The Human CYPIA2 Gene and Induction by 3"ethylcholanthrene A REGION OF DNA THAT SUPPORTS AH-RECEPTOR BINDING AND PROMOTER-SPECIFIC INDUCTION*

The gene for cytochrome P4501A2 is constitutively ex- pressed in the liver of vertebrates and shows induced expression when an organism is exposed to polycyclic aromatic hydrocarbons and halogenated hydrocarbons. To identify DNAelements regulating transcription of the human CYPlA2 gene, transient transfection experiments were conducted in the human hepatoma cell line HepG2. Dissection of the S'-flanking portion of the CYPlA2 gene identified two regions that contributed to the overall induction by 3-methylcholanthrene. One region located at -2532l-2423 contains an xenobiotic-re- sponsive element-like sequence, termed X1, that binds a nuclear 2,3,7,8-tetrachlorodibenzo-p-dioxin-inducible protein in HepG2 and wild type mouse Hepa-1 cells, but not in the Ah receptor nuclear translocation defective mouse C- mutant c4 cells. In addition, deletion of this region of the CYPIA2 gene reduces the 3-methylcholan- threne (3-MC)hitiated induction of chloramphenicol acetyltransferase activity in both promoter- and en- hancer-specific constructs. The second responsive region is located at -2259/-1987. This region of the gene contains a second xenobiotic-responsive element-like element, but this element does not the there exist a A was shown These results suggest that Ah and elements regulate the expression of the


The nucleotide sequence(s) reported in this paper has been submitted
Health Sciences Center, 4200  ants, such as halogenated aromatic hydrocarbons (HAH)' and polycyclic aromatic hydrocarbons (PAH) (3). The induction of cytochrome P4501A1 in mice and rats has been shown to be at the level of transcriptional regulation (4)(5)(6). The mechanism by which PAHs, such as 3-methylcholanthrene (3°C) and halogenated aromatic hydrocarbons, such as 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), transcriptionally activate the CYPlAl gene involves the binding of ligand to the Ah-receptor (AhR). The ligand-bound AhR translocates to the nucleus (7-91, where it associates with enhancer elements in the 5"flanking region of the CYPlAl gene referred to as dioxin-responsive elements (DRE) or xenobiotic-responsive elements (XREs). These elements are conserved among mice, rat, rabbit, and humans with respect to sequence and location within the CYPlAl gene (10, 11). Multiple copies of the consensus DRE sequence are found in the 5'-flanking region of the CYPlAl gene and have been shown to be required for inducer-dependent transcription when used in DNA transfection experiments (10- 14).
Alignment of the functional DRE sequences from the mouse CYPlAl gene has resulted in the identification of a consensus DRE having the invariant core sequence of T-GCGTG, flanked by several conserved nucleotides (12,13). Those nucleotides important for AhR binding have also been defined through detailed studies using mutant DRE oligomers in gel mobility shift assays and methylation protection and interference experiments (14)(15)(16)(17). In addition to defining the nucleotides required for high-affinity binding, those nucleotides required for function have also been defined (12).
For the most part, the molecular mechanisms that control the expression of the CYPIA2 gene are unknown. The regulation of the CYPlA2 gene by TCDD and 3°C in mice (4,5) and isolated rat hepatocytes (18) occurs through transcriptional activation. DNA transfection experiments demonstrated that the human CYPlA2 gene contains sequences within 3.2 kb of the 5"flanking gene that are responsive to 3°C in human hepatoma cells, suggesting transcriptional activation of this gene (19). DNA sequence analysis, however, has not identified DREs in the flanking sequences of the CYPlA2 gene from any species. Although both CYPlAl and CYPlA2 are transcriptionally activated by PAHs, they are also controlled by other modes of regulation. For example, they differ in their pattern of expression in that 1A2 is found predominantly in the adult liver, whereas 1Al is found in the liver, as well as in extrahepatic tissue of all age groups (20,21). It has been shown recently that an auxiliary protein, ARNT (AhR nuclear translocator), is required for the nuclear translocation of the AhR and for AhR ~ The abbreviations used are: H A H , halogenated aromatic hydrocarbon; PAH, polycylic aromatic hydrocarbon; 3-MC, 3-methylcholanthrene; AhR, dioxin Ah-receptor; CAT, chloramphenicol acetyltransferase; D m , dioxin responsive element; TCDD, 2,3,7,8-tetrachlorodibenzo-p-dioxin; XRE, xenobiotic-responsive element; kb, kilobase pairs. 6949 binding to DREs (22,23). It is plausible, based on the differences in regulatory patterns between 1Al and 1A2, that other trans-activating factors (e.g. other ARNT-like proteins) act in concert with the AhR in mediating the induced expression of the CYPlA2 gene (24). Such interactions may result in an alternate binding site for the AhR at a responsive element. The present study was designed to localize the responsive element in the human CYPlA2 promoter and determine if related DRE sequences bind the AhR and if they are required for inducerdependent transcriptional activation of the human C Y P I A 2 gene. The results of these experiments implicate the AhR as well as AhR independent mechanisms underlying the induction of the C Y P l A 2 gene.

MATERIALS AND METHODS
Plasmid Constructions-% obtain 5'-progressive deletions of the human CYPlA2 gene (19), sequences from -3201 to +53 (relative to the start of transcription), containing the promoter, exon one, and 5'-flanking sequences, were removed from a genomic clone with a KpnI restriction digest. The KpnI sites were made blunt ended and cloned into the EcoRV site of the plasmid pBSCAT, generating the plasmid plA2CAT. Deletions were generated using exonuclease 111 and mung bean nuclease digestion (Stratagene, San Diego, CA). The plasmid was first linearized at the KpnI site in the vector and then digested with XhoI to create a 5"overhang for directional exonuclease 111 digestion. Following incubations with exonuclease I11 for various periods of time, mung bean nuclease was used to repair the ends and the plasmid was re-ligated. This technique allowed for the generation of deletion clones without having to subclone deleted sequences. Deletion end points were determined by double-stranded DNA sequencing (25).
To generate fragments to test for enhancer activity, the DNA containing the 3-MC-responsive element was removed from the human CYPlA2 gene by a KpnI (-3201) and PstI (-1595) restriction digest. This fragment was made blunt ended and cloned into the EcoRV site of pSVCAT-BS and is called plA2SVCAT. This vector contains the SV40 enhancerless promoter driving the expression of the CAT gene. 5'-progressive deletions of this responsive element were generated as described above. Additional deletions were generated using the polymerase chain reaction.
Cell Culture, DNA Dansfection, and CAT Assay-The human hepatoma cell line, HepG2, was obtained from ATCC, and the mouse wild type hepatoma cell Hepa-lclc7 and the AhR nuclear translocation defective class I1 cell (26) were provided by Dr. Whitlock (Stanford University). All cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. The DNA transfections were performed by the calcium phosphate precipitation technique (27). Cells were transfected with 10 pg of plasmid DNA and in cotransfected experiments with 0.1 pg of pSV2LUC (28) as an internal control for transfection efficiency. Twenty-four hours after transfection, cells were treated with 10 y 1~ 3°C or 0.1% dimethyl sulfoxide, and cells were harvested 16-18 h after treatment.
The CAT assays were performed by thin layer chromatography as described by Gorman et al. (29). The quantitation was done by cutting the chromatogram corresponding to the acetylated and nonacetylated [14C]chloramphenico1 and counting by liquid scintillation. CAT activity is expressed as the percentage of conversion calculated as the ratio of the acetylated form to the total. The -fold induction refers to the ratio of CAT activity of induced cells to the uninduced cells. Protein concentrations were measured by the method of Bradford (30).
Isolation of Nuclear and Cytosolic Protein Extracts and Gel Mobility Shift Assays-Nuclear extracts were prepared from control and TCDDtreated HepG2, mouse Hepa lclc7 (wild type), and Class I1 variant cells as described by Denison (31). Cells were treated with 10 I~M TCDD for 2 h prior to isolation of nuclear extract. Mouse Hepa cell cytosol was prepared in ice-cold EPGM buffer (1 m EDTA, 20 m M potassium phosphate, 10% (w/v) glycerol, and 2 m 2-mercaptoethano1, pH 7.2) as described by Wilhelmsson (32). For gel mobility shift assays, 10 pg of nuclear extracts or 20 pg of activated cytosol were used. Cytosol was activated by the addition of 10 I~M TCDD for 4 h at 28 "C. Gel mobility shift assays were performed as described by Denison (31), with the exception that gel electrophoresis was conducted using 1 x TBE buffer (0.089 M Tris borate, 0.089 M boric acid, 0.002 M EDTA). The DNA probes, double-stranded oligonucleotides, were labeled with T4 polynucleotide kinase and [y-32PlATP. Sequences of probes are as follows: mouse DRE3, 5' GAGCTCGGAGTTGCGTGAGAAGAGCC 3'; human

RESULTS
Identification of cis-Acting Regulatory Elements in the Human CYPlA2 Promoter and 5'-Flanking Sequences-In previous work (19), we reported that there existed cis-acting elements within approximately 3 k b of the human C Y P I A 2 gene that were responsible for 3-MC-induced transcription of heterologous gene promoters. To further characterize these sequences, the plasmid, plMCAT, containing the human CYPlA2 promoter and 3.2 kb of 5'-flanking sequences was used in transient transfection experiments in HepG2 cells to determine the dose-dependent induction by 3°C. As shown in Fig. 1, increasing concentrations of 3°C resulted in increased expression of CAT. A concentration of 10 resulted in approximately 8-fold induction of CAT expression, and this concentration was used in subsequent experiments. CAT constructs containing the CYPlA2 promoter and flanking sequences in an orientation opposite to the transcription of the reporter gene did not show CAT activity above background in transfected control and treated cells, providing further evidence that the observed induced CAT activity was directed by trans-activation of C Y P l A 2 sequences.
To identify the region responsible for the 3-MC-inducible expression of the human CYPlA2 gene, the 5"flanking region of the gene was dissected by progressively deleting sequences in the 5' to 3' direction of the plA2CAT plasmid. The ability of these sequences to drive the expression of CAT was determined by transient transfection experiments in HepG2 cells. Progressive 5'-deletions resulted in a decrease of induced CAT activity ( Fig. 2A). Deletions to -2195 kb (relative to the CYPIA2 gene transcriptional start site) resulted in an approximately 4 0 4 0 % decrease in induced CAT expression, whereas further deletions to -1716 kb completely abolished both constitutive and induced expression. These results confirmed our earlier observation that a 3-MC-inducible element was contained between -3. -1595 (PstUKpnI), encompassing the region determined from deletional analysis of the promoter to be 3-MC-responsive, was cloned into an enhancer vector in which the CAT gene is under the control of the SV40 enhancerless promoter (Fig. 2 B ) . Having shown that the region from -3201 to -1595 was responsive to 3-MC, we generated both 5'to 3'-and 3'-to 5"progressive deletions. As shown in Fig. 2 B , 5"deletions to -2532 did not change the induced response, but deletions to -2423 resulted in approximately a 40% decrease in inducible CAT activity. Results of these deletional clones are consistent with the deletional data from promoter constructs (Fig. 2 A ) that showed a similar decrease in induced CAT activity at about 2.2 kb. A further decrease in sequences to -1987 completely eliminated induction of CAT activity by 3°C. These data establish the importance of sequences that lie within the region between -2532 and -1987 and suggest that two domains may be involved in 3-MC-mediated CAT activity.
To determine if sequences within the 3'-end of the responsive element are also required for inducer activity, deletions were generated from the 3'-end, keeping the 5"region constant. Results of these deletions are also shown in Fig. 2 B . Deletion of the 3'-end to -1762 had no effect on the inducible CAT activity. Deletions to -2259 resulted in approximately 60% decrease, whereas further deletions to -2847 completely abolished activity. These deletions are consistent with the promoter and other 5"enhancer deletions, demonstrating that two regions appear to be required for full inducer activity. One region encompasses the sequences defined by the end points -25321-2423 and the other region by the end points defined from the promoter construct -2195 (Fig. 2 A ) to -1987 which is defined by the en-hancer construct (Fig. 2 B ) .
The 3"region was further characterized by generating several additional DNA fragments by polymerase chain reaction and cloning these fragments into the SV4OCAT plasmids. As shown in Fig. 2 B , a DNA fragment from bases -2752 to -1987 exhibits a 5.6-fold increase in CAT activity following treatment with 3-MC, whereas a fragment from -2752 to -2259 exhibits only a 1.9-fold increase. When we generated a fragment from -2546 to -2095, a consistent 2.0-fold increase in CAT activity was observed. Since inducible CAT activity drops from nearly 6to 2-fold by eliminating bases from -2095 to -1987, these data indicate that DNA within this region (-2095 to -1987) is important for 3-MC-inducible activity.
When the DNA sequence was analyzed from -2546 to -1987, two XRE-like sequences were identified in this region. These sequences, referred to as X1 and X2, are shown in comparison to the mouse DRE3 (12) and human XREl(10) sequences and the consensus XRE (Fig. 3). The X1 sequence is 84% similar to the consensus XRE and the sequence X 2 is 79% similar. In addition, X1 is 75% similar to the functional consensus (12) and 100% similar to the consensus binding sequence (171, whereas X2 is 88 and 71% similar, respectively.
Gel Mobility Shift Assays of XRE-like Sequences-The fact that both XRE-like sequences appear in the 3-MC-responsive element suggested that X1 and X 2 may be recognized by the AhR and that both sequences may be required for inducermediated activation of CAT. To determine if either of the XRElike sequences were able to serve as a target sequence for the AhR, gel mobility shift assays were employed. Initial experiments were conducted to determine if these two XRE-like se-

A G G C G T T G C G T G A G A A G G A CONSENSUS
HumanX1

A W A G T G C G T G T C A G G T C -2142
HumanX2 quences could compete for binding to the mouse DRE3. Using HepG2 nuclear extracts from control and TCDD-treated cells, it is demonstrated that only the X1 sequence competed for AhR binding (Fig. 4). An SV40 promoter fragment, used as a nonspecific competitor, did not reduce binding of the inducible band to the DRE3 sequence. To show directly that this X1 sequence binds the AhR, gel mobility shift assays were performed using nuclear extracts prepared from wild type mouse Hepa-lclc7 cells in addition to the Class I1 variant cells. Cells were treated for 2 h with 10 n~ TCDD as described under "Materials and Methods." As shown in Fig. 5, a TCDD-inducible band formed with the X1 sequence. The X1-inducible band was specific in that it was eliminated by the presence of cold DRE in the reaction, but not by a nonspecific DNA, which lacks a DRE consensus sequence. In addition, in vitro transformation of the cytosolic AhR, followed by gel shift analysis, resulted in specific binding of the activated receptor to the X1 sequence (Fig. 6).

T T T C C T T G C G T T T T A C C T
Differences in the intensities of the TCDD-inducible bands between the DRE and X1 binding probably result from a lower affinity of the AhR for the X1 sequence, implying that nucleotides other than those described by Denison (17) contribute to tight binding. However, the presence of specific AhR binding at the X1 sequence is consistent with the functional response observed with the fragment of DNA that contains this element.
Location of a Second TATA Box-While the sequence spanning the X1 region is important for promoter and enhancer activity, DNA sequence between -2095 and -1987 also contributes to 3°C induction. The DNA sequence from the X1 region to -1970 is shown in Fig. 7. The sequence from -2221 to -1970 shows several regions of DNA that have potential to associate with known transcriptional factors. There exists two potential A P 1 binding sites which are characterized by the consensus sequence STGACTMA (33), a half-binding site for the liverspecific protein HNFl (34) and a conserved TATAA box (35). Interestingly, a DNA sequence search of GenBank to determine the fidelity of eukaryotic promoters identified the sequences within this region as a probable promoter. To examine the hypothesis that the second TATAA sequence, identified as TATA2, can serve as an efficient promoter, an EcoRVIHaeIII digest was performed on the -3201 promoter CAT construct and the DNA from bases -2259 to -1970 cloned into the promoterless CAT construct (Fig. 7B ). This DNA construct contains a fragment of DNA with the X2 sequence, the potential AP1 sites, and the TATA2 sequence. When this DNA was transfected into HepG2 cells, CAT activity was detected, which indicates that the DNA is able to support promoter activity. In addition, when the cells were treated with 3-MC, nearly a 4-fold increase in CAT activity was observed. However, when this element was placed in the opposite orientation, there was no constitutive or 3°Cinducible CAT activity. This result indicates that this fragment supports promoter activity in an orientation-specific fashion and that other regions of the DNA enhance the promoter activity following treatment with 3°C. When the DNA fragment from -25461-2095 (Fig. 2 B ) , which does not contain the TATA2 element, was cloned into the promoterless CAT construct, no expression or inducible activity was observed. This result suggests that the TATA2 sequence, in addition to other DNA sequences such as the AP1 sites, may play an important role in supporting inducible CAT activity.

DISCUSSION
The in vivo regulation of the mouse CyplaZ gene is controlled through transcriptional gene activation (5). Genetic studies, in which 1A2 expression cosegregates with the AhR, implicates the AhR as the mediator for the induced expression of 1A2 (36). Primary rat hepatocytes have been used to show by nuclear run-on studies that the rat CYPlA2 gene is also transcriptionally activated by PAHs (18). Taken together, these data strongly suggest that the AhR binds sequences in the flanking gene to activate transcription.
Transient transfection studies in transformed human cell lines helped identify a region in the 5'-flanking gene of the CYPlA2 gene that was responsive to 3°C in the human hepatoma cell line, HepG2, but not in other non-hepatic cell lines (19). The present study was designed to localize the sequences responsible for the induced response and to determine if they bind the AhR. Deletional analyses of CYPlA2 promoter-CAT constructs and SV40 promoter-CAT constructs identified two regions in the 5"flanking gene that appear to be required for full inducer activity. One region contained an XRE-like sequence (Xl) that was able to bind a specific TCDD-inducible complex in receptor-competent cells, but not in receptor defective cells. Binding could be displaced by the DRE sequence, confirming that the inducible protein associated with X1 was the AhR. It is important to appreciate that binding of the AhR to X1 was much weaker than that observed for binding to the DRE element, which correlated with weak enhancer activity of DNA that contained the X1 sequence. However, when the X1 sequence was eliminated, promoter-and enhancer-specific transcriptional activation dropped up to 50%, indicating that the region of DNA containing X1 was important for the maximal 3-MC-induced activity of the CYPlA2 gene.
A second region of DNA, which contained another XRE-like sequence that we have called X 2 , did not bind the receptor as determined by gel mobility shift assays. However, when DNA fragments were prepared that resulted in removal of the X1 sequence but not the X2 sequence, 3°C was still able to initiate a 3-fold induction of CAT activity. DNA fragments that encompassed a region from -2259 to -1987 served as efficient enhancers in either orientation when directing 3-MC-induced transcription of the heterologous SV40 promoter CAT plasmids, demonstrating there existed cis-acting elements within this region that supported induction by PAHs. Interestingly, the X 2 sequence, which resembles the known DRE sequence, does not bind the AhR and most likely plays little role in this induction process. However, there are two stretches of DNA that are similar to the consensus AP1or 12-0-tetradecanoylphorbol-13-acetate-responsive elements. These potential AP1 elements could play a significant role in C Y P l A 2 induction by PAHs, since it has recently been demonstrated that an AP1 binding site is crucial for the P-naphthoflavone-induced transcriptional activation of the human NAD(P)H:quinone oxidoreductase gene (37). In addition, when we constructed by polymerase chain reaction a fragment of DNA that contained the X1 and X2 sequence, but not the 3' AP1 site (-25461-2095), a significant reduction in inducible CAT activity was observed. Since it is known that a complex assortment of proteins from the Fos and Jun family bind AP1 sites (38) and TCDD actually induces binding to AP1 sites, (39) it is very likely that the AP1 sites on this fragment of DNA contribute to the 3-MC-induced transcriptional activation that is observed.
During the course of these experiments, it became apparent that a second TATA box existed within the responsive region. When DNA containing the AP1 sites and the TATA box were cloned upstream of the promoterless CAT gene, the plasmid supported constitutive and 3-MC-induced CAT activity. It is unclear what role the second TATA box plays in the expression and inducibility of the human CYPlA2 gene, but it is conceivable that along with the other regulatory elements located in the same region of the gene, this second TATA box participates in transcriptional control of the gene.
Combined, these results suggest that induction of the C Y P I A 2 gene is controlled by a number of different regulatory factors. The X1-AhR binding region is important for the overall expression of the gene, but removal of this portion of DNA does not completely eliminate the induction response. Although the DRE elements that flank the CYPlA1 gene serve as enhancer sequences and are responsible for most of the PAH-and TCDDinitiated transcriptional activation, additional regulatory elements appear to support PAH induced transcriptional activation of the C Y P I A 2 gene. Similar results have been observed in both the NAD(P)H:oxidoreductase gene (37,40) and the glutathione S-transferase Ya subunit gene (4143), which contain both the AhR-specific XRE elements (12,44) as well as the antioxidant-responsive elements that are encoded by the sequence 5'-ggTGACaaaGC-3' (42,43). Since the antioxidantresponsive element reveals sequence similarity to the motifs that are recognized by the Jun and Fos family of proteins, experiments have been conducted to examine the contribution of these proteins toward gene regulation. These results indicate that these proteins are involved in the induction of the NAD(P)H: quinone oxidoreductase gene (37), but most likely do not participate in the induction of the glutathione S-transferase Ya