rabbit or goat IgG were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Oligonucleotides were commercially synthesized at Hokkaido System Sciences (Sapporo, Japan). Restriction enzymes were purchased from Takara (Shiga, Japan), TOYOBO (Osaka, Japan), and New England Biolabs (Beverly, MA). All other reagents were of the highest grade commercially available.

Human Hepatocyte Culture. Human cryopreserved hepatocytes, lots 82 (Hispanic, female, 23 years) and 100 (Caucasian, female, 74 years), were purchased from In Vitro Technologies (Baltimore, MD) and maintained in hepatocyte culture medium (Cambrex, East Rutherford, NJ) on collagen-coated plates at 37°C under 5% CO₂. Human hepatocytes were seeded into collagen-coated six-well plates at 1.5 × 10⁶ cells/well. After 24 h, the culture medium was changed to hepatocyte culture medium (epidermal growth factor- and antibiotics-free) containing each drug; 10 μM rifampicin, 10 μM clotrimazole, 1 mM phenobarbital, 100 nM CITCO, or 0.1% (v/v) DMSO (vehicle alone). Hepatocytes were maintained for 48 h until harvesting.

Isolation of Total RNA and Real-Time RT-PCR Analysis. Total RNA was extracted using ISOGEN (Invitrogen, Carlsbad, CA) following the manufacturer's protocol, and cDNA was synthesized as described previously (Tsuchiya et al., 2004). Human GAPDH mRNA was quantified by real-time RT-PCR using the Smart Cycler (Cepheid, Sunnyvale, CA) as described previously (Tsuchiya et al., 2004). Human CYP2A6 mRNA was also quantified under the same condition. The primers for human GAPDH (Tsuchiya et al., 2004) and CYP2A6 (Yoshida et al., 2003) were from our previous studies except for the forward primer of CYP2A6, 5′-aaagagttcctgtcactgttgc-3′.

Electrophoretic Mobility Shift Assays. Expression vectors for human CAR (pCR3/hCAR) and human RXRα (pGEM-3Z/hRXRα) were kindly provided by Dr. Masahiko Negishi (National Institutes of Environmental Health Sciences, National Institutes of Health, Research Triangle Park, NC). The human PXR expression vector (pCMV4/hPXR) was previously constructed (Takada et al., 2004). Human PXR cDNA digested from pCMV4/hPXR was subcloned into the pTNT vector (Promega, Madison, WI) for in vitro transcription/translation. Human PXR, human CAR, and human RXRα proteins were synthesized in vitro using the TNT T7 Quick Coupled Transcription/Translation System (Promega) following the manufacturer's protocol. The oligonucleotide sequences are shown in Table 1. Double-stranded oligonucleotides were labeled with [γ-³²P]ATP using T4 polynucleotide kinase (TOYOBO) and purified by Microspin G-50 columns (GE Healthcare Bio-Sciences, Piscataway, NJ). The labeled probe (40 fmol, ∼10,000 cpm) was applied to each binding reaction in 25 mM HEPES-KOH, pH 7.9, 0.5 mM EDTA, 10% glycerol, 50 mM KCl, 0.5 mM dithiothreitol, 0.5 mM (p-amidinophenyl) methanesulfonyl fluoride, 2 μg of poly(dI-dC), and 3 μl of in vitrotranscribed/-translated proteins to a final reaction volume of 15 μl. To determine the specificity of the binding to the oligonucleotides, competition experiments were conducted by coincubation with 10-, 50-, and 200-fold excesses of unlabeled competitors. For supershift experiments, 1 μg of anti-RXRα antibodies or normal rabbit IgG was preincubated with in vitro-transcribed/-translated products on ice for 30 min. The reactions were incubated at room temperature for 15 min and then loaded on 4% acrylamide gels in 0.5× Tris-borate EDTA buffer. The gels were dried and exposed to film for 3 to 15 h. The DNA-protein complexes were detected with a Fuji Bio-Imaging Analyzer BAS 1000 (Fuji Film, Tokyo, Japan).

Expression Vectors for Human HNF-4α and PGC-1α and Reporter Constructs. Human HNF-4α cDNA was amplified by PCR using cDNA from HepG2 as a template with the forward primer adapted with a NheI site, 5′-aaacgctagccgacatggacatggcc-3′, and the reverse primer adapted with a KpnI site, 5′-ccaggtaccagcggcttgctagataac-3′. The NheI/KpnI-digested PCR fragment was subcloned into the pTARGET vector (Promega). Human PGC-1α cDNA was amplified by PCR using cDNA from human normal kidney with the forward primer 5′-ggatggcgtgggacatgtg-3′ and the reverse primer 5′-tcagctagggaacatgttac-3′. The PCR fragment was subcloned into the pTARGET vector. pCYP3A4–362-7.7K, the pGL3-basic vector containing the core promoter of CYP3A4 (–451/–79) and the distal xenobiotics-responsive enhancer module (–7925/–7297), was previously constructed (Takada et al., 2004). Throughout this manuscript, the base A in the initiation codon ATG is denoted +1 and the base before A is numbered –1.

The 5′-flanking region from –6754 to –1 of the human CYP2A6 gene was amplified by PCR with the forward primer adapted with a NheI site 5′-aacagagctagccgggcac-3′ and the reverse primer 5′-tctatcatcccactaccacc-3′. The NheI-digested PCR fragment was subcloned into the pGL3-basic vector. This plasmid termed pGL3–6754 was used for construction of the other reporter plasmids by digestion with restriction enzymes or PCR and subcloning. Nucleotide sequences were confirmed using Thermo Sequenase Cy5.5 Dye Terminator Cycle Sequencing kit or Thermo Sequenase Cy5 Dye Terminator Cycle Sequencing kit with a Long-Read Tower DNA sequencer (GE Healthcare Bio-Sciences).

Transfection and Luciferase Assay. HepG2 cells obtained from American Type Culture Collection (Manassas, VA) were cultured in Dulbecco's modified Eagle's medium (Nissui Pharmaceutical, Tokyo, Japan) supplemented with 10% fetal bovine serum (Invitrogen) and 0.1 mM nonessential amino acids (Invitrogen) at 37°C under 5% CO₂. HepG2 cells were seeded into 24-well plates at 1 × 10⁵ cells/well. Transfection was performed using Tfx-20 reagent (Promega). In brief, the transfection mixes consisted of 200 ng of reporter constructs, 30 ng of human PXR or human CAR expression plasmid, and 60 ng of human HNF-4α or human PGC-1α expression plasmid. Forty-eight hours after the transfection, cells were harvested and lysed to measure the luciferase activity using a Dual Luciferase Reporter Assay System (Promega). The relative luciferase activities were normalized with the Renilla luciferase activities. When the expression vector of HNF-4α or PGC-1α was cotransfected, the luciferase activities were normalized with the protein contents because HNF-4α or PGC-1α affected the Renilla luciferase activity.

Site-Directed Mutagenesis. The plasmids mutated in the DR4-like elements, pGL3–6698mut, pGL3–5476mut, pGL3–4618mut, pGL3di-mut, and pGL3tri-mut, were constructed by site-directed mutagenesis with a QuikChange II XL site-directed mutagenesis kits (Stratagene, La Jolla, CA) using a pGL3–6.7/-4.5 plasmid as a template. The primers used were: –6698mut, 5′-gaggcgggcAGATCAcctgACGGGAggagttcgagacc-3′; –5476mut, 5′-gggcGGATCActtaAGGGGGggagttcaagaccagc-3′; and –4618mut, 5′-cgaagtgggcAGATCAcctgAGGGGGggagtttgaaac-3′, in which hexamer half-sites are indicated by capital letters, and mutated nucleotides are underlined. Nucleotide sequences were verified by DNA sequencing.

Chromatin Immunoprecipitation Assay. HepG2 cells were cultured in a 100-mm dish to 60% confluence. Cells were transfected with human PXR (1.2 μg) and human PGC-1α (2.4 μg) expression plasmids and incubated for 24 h. Human hepatocytes (lot 82) were treated with 10 μM rifampicin or vehicle (DMSO) for 24 h. ChIP assays were performed using a chromatin immunoprecipitation (ChIP) assay kit (Upstate, Lake Placid, NY) according to manufacturer's protocol. Goat anti-human PXR and goat anti-human PGC-1α antibodies and normal goat IgG (control) were used for immunoprecipitation of protein-DNA complexes. PCR was performed using primer sets: region 1, forward, 5′-cgacagaacgagactccttc-3′ and reverse, 5′-gtgcaatctcggctcgctg-3′; region 2, forward, 5′-gacgattgaatcagggggcag-3′ and reverse, 5′-cctctcgggttcaagcaattc-3′; region 3, forward, 5′-aaatacaagggagtacaagcag-3′ and reverse, 5′-gacacagctcatttttctatt-3′; and region 4, forward, 5′-gtctgttttctgtcctctgta-3′ and reverse, 5′-atagaacctccactgcccatc-3′. The PCR products were electrophoresed on a 2% agarose gel and visualized by ethidium bromide staining.

Statistical Analyses. Data are expressed as mean ± S.D. Comparison of two groups was made with two-tailed Student's t test. Comparison of multiple groups was made with ANOVA followed by Dunnett or Tukey test. P < 0.05 was considered statistically significant.

Results

Induction of CYP2A6 mRNA in Human Hepatocytes. The expression levels of CYP2A6 mRNA in human hepatocytes treated with 10 μM rifampicin, 10 μM clotrimazole, 1 mM phenobarbital, or 100 nM CITCO for 48 h were determined by real-time RT-PCR (Fig. 1). In hepatocytes of lot 82, CYP2A6 mRNA was significantly induced by phenobarbital (5.2-fold), rifampicin (3.2-fold), and CITCO (2.4-fold). Although the extent was lower, the induction was observed by phenobarbital (2.9-fold), rifampicin (2.1-fold), and CITCO (1.6-fold) in hepatocytes of lot 100. Although it was statistically insignificant, clotrimazole tended to induce the CYP2A6 mRNA. Thus, these results suggest that CYP2A6 mRNA is induced by ligands of human PXR and activators of human CAR.

Binding Sites of PXR/RXRα and CAR/RXRα on theCYP2A6Gene. To find potential PXR/RXRα and CAR/RXRα binding sites, the hexamer sequence (A/G)G(G/T)TCA in the ER6 or DR3 on the CYP3A4 gene was used to search ERn (everted-repeat separated by n nucleotides; n, 3–9) or DRn (direct-repeat separated by n nucleotides; n, 3–6) on the 5′-flanking region of CYP2A6 gene. A computer-assisted homology search identified three DR4-like elements with up to –8 kb of the CYP2A6 gene (Table 2). These DR4-like elements located at –6698, –5476, and –4618 have a 1-bp mismatch with a consensus DR4 sequence.

To examine whether a heterodimer PXR/RXRα or CAR/RXRα can bind to these DR4-like elements on the human CYP2A6 gene, electrophoretic mobility shift assays were performed (Fig. 2; Table 1). The binding of PXR/RXRα to a consensus ER6 sequence was confirmed using a probe of ER6/3A4 (Fig. 2A). The band was competed out by three probes of –6698/2A6, –5476/2A6, and –4618/2A6 containing DR4-like elements as a cold competitor. The binding of CAR/RXRα to a consensus DR4 sequence was confirmed using a probe of DR4/2B6 (Fig. 2B). The band was also competed out by three probes of –6698/2A6, –5476/2A6, and –4618/2A6 as a cold competitor. When the probes of –6698/2A6, –5476/2A6, and –4618/2A6 were used, the binding of PXR/RXRα was observed (Fig. 2C). The shifted bands were supershifted with anti-RXRα antibodies and were competed out by ER6/3A4 as a cold competitor. Furthermore, the binding of CAR/RXRα to the probes of –6698/2A6, –5476/2A6, and –4618/2A6 was also observed (Fig. 2D). These results suggest that PXR/RXRα and CAR/RXRα can bind to the DR4-like elements on the human CYP2A6 gene.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Induction of CYP2A6 mRNA level in human hepatocytes. Human hepatocytes were treated with 10 μM rifampicin (RIF), 10 μM clotrimazole (CTZ), 1 mM phenobarbital (PB), 100 nM CITCO, or vehicle alone (0.1% DMSO) for 48 h. Total RNA was obtained from human hepatocytes and real-time RT-PCR was performed. To normalize RNA loading and PCR variations, the CYP2A6 mRNA levels were corrected with the GAPDH mRNA levels. Each column represents the mean ± S.D. of triplicate determinations. *, P < 0.05; **, P < 0.01 compared with control (vehicle) samples.

CYP2A6 Promoter Activity with Coexpression of PXR or CAR in HepG2 Cells. To examine whether PXR and CAR can activate the transcriptional activity of the CYP2A6 promoter, luciferase assays were performed with a series of reporter plasmids containing the 5′-flanking region of CYP2A6 in HepG2 cells. Exogenously expressed PXR and CAR accumulate spontaneously in the nucleus regardless of the activation state (Kawamoto et al., 1999; Kawana et al., 2003). When the pCYP3A4–362-7.7K plasmid was used as a positive control, coexpression of PXR could increase the luciferase activity (38-fold), and rifampicin enhanced the activity (Fig. 3A). The basal transcriptional activity of CYP2A6 gradually increased with the deletion from –6754 to –1013 (approximately 4-fold), and the pGL3–185 plasmid showed the highest basal transcriptional activity (approximately 30-fold of that of the pGL3–6754 plasmid). However, PXR or rifampicin did not activate the transcriptional activities of any CYP2A6 constructs. As shown in Fig. 3B, when the pGL3-tk-PBREM plasmid was used as a positive control, coexpression of CAR increased the luciferase activity (2.5-fold). However, CAR did not activate the transcriptional activities of any CYP2A6 constructs. Coexpression of RXRα did not enhance the transcriptional activities of any CYP2A6 constructs in the presence of PXR or CAR (data not shown). These results suggest that only PXR/RXRα or CAR/RXRα was not enough for induction of CYP2A6.

Effects of Coexpression of HNF-4α and PGC-1α on Transcriptional Activity of CYP2A6. We examined the effects of cofactors on the transcriptional activity in HepG2 cells. When pCYP3A4–362-7.7K plasmid was used, HNF-4α could activate the basal transcriptional activity (2.0-fold), although the change was insignificant (Fig. 4A). HNF-4α could synergistically activate the PXR-dependent activation (4.1-fold of the activity with PXR and 157-fold of control) but not the CAR-dependent activation. In contrast, the transcriptional activity of pGL3–6754 was not affected by HNF-4α as well PXR and CAR. As shown in Fig. 4B, PGC-1α could activate the transcriptional activity of the CYP3A4–362-7.7K plasmid (2.9-fold), although the change was insignificant. PGC-1α could synergistically activate the PXR-(3.6-fold of the activity with PXR and 138-fold of control) and CAR-dependent (3.7-fold of the activity with CAR and 93-fold of control) activation. Interestingly, PGC-1α could significantly activate the basal transcriptional activity of pGL3–6754 (1.5-fold), and the activity was synergistically activated by PXR (1.6-fold of the activity with PGC-1α and 2.5-fold of control) but not by CAR. These results suggest that the CYP2A6 gene is regulated by PXR as well as PGC-1α.

Role of DR4-Like Element on PXR/PGC-1α-Dependent Transactivation of CYP2A6. To evaluate the functional role of the DR4-like elements on the regulation of CYP2A6 via PXR and PGC-1α, a series of deleted reporter constructs were used (Fig. 5A). The transcriptional activity of pGL3–6754 was increased by the coexpression of PXR and PGC-1α up to 3.1-fold. However, the activation by PXR and PGC-1α was decreased to 1.4-fold with the pGL3–4907 plasmid. The deletion from –4907 to –2157 abolished the activation. As shown in Fig. 3, a potential suppressor region was suggested from –1013 to –185. To exclude their effects, reporter plasmids in which the region from –4532 to –186 was deleted were constructed. The pGL3–6.7/-4.5 plasmid was potently activated by PXR/PGC-1α (6.9-fold) (Fig. 5B). Although the pGL3–6.7/-5.4 plasmid retained the responsiveness to PXR/PGC-1α (4.4-fold), the pGL3–6.7/-5.8 plasmid lost the responsiveness. Furthermore, the pGL3–5.5/-4.5 plasmid responded to PXR/PGC-1α (3.5-fold), but pGL3–4.9/-4.5 did not. The pGL3–5.5 plasmid was weakly activated by PXR/PGC-1α (1.8-fold). These results suggest that a DR4-like element at –5476 would be essential for activation by PXR/PGC-1α, and an additional DR4-like element would be cooperatively associated.

To confirm the role of DR4-like elements on the transactivation by PXR/PGC-1α, mutations were introduced into the DR4-like elements on the pGL3–6.7/-4.5 plasmid (Fig. 6). The fold activation of pGL3–6698mut plasmid by PXR/PGC-1α was increased compared with that of pGL3–6.7/-4.5, suggesting that the DR4-like element at –6698 might not be important for the transactivation. The activation of pGL3–4618mut by PXR/PGC-1α was significantly decreased to 4.5-fold. An additional mutation in the DR4-like element at –5476 resulted in a further decrease of the fold activation (2.8-fold). These results suggest that DR4-like elements at –4618 and –5476 are essential for PXR/PGC-1α-dependent transactivation of the human CYP2A6 gene.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Electrophoretic mobility shift assays of binding of PXR/RXRα or CAR/RXRα to consensus ER6 (CYP3A4), consensus DR4 (CYP2B6), or DR4-like elements on the CYP2A6 gene. Electrophoretic mobility shift assays were performed with oligonucleotide probes labeled with ³²P and bound with in vitro-transcribed/t-ranslated proteins. Oligonucleotides of ER6/3A4 (A), DR4/2B6 (B), –6698/2A6 (C and D), –5476/2A6 (C and D), and –4618/2A6 (C and D) used as probes are shown in Table 1. Cold oligonucleotides were used as a competitor in 10-, 50-, and 200-fold molar excess. For supershift analyses, 1 μg of anti-RXRα antibodies or normal rabbit IgG was preincubated with the RXRα proteins on ice for 30 min. Lower arrow, position of the PXR/RXRα-dependent (A and C) or CAR/RXRα-dependent (B and D) shifted band; upper arrow, supershifted complex by anti-RXRα antibodies. Due to the different exposure times, the band intensities could not be accurately compared.

Binding of PXR and PGC-1α in the Native CYP2A6 Chromatin. To further test the binding and interaction of PXR and PGC-1α to CYP2A6 gene in vivo, ChIP assays using HepG2 cells and human hepatocytes were performed. Because the PGC-1α level is low in HepG2 cells, PGC-1α as well as PXR was overexpressed to increase their expression levels in HepG2 cells. PCR primers were designed to amplify DR4-like elements at –6698 (region 1), –5476 (region 2), and –4618 (regions 3) as well as a negative control (region 4). Normal goat IgG was used as a negative control of immunoprecipitation for ChIP assays. As shown in Fig. 7, the immunoprecipitant of HepG2 cells obtained with anti-PXR antibodies generated a distinct PCR product for regions 2 and 3 and a weak PCR product for region 1. The immunoprecipitant of HepG2 cells obtained with anti-PGC-1α antibodies generated a distinct PCR product for regions 1 to 3. The immunoprecipitant of human hepatocytes obtained with anti-PXR antibodies generated a distinct PCR product for region 3 and a weak PCR product for regions 1 and 2. Rifampicin increased the amount of PXR recruited to regions 1 to 3. The immunoprecipitant of human hepatocytes obtained with anti-PGC-1α antibodies generated a distinct PCR product for regions 2 and 3 and a weak PCR product for region 1. Rifampicin increased the amount of PGC-1α recruited to region 1 but decreased for region 2 and did not affect for region 3. These results suggest that PXR and PGC-1α bind to the native CYP2A6 chromatin.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Relative promoter activities of CYP2A6 constructs and effects of coexpression of PXR or CAR. A series of reporter constructs containing the 5′-flanking region of the CYP2A6 gene were transiently transfected into HepG2 cells with the expression plasmids for human PXR (A) or human CAR (B). After 24 h, the cells were treated with 10 μM RIF or DMSO vehicle (0.1%) for 24 h in some cases. The cells were harvested and assayed for the luciferase activities according to the manufacturer's protocol. Each column represents the mean ± S.D. of three independent experiments. –, pCMV4 (A) or pCR3 (B) empty vector (control).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Effects of HNF-4α or PGC-1α in the presence of PXR or CAR on the CYP2A6 transcriptional activity. Different CYP2A6 constructs generated as described under Materials and Methods. HepG2 cells were transiently transfected with reporter constructs and expression plasmids for human PXR, human CAR, human HNF-4α, or human PGC-1α. Forty-eight hours after transfection, cells were harvested and assayed for the luciferase activities according to the manufacturer's protocol. Each column represents the mean ± S.D. of three independent experiments. –, pCMV4 or pTARGET empty vector. **, P < 0.01; ***, P < 0.001 compared with control. †††, P < 0.001 compared with single-factor transfection.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Luciferase assays of CYP2A6 constructs with coexpression of PXR and PGC-1α. Left, schematic representation of CYP2A6 constructs. Reporter plasmids with deletion from the 5′ direction (A) or a variety of deleted (–4532 to –186) plasmids (B) were cotransfected with human PXR and PGC-1α into HepG2 cells. After 48 h, the cells were harvested and assayed for luciferase activities. Each column represents the mean ± S.D. of three independent experiments. –, pCMV4 or pTARGET empty vector. *, P < 0.05; ***, P < 0.001 compared with control (pCMV4 and pTARGET empty vectors).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Effects of mutation of three DR4-like elements on PXR/PGC-1α-dependent transactivation of the CYP2A6 gene. Mutations were introduced into pGL3–6.7/-4.5 by site-directed mutagenesis. A schematic representation of the mutated constructs is shown. Reporter plasmids were cotransfected with human PXR and human PGC-1α into HepG2 cells. After 48 h, the cells were harvested and assayed for luciferase activities. Fold activation compared with the control (pCMV4 and pTARGET empty vectors) is shown. Each column represents the mean ± S.D. of three independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Discussion

P450 enzymes are extremely important in phase I drug metabolism, and their activity affects the clearance and disposition of the majority of therapeutic agents. The expression of these enzymes is altered by a variety of compounds, which may or may not be substrates of the enzymes, through the activation or deactivation of various nuclear receptors and/or transcriptional factors. Although data concerning the mechanism of the induction of CYP1, CYP2B, and CYP3A are accumulating, the molecular mechanism of the induction of CYP2A6 has never been determined. In the present study, we identified regulatory elements in the CYP2A6 gene and demonstrated functional roles for PXR and PGC-1α in the transcriptional regulation of CYP2A6.

First, it was confirmed that CYP2A6 mRNA is induced by rifampicin, phenobarbital, and CITCO in human hepatocytes, in accordance with previous studies (Pichard-Garcia et al., 2000; Maglich et al., 2003). Many earlier studies reported that these compounds induce CYP3A4 and CYP2B6. Rifampicin and phenobarbital are known to be activators of both PXR and CAR. Clotrimazole is a PXR activator but is a CAR deactivator (Pascussi et al., 2003). CITCO, a specific activator of human CAR, can also activate human PXR, although the potency of PXR activation is >50-fold lower than that of CAR activation (Maglich et al., 2003). Therefore, the induction of CYP2A6 can be assumed to involve PXR and/or CAR. It should be noticed that the induction levels of CYP2A6 were at most severalfold, and in some cases, the induction was not observed (Donato et al., 2000; Edwards et al., 2003), whereas the induction levels of CYP3A4 and CYP2B6 were as high as 50- to 80-fold (Goodwin et al., 2001; Rae et al., 2001; Wang et al., 2003).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

ChIP assays of PXR and PGC-1α binding to human CYP2A6 gene. Schematic diagram of the CYP2A6 gene is shown at the top of the figure. ChIP assays were performed as described under Materials and Methods. HepG2 cells were transfected with PXR and PGC-1α expression plasmids. Human hepatocytes (lot 82) were treated with 10 μM RIF or vehicle (DMSO) for 24 h. Anti-PXR or anti-PGC-1α antibodies were used to precipitate the DNA-protein complexes. DNA fragments amplified by PCR were analyzed on a 2% agarose gel.

Until now, the transcriptional activity of CYP2A6 was investigated at up to –1.0 kb by luciferase assay (von Richter et al., 2004). Pitarque et al. (2005) demonstrated putative response elements for the transcriptional factors HNF-4α, C/EBPα and Oct-1 in the 5′-flanking sequence from –112 to –61. The constitutive expression of CYP2A6 is governed by the interplay between these factors (Pitarque et al., 2005). In the present study, we first investigated the transcriptional activity of a far upstream region, up to –6.8 kb of the CYP2A6 gene. The basal transcriptional activity of CYP2A6 was prominently decreased with the inclusion of the upstream region from –1013 to –185. The results were in accordance with a previous study by Pitarque et al. (2005) showing that the transcriptional activity was decreased by the inclusion of the upstream from –200. These results suggested that there might be a suppressor region(s) upstream from –200.

PXR and CAR bind to elements containing a direct repeat of the hexamer (A/G)G(G/T)TCA separated by three to five nucleotides such as DR3, DR4, and DR5 or an everted repeat separated by a six-nucleotide spacer (ER6). The binding elements for PXR and CAR in various CYP genes are shown in Table 2. In some cases, a substitution in the third or forth nucleotide of the half-site is observed, but a distinct but related half-site allows the binding of the PXR/RXRα or the CAR/RXRα heterodimer. However, a single nucleotide difference is able to change the binding, since the receptors could bind to DR5 of CYP2C9 but not to DR5 of CYP2C19 (Chen et al., 2003). Three DR4-like elements on the CYP2A6 gene found in this study have a single-nucleotide difference from the consensus DR4. Electrophoretic mobility shift assays have demonstrated that in vitro-translated PXR/RXRα or CAR/RXRα heterodimers bind to the DR4-like elements. However, the binding was readily displaced by ER6/3A4 or DR4/2B6, suggesting that the DR4-like elements have weak binding affinity toward PXR or CAR. Adenosine has never been found as the third nucleotide in the consensus half-site of any genes. Thus, the difference in the DR4-like sequence might be responsible for the weak binding affinity of the receptors. That there was no transactivation of CYP2A6 by the transfection of PXR or CAR alone might be due to the weak binding affinity. Consequently, the inducibility of CYP2A6 would be lower than CYP3A4 and CYP2B6. Accumulating evidence suggests that cross-talk between PXR and CAR has a role in the transactivation of target genes. Although the contribution of CAR was not directly demonstrated, we cannot exclude the role of CAR in the regulation of CYP2A6.

Several studies suggested that HNF-4α had synergistic effects on the regulation of the CYP3A4, CYP2C8, and CYP2C9 genes by PXR (Tirona et al., 2003; Chen et al., 2005; Ferguson et al., 2005). In the CYP3A4 gene, an HNF-4α binding site and a distal PXR response element are closely located at –7872 and –7822, respectively. In the CYP2C9 gene, HNF-4α binding sites are located at –152 and –185, and a PXR response element is located at –1839. In the CYP2A6 gene, an HNF-4α binding site is located on –81 for basal transcription (Pitarque et al., 2005). In contrast to CYP3A4 and CYP2C, synergistic effects between PXR and HNF-4α were not observed for CYP2A6. The distance between DR4-like elements and the HNF-4α binding site would be critical for cooperative transcriptional activation.

PGC-1α was originally identified as a peroxisome proliferator-activated receptor-γ-interacting coactivator in brown adipose tissue (Puigserver et al., 1998) and is a versatile coactivator for numerous nuclear receptors. PGC-1α interacts with both PXR and HNF-4α. It has been reported that PXR interferes with HNF-4α signaling for the CYP7A1 gene by targeting PGC-1α (Bhalla et al., 2004). Recently, Li and Chiang (2006) reported the interaction of PXR with HNF-4α and that its coactivators PGC-1α and steroid receptor coactivator-1 contribute to the strong induction of CYP3A4 by rifampicin. We found that PXR and PGC-1α synergistically up-regulate the CYP2A6 gene. Endogenous HNF-4α expressed in HepG2 cells might also contribute to the regulation. The transactivation by PGC-1α alone might be due to the endogenous ligand-activated PXR or other unknown factors in HepG2 cells.

It has been reported that the expression levels of PXR or CAR are significantly correlated with the expression levels of mRNA of the target genes, such as CYP2B6 (Chang et al., 2003), CYP3A4 (Vyhlidal et al., 2006), CYP3A5 (Burk et al., 2004), and CYP3A7 (Vyhlidal et al., 2006). Because we could demonstrate the induction of CYP2A6 via PXR, it would be of interest to investigate whether the expression level of CYP2A6 may be correlated with the PXR expression level. The variability of the PXR level might be one of the factors in the large interindividual variability in CYP2A6 mRNA. Recently, we found that in vivo nicotine metabolism catalyzed by CYP2A6 is higher in females than in males (Nakajima et al., 2006). This is reminiscent of the finding that the expression levels of PXR, CAR, CYP2B6, and CYP3A4 were higher in female than in male livers. The association is intriguing, although the factors regulating these sex differences in human liver are unknown.

The induction of P450 results in enhanced metabolism and clearance of the substrate/inducer itself or other coadministered drugs. CYP2A6 is a metabolic enzyme of nicotine (Nakajima et al., 1996). Recently, it has been reported that nicotine would be a ligand of human PXR (Lamba et al., 2004). Nicotine readily penetrates the brain and induces CYP2B in rodent and human brain (Miksys et al., 2003) and Cyp3a in mouse brain (Hagemeyer et al., 2003). Thus, nicotine might also induce CYP2A6 expression in human brain, resulting in increased tolerance to nicotine.

In summary, we have identified three DR4-like elements on the CYP2A6 gene to which PXR and CAR can bind. Among them, DR4-like elements at –5476 and –4618 are essential for PXR- and PGC-1α-dependent transactivation of the CYP2A6 gene. This is the first study to demonstrate the molecular mechanism of the induction of CYP2A6.