Regulation of Cyp2a5 transcription in mouse primary hepatocytes: roles of hepatocyte nuclear factor 4 and nuclear factor I

Abstract

The cytochrome P4502a5 (Cyp2a5) gene is expressed principally in liver and olfactory mucosa. In the present study, the transcriptional mechanisms of hepatocyte-specific expression of Cyp2a5 were studied in mouse primary hepatocytes. The Cyp2a5 5′-flanking region −3033 to +10 was cloned in front of a luciferase reporter gene and transfected into hepatocytes. Deletion analysis revealed two major activating promoter regions localized at proximal 271 bp and at a more distal area from −3033 to −2014 bp. The proximal activation region was characterized further by DNase I footprinting, and a single clear footprint was detected in the studied area centred over a sequence similar to the NF-I (nuclear factor I)-binding site. The binding of NF-I was confirmed using an EMSA (electrophoretic mobility-shift assay). A putative HNF-4 (hepatocyte nuclear factor 4)-binding site was localized at the proximal promoter by computer analysis of the sequence, and HNF-4α was shown to interact with the site using an EMSA. The functional significance of HNF-4 and NF-I binding to the Cyp2a5 promoter was evaluated by site-directed mutagenesis of the binding motifs in reporter constructs. Both mutations strongly decreased transcriptional activation by the Cyp2a5 promoter in primary hepatocytes, and double mutation almost completely abolished transcriptional activity. Also, the functionality of the distal activation region was found to be dependent on the intact HNF-4 and NF-I sites at the proximal promoter. In conclusion, these results indicate that HNF-4 and NF-I play major roles in the constitutive regulation of hepatic expression of Cyp2a5.

INTRODUCTION

Cytochrome P450 (CYP) enzymes constitute a superfamily of haemoproteins that play a major role in the detoxification of xenobiotics [1]. Xenobiotic-metabolizing CYP enzymes are expressed in a tissue-selective manner, and the main site of detoxification is the liver. The CYP2A5 enzyme is the major catalyst of coumarin 7-hydroxylation in mouse liver [2,3]. As far as its regulation is concerned, Cyp2a5 seems to differ from most other CYP enzymes: it can be induced by structurally unrelated compounds and also by several chemicals that usually repress other CYP forms. Cyp2a5 is induced by classical inducers, such as phenobarbital [4], and by various hepatotoxic agents, such as cocaine, pyrazole, solvents, heavy metals, hexachlorobutadiene and carbon tetrachloride [5–8]. Cyp2a5 is also up-regulated by cAMP [9]. Elevated CYP2A5/CYP2A6 protein levels in hepatocytes are associated with the development of liver tumours in mice [10] and in humans [11] respectively. In addition, inflammation of the liver caused by hepatitis B virus induces Cyp2a5 [12]. In the human liver, the orthologous form CYP2A6 increases in conjunction with liver cirrhosis [13]. It has been proposed that CYP2A5 may be one of the major xenobiotic-metabolizing CYP forms in a damaged liver [14].

The mouse Cyp2a5 and the human CYP2A6 genes are both predominantly expressed in hepatocytes, but they are also present in some extrahepatic tissues, mainly olfactory mucosa [15,16]. On the other hand, the orthologous rat isoform CYP2A3 is not expressed in liver, and CYP2A3 has been detected only in olfactory mucosa and lung [15,16]. The mechanisms of liver-selective expression of Cyp2a5 are poorly understood. Cyp2a5 displays circadian expression in mouse liver, with circadian accumulation of CYP2A5 protein in mouse liver microsomes [9,17]. This circadian expression is regulated by the PAR family transcription factor DBP (D-binding protein) [17]. Otherwise, there is no information on the factors that mediate Cyp2a5 activation in liver.

In the present study, we studied the transcriptional mechanisms of liver-selective expression of Cyp2a5 in a mouse primary hepatocyte model. We localized the main regulatory regions and showed that the transcription factors HNF (hepatocyte nuclear factor)-4 and NF-I (nuclear factor I) play a central role in the activation of Cyp2a5 transcription in hepatocytes.

EXPERIMENTAL

Isolation and culturing of hepatocytes

Mouse primary hepatocytes were isolated from DBA/2 mouse liver as described previously [18]. The mouse liver was perfused with collagenase solution (Worthington Biochemical, Freehold, NJ, U.S.A.), and liver cells were collected. After filtration and centrifugation, the isolated hepatocytes were dispersed in William's medium E (Sigma, St. Louis, MO, U.S.A.) containing 20 ng/ml dexamethasone (Sigma), ITS (5 mg/l insulin/5 mg/l transferrin/5 μg/l sodium selenate; Sigma), 10 μg/ml gentamicin (Invitrogen, Carlsbad, CA, U.S.A.) and 10% (v/v) foetal bovine serum (Invitrogen) at a density of 300000 cells/well in a twelve-well plate or 5×10⁶ cells/90-mm-diameter dish. The cultures were maintained at 37 °C in a humidified incubator for 2–3 h, after which non-attached cells were discarded by aspiration, and the medium was replaced by serum-free William's medium E. The cultures were maintained for additional 24 h before transient transfection or nuclear extract preparation.

Cell lines

HepG2 cells and COS-1 cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen) containing 10% (v/v) foetal bovine serum, 100 units/ml penicillin and 100 μg/ml streptomycin (Invitrogen).

Cloning of Cyp2a5 5′-flanking region

Genomic DNA was prepared from male DBA/2J mouse spleens. DNA was partially digested with MboI, treated with calf intestinal phosphatase and ligated to the arms of the lambda EMBL 3 vector. The library was screened with a ³²P-labelled 2 kb fragment of the Cyp2a5 gene (clone 29) [19], and 17 positive clones were obtained. The clone X1412-1 contained a 16.7 kb insert and was the longest among the Cyp2a5-containing clones. The identity of the insert was verified with the presence of a HindIII site in exon 4. The clone contained the Cyp2a5 exons 1 to 4 and more than 13 kb of the 5′-flanking region. A part of the insert was subcloned and sequenced.

Plasmids

Reporter plasmid: the Cyp2a5 5′ −3033 to +10 (from the transcriptional start site) fragment was prepared by PCR amplification from DBA/2 mouse genomic DNA and cloned into the pGL3-basic vector (Promega, Madison, WI, U.S.A.) in front of the luciferase reporter gene. This construct was designated as Cyp2a5 −3033-Luc. In addition, several shorter 5′ deletion constructs were prepared by PCR using the Cyp2a5 −3033-Luc plasmid as a template and subcloning the PCR products. The correct identity of the constructs was verified by sequencing. The rat HNF4α1 expression plasmid [20] was kindly provided by Dr Mary C. Weiss (Unité de Génétique de la Différenciation, Département de Biologie de Developpement, Institut Pasteur, Paris Cedex 15, France).

Site-directed mutagenesis

Site-directed mutagenesis was performed with the Quik-Change™ Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, U.S.A.), according to the manufacturer's instructions. Briefly, the mutations were introduced into Cyp2a5 −3033-Luc and Cyp2a5 −271-Luc plasmids using mutated oligonucleotides (NF-I mutation, −141 5′-CCTCCTCCTCCAGTGTTAACAATGTCAAAAACTTGGTGCAC-3′ −101; HNF-4 mutation, −73 5′-GTTGCATAATCAAGACCAAAGTAAGTCCTTCTGTCTCTGGATG-3′ −31; the mutated nucleotides are indicated in bold) and PfuTurbo DNA polymerase followed by selective digestion of the original template plasmid with the restriction enzyme DpnI. Correct assembly of the mutations was confirmed by sequencing.

Transient transfection

After 24 h in culture, the mouse primary hepatocytes were washed with William's medium E and transiently transfected with experimental Cyp2a5-Luc plasmid (0.5 μg/well) using Tfx™-20 lipid-reagent (Promega) and OptiMEM I medium (Invitrogen). The pRL-TK plasmid (0.1 μg/well) (Promega) containing the Renilla luciferase reporter gene was co-transfected with the experimental Cyp2a5-Luc constructs to provide an internal control for transfection efficiency. After a transfection period of 1 h, 1 ml of William's medium E containing dexamethasone, ITS and gentamicin was added. The cells were maintained for an additional 48 h, after which they were lysed and their luciferase activity was assayed with the Dual-Luciferase Reporter Assay System (Promega).

The COS-1 cells were seeded on 24-well plates 1 day before transfection. Cells were transiently transfected with 0.65 μg of reporter plasmid, 50 ng of pRL-TK control plasmid and 50 ng of expression vector per well using Tfx™-20 lipid-reagent and OptiMEM I medium. The cells were maintained for an additional 24 h, after which they were lysed and their luciferase activity was measured with the Dual-Luciferase Reporter Assay System.

Preparation of nuclear extracts

Nuclear extracts were prepared according to Schreiber et al. [21]. Cultured mouse primary hepatocytes, or HepG2 cells, (90-mm-diameter dish) were first washed with 10 ml of PBS, then scraped into another 10 ml of PBS, and pelleted by centrifugation at 110 g for 5 min, suspended in 1 ml of PBS and centrifuged at 660 g for 15 s. The cell pellet was resuspended in cold hypotonic buffer A (10 mM Hepes, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol and 0.5 mM PMSF), and the cells were allowed to swell on ice for 15 min. The cell membranes were lysed by adding 10% solution of Tergitol (type Nonidet P40) (Sigma) to a final percentage of 0.6%. The homogenate was centrifuged at 660 g for 30 s, and the nuclear pellet was resuspended in cold buffer C (20 mM Hepes, pH 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol and 1 mM PMSF). The nuclear proteins were extracted by incubation at 4 °C for 15 min on a shaking platform. The samples were centrifuged at 15000 g for 5 min, and the supernatant fractions containing the nuclear proteins were collected. The protein content of the nuclear extract was determined using the Bradford protein analysis method [22].

Nuclear extracts from DBA/2 mouse liver were prepared similarly, except that the tissue was first homogenized in lysis buffer [20 mM Tris/HCl, pH 7.5, 10 mM NaCl, 0.1 mM EDTA, 0.1 mM EGTA and Complete™ Mini protease inhibitor cocktail (Roche Diagnostics GmbH, Mannheim, Germany)] (4 ml/g of tissue).

DNase I footprinting

The −271 to +10 fragment of the Cyp2a5 5′-flanking region was PCR-amplified from the plasmid Cyp2a5 −271-Luc. The primers were designed to include 12 and 39 nucleotides of the vector sequence from the pGL3-basic at its 5′ and 3′ end respectively. The PCR product was then subcloned to the pCR 2.1-TOPO vector (Invitrogen). The DNA probe for footprinting was then prepared from the Cyp2a5 5′ −271 to +10 in pCR 2.1-TOPO by digestion with EcoRI. The fragment was treated with alkaline phosphatase and end-labelled with T4 polynucleotide kinase using [γ-³²P]ATP. To produce a probe labelled at one end only, one of the labelled ends was cleaved either with HindIII or with KpnI. The probe was finally purified using the QIAquick gel extraction kit (Qiagen, Venlo, The Netherlands).

DNase I footprinting analyses were performed with the Core Footprinting System (Promega), essentially according the manufacturer's instructions. Labelled probe (3.5 ng of 20000 c.p.m.), 0.4 μg of poly(dI-dC)·(dI-dC), 5 μg of mouse hepatocyte nuclear extract and 25 μl of binding buffer [50 mM Tris/HCl, pH 8.0, 100 mM KCl, 12.5 mM MgCl₂, 1 mM EDTA, 20% (v/v) glycerol and 1 mM dithiothreitol] were incubated at room temperature (22 °C) for 30 min in a final volume of 50 μl. DNase reaction buffer (containing 5 mM CaCl₂ and 10 mM MgCl₂) (50 μl) was added, and the reaction mixtures were incubated at room temperature for an additional 1 min. RQ1 RNase-free DNase (0.15 unit) was then added, and the DNase digestion was allowed to proceed for 1 min at room temperature. The reactions were terminated by adding 90 μl of the DNase stop solution [200 mM NaCl, 30 mM EDTA, 1% (w/v) SDS and 100 μg/ml yeast RNA]. DNA was recovered from the reaction mixtures by phenol/chloroform/isopentanol extraction and ethanol precipitation. The samples were analysed by electrophoresis through 6% (w/v) polyacrylamide DNA sequencing gel and detected by autoradiography. The region protected from DNase I cleavage was localized by using a chemical A+G sequencing reaction [23].

EMSA (electrophoretic mobility-shift assay)

Single-stranded oligonucleotides were purchased from Sigma Genosys. Double-stranded oligonucleotides were prepared by annealing the desired sense and antisense oligonucleotides (Table 1). Double-stranded oligonucleotides were 5′-end-labelled with [γ-³²P]ATP and T4 polynucleotide kinase and then purified using the QIAquick nucleotide removal kit.

Table 1. Sequence of the double-stranded oligonucleotides used in EMSA.

The core sequence is underlined and the mutations are indicated in bold.

Oligonucleotide	Sequence	Location
NF-I EMSA
Cyp2a5	5′-CCTCCTCCAGTGTTGGCAATGTCCCAAACTTGGTGCAC-3′	−138 to −101
Cyp2a5 mutation	5′-CCTCCTCCAGTGTTAACAATGTCAAAAACTTGGTGCAC-3′	−138 to −101
CYP2A3	5′-CCTCCTTGAGTGTTGGCTATGTCCCAAACTAGG-3′	−138 to −106
NF-I consensus	5′-GGCACCTGTTTCAATTTGGCACGGAGCCAACAG-3′
NF-I consensus mutation	5′-GGCACCTGTTTCAATTTGTTACGGAGTTAACAG-3′
HNF-4 EMSA
Cyp2a5	5′-GTTGCATAATCAAGACCAAAGTCCGTCCTTCTGTCTCTGGATG-3′	−73 to −31
Cyp2a5 mutation	5′-GTTGCATAATCAAGACCAAAGTAAGTCCTTCTGTCTCTGGATG-3′	−73 to −31

EMSAs for NF-I were performed with the Nushift™ kit (Geneka Biotechnology, Montreal, Canada), essentially according to the manufacturer's instructions. Extract pre-mixtures containing 14 μg of mouse liver nuclear extract, binding buffer B1 and stabilizing solution D were pre-incubated at 4 °C for 20–45 min in a final volume of 16 μl. Probe pre-mixtures containing 0.08 pmol (16000 c.p.m.) of labelled double-stranded Cyp2a5 oligonucleotide, binding buffer C1, stabilizing solution D and, for competition experiments, unlabelled competitor oligonucleotide (at 5–100-fold excess) in a final volume of 8 μl, were added to the extract pre-mixtures and incubated at 4 °C for an additional 20 min. For supershift experiments, 3 μl of rabbit anti-NF-I polyclonal antibody [α-CTF (CAAT box transcription factor) antiserum] (a gift from Dr Naoko Tanese, Department of Microbiology, New York University School of Medicine, New York, NY, U.S.A.) was added to the extract pre-mixture before pre-incubation. The samples were separated by electrophoresis on a 6% (w/v) polyacrylamide gel, and the retarded complexes were detected by autoradiography.

The HNF-4 EMSAs were performed using 10–20 μg of mouse liver nuclear extract or HepG2 nuclear extract, binding buffer [25 mM Hepes, pH 7.9, 10% (v/v) glycerol, 50 mM KCl, 0.5 mM EDTA and Complete™ Mini protease inhibitor cocktail], 1 μg/μl ssDNA and 0.04 pmol (30000 c.p.m.) of labelled double-stranded Cyp2a5 oligonucleotide. The reaction mixtures were incubated at room temperature for 30 min in a final volume of 15 μl. For competition experiments, unlabelled competitor oligonucleotides (at 10–100-fold excess) were included in the mixtures. For supershift experiments, nuclear extracts were pre-incubated with 0.2 μg of goat anti-HNF-4α polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA, U.S.A.) on ice for 20 min. The samples were separated by electrophoresis on a 6% (w/v) polyacrylamide gel, and the retarded complexes were detected by autoradiography.

RESULTS

Cloning of the Cyp2a5 5′-flanking region

The DBA/2J mouse genomic DNA library was screened with the Cyp2a5 genomic fragment 29 [19], and a clone containing at least 13 kb of the Cyp2a5 5′-flanking region was obtained. A part of the insert was subcloned, and 3033 bp upstream from the transcription start site of the 5′-flanking region was sequenced. With the NCBI Blast program (http://www.ncbi.nlm.nih.gov), the Cyp2a5 5′ −3033 to −1 sequence was found to be 85% similar to the rat CYP2A3 gene (GenBank® accession number M33190) 5′-flanking region.

Localization of the upstream activation regions of the Cyp2a5 gene

To identify the sequences responsible for the transcriptional activation of the mouse Cyp2a5 gene in primary hepatocytes, a series of 5′-truncated mouse Cyp2a5 promoter–luciferase reporter plasmids were constructed and transfected into mouse primary hepatocytes (Figure 1). The highest luciferase activity was produced by the longest Cyp2a5 −3033-Luc construct. Deletion of the sequence from −3033 to −2014 decreased the transcriptional activity by 86%, indicating the presence of activating enhancer elements in this region. Removal of the sequence from −2013 to −1021 did not markedly affect luciferase activity, but deletion of the sequence from −1020 to −272 increased this activity >3-fold relative to the Cyp2a5 −1020-Luc construct, implying the presence of repressing sequence elements in the region of −1020 to −272. A 5′ deletion from −271 to −64 decreased transcriptional activity by 90%, bringing it close to the basal level of the promoterless control plasmid pGL3-basic. Thus the proximal promoter region from −64 to −271 contains activating cis-elements and, in addition to the distal enhancer at −3033 to −2014, forms a second activation region for the Cyp2a5 gene.

Figure 1. Transcriptional activity of Cyp2a5 5′-luciferase constructs transfected to mouse primary hepatocytes.

The cells were harvested and luciferase activities were measured 48 h after transfection. The activities produced by the studied constructs were normalized against co-transfected control plasmid (pRL-TK) activities. The values represent means+S.D. of four individual samples. The experiment was repeated twice with similar results.

DNase I footprinting analysis of the proximal promoter

DNase I footprinting was used to locate the protein-binding sites within the proximal activation region. A −271 to +10 Cyp2a5 probe was labelled and incubated with hepatocyte nuclear proteins, after which the reaction mixture was treated with DNase I. A single clear, protected footprint was detected in the non-coding strand and located by sequence analysis at −132 to −107 (Figure 2). Analysis of the coding strand produced a similar result (results not shown). A search with the MatInspector professional program (http://www.gene-regulation.com/pub/databases.html#transfac) revealed this region to be centred over a sequence similar to the consensus binding site of transcription factor NF-I.

Figure 2. DNase I footprinting of the Cyp2a5 proximal promoter.

A −271 to +10 Cyp2a5 probe was labelled and incubated with hepatocyte nuclear proteins, after which the reaction mixture was treated with DNase I. The resulting fragments were separated by PAGE and detected by autoradiography. The footprint was localized by chemical A+G sequencing reaction (results not shown). The footprint was identified to be similar to the NF-I motif with the MatInspector program using TRANSFAC 4.0 matrices.

Characterization of the protein binding to the −132 to −107 element

The proteins binding to the −132 to −107 footprinted sequence were elucidated further by EMSA. A 38 bp double-stranded, end-labelled oligonucleotide from position −138 to −101 of the Cyp2a5 promoter was incubated with nuclear extract from mouse liver, and the resulting DNA–protein complexes were separated by gel electrophoresis. Two major retarded complexes were detected (Figure 3). The upper complex was competed by a 10–100-fold excess of unlabelled probe itself and also by a 5–50-fold excess of oligonucleotide-containing consensus binding site for the transcription factor NF-I (Table 1). Mutation of the nucleotides known to be critical for NF-I binding abolished competition by the NF-I consensus oligonucleotide. Furthermore, mutation of the NF-I core sequence in the Cyp2a5 oligonucleotide also abolished competition. Thus the competition reactions of the EMSA assay suggest that the detected footprint results from interactions with an NF-I-like transcription factor or factors. The lower band was not competed by the NF-I consensus oligonucleotide and was not affected by the mutation of the Cyp2a5 oligonucleotide, suggesting that this retarded complex does not involve NF-I-like proteins.

Figure 3. Characterization of NF-I binding to the Cyp2a5 promoter by EMSA.

(A) Interaction of mouse liver nuclear proteins with the Cyp2a5 5′ −138 to −101 sequence in EMSA. Lane 1 represents a binding reaction with no protein and lane 2 is a control reaction with no competing oligonucleotides. The other lanes represent competition reactions with different unlabelled oligonucleotides as indicated. The sequences of the oligonucleotides are presented in Table 1. (B) Supershift analysis of Cyp2a5 promoter −138 to −101 binding proteins with anti-NF-I antibody. The mouse liver nuclear extract was pre-incubated with anti-NF-I antibody before the addition of 5′-end-labelled DNA oligonucleotide as described in the Experimental section. The effect of the antibody was compared with a control reaction with no antibody. The retarded protein–DNA complex competed by excess of unlabelled NF-I consensus oligonucleotides is indicated with an arrow on the left-hand side of the Figure. The anti-NF-I antibody supershifted complex is indicated with an arrow on the right-hand side of the Figure.

Involvement of NF-I in the formation of the detected DNA–protein complex was verified further by using the anti-NF-I antibody. A clear supershifted band was detected after incubation with the anti-NF-I antibody, and the upper complex was shifted (Figure 3). Thus the supershifting assay confirmed that an NF-I protein really binds to the Cyp2a5 promoter region detected by DNase I footprinting.

Identification of the HNF-4 binding site in the Cyp2a5 proximal promoter

Computer analysis (MatInspector) of the Cyp2a5 gene revealed a putative binding site for the transcription factor HNF-4α in the proximal promoter at position −63 to −47. We were unable to detect any footprint at that position in our DNase I footprinting analysis of the proximal promoter. However, HNF-4 is involved in transcriptional activation of several CYP genes, including the closely related Cyp2a4 [24], and we therefore examined further whether HNF-4 plays any role in the regulation of Cyp2a5.

The putative Cyp2a5 HNF-4 DNA element was subjected to EMSA analysis using nuclear extracts from mouse liver and HepG2 cells (that constitutively express HNF-4) (Figure 4). Strong retarded complexes were formed with both liver and HepG2 extracts. The bands were abolished by mutation of the nucleotides known to be critical for HNF-4 binding [24]. The bands were supershifted by HNF-4-specific antibody confirming that HNF-4 is able to bind to the Cyp2a5 HNF-4-binding element. We next compared the protein binding to the Cyp2a5 HNF-4 element using nuclear extracts from primary hepatocytes, liver and HepG2 cells. The strongest binding was detected with the HepG2 extract followed by the liver extract. The primary hepatocyte extract produced a very faint band (clearly visible only upon extended exposure) compared with liver or HepG2, indicating that low levels of HNF-4 are present in primary hepatocytes.

Figure 4. Characterization of HNF-4 binding to the Cyp2a5 promoter by EMSA.

(A, B) Interaction of mouse liver (A) and HepG2 (B) nuclear proteins with the Cyp2a5 5′ −73 to −31 sequence in EMSA. The retarded protein–DNA complexes are indicated with arrows. Lane 1 represents a binding reaction with no protein, and lane 2 is a control reaction with no competing oligonucleotides. The other lanes represent competition reactions with different unlabelled oligonucleotides as indicated. The sequences of the oligonucleotides are presented in Table 1. (C) Supershift analysis of the Cyp2a5 promoter −73 to −31 binding proteins with anti-HNF-4α antibody. The mouse liver and HepG2 nuclear extracts were pre-incubated with anti-HNF-4α antibody before the addition of 5′-end-labelled DNA oligonucleotide as described in the Experimental section. The effect of the antibody was compared with a control reaction with no antibody. The retarded protein–DNA complexes formed in the control reaction are indicated with arrows on the left-hand side of the figure. The anti-HNF-4 antibody supershifted complex is indicated with an arrow on the right-hand side of the figure. (D) Comparison of HNF-4 binding to the Cyp2a5 promoter HNF-4-binding element using nuclear extracts from mouse primary hepatocytes, mouse liver and HepG2 cells. Extract protein (20 μg) was used for the experiment.

HNF-4 activates Cyp2a5 transcription

We next studied whether or not HNF-4 is able to trans-activate Cyp2a5. The Cyp2a5 promoter construct Cyp2a5 −271-Luc was transfected together with the HNF-4α expression plasmid to HNF-4-deficient COS-1 cells. HNF-4 activated this construct 5.2-fold compared with the control without HNF-4 expression. Furthermore, mutation of the Cyp2a5 −63 to −47 HNF-4-binding site abolished activation (Figure 5).

Figure 5. Effect of HNF-4 co-transfection on Cyp2a5 promoter activity in COS-1 cells.

The expression vector for HNF-4α was co-transfected with the Cyp2a5 −271-Luc reporter plasmid or with the reporter plasmid containing a mutated HNF-4-binding site (Cyp2a5 −271-Luc HNF-4 mut) into COS-1 cells. The cells were harvested and luciferase activities were measured 24 h after transfection. The activities produced by the studied constructs were normalized against co-transfected control plasmid (pRL-TK) activities. The results are presented as fold activation of the control experiment with no expression vector. The values represent means+S.D. of four individual samples. ***P<0.001 compared with the control without HNF-4α expression vector co-transfection (independent-samples t test). The experiment was repeated twice with similar results.

Regulation of Cyp2a5 transcription in primary hepatocytes by NF-I and HNF-4

The contribution of NF-I and HNF-4 to the transcriptional regulation of Cyp2a5 in primary hepatocytes was examined using mutations of the Cyp2a5 promoter–luciferase reporter plasmids. Mutation of the HNF-4 site strongly decreased the transcription of the Cyp2a5 −3033-Luc and Cyp2a5 −271-Luc constructs to only 1.6% and 4.5% of the control activities respectively. NF-I site mutation also very efficiently decreased the transcription of Cyp2a5 −271-Luc to 14% of the control and that of Cyp2a5 −3033-Luc to 15% of the control. The combined effect of HNF-4 and NF-I mutations was studied using double mutation of the Cyp2a5 −271-Luc construct. Mutation of both transcription-factor-binding sites almost completely prevented transcription (Figure 6).

Figure 6. Effect of NF-I- and HNF-4-binding sites on Cyp2a5 transcription in mouse primary hepatocytes.

The Cyp2a5 5′-Luc plasmids Cyp2a5 −3033-Luc (A) and Cyp2a5 −271-Luc (B) with mutated NF-1- or/and HNF-4-binding sites were prepared and transfected into mouse primary hepatocytes. The activities produced by the studied constructs were normalized against co-transfected control plasmid (pRL-TK) activities. The luciferase activities produced by the mutant constructs were compared with the activities produced by the construct with intact NF-I- and HNF-4-binding sites. The values are presented as percentages of control (unmutated construct) and represent the means+S.D. of four individual samples. ***P<0.001 compared with the unmutated construct (one-way ANOVA, followed by least-significant difference post-hoc test). The experiment was repeated twice with similar results.

DISCUSSION

Many genes display tissue-specific regulation or are expressed at variable levels in different cell types. Tissue-specific gene expression is regulated primarily by tissue-enriched transcription factors. This tissue-specific regulation has been most extensively studied in liver. Six families of liver-enriched transcription factors [HNF-1, HNF-3, HNF-4, HNF-6, C/EBP (CCAAT/enhancer-binding protein) and DBP] play an important role in liver-specific gene regulation [25]. However, the functional analysis of these transcription factors has led to the conclusion that the co-operation of liver-enriched transcription factors with ubiquitous transactivating factors (such as Sp1 and NF-I) is necessary for the maintenance of liver-specific gene expression [26]. These two groups of transcription factors have also been shown to act in concert in the activation of the liver-specific gene transcription of many genes, including albumin [27].

The transcription factor family HNF-4 consists of three members, i.e. HNF-4α, HNF-4β and HNF-4γ, and controls numerous important metabolic functions, including the metabolism of glucose, cholesterol and fatty acids [25]. The three members of the HNF-4 family display overlapping tissue expression, but HNF-4α plays the major role in liver. HNF-4 belongs to the large family of nuclear receptors and is considered to be an orphan receptor, but fatty acyl-CoA thioesters have been suggested as ligands for HNF-4α [28]. HNF-4α participates in the development and maintenance of the hepatic phenotype [29]. Interestingly, the transcription of some hepatic genes requires the constant presence of HNF-4α, while others, such as PXR (pregnane X receptor), only need to be switched on during development [30,31]. Along with many other liver-expressed genes, several members of the cytochrome P450 superfamily are also under the direct regulation of HNF-4α [32]. Especially, many members of the CYP 2 family have putative HNF-4α binding sites in the proximal promoter. Some of these genes, such as the rabbit 2C1, CYP2C2 and CYP2C3, and the human CYP2C9 and CYP2D6, have actually been found to be regulated by HNF-4α [32]. However, rat CYP2C7, CYP2C11, CYP2C12 and CYP2C13, for example, are not regulated to any major extent by HNF-4α, despite the putative binding site in the promoter [32].

HNF-4 participates in the hepatic regulation of some CYP2A forms. Yokomori et al. [24] found that HNF-4 is able to bind to the Cyp2a4 proximal promoter and activates the Cyp2a4 gene. Furthermore, Jover et al. [33] found that adenoviral transfection of HNF-4α antisense to human hepatocytes strongly down-regulated CYP2A6 expression, suggesting that HNF-4α plays an important role in the regulation of CYP2A6. In the present study, we showed that the Cyp2a5 gene is also regulated by HNF-4α. This is supported by several lines of evidence. The putative HNF-4-binding site in the Cyp2a5 proximal promoter is able to bind HNF-4α. Co-transfection of the HNF-4α expression plasmid to COS-1 cells activates the transcription of the Cyp2a5 promoter construct, but this activation is abolished by mutation of the HNF-4-binding site. Furthermore, in mouse primary hepatocytes, mutation of the Cyp2a5 promoter HNF-4-binding site strongly decreases the transcriptional activation of the Cyp2a5 promoter at both −271 bp and −3033 bp of the 5′-flanking region. The HNF-4-binding site is located at position −63 to −47. However, the promoter region −63 to +10 produced very little transcriptional activity. Apparently, the sequence immediately upstream from the binding site is necessary for proper functioning of HNF-4. This sequence may be required for binding, or alternatively, co-operation with factors that interact with a region further upstream may be necessary.

NF-I proteins mediate both the initiation of transcription and DNA replication, and they are encoded by four distinct genes (NF-IA, NF-IB, NF-IC and NF-IX), which are conserved in every vertebrate species examined, including mouse [34] and humans [35,36]. NF-I proteins have a conserved N-terminal region that mediates DNA binding and dimerization [37]. Each NF-I gene transcript can be alternatively spliced, generating various isoforms of NF-I proteins [38]. Different NF-I isoforms contain variable C-terminal transactivating domains [37]. Some isoforms can also repress transcription [39]. NF-I isoforms form stable homo- and hetero-dimers in all possible combinations [40], but homodimers have been shown to be transcriptionally more active than heterodimers [41].

Different levels of alternatively spliced NF-I transcripts are present in different cell types and tissues, suggesting a potential role for these transcription factors in tissue-specific gene expression [34,35]. In concert with liver-enriched transcription factors, NF-I proteins have been shown to regulate the liver-specific transcription of many genes, including albumin [27] and vitellogenin [42]. NF-I proteins have also been shown to participate in the regulation of CYP gene expression. NF-IC is essential for the transcription of the CYP17 gene in human adrenal NCI-H295A cells [27,43]. Moreover, the PBREM (phenobarbital-responsive enhancer module) of the mouse Cyp2b10 gene contains an NF-I-binding site flanked by two nuclear-receptor-binding sites. This NF-I site is important for both the induction response and the basal expression of the Cyp2b10 gene [44]. The expression of the rat CYP1A2 gene, expressed selectively in liver and olfactory mucosa, also seems to be regulated by NF-I [45].

The present paper provides evidence that the expression of the Cyp2a5 gene in mouse primary hepatocytes is under the control of transcription factor NF-I. DNase I footprinting of the proximal promoter region from −271 to the transcriptional start site revealed that mouse primary hepatocyte nuclear proteins interacted with the promoter region from −132 to −107 corresponding to a putative binding site of the transcription factor NF-I. Using an EMSA, a protein complex, which was supershifted by an antibody specific against NF-I, was found to interact with the identified DNA sequence region. Site-directed mutagenesis of the NF-I-binding site in the Cyp2a5 promoter constructs confirms the functional contribution of NF-I to the activation of Cyp2a5 gene expression in hepatocytes.

In 1998, Zhang and Ding [46] identified a conserved NF-I-like-binding site in the promoter region of the rat CYP2A3 gene, designated NPTA (nasal predominant transcriptional activating element), and proposed that this conserved binding site contributes to tissue-selective expression of the rat CYP2A3, mouse Cyp2a5 and human CYP2A6 genes in olfactory mucosa. In the present study, we showed that NF-I proteins are of importance for liver-enriched expression of Cyp2a5, and that an NF-I sequence motif in the Cyp2a5 promoter similar to the CYP2A3 NPTA element is involved. However, it seems probable that different NF-I isoforms are responsible for the transcriptional activation of Cyp2a5 in olfactory mucosa and liver. The NPTA-binding site of the rat CYP2A3 gene was reported to interact only with nuclear proteins from olfactory mucosa and not with liver nuclear extracts [46]. This is contradictory to our present observation showing that, in the EMSAs, the NPTA element of the CYP2A3 gene was able to compete with the Cyp2a5 promoter NF-I site for binding of the mouse liver nuclear proteins. The CYP2A3 gene is not expressed in liver to any appreciable extent, and the NF-I/NPTA element may therefore not be the sole determinant of liver-specific expression, but interaction with additional factors is possibly also required.

In the present study, we identified the two major factors activating the Cyp2a5 promoter. Both of these factors, HNF-4 and NF-I, act on the proximal promoter. An additional activation region was located at a further upstream region −3033 to −2014. Nevertheless, the function of HNF-4 appears to be indispensable also for the function of the distal activation region, as mutation of the proximal HNF-4 site decreased transcription to only 1.6% of the control activity in the context of the −3033 to +10 promoter. Mutation of the NF-I site decreased the transcriptional activity from the Cyp2a5 −3033-Luc construct significantly, but to a lesser extent than the HNF-4 site mutation. While separate mutations of the HNF-4- and NF-I-binding sites both strongly decreased transcription, the double mutation almost completely abolished transcription, indicating the vital role of these two transcription factors for constitutive regulation of Cyp2a5 gene expression. The crucial role of HNF-4 is emphasized further by the fact that the low levels of this transcription factor detected in primary hepatocytes are necessary and sufficient for Cyp2a5 transcription. In intact liver expressing higher levels of HNF-4, the contribution of HNF-4 to transcriptional activation of Cyp2a5 is probably also quantitatively even more important. The promoters of mouse Cyp2a5 and human CYP2A6 display significant sequence similarity (Figure 7). The HNF-4- and NF-I-binding sites are conserved in both genes, suggesting that these binding sites may also be important for the transcriptional regulation of CYP2A6 in human liver.

Figure 7. Alignment of the Cyp2a5 and CYP2A6 promoter sequences.

The conserved TATA box, HNF-4 and NF-I elements are indicated by shading.

In addition to HNF-4 and NF-I, several additional factors may participate in the regulation of Cyp2a5. Indeed, it has been shown that the DBP factor regulates Cyp2a5 transcription in a circadian fashion [17]. Furthermore, induction of Cyp2a5 transcription by such factors as phenobarbital and cAMP [4,18] may require the action of additional transcription factors. Nevertheless, HNF-4 and NF-I appear to be the major factors controlling the constitutive expression of Cyp2a5 and play essential roles in hepatic expression.