Introduction

Top Summary Introduction Materials and methods Results Discussion References

The cytochrome P-450 (CYP) superfamily consists of various isozymes that are classified into gene families and subfamilies based on the similarity in their primary amino acid sequences (Nelson et al., 1996). The isozymes catalyze xenobiotics including drugs and environmental chemicals as well as endogenous steroids (Gonzalez, 1989; Porter and Coon, 1991; Guengerich, 1997). Several members of CYP families are known to be induced markedly by these substances, including aromatic hydrocarbons, barbiturates, peroxisome proliferators, and steroids (Denison and Whitlock, 1995). Aromatic hydrocarbons such as 3-methylcholanthrene (3-MC) and 2,3,7,8-tetrachlorodibenzo-p-dioxin induce several CYPs including 1A1, 1A2, 1B1, and 2A8 (Fukuhara et al., 1989b; Denison and Whitlock, 1995). Although the induction mechanisms of these CYPs have not yet been fully elucidated, transcriptional activation mechanism of one of these genes, CYP1A1, has been studied extensively (Whitlock et al., 1996). The inducer binds to aryl hydrocarbon receptor (AhR) and it heterodimerizes with AhR nuclear translocator (Arnt). AhR-Arnt complex then binds to the enhancer element termed xenobiotic responsive element (XRE) located in the 5'-flanking region of the CYP1A1 gene.

In our laboratory, we have studied CYPs of the Syrian hamster and cloned and characterized CYP2A8, 3A31, and 2A9 (Fukuhara et al., 1989a; Alabouch et al., 1998; Kurose et al., 1998). These studies showed that induction mode of CYPs in the Syrian hamster is markedly different from that of the rat and mouse. Notably, the induction of CYP2A subfamily of the hamster is unique. In the rat and mouse liver, CYP1A1 and 1A2 are the major forms induced by 3-MC-type inducers (Conney, 1982; Nebert and Gonzalez, 1987; Whitlock, 1987), whereas in the hamster liver, it is not CYP1A1 but CYP2A8 that is a major form induced by 3-MC-type inducers (Sunouchi et al., 1988; Fukuhara et al., 1989b; M. Fukuhara, unpublished data). Expression of CYP2A isozymes is known to differ depending on species and isozymes (Honkakoski and Negishi, 1997). However, the mechanisms by which CYP2A isozymes are induced have been scarcely studied and any studies have not identified transcriptional elements on the CYP2A genes that may regulate the expression, except for nasal transcriptional activating element on CYP2A3 gene (Zhang and Ding, 1998).

To understand the mechanism of induction of the hamster CYP2A8, we studied the induction mode of CYP2A8 in primary hepatocyte cultures, which suggested the involvement of XRE-mediated CYP2A8 gene expression (Tohkin et al., 1996). Hence, in the present study, to further elucidate the regulation mechanism of the CYP2A8 gene expression, we analyzed the 5'-flanking region of the gene. We identified the functional XRE and furthermore, a novel positive regulatory element (PREX) that enhanced the XRE-mediated promoter activity of the CYP2A8 gene. We also demonstrated the presence of specific nuclear factors involved in the activation by PREX.

	Materials and Methods

Top Summary Introduction Materials and methods Results Discussion References

Animals, Cells, and Materials. Female Slc:Syrian hamsters (Nippon SLC Inc., Hamamatsu, Japan), aged between 6 to 8 weeks, were used for the preparation of primary cultures of hepatocytes. Drosophila Schneider's line 2 (SL2) cells were obtained from American Type Culture Collection (Rockville, MD). Schneider's Drosophila medium, fetal bovine serum, OPTI-MEM, Lipotectin and 5' rapid amplification of cDNA ends (RACE) System kit were purchased from GIBCO BRL (Gaithersburg, MD). Synthetic oligonucleotides were purchased from Greiner Japan Inc. (Tokyo, Japan). Luciferase reporter vector pGL2-Basic was obtained from Promega (Madison, WI) and EMBL3 vector and pBluescript SK- vector were from Stratagene (La Jolla, CA). Human AhR, Arnt, and Sp1 expression plasmids (pGEMAct-hAhR, pGEMAct-Arnt, pGEMAct-Sp1) (Kobayashi et al., 1996) and anti-rat AhR IgG (Matsushita et al., 1993) were kindly donated by Dr. Fujii-Kuriyama (Tohoku University, Sendai, Japan). All other reagents used were as described previously (Tohkin et al., 1996).

Cloning of CYP2A8 Genomic DNA. Syrian hamster genomic DNA prepared from spleens was partially digested with Sau3A I and ligated into the BamHI site of EMBL3 vector. Approximately 10⁶ phage plaques were screened initially with a 5'-end EcoRI fragment (0.6 kb) of CYP2A8 cDNA (Fukuhara et al., 1989a) as a probe. Four independent clones were isolated, and the genomic clone DNAs were further analyzed by restriction mapping and Southern hybridization with a CYP2A8 specific 5'-end oligonucleotide (5'-TGCCACCATGCTGGTGTCC-3'). One of the clones, designated L4, contained a 1.85-kb SalI fragment that hybridized to the 5'-end oligonucleotide probe. This fragment was then subcloned into the pBluescript SK- vector and sequenced. This sequence showed that the fragment had contained the first coding exon. Because the L4 phage also contained a further 15.6-kb upstream region, some fragments of the phage DNA were subcloned into the pBluescript SK- vector, which were used in subsequent promoter studies. To determine the DNA sequence and to construct CYP2A8-luciferase reporter gene fusion plasmids, a series of deletions of the CYP2A8 gene 5'-flanking region were generated by the exonuclease III/mung bean nuclease method. The 5.6-kb XbaI-SalI fragment (6.8 k to 1.2 k) was subcloned into the pBluescript SK-. After digestion with BstXI (in the multiple cloning site located at the upstream of the XbaI site) and XbaI, the DNA was incubated with exonuclease III for appropriate time intervals at 30°C. The remaining single-stranded DNAs were removed by mung bean nuclease treatment followed by self-ligation. The resulting deletion clones were then sequenced and used for construction of luciferase fusion plasmids.

DNA Sequencing and Analysis. Nucleotide sequences were determined by the dideoxy chain termination method using a DSQ1000 DNA sequencer (Shimadzu, Kyoto, Japan). The nucleotide sequences were analyzed and compared using GeneWorks version 2.5 sequence analysis software (Oxford Molecular Group Inc., CA). Transcriptional regulatory elements were searched through TRANSFAC database (Heinemeyer et al., 1998).

Nucleotide Sequence Accession Number. The nucleotide sequence data reported in this article has been submitted to the DDBJ/Gene Bank/EMBL Data Bank with the accession number AB001516.

Determination of Transcription Start Site. To determine the transcription start site of the CYP2A8 gene, the 5'-RACE experiment was performed using the 5' RACE System kit essentially according to the manufacturer's instructions. Forty nanograms of poly(A)⁺ RNA from female Syrian hamster liver was reverse-transcribed by a CYP2A8 cDNA-specific antisense primer (5'-GTATGAGAAAGCTGGTCTC-3'). The first-strand cDNA was amplified by PCR using the anchor primer provided with the system and a nested CYP2A8 cDNA-specific antisense primer (5'-ACTTCTCTGTGTCCAGCTCC-3'). A single band of about 230 bp was gel purified and sequenced directly with a upstream CYP2A8 cDNA-specific antisense primer (5'-TCTCCTCTGCCTCCACACAGAC-3').

Luciferase Reporter Gene Construction. A series of constructs were prepared, in which various lengths of the 5'-flanking region of the CYP2A8 gene were cloned upstream of the luciferase reporter gene in plasmid pGL2'-Basic. This plasmid was constructed by digestion of the luciferase reporter vector pGL2-Basic with SalI followed by self-ligation of the blunt-ended site. First, we subcloned the 1.2 kb SalI -BstXI fragment of the 1.85 kb SalI fragment of the L4 clone into pGL2'-Basic. Because the BstXI site includes translation initiation codon ATG, we eliminated the ATG codon as follows. The plasmid that contained the 1.85-kb SalI fragment was digested with BstXI and blunt ended with T4 DNA polymerase to remove the ATG, then digested with SalI. The resulting 1.2-kb fragment (1199 to +26, relative to the CYP2A8 gene transcriptional start site) was inserted into the SalI-blunt ended XhoI site of pBluescript SK-. Then, plasmid pGL2A8-1.2 was constructed by insertion of the 1.2-kb fragment, which was cut out with SacI and XhoI into the SacI-XhoI site of pGL2'-Basic. Plasmid pGL2A8-2.1 was constructed by insertion of the SacI-SalI fragment (2073 to 1198) of the 5'-flanking region into the SacI-SalI site of the pGL2A8-1.2. Plasmid pGL2A8-2.2 was constructed by insertion of the PstI-SalI fragment (2137 to 1198) of the 5'-flanking region into the PstI-SalI site of the pGL2A8-1.2. To prepare a series of deletions of the 5'-flanking region-luciferase fusion constructs, the deletion constructs described in the section Cloning of CYP2A8 Genomic DNA above were excised by digestion with SacI (in the vector and at 2073) or with SacI (in the vector) and SalI (at 1198). The series of deletion fragments were then subcloned into the SacI site or SacI-SalI site of pGL2A8-2.1. Plasmid pGL2A8-1 was constructed by deletion of the PvuII fragments of one of the deletion constructs pGL2A8-2.6 (2668) followed by self-ligation. Plasmid pGL2A8-2 was constructed by insertion of double-stranded synthetic oligonucleotide OL3 (2452 to 2421) into the PvuII site of pGL2A8-1. Plasmid pGL2A8-3 was constructed by insertion of a PvuII fragment (2437 to 2208) into the PvuII site of pGL2A8-1. Plasmid pGL2A8-4 was constructed by insertion of a PvuII fragment (2437 to 2208) into the PvuII site of pGL2A8-1.2. Plasmid pGL2A8-5 was constructed by insertion of a PstI fragment (2626 to 2138) into the PstI site of pGL2A8-1.2.

Transient Transfection of Plasmids into Hepatocytes. Primary cultures of Syrian hamster hepatocytes were prepared as described previously (Tohkin et al., 1996) and 1 × 10⁶ cells were plated in 35-mm collagen-coated plastic culture dishes in 2 ml of Waymouth's MB752/1 medium containing bovine serum albumin (2%), insulin (0.5 mg/liter), transferrin (0.5 mg/liter), selenium (0.5 µg/liter), and dexamethasone (1 nM). After cultivation for 24 h, the dishes were washed with 1 ml of serum free OPTI-MEM and then exposed to a mixture consisting of 5 µl of Lipofectin with 2.5 µg of pGL2A8 and 5 µg of pSV--galactosidase plasmids in 1 ml of serum free OPTI-MEM. After 17 h of incubation, the Lipofectin-plasmid mixture was removed and the cells were cultured with 2 ml of Waymouth's MB752/1 medium for 5 h, and then the medium was replaced with Waymouth's MB752/1 medium containing 1 µM 3-MC in dimethyl sulfoxide. The final concentration of dimethyl sulfoxide in the culture medium was 0.1%. After 24 h of treatment with 3-MC, luciferase activities in cell extracts were measured by a luciferase assay kit. Differences in transfection efficiencies between dishes within given experiments were normalized by the -galactosidase activity.

Transient Transfection of Plasmids into Drosophila SL2 Cells. SL2 cells were maintained in the Drosophila Schneider's medium supplemented with 10% fetal bovine serum. Plasmids (3 µg of DNA) were introduced into SL2 cells (5 × 10⁶ cells/2 ml of culture medium/35-mm plastic culture dish) by the calcium-phosphate method as described by Di Nocera and Dawid (1983). After incubation for 72 h, 3-MC or dimethyl sulfoxide alone was added to the cell culture. Cells were harvested 24 h later and collected by centrifugation. Cells were washed two times with phosphate-buffered saline and luciferase activities in cell extracts were measured.

Nuclear Extract Preparation and Gel Mobility Shift Assay. Nuclear extracts from Syrian hamster livers and SL2 cells were prepared according to the procedure of Sierra et al. (1993). The extracts were dialyzed against 25 mM HEPES-KOH (pH 7.6) containing 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 10% glycerol (v/v), 40 mM KCl, and 0.1 mM EDTA. Double-stranded synthetic oligonucleotides were end-labeled by filling the recessed 3'-termini with [-³²P]dCTP using a large fragment of DNA polymerase I and purified using a Sephadex G-25 column. The labeled oligonucleotide probes (20,000-30,000 cpm) were mixed with 2 µg of poly(dI-dC)poly(dI-dC), 1 µg of nuclear extract, and competitor oligonucleotides, if necessary, in a final volume of 20 µl of solution containing 25 mM HEPES-KOH (pH 7.8), 0.5 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 10% glycerol (v/v), 50 mM KCl, and 0.5 mM EDTA. The solutions were incubated at 25°C for 30 min and electrophoresed on 4% nondenaturing polyacrylamide gels using 50 mM Tris-HCl (pH 8.5), 0.38 M glycine, and 2 mM EDTA as a running buffer. Gels were dried and exposed to X-ray film for autoradiography.

	Results

Top Summary Introduction Materials and methods Results Discussion References

Analysis of the 5'-Flanking Region of CYP2A8 Gene. We constructed a Syrian hamster genomic library, and isolated a genomic clone containing about 16.8 kb of 5'-flanking region, the first and second exons, the first intron, and part of the second intron of the CYP2A8 gene. The start site of transcription on the CYP2A8 gene was determined by the 5'-RACE method with poly (A)⁺ RNA prepared from hamster livers. Sequence analysis identified the transcription start site at 27 bp upstream of the ATG start codon. The nucleotide sequences of the 5'-flanking region were analyzed up to 3322 (relative to the CYP2A8 gene transcription start site) and the sequences from 2672 to 2123 and 172 to +128 are shown in Fig. 1. A putative TATA box, TATAAA, was located at 30 bp upstream of the transcription start site. Upstream of the TATA box, a putative basic transcription element (BTE; GC box-like sequence) was found at position 50 to 36. We found a consensus sequence for XRE locating between 2366 and 2349 that contained the full AhR-Arnt responsive element including two core sequences, CACGC, repeated in tandem.

View larger version (48K): [in this window] [in a new window]   Fig. 1.   Nucleotide sequence of 5'-flanking region of CYP2A8. Sequence of the 5'-flanking region up to 172 and the sequence from 2672 to 2123 are shown. Nucleotide sequences of TATA box, BTE (GC box-like sequence), XRE, and positive regulatory elements (PREX and -) are enclosed in boxes. Arrows show 5'-end point of deletion mutants.

XRE-Mediated Gene Regulation. We previously showed that the expression of the CYP2A8 gene by 3-MC was observed both in the primary hepatocyte cultures (Tohkin et al., 1996) and in in vivo experiments (Fukuhara et al., 1989b). Using this hepatocyte culture, we studied the regulatory elements that participate in the transcriptional activation by 3-MC. Various luciferase reporter constructs containing successive 5'-deletions of the CYP2A8 regulatory region were introduced into cultured hepatocytes, and the transient expression of the luciferase enzyme driven by the constructs was determined in the presence or absence of 3-MC. The plasmids carrying 5'-flanking sequences longer than 2.6 kb had high transcriptional inducibilities by 3-MC showing approximately 20-fold induction, whereas those with length shorter than 2.3 kb failed to respond to 3-MC (Fig. 2). These results suggest that the sequence between 2.6 kb and 2.3 kb is necessary to the transcriptional activation of the gene by 3-MC treatment.

View larger version (29K): [in this window] [in a new window]   Fig. 2.   Structure of successive 5'-deletion mutants of CYP2A8 regulatory region and their luciferase activities in transfected primary hepatocyte cultures. Illustration shows luciferase reporter constructs containing various 5'-deletions. Numbers given to deletion plasmids indicate 5'-ends of the 5'-flanking sequences of the CYP2A8 gene from the transcription start site. These plasmids were introduced into primary cultures of hepatocytes and luciferase assays were performed after 17 h of 3-MC treatment. Fold induction refers to ratio of luciferase activity of 3-MC-treated cells to untreated cells. Values are means ± S.E. of three independent experiments.

View larger version (30K): [in this window] [in a new window]   Fig. 3.   Structure of deletion mutants of CYP2A8 regulatory region and their luciferase activities in transfected primary hepatocyte cultures. Illustrations show the pGL2 fusion construct of several successive 5'-deletions (A) and internal deletion mutants (B). Numbers given to deletion plasmids indicate 5'-ends of the 5'-flanking sequences of the CYP2A8 gene from the transcription start site. These plasmids were introduced into primary hepatocyte cultures and luciferase assays were performed after 17 h of 3-MC treatment. Fold induction refers to ratio of luciferase activity of 3-MC-treated cells to untreated cells. Values are means ± S.E. of three independent experiments.

Positive Regulatory Element for PREX. As described above (Fig. 3A), the construct starting from 2409 lost more than 50% of the induction compared with the full activity. This indicated the existence of positive regulatory elements other than XRE that are necessary for the full activation of the gene expression. As shown in Fig. 3B, pGL2A8-4, which has the XRE but lacks the sequence upstream of 2437, had also less than half the ability of pGL2A8-3 and -5 to activate transcription. This result restricts the location of the positive regulatory element between 2437 to 2626. On the other hand, pGL2A8-2, which contains the positive regulatory region but not the consensus XRE, did not show induction of the gene expression (Fig. 3B). Furthermore, the positive regulatory region exerted no influence to basal transcription level (Fig. 3). These results mean that the positive regulatory effect of the element depends on the XRE. Therefore, we designated the novel positive regulatory element for the XRE-mediated CYP2A8 gene expression as PREX. The fact that pGL2A8-2468 had the full ability of the gene activation (Fig. 3A), whereas pGL2A8-4, which lacks the regions upstream of 2437, had less than half of ability of the gene activation (Fig. 3B), suggests that the region between 2468 and 2437 includes the essential part of PREX that potentiates the XRE-mediated promoter activity.

Cooperative Transcriptional Enhancement of AhR-Arnt with Sp1. We found XRE and BTE in the 5'-flanking region of the CYP2A8 gene. Using a model CYP1A1 promoter, Kobayashi et al. (1996) reported that AhR-Arnt complex, the trans-acting factor on XRE, enhanced the transcription activity in cooperation with Sp1, the trans-acting factor on BTE. Therefore, we studied whether these factors are certainly involved in the activation of the CYP2A8 gene transcription or not. We examined the transcriptional activity of CYP2A8-luciferase fusion genes (pGL2A8-2668, pGL2A8-3, pGL2A8-4, and pGL2A8-5) in the transient expression system using Drosophila SL2 cells, which are devoid of AhR, Arnt, and Sp1 proteins (Kobayashi et al., 1996). As shown in Fig. 4, 3-MC did not induce luciferase expression from each construct in SL2 cells transfected with Sp1 expression plasmid alone. The expression of AhR and Arnt activated the transcription only slightly in all cases. In contrast, cotransfection of the AhR and Arnt expression plasmid with the Sp1 expression plasmid enhanced luciferase expression by 6-fold in the presence of 3-MC in all cases. This means that AhR-Arnt activates the gene expression by 3-MC in cooperation with Sp1. The induction of luciferase expression was observed not only in the SL2 cells transfected with pGL2A8-2 668 but also those transfected with pGL2A8-3, which lacked most of the internal region (2207/131) of pGL2A8-2668. This indicates that the internal region deleted is not essential for the enhancement by AhR-Arnt and Sp1. Although the induction of luciferase expression of pGL2A8-4 by 3-MC was less than half that of pGL2A8-5 in hepatocytes (Fig. 3B), induction of the two constructs by 3-MC was almost equivalent in SL2 cells transfected with AhR, Arnt, and Sp1 expression plasmid (Fig. 4, C and D). These results indicate that PREX could not enhance the XRE-mediated CYP2A8 gene in SL2 cells and that the nuclear factors for PREX would be deficient in SL2 cells.

View larger version (26K): [in this window] [in a new window]   Fig. 4.   Functional property of AhR-Arnt on enhancement of transcriptional activity of CYP2A8 in cooperation with Sp1. AhR, Arnt, and/or Sp1 expression plasmids (each 3 µg) were introduced into SL2 cells with different deletion mutants of CYP2A8-luciferase fusion plasmids. Fusion plasmids are schematically shown in Fig. 3. Cells were cultured with or without 3-MC for 24 h and luciferase activities of cell extracts were measured. Open columns represent untreated cells and closed columns represent 3-MC-treated cells. Values are means ± S.E. of three independent experiments.

Gel Mobility Shift Assays of PREX. To examine whether any nuclear factors could bind to PREX, we performed gel mobility shift assays using hamster liver nuclear extracts with the labeled PREX as a probe. Because the region between 2468 and 2437 included the essential part of PREX, we used three double-stranded synthetic oligonucleotides OL1 (2481/2453), OL2 (2470/2435) and OL3 (2452/2421) as probes (Fig. 5A). As shown in Fig. 5B, two DNA-protein complexes were observed only when OL3 was used as a probe. The addition of a 50-fold excess of unlabeled OL3 as a competitor prevented the formation of the DNA-protein complexes. To further determine the core element necessary for the protein binding, competition experiments were carried out using a series of competitors. As shown in Fig. 6, the m1 competitor failed to compete absolutely, indicating that the sequence region mutated in the m1 is essential for the formation of the specific complexes. The regions mutated in the m4 and m5 are also involved in the complex formation to some extent. From these results, we conclude that the sequences from 2450 to 2446 and from 2435 to 2426 are both essential for the complex formation. Thus, we named these sequences as PREX and PREX, respectively. The involvement of both the sequences for the binding was also supported by the fact that no shifted bands were observed using OL2, which contains PREX alone and not PREX, as a probe (Fig. 5).

View larger version (48K): [in this window] [in a new window]   Fig. 5.   Binding of hamster liver nuclear factors to PREX. A, nucleotide sequences used for probes and competitors. Numbers given to oligonucleotides indicate position of 5'-flanking sequences of the CYP2A8 gene from the transcription start site. B, labeled oligonucleotides were incubated with Syrian hamster liver nuclear extracts in the presence or absence of a 50-fold excess of unlabeled oligonucleotides as shown on the top of each lane.

View larger version (53K): [in this window] [in a new window]   Fig. 6.   Competition by mutant PREXs for binding of hamster liver nuclear factors to PREX. A, nucleotide sequences of wild-type and mutant PREXs. Numbers given to oligonucleotide indicate position of 5'-flanking sequences of the hamster CYP2A8 gene from the transcription start site. Underlines indicate protein-binding elements. Asterisks indicate the same nucleotides as OL3. B, labeled OL3 was incubated with Syrian hamster liver nuclear extracts in the presence or absence of a 50-fold excess of unlabeled mutated PREXs as shown in A.

View larger version (57K): [in this window] [in a new window]   Fig. 7.   Gel mobility shift assay using hamster liver nuclear extracts and SL2 cell nuclear extracts with PREX. Labeled OL3 was incubated with Syrian hamster liver nuclear extracts or SL2 cell nuclear extracts in the presence or absence of a 50-fold excess of unlabeled OL3 as shown on the top of each lane.

	Discussion

Top Summary Introduction Materials and methods Results Discussion References

The present study has partly elucidated the mechanism by which the CYP2A8 gene is transcriptionally activated by 3-MC in the Syrian hamster hepatocytes. The mechanism was shown to involve the XRE-mediated induction as observed in other 3-MC-inducible gene expressions. In addition, we found that the XRE alone was not sufficient for the full activation of the gene expression but that another element, PREX, which we found, was necessary for it. This was shown first by luciferase reporter gene assays that demonstrated that the transcription enhancing activity decreased to less than a half when PREX or 5'-part of PREX (PREX) was deleted (Fig. 3). Secondly, gel mobility shift assays demonstrated that nuclear factors bound to PREX (Fig. 5) and two sites in this region (PREX and PREX) were essential for the binding because mutation of either one of the two sites resulted in loss of the binding activities of the nuclear factors to PREX (Fig. 6). Moreover, PREX did not enhance the XRE-mediated promoter activation in SL2 cells (Fig. 4) that are lacking in binding proteins for PREX (Fig. 7). These results indicate that the binding proteins would function as trans-acting factors for PREX in hamster livers. Because PREX did not activate the transcription of the CYP2A8 gene without XRE (Fig. 3B), it is likely that the nuclear factors for PREX interacted with AhR-Arnt complex and/or Sp1 directly or indirectly, resulting in the enhancement of the XRE-mediated CYP2A8 gene expression.

To know whether some of the known regulatory elements are involved in PREX region, we analyzed the sequence of PREX region. A search of TRANSFAC database revealed that three regulatory elements for AP-4, HSF, and C/EBP were found in PREX (OL3) region (11-20, 12-8, and 16-29 of OL3 sequence, respectively). However, it is obvious that the PREX site is essential for the protein-binding (Fig. 6), whereas none of the three sites in PREX have PREX site. Therefore, the three sites would not participate in the binding of the PREX-binding proteins.

Two specific protein-PREX complex formations were observed in gel mobility shift assays (Fig. 5 and 6). We can postulate the mode of binding of nuclear factors to PREX as follows. First, there might be two nuclear factors that can bind to PREX. Second, these factors could form a dimer complex and one factor could bind to PREX and the other could associate with the DNA-binding factor, which might dissociate during electrophoresis. Third, one nuclear factor that is easily degraded might bind to PREX. In this case, the nuclear factor might be degraded at the position that was not correlated to the DNA binding region. Another possibility is that each nuclear factor might recognize the PREX and PREX, respectively. But this possibility is negligible because the mutation of one of the two binding sites resulted in the complete loss of the bindings of the nuclear factors.

In the analysis of the 5'-flanking region of the CYP2A8 gene, we found BTE in addition to XRE, both of which were also reported in the genes of other 3-MC-inducible drug-metabolizing enzymes including CYP1A1, CYP1B1, NAD(P)H:quinone oxidoreductase, glutathione S-transferase Ya subunit, class 3 aldehyde dehydrogenase, and UDP-glucuronosyltransferase family 1 (Fujisawa-Sehara et al., 1987; Rushmore et al., 1990; Favreau and Pickett, 1991; Asman et al., 1993; Tang et al., 1996; Emi et al., 1996). As observed in CYP1A1 promoter, we also demonstrated that the AhR-Arnt was involved in the induction of CYP2A8 gene expression in cooperation with Sp1 using Drosophila SL2 cells (Fig. 4). The distance of XRE from the transcription start site of CYP2A8 is quite far compared with that of the other 3-MC-inducible genes. However, it was shown to be not important for XRE-mediated transcriptional activation because the internal deletion mutant of CYP2A8-luciferase fusion plasmids (pGL2A8-3 and pGL2A8-5), in which the XRE was located near the transcription start site, had transcriptional activities equivalent to that of the pGL2A8-2668, in which the XRE located quite far from the start site (Fig. 3).

Various factors have been reported to be involved in the transcriptional regulation of CYP genes. We previously reported that activator protein-1 (AP-1) was involved in the 3-MC-induced CYP2A8 expression in hamster hepatocytes (Tohkin et al., 1996). In the present study, sequence analysis up to 3322 revealed seven possible AP-1 binding sites in the 5'-flanking region (data not shown). Therefore, to know how AP-1 would act on the transcriptional regulation of CYP2A8, we examined the effects of okadaic acid, which is an AP-1 inducer (Thevenib et al., 1991; Haby et al., 1994), and c-Jun, which acts as AP-1 with a homodimer, on 3-MC-inducible luciferase expressions using several reporter gene constructs. Although both okadaic acid and c-Jun potentiated 3-MC-inducible CYP2A8 expression in the hepatocytes (Tohkin et al., 1996), we were unfortunately unable to obtain the activation effects of AP-1 on luciferase induction (data not shown). Thus, it is likely that AP-1 plays a role at the upstream region of more than 5 kb, although the mechanism by which AP-1 participates in the transcription of CYP2A8 remains to be elucidated. It has been reported that a negative regulatory element located in the 5'-flanking region of hamster, rat, and human CYP1A1 genes contains an octamer binding motif (Boucher et al., 1993; Sterling et al., 1993; Sagami et al., 1994), and that Oct-1 is a negative regulator of rat CYP1A1 expression via the octamer sequence (Sterling and Bresnick, 1996). Although we found analogous sequences to the Oct-1 motif on 2561 to 2552 of CYP2A8 upstream region, transient transfection experiments indicated that this element did not function as a negative regulatory element for 3-MC-inducible CYP2A8 expression in hamster hepatocytes (Fig. 3A).

All these results indicate that the induction mechanism mediated by XRE and BTE seems to be common in the 3-MC-inducible genes, however, the mechanism of CYP2A8 differs from that in other 3-MC-inducible genes such as CYP1A1. Notably PREX, a positive regulatory element found in the 5'-regulatory region in the CYP2A8 gene, is unique among those so far identified in the activation of 3-MC-inducible genes.

In summary, we found a new positive regulatory element, PREX that potentiated XRE-mediated CYP2A8 gene expression by 3-MC in Syrian hamster hepatocytes. As for the mechanism of the gene expression, at least four regulatory factors are involved, which include AhR, Arnt, Sp1, and nuclear factor(s) for PREX. The mechanism by which PREX enhances XRE-mediated gene activation would be further clarified by characterization of the nuclear proteins that interact with PREX.