Characterization of phenobarbital-inducible mouse Cyp2b10 gene transcription in primary hepatocytes.

The mouse phenobarbital (PB)-inducible Cyp2b10 gene promoter has been isolated and sequenced, and control of its expression has been characterized. The 1405-base pair (bp) Cyp2bl0 promoter sequence is 83% identical to the corresponding region from the rat CYP2B2 gene. In addition to the lack of CA repeats, differences include insertion of 42 base pairs (-123/-82 bp) into the middle of a consensus sequence to the so-called "Barbie box." In this report, we have developed a primary mouse hepatocyte culture system in which endogenous 2B10 mRNA as well as Cyp2b10-driven CAT activity were induced by PB and 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene (TCPOBOP), but not by the 3-chloro derivative of TCPOBOP. Deletion analysis of the Cyp2b10 promoter identified a basal transcription element at -64/-34 bp and a negative element at -971/-775 bp. Sequences contained within the -1404/-971 bp region are responsible for the induced CAT activity. DNase I protection and gel shift assays detected five major protein binding sites within the -1404/-971 bp fragment, one of which shared high sequence identity with a portion of a regulatory element in CYP2B2 gene (Trottier, E., Belzil, A., Stoltz, C., and Anderson, A. (1995) Gene 158, 263-268). Our results indicate that sequences important for PB-induced transcription of Cyp2b10 gene are located in the distal promoter.

the lack of PB-responsive cell lines or other reliable in vitro assay systems.
An extensive study of the barbiturate-regulated induction mechanism of bacterial CYP102 gene led Fulco and associates to identify a 17-bp sequence as the binding site for a barbiturate-regulated factor (He and Fulco, 1991;. Based on sequence comparisons and in vitro protein binding studies (He and Fulco, 1991), it was suggested that a similar sequence located in the CYP2B promoter (Ϫ89/Ϫ73 bp) could be the PB-responsive element for this and other mammalian P450 genes (Shaw and Fulco, 1993;. Some studies indicated that this so-called Barbie box (consensus: 5Ј-ATCAAAAGCTGGAGG) may play a role in PB-induction of mammalian genes. The element required for in vitro transcription of a CYP2B2 minigene (Ϫ88/Ϫ56 bp) includes the Barbie box (Rangarajan and Padmanaban, 1989;Upadhya et al., 1992). The deletion or mutation of a Barbie box-like sequence (Ϫ140/Ϫ124 bp) in rat ␣ 1 -acid glycoprotein (AGP) promoter suppressed both basal and dexamethasone-or PB-induced CAT expression (Fournier et al., 1994). Other studies argue against a PB-inducible role for the Barbie box. A transgenic mouse line carrying up to Ϫ800 bp of CYP2B2 5Ј-flanking sequence expressed CYP2B2 mRNA only constitutively. PB-dependent induction of 2B2 mRNA was evident when additional 5Ј sequences were present (Ramsden et al., 1993). Furthermore, in rat Qsj:SD strain CYP2B2 induction is defective even though CYP2B1 is activated by PB. Basal expression from both genes can be detected, and both CYP2B2 and CYP2B1 promoter sequences up to Ϫ800 bp matched those of the parent strain, containing intact Barbie box sequences (Hashimoto et al., 1988). As for AGP gene, the kinetics of PB induction are much slower than for CYP2B (Fournier et al., 1994), other studies do not demonstrate any protein binding to Barbie boxlike sequences in rat or mouse AGP promoter (Ratajczak et al., 1992;Lee et al., 1993), and basal and dexamethasone-dependent expression are governed by another elements at Ϫ155/143 and Ϫ120/Ϫ105 bp (Ingrassia et al., 1994). These conflicting data indicate that the role of Barbie box in CYP2B regulation is not yet resolved.
The transcription of hepatic P450 genes is rapidly activated by PB in vivo. While many hepatoma cell lines have lost their ability to express P450s or respond to PB, in some primary cultures CYP2B mRNAs can be increased (Schuetz et al., 1990;Waxman et al., 1990;Akrawi et al., 1993;Aubrecht et al., 1993;Sidhu et al., 1993). However, there is a paucity of data on CYP2B transcriptional regulation or reporter gene assays in primary cells. Clearly, the development of a faithful PB-inducible system would foster a better understanding of induction processes involved.
In the present report, we have now developed a PB-responsive primary hepatocyte culture method suitable for reporter gene assays and mRNA analyses. We have cloned and charac-terized the promoter element from the mouse Cyp2b10 gene, the major PB-inducible P450 (Honkakoski and Lang, 1989;Aida and Negishi, 1991;Honkakoski et al., 1992aHonkakoski et al., , 1992b. Our results suggest that Barbie box-like sequences do not have a major transcriptional role, and that sequences important for PB-induced transcription are located in the distal part of Cyp2b10 gene.
Correspondence between Genomic DNA and mRNA Sequence-We isolated total liver RNA from PB-induced male DBA/2J mice to minimize the expression of related, female-specific CYP2B9 mRNA (Noshiro et al., 1988) and used sequence information and conserved exon-intron junctions of Cyp2b genes (Lakso et al., 1991) for designing specific primers for 2B10. Ten micrograms of RNA was reverse-transcribed using an oligo(dT) primer and Moloney murine leukemia virus reverse transcriptase (Boehringer Mannheim). The cDNA was amplified for 20 cycles using primers A (5Ј-CGTGAATTCCTTGAAGGTTGGCTCAAC-GACAG, ϩ314/ϩ341 nucleotide of pf3/46 cDNA) and B (5Ј-CGTGAAT-TCAACATTGGTTAGACCAGGACCATGG, 5Ј end of CYP2B10 mRNA, deduced from primer extension) and Taq DNA polymerase (Boehringer Mannheim). The amplified DNA was digested with EcoRI and ligated into the EcoRI site of M13mp18. Single-stranded phage DNA from 10 independent colonies was isolated and sequenced using M13 primers. Each sequence matched the deduced first exon sequence of the genomic DNA, confirming that we had isolated the Cyp2b10 gene.
Determination of the Transcription Start Site-One hundred micrograms of total liver RNA from untreated and PB-induced male DBA/2J mice were reverse-transcribed at 42°C for 1 h using 32 P-end-labeled primer C (5Ј-GAAGTTGCCACGGGACTTTGGG; ϩ78/ϩ99 nucleotides of pf3/46 cDNA). The residual RNA was destroyed by RNase A; resulting cDNAs, and G and GϩA Maxam-Gilbert reactions of the amplified genomic fragment (primers C and D (5Ј-CAGCACACCCGCAGTCTCT-TGT; nucleotides Ϫ306/Ϫ285) were electrophoresed on 8% acrylamide, 6 M urea gels and autoradiographed.
Tissue Distribution and Induction of CYP2B10 mRNA-To avoid problems with cross-hybridization with the cDNA probe due to large number of Cyp2b genes and low levels of CYP2B10 mRNA in extrahepatic tissues, we chose RT-PCR method instead of Northern blotting. The cDNAs were synthesized as above using 50 g of total tissue RNA from control and induced male mice. CYP2B10 cDNA was amplified using Taq DNA polymerase and primers A and B (20 cycles for liver, 30 cycles for other tissues). The products were electrophoresed through a 2% agarose gel and photographed. Control experiments proved that RT-PCR was done under conditions where the product yield was proportional to the cycle number and linearly dependent on the amount of reverse-transcribed RNA. Southern blotting and probing with EcoRI-StuI cDNA fragment and sequencing of the 360-bp product were also done to verify the CYP2B10 message.
DNase I Protection and Gel Shift Assays-Nuclear extracts from untreated and induced mouse livers were further purified by heparinagarose (Sigma) as described by Yoshioka et al. (1990). The eluted proteins were dialyzed against 10 mM Hepes, pH 7.6, 100 mM KCl, 10% glycerol, 0.1 mM EDTA, 1 mM dithiothreitol, 0.4 mM phenylmethylsulfonyl fluoride, and 1 g/ml each of leupeptin, aprotinin, and pepstatin, and concentrated to 4 -7 mg/ml protein. DNase I protection assays were carried out according to instructions in the SureTrack kit (Pharmacia) using 32 P-end-labeled, appropriately cut fragments from Hind/Bam-4 or Hind/Pst-4 plasmid DNA or generated by amplification using a 32 P-endlabeled primer. The DNA fragments were separated on standard sequencing gels, which were then dried and autoradiographed. Gel shift assays were performed using 1-5 g of crude nuclear extract in 10 -20 l of 10 mM Hepes, pH 7.6, 15% glycerol, 2 g of poly(dI-dC), 0.05% Nonidet P-40, 50 mM NaCl, and about 30,000 cpm of 32 P-end-labeled oligonucleotide or DNA fragment probe. The free and protein-bound probes were separated on 4.5% or 7% acrylamide gels in 0.5 ϫ TBE buffer prior to autoradiography.
CAT Reporter Gene Construction-Based on the detected transcription start sites, we chose HincII site (ϩ4) as the 3Ј end for the promoter construct. Other appropriate restriction enzymes were used to cut fragments out of Hind/Bam-4 pDNA. The fragments and SalI-digested pCAT-Basic vector (Promega, Madison, WI) were treated with T4 or Klenow DNA polymerase (Life Technologies, Inc.) and ligated using T4 DNA ligase (Stratagene, La Jolla, CA). Following transformation into Escherichia coli HB101 cells (Life Technologies, Inc.), the recombinant plasmids were purified twice on CsCl gradients, verified by restriction mapping and sequencing over the junction sites, and checked for content of supercoiled plasmid on agarose gels. The plasmids were named according to distance of 5Ј-end of the insert from the transcription start site. Ϫ4300CAT was generated by ligation of 3.7-kb PstI-PstI fragment from Bgl/Pst-4 plasmid into PstI-digested Ϫ1404-CAT plasmid. Other constructs were made by amplification of specific regions of Ϫ1404-CAT using primers into which HindIII (5Ј) or PstI (3Ј) were incorporated. After digestion of DNAs with HindIII and PstI, the fragments were ligated into corresponding sites of Ϫ1404-CAT plasmid. The purified plasmids were verified by sequencing over the amplified region.
Transient Transfection-Electroporation was chosen because DNA transfection takes place prior to plating, and equal cell aliquots from same pool are taken for treatments, eliminating differences in transfection efficiencies (see LeCam et al. (1994), and references therein). It is also advantageous to be able to perform transfection and start treatments quickly, limiting the transcription taking place during transfection. Our preliminary experiments with ␤-galactosidase and pCAT-Control plasmids (Promega) proved that with the same cell preparation, transfection efficiencies were similar regardless of DNA construct, even though daily variations were observed. Primary hepatocytes were isolated from male C57BL/6 mice using a two-step collagenase perfusion (0.4 -0.5 mg/ml in HBSS supplemented with 10 mM Hepes, pH 7.4, and 1 M porcine insulin; Sigma) and low speed centrifugations. They were further purified by Percoll (Pharmacia) density centrifugation to a viability of at least 90% . The cells were washed and suspended (20 -30 ϫ 10 6 /ml) in ice-cold PBS containing 20 mM Hepes, pH 7.4. The cells were electroporated at 960 microfarads and 150 -170 V using 300 g/ml sonicated herring sperm DNA as carrier and up to 100 g/ml CAT construct. The electroporated cells were diluted into prewarmed Williams' E medium supplemented with 7% fetal bovine serum, ITS supplement (Sigma), and streptomycin-penicillin G (100 units/ml) and incubated for 5 min. From each pool, 3-ml aliquots were dispensed on 60-mm cell culture dishes, and the cells were allowed to attach for 30 min at 37°C under 5% CO 2 . The unattached cells were removed, and the dishes were washed with PBS. The cells (2-3 ϫ 10 6 /dish) then received fresh medium without serum, containing 5 nM dexamethasone, and inducers or vehicles (1 mM PB in saline; 50 nM TCPOBOP or 1,4-bis[2-(3-chloropyridyloxy)]benzene in dimethyl sulfoxide). In some instances, cycloheximide (10 M), ␣-amanitin (1 M), actinomycin D (0.5 M), GH (50 ng/ml), or EGF (20 ng/ml) were added 30 min prior to the inducer. The transfection and CAT assay conditions were optimized in preliminary experiments using pCAT-Control plasmid. Culture conditions, treatment times, and doses were defined by analyzing CYP2B10 mRNA and protein, and CAT expression from the Ϫ1404-CAT and Ϫ64-CAT plasmids. Alternatively, cells were plated without electroporation, allowed to attached, and washed with PBS. Cells were transfected using 7.5 g of DNA and 50 g of Lipofectin reagent (Life Technologies, Inc.) in 3 ml of medium (Shih and Towle, 1995) without dexamethasone or serum. After an overnight incubation, the medium was changed to additionally contain 5 nM dexamethasone and 0.5 mg/ml Matrigel, and inducers were added for 24 h.
Analysis of Cellular mRNAs and CAT Activity-The cell medium was removed at 8 h, and cells were lysed using Trireagent (Molecular Research Center, Inc., Cincinnati, OH). The CYP2B10 mRNA was analyzed using the methods described above, and mouse albumin 180-bp cDNA fragment was amplified using primers E (5Ј-AGACATC-CTTATTTCTATGCCC) and F (5Ј-CTGCATACTGGAGCACTTCATT). The cellular mRNA-derived PCR products were cloned into pCRII vector (Invitrogen) and sequenced. The verified cDNA fragments were used as probes in Northern blotting of cellular RNA (Maniatis et al., 1989). For CAT assay, 22 h after medium change, cells were washed with ice-cold PBS containing 1 mM EDTA, scraped off, and pelleted. The cell pellets were lysed as described by Pothier et al. (1992) and assayed for protein (Bradford, 1976) and CYP2B10 protein (Honkakoski et al., 1992b). Equal amounts of cell extracts (150 -200 g of protein) were then heat-treated and assayed for CAT activity (Gorman et al., 1982) with the exceptions that acetyl coenzyme A concentration was 1.33 mM and incubation time was extended to 4 h.

RESULTS
Sequence Analysis of the Cyp2b10 Gene-We isolated a genomic clone containing at least 13 kb of 5Ј-flanking region, the first exon, and part of the first intron of the Cyp2b10 gene. The other 21 genomic clones did not contain any portions of the Cyp2b10 gene, and they could be classified into seven groups, confirming the large size of Cyp2b gene family (Lakso et al., 1991). The first exon 197-bp sequence (Fig. 1) was 94%, 82%, and 79% similar to that of rat PB-inducible CYP2B2, and those of non-inducible mouse Cyp2b9 and rat CYP2B3 (Jean et al., 1994), respectively. In 5Ј-flanking sequences, the overall homology of Cyp2b10 to CYP2B2 was 83%. The Cyp2b10 promoter was 62% and 70% similar to that of CYP2B3 and Cyp2b9, respectively, within the proximal 400 bp; upstream of the (CA) repeat, the similarity to Cyp2b9 dropped to only 36%. Only a few differences from the CYP2B2 gene were noted; first, the (CA) repeats (Suwa et al., 1985;Lakso et al., 1991) were very short in Cyp2b10 ( Fig. 1; Ϫ303/Ϫ300 bp). Second, the so-called Barbie box sequence (Shaw and Fulco, 1993)

was interrupted in
Cyp2b10 by an insertion of 42 nucleotides, the result being that sequence homology to the Barbie box consensus decreased to 67% (Fig. 1, wavy lines). However, the insertion was shared by mouse Cyp2b9 and rat CYP2B3 genes (Jean et al., 1994), suggestive of loss of the 42-bp DNA element from CYP2B1/2 genes after the rat-mouse divergence. An atypical TATA box (CATA-AAAG) and sequences resembling the binding sites for transcription factors including Sp1, AP-1, C/EBP, Ets-1, HNF-5, and GR were also found ( Fig. 1).
Transcription Start Site-We determined the transcription start sites of Cyp2b10 gene by primer extension analysis. Male mice were used to minimize the expression of CYP2B9 mRNA and its possible cross-hybridization with the primer. Three bands represent the start sites at adenines at 29, 30, and 34 bp downstream of the putative TATA box, respectively (Fig. 2). Treatment by PB did not change the location of this site, since the same fainter bands were detected in control RNA. The observed start sites matched closely to those of CYP2B1/2 in rat liver (Suwa et al., 1985).
Cyp2b10 Regulation in Mouse Tissues-Previous Northern blot analysis showed that hepatic CYP2B10 mRNA was highly induced within 3 h after PB or TCPOBOP injection (Aida and Negishi, 1991;Honkakoski et al., 1992a). To minimize detection and cross-hybridization problems, CYP2B10 mRNA expression was measured by RT-PCR. Although CYP2B10 mRNA was detected in lung and intestine after longer amplification, PB-inducible expression was observed only in the liver (Fig. 3). Sequencing of the amplified product yielded only CYP2B10 mRNA in liver (not shown). The nuclear run-on assays showed that both PB and TCPOBOP induced nascent 32 P-labeled CYP2B10 transcripts (detected by pf3/46 cDNA), while that of non-inducible CYP2D9 plus 2D10 (p16␣-14 cDNA) was relatively unchanged. Furthermore, this induced transcription was largely unaffected by the presence of GH, since the pattern was quite similar between intact and hypophysectomized mice (Fig.  4). This confirms that pituitary factors are not critical for Cyp2b10 expression in mice (Smith et al., 1993), as observed for CYP2B1/2 in rats (Yamazoe et al., 1987;Schuetz et al., 1990). The transcription of ␥-actin was variable for unknown reasons. The amount of CYP2D9 plus 2D10 transcripts was decreased by hypophysectomy (Fig. 4), consistent with our previous reports (Wong et al., 1989;Yoshioka et al., 1990).
Cyp2b10 Regulation in Mouse Primary Hepatocytes-We then developed a primary cell culture system, in which endogenous CYP2B10 mRNA could be rapidly induced. Northern hybridization (Fig. 5A) and RT-PCR assays (Fig. 5B) displayed more than 10-fold increases of CYP2B10 mRNA in hepatocytes within 8 h by treatment with PB, 100 nM dexamethasone, or TCPOBOP, but not by 3-chloro derivative of TCPOBOP. Time course studies indicated CYP2B10 mRNA was elevated already after 1 h, reaching a maximum level at 8 h, as in vivo (data not shown). Most importantly, the increase in CYP2B10 mRNA was blocked by pretreatment with the transcriptional inhibitors ␣-amanitin and actinomycin D (Fig. 5, A and B, lanes 7 and 8) but not by cycloheximide, GH, or EGF (lane 6 and data not shown). This indicates that Cyp2b10 gene transcription is rapidly activated also in primary hepatocytes. Intriguingly, Northern hybridization detected a non-inducible 2.7-kb mRNA, in addition to 2.2-kb CYP2B10 mRNA (Fig. 5A). Since CYP2B9 or CYP2B13 mRNAs could not be amplified from these cells (not shown), the 2.7-kb mRNA may be a novel member of the Cyp2b family or an alternatively spliced variant. Nevertheless, it served as an excellent internal control since both the 2.7-kb and 2.2-kb RNAs are newly transcribed and present in similar levels. FIG. 4. Transcriptional regulation of the Cyp2b10 gene. Hypophysectomized (Hypox) and sham-operated (Sham) DBA/2J male mice were injected with corn oil (lanes 1), PB (lanes 2), or TCPOBOP (lanes 3). Purified nuclei (400 g of DNA) were incubated with [␣-32 P]UTP for 20 min, and the isolated RNA transcripts (20 ϫ 10 6 cpm) were hybridized to indicated immobilized linear plasmid DNAs (5 g) detecting CYP2B10 (pf3/46), CYP2D9 plus CYP2D10 (p16␣-14), ␥-actin mRNAs, or pUC according to Legraverend et al. (1992). The filters were washed (3 ϫ 20 min) at 68°C with 0.1 ϫ SSC, 0.1% SDS, and autoradiographed. The abundant albumin mRNA was used to control RNA loading and quality, and it was not changed by the treatments. We also found that low concentrations (Ͻ10 nM) of dexamethasone in the medium are essential for PB induction of CYP2B10 mRNA while higher levels (Ն100 nM) increase the basal level.
Functional Analysis of Cyp2b10 Promoter-We transfected primary hepatocytes by electroporation with various CAT reporters containing successive 5Ј-deletions of the Cyp2b10 promoter. The promoters carrying 5Ј-flanking sequences up to Ϫ775 bp had high basal transcriptional activities, mostly contributed by sequences between Ϫ64 and Ϫ34 bp. Elimination of the Barbie box-like sequences had no effect (Ϫ376CAT versus Ϫ64CAT; Fig. 6A). Sequences upstream of Ϫ775 bp decreased the basal activity considerably, which suggested a negative regulatory element between Ϫ971 bp and Ϫ775 bp of Cyp2b10 gene (Fig. 6A). Inducible transcription was observed only with the two longest 5Ј-flanking sequences: PB or TCPOBOP treatments resulted in 2.0 -3.3-fold increases of CAT activity with the Ϫ4300-CAT and Ϫ1404-CAT reporters, respectively (Fig.  6B, Table I). This indicates that the Ϫ1404/Ϫ971 bp region may contain DNA elements important for PB induction, since further addition of 5Ј sequences did not enhance induction of CAT. Importantly, 1,4-bis[2-(3-chloropyridyl-oxy)]benzene, which did not induce 2B10 mRNA (Fig. 5, A and B), also failed to induce CAT activity from the Ϫ1404CAT reporter (Fig. 6C), whereas PB and TCPOBOP increased both 2B10 mRNA and CAT activity. The lipofection protocol gave qualitatively similar results with 4 -5-fold increase by TCPOBOP but somewhat higher basal CAT activity (Fig. 6D).
Nuclear Protein Binding to Cyp2b10 Promoter-In the light of the reported importance of the Barbie box to PB induction and results of our functional studies, we performed DNase I protection with heparin-agarose-enriched nuclear extracts. In the proximal promoter, regions at Ϫ64/Ϫ45 and Ϫ235/Ϫ215 bp (Fig. 7A) appear to match the regions displaying PB-enhanced binding within the CYP2B2 gene (Shephard et al., 1994); however, no differences in protein binding was evident with Cyp2b10 gene as confirmed by gel shift assays (not shown). Intriguingly, the regions resembling the Barbie box (Fig. 1) were not protected in any of the samples (Fig. 7A). This finding was confirmed by gel shift assays; furthermore, addition of PB directly (He and Fulco, 1991) or together with cytosolic fractions to nuclear extracts in vitro did not produce any inducer- Williams' E supplemented with ITS. Thereafter, the cells were washed with PBS, and medium containing additionally 5 nM dexamethasone and 0.5 mg/ml Matrigel was added without (Ϫ) or with (ϩ) 50 nM TCPOBOP for another 24 h. CAT activities and protein were assayed as described under "Experimental Procedures." dependent complexes (not shown). This indicates that Barbie box-like sequences of Cyp2b10 do not bind nuclear proteins. In the Ϫ1404/971 bp region, we could not detect any reproducible differences in the protection pattern between the samples (Fig.  7, B and C), further confirmed in gel shift assays (not shown). The major distal protected segments included Ϫ1354/1330 bp, FIG. 7. DNase I-protected elements of the proximal (panel A) and distal (panels B and C) regions of Cyp2b10 gene. Heparin-agaroseenriched liver nuclear proteins (2-20 g) from untreated or PB-treated DBA/2J male mice adjusted to 20 g of total protein with bovine serum albumin were incubated at room temperature with 32 P-end-labeled XbaI-HincII (panel A), HindIII-FokI (panel B), or FokI-NcoI (panel C) DNA fragments for 20 min, and digested with 0.5-1 unit of DNase I for 30 s. Reaction products were ethanol-precipitated after proteinase K digestion and phenol-chloroform extraction, separated on a 8% sequencing gel with GϩA reaction ladders, and autoradiographed. Protected regions are indicated with black bars and distances from the transcription start site.

TABLE I Basal and induced CAT activity of various Cyp2b10 promoter/CAT reporter gene constructs in primary hepatocytes
The CAT activities (mean Ϯ S.D.) are expressed relative to Ϫ64CAT (ϭ100), assayed from several sets (n) of independent hepatocyte transfections. The conversion rate for Ϫ64CAT ranged from 6 to 22% depending on the cell batches. The extent of induction is the ratio of PB-or TCPOBOP-induced CAT activity to basal CAT activity (mean Ϯ S.D.) for each construct.

Construct
(n) Relative basal CAT activity Fold-induction by PB or TCPOBOP

DISCUSSION
Cell Culture-The signaling mechanisms and DNA elements involved in PB-induced transcription of CYP2B genes have not been elucidated in detail. Hepatoma cell lines in general have lost either their ability to express CYP2B genes or their responsiveness to PB, reflecting the highly differentiated nature of the CYP expression. Intensive efforts, therefore, have been made to develop primary hepatocyte cultures responsive to PB (see, e.g., Schuetz et al. (1990), Waxman et al. (1990), Akrawi et al. (1993), Aubrecht et al. (1993), Sidhu et al. (1993), and ). In these systems, CYP2Bs mRNAs and proteins are highly inducible, and a great deal has been learned about inhibitory effects of serum, growth factors, cytokines, and cyclic AMP (Waxman et al., 1990;Schuetz et al., 1990;Aubrecht et al., 1993;Abdel-Razzak et al., 1995;Clark et al., 1995;Sidhu and Omiecinski, 1995), as well as the dexamethasone requirement for optimal CYP2B induction (Waxman et al., 1990;Kocarek et al., 1994;. However, the presence of feeder cells or Matrigel underlay, or the formation of cell clusters, may compromise DNA transfection into the cells (Pasco and Fagan, 1989). Similarly, Matrigel overlay limits the transfection period to the first 4 h of culture, whereas PB treatment is initiated usually 48 h later (Schuetz et al., 1990;Sidhu et al., 1993). This may result in destabilization of transfected DNA, in loss of transcription factors, and in uncontrolled transcription from the plasmid prior to treatment. In some instances, induction kinetics seem to differ from the rapid response in vivo. Given the long PB treatment (24 -96 h), some co-treatments may also affect CYP2B mRNA stability as well.
With these concerns in mind, we have now developed a primary hepatocyte culture in which endogenous Cyp2b10 gene is maximally activated within 8 h. This activation was blocked by pretreatment by transcriptional inhibitors but not by cycloheximide or growth factors, and the activation kinetics mirrors the in vivo responses (Aida and Negishi, 1991;Honkakoski et al., 1992a). Similar protocols were developed by Salonpää et al. (1994) and Lecam et al. (1994) for CYP2A5 and GH-regulated rat Spi 2.1 gene, respectively. Protocol differences include shorter cell attachment time with serum, decreased dexamethasone levels, and plating to higher density, all of which were found necessary for optimal CYP2B10 mRNA expression.
Reporter Gene Studies-We then sought after DNA transfection methods compatible with our cell system. We used electroporation to avoid differences in transfection efficiency by using the same pool of transfected cells for aliquoting for different treatments (Paquereau and LeCam, 1992;LeCam et al., 1994). Our observations with various Cyp2b10 deletion constructs correlate well with the data from CYP2B2 transgenic mice. In both cases, promoters containing about 800 bp of 5Ј-flanking sequence had high basal expression but showed no PB-inducibility in hepatocytes. Similarly, additional 5Ј-flanking sequences decreased the basal activity but conferred PBinducible transcription (Ramsden et al., 1993). We found that the Ϫ1404/Ϫ971 bp region was critical for Cyp2b10 inducibility. The importance of this element was strengthened by our finding that another Cyp2b10 inducer TCPOBOP (but not its inactive 3-chloro derivative) also increased CAT activity from Ϫ1404CAT reporter. Two other studies have also indicated that a PB-responsive element (PBRE) might be located in distal parts of PB-inducible genes. Hahn et al. (1991) identified a PBRE between Ϫ5.9 and Ϫ1.1 kilobase pairs of the chicken CYP2H1 gene that exhibited 2.4-fold increases. Trottier et al.
(1995) located a PBRE of CYP2B2 gene within Ϫ2318 and Ϫ2155 bp, showing 3.5-to 6.6-fold. The 163-bp CYP2B2 and 4800-bp CYP2H1 fragments could also confer PB inducibility to heterologous promoters. Despite our attempts to link Ϫ1404/ Ϫ971 bp element and several overlapping fragments to heterologous and proximal Cyp2b10 promoters, we could not confer any PB inducibility to these constructs (not shown). This may be due to disruption of interactions between the enhancerbinding factor(s) and factors binding elsewhere on the Cyp2b10 gene, such as the repressor element at Ϫ971/Ϫ775 bp.
Apparently, these reporter genes are expressed differently, but so are the endogenous PB-inducible CYP genes between species. For instance, rat and mouse CYP2B forms differ in their response to dexamethasone, TCPOBOP, and GH (Meehan et al., 1988;Poland et al., 1981;Honkakoski et al., 1992a;Smith et al., 1993), and there are differences in tissue specificity, basal liver expression, and sensitivity to cycloheximide within PB-inducible CYP2B genes (Omiecinski, 1986;Ohmori et al., 1993;Dogra et al., 1993). It is therefore possible that some aspects of the induction process may differ between species, and activation of Cyp2b10 gene may include a unique step in addition to the principal mechanism common to PB-inducible P450 genes.
Protein-DNA Interactions-The well established function of Barbie box in bacterial CYP102 gene transcription and its homology to other PB-inducible gene promoters (He and Fulco, 1991;Shaw and Fulco, 1993; has focused much attention on the role of Barbie box-like sequences in mammalian gene regulation (Upadhya et al., 1992;Fournier et al., 1994). Intriguingly, a 42-bp insertion into Cyp2b10 sequence not only splits the Barbie box, it also decreases its degree of sequence identity. We did not detect any nuclear protein binding to Cyp2b10 Barbie box-like sequences. They did not mediate PB-induction, nor did their deletion have any effect on basal promoter activity, and importantly, the presence of 42-bp insertion had no effect on the inducibility of the endogenous Cyp2b10 gene. These findings imply that Barbie box may not be a common regulatory element for PB-inducible expression. In support of this view, PB-responsive mouse Ϫ1404/Ϫ971 bp and rat Ϫ2318/Ϫ2155 bp fragments do not contain any Barbie box-like sequences.
In distal Ϫ1404/Ϫ971 bp region, none of the protein binding sites displayed any major difference between control and PB treatment, a finding identical to rat 2B2 163-bp region (Trottier et al., 1995). There are several explanations for this apparent difference between binding and functional studies. First, a pre-existing factor might be modified by a signal derived from the inducer, e.g. by phosphorylation, a mechanism known to activate nuclear factors (Hunter and Karin, 1992). This would be quite consistent with the rapid PB induction and its insensitivity to cycloheximide. Other possibilities include that a factor is being displaced by a similar, but functionally different factor (Isshiki et al., 1991), or that the induction is mediated by protein-protein interactions (Konig et al., 1992) rather than by a mechanism involving primarily DNA-protein recognition. Even though it is unclear why a putative PBRE would be located in different regions of the homologous rat and mouse CYP2B genes, we found similarities between the rat 163-bp sequence and the mouse Ϫ1404/Ϫ971 bp region. A 25-bp DNA fragment at Ϫ1.2 kilobase pairs of Cyp2b10 gene is 80% similar to a portion of the 163-bp sequence, and an overlapping segment (Ϫ1228/Ϫ1195 bp) can bind nuclear protein(s). Both rat and mouse 25-bp fragments carry core binding sites (AGGTCA) for members of ligand-dependent transcription factor superfamily. This is very interesting in view of reports that some endocrine-sensitive factors, steroid hormones, and oxysterols might modulate basal and/or PB-induced CYP2B expression (Larsen and Jefcoate, 1995;Kocarek et al., 1993). It has been suggested that PB-like inducers may act on CYP2B genes indirectly by inhibition of endogenous sterol metabolism and interfering with the natural sterol signaling of CYP2B expression (Waxman and Azaroff, 1992). Our preliminary studies indicate that several nuclear factors are able to bind the 25-bp segment, and we are currently trying to purify and characterize these factors.