Identification and Characterization of a cAMP-Responsive Element in the Region Upstream from Promoter 1.3 of the Human Aromatase Gene

Archives of Biochemistry and Biophysics Vol. 371, No. 2, November 15, pp. 179 –190, 1999 Article ID abbi.1999.1454, available online at http://www.idealibrary.com on

Identification and Characterization of a cAMP-Responsive Element in the Region Upstream from Promoter 1.3 of the Human Aromatase Gene Dujin Zhou and Shiuan Chen 1 Division of Immunology, Beckman Research Institute of the City of Hope, Duarte, California 91010

Received March 5, 1999, and in revised form August 25, 1999

Aromatase converts androgens to estrogens. The expression of this enzyme is driven by multiple tissuespecific promoters which are differentially regulated. Aromatase expression in breast cancer and the surrounding adipose cells is directed mainly by promoters I.3 and II, while its expression in the normal breast adipose tissue is driven by promoter I.4. Like promoter II, promoters I.3 is thought to be a cAMP-driven promoter, demonstrated previously by cell culture experiments. In the present study, we have identified and characterized a cAMP-responsive element (CREaro) upstream from promoter 1.3. This positive element, TGAAGTCA, between 266 and 259 bp relative to the transcriptional start site of promoter 1.3 was identified by DNA deletion and mutation analyses. The sequence of CREaro is one base different from the consensus CRE sequence (CREpal; TGACGTCA), and the mutational analysis revealed that CREaro had a higher enhancer activity to promoter I.3 than CREpal. Nuclear proteins from both WS3TF breast tumor fibroblasts and SK-BR-3 breast cancer cells bound to this CREaro, as demonstrated by DNA mobility shift assay. The molecular weight of the major binding protein in fibroblasts was determined to be approximately 60 kDa, as shown by UV crosslinking, which is different from those of known CRE-binding proteins. It is thought that CREB1 is not expressed in tumor fibroblasts because the Western blot analysis using antiCREB1 antibody was not able to detect CREB1 in the nuclear protein extract from these cells. DNA mobility shift analysis using a nuclear protein extract from SKBR-3 cells revealed that at least two proteins bound to the CREaro and that one of these proteins was identified to be CREB1. These studies provide direct evidence that promoter I.3 is a cAMP-responsive promoter. © 1999 Academic Press

1 To whom correspondence should be addressed. Fax: (626) 3018186. E-mail: [email protected].

0003-9861/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.

Key Words: aromatase; gene expression regulation; cAMP-responsive element.

Human breast cancers can be divided into hormonedependent and hormone-independent subtypes. About two-thirds of breast cancers are hormone-dependent. Estrogens are considered to be the major hormonal stimulus for growth of the hormone-dependent type of breast carcinoma (1). The major sites of estrogen biosynthesis are the ovarian granulosa cells in premenopausal women and adipose tissue in postmenopausal women (2). The conversion of androgens to estrogens, the rate-limiting step in estrogen production, is catalyzed by the enzyme complex aromatase. Aromatase plays a key role in the pathogenesis of hormone-dependent breast cancer. James et al. (3) reported that aromatase activity, when measured in vitro, was found to be higher in breast tumors than in the fat next to the tumor or in normal breast fat. Studies from our and other laboratories have demonstrated that the expression levels of both aromatase mRNA and aromatase protein in breast cancer tissues were significantly higher than those in regions distal to tumors or in noncancerous breast tissues (4 –9). Estrogens, locally produced by aromatase in breast cancer cells, play an important role in promoting breast tumor growth in both an autocrine and a paracrine manner (10 –12). In addition, results obtained from studies using a transgenic mouse model in which aromatase is overexpressed in mammary tissues indicate that in situ produced estrogen plays a more important role than circulating estradiol in breast tumor promotion (13). The control of human aromatase gene expression is complex in that several promoters direct aromatase gene expression in a tissue specific manner (14 –18). This conclusion is based on the fact that there are 179

180

ZHOU AND CHEN

multiple tissue-specific exons 1, exons 1.1, 1.2, 1.4, 1.5, 1.3, and pII, existing in the mRNAs isolated from placenta, placenta, adipose tissue, skin fibroblasts or fetal liver, adipose tissue, and ovary, respectively. Aromatase mRNA in the noncancerous breast regions is exon 1.4-dominant, indicating that aromatase expression in the normal breast tissue is mainly driven by promoter 1.4 which has been shown to be glucocortcoiddependent (19). Our previous studies on 70 breast tumor specimens using RT-PCR technique has revealed that exons 1.3 and pII are the two major exon 1s present in aromatase mRNAs isolated from breast tumors, suggesting that promoters 1.3 and II are the major promoters directing aromatase expression in breast cancer and surrounding adipose stromal cells (20). The switching from promoter 1.4 to 1.3 in breast cancer cells was also shown from a study on 49 Japanese breast cancer samples (9) and a study on 18 breast cancer patients (21). It has been demonstrated that both aromatase activity and aromatase transcripts in adipose cells were increased dramatically upon exposure to (Bu) 2cAMP or forskolin (17). Furthermore, the exon 1.3-specific and exon II-specific aromatase transcripts, but not 1.4-specific sequences, were detected in cells cultured in the presence of cAMP (17), implying a cAMP-dependent mechanism for the transcriptional regulation of promoters 1.3 and II. An Ad4BP/SF1 element and recently a CRE-like sequence (CLS) in the aromatase promoter II region have been reported to be critical for the transcriptional response of the promoter II to cAMP (22, 23). We have previously identified and characterized the promoter 1.3 (24) and a silencer element that is situated between promoters I.3 and II and downregulates the function of these promoters (25). In this study, we identified a cAMP-responsive element, CREaro, which is situated upstream from promoter I.3. Functional analysis of this enhancer element was performed in both breast cancer SK-BR-3 cell line and breast tumor fibroblasts WS3TF. MATERIALS AND METHODS Materials. Restriction endonucleases, T4 kinase and T4 DNA ligase were purchased from New England Biolabs, Inc. (Beverly, MA), Boehringer Mannheim Biochemicals (Indianapolis, IN), or Promega (Madison, WI). Radiolabeled nucleotides were from New England Nuclear (Boston, MA). [ 14C]Chloramphenicol (D-threo-[dichloroacetyl-l- 14C] chloramphenicol; sp. act. 55 mCi/mmol) was from Amersham (Arlington Heights, IL). Forskolin and other reagents were bought from Sigma (St. Louis, MO). DNA sequencing kits were from United States Biochemical (Cleveland, OH). CREB consensus oligonucleotide, CREB mutant oligonucleotide, human CREB1 DNA binding domain, and CREB antibody were bought from Santa Cruz Biotechnology Co. (Santa Cruz, CA). The CAT expression vector, pUMSVOCAT, is the gift from Dr. K. Kurachi at University of Michigan (Ann Arbor, MI). Cell culture and forskolin induction. SK-BR-3 cell line (human breast adenocarcinoma) was from ATCC (Rockville, MD). SK-BR-3

cells were cultured in McCoy’s 5a medium supplemented with 10% fetal bovine serum, penicillin, and streptomycin at 37°C and 5% CO 2. WS3TF, a tumor fibroblast line derived from breast cancer area, was the gift of Dr. R. J. Santen at University of Virginia Health Science Center (Charlottesville, VA) and maintained in Waymouth’s MB 752/1 medium with 15% fetal calf serum. The cultured cells at 70% confluence were treated with either dimethyl sulfoxide (for vehicle control) or forskolin in dimethyl sulfoxide to a final concentration of 20 mM for 24 h. The cells from each treatment were subjected to nuclear extract preparation or RNA isolation. RNA preparation and RT-PCR analysis of aromatase transcripts. Total RNA from different cultured cells was extracted using the procedures as previously described (20). Briefly, cells were washed with ice-cold phosphate-buffered saline. The pellets of harvested cells were resuspended in a lysis buffer (10 mM Tris–HCl, pH 7.8, 1.5 mM MgCl 2, 150 mM NaCl, and 0.65% NP-40) and incubated for 5 min on ice. The intact nuclei were removed by a brief microcentrifuge spin, and an equal volume of a urea-containing buffer (100 mM Tris–HCl, pH 7.4, 7 M urea, 0.35 M NaCl, 1% SDS, and 10 mM EDTA) was added to the cytoplasmic supernatant to denature proteins. Proteins were removed by extractions with phenol/chloroform. The cytoplasmic RNAs were recovered by ethanol precipitation and quantitated by measuring their absorbance at 260 nm. The RNA preparations were treated with RNAse-free DNAse to remove the contaminant DNA. Primer-specific RT-PCR analyses were performed to examine the alternative utilization of exon Is of the human aromatase gene in the forskolin-treated cultured cells as described previously (20) using the exons 1-specific primers designed by Harada et al. (16). Specifically, RT-PCR reactions were carried out in a total volume of 50 ml, containing 20 mM Tris–HCl, pH 8.3, 50 mM KCl, 15 mM MgCl 2, 0.01% gelatin, 200 mM dNTPs, 20 pmol of each primer, 1.5 mg of total RNA, 2.5 units of Taq DNA polymerase, and 2 units of AMV reverse transcriptase. The reverse transcription reaction was carried out at 37°C for 7 min, and PCR was performed for 25 cycles using the following temperature profile: 51°C, 2 min (primer annealing); 72°C, 2 min (primer extension); and 94°C, 1 min (denaturation). An additional extension cycle was performed for 8 min at 72°C before cooling the reaction mixture to 4°C. To examine total levels of aromatase transcripts, we also performed RT-PCR using primers 2a and 2d to amplify the exon II region. All aromatase mRNA contains exon II regardless of which exon I is present. In order to eliminate artifacts, after separation of the PCR product on agarose gel (1.8%), the DNA products are transferred to Zetaprobe membranes (Biorad), followed by hybridization using a 32P-labeled probe derived from the middle of exon II (59-ATGGTTTTGGAAATGCTGAA-39). Hybridization using such a third probe further ensures that the PCR products are those expected. The conditions for hybridization are according to the BioRad instruction manual. For quantification, Zetaprobe membranes hybridized with an exon II probe were exposed to a PhosphorImager (Molecular Dynamics, Sunnyvale, CA) and the radioactivity of PCR products was determined. We have performed PCR at a nonsaturating condition, 25 cycles. In this way, at least we can compare the relative usages of different exon Is in RNAs from the same sample. Our RT-PCR analysis was verified with two control experiments (20). With 25 cycles, product accumulation was found to be exponential. In addition, the quantity of PCR products obtained with these cells (comparing the intensities of hybridized bands) increased in an aromatase mRNA concentration-dependent manner. In order to verify the equal loading of total RNA in each sample, the transcripts of a housekeeping gene b-actin in 1.5 mg total RNA from forskolin-treated and control SK-BR-3 cells and fibroblasts were amplified and hybridized by a similar RT-PCR/ southern hybridization analysis. The sequence of b-actin sense primer is 59-AGGAGCACCCCGTGCTGCTGA-39, and the antisense primer is 59-CTAGAAGCATTTGCGGTGGAC-39. The oligonucleo-

CHARACTERIZATION OF cAMP-RESPONSIVE ELEMENT tide derived from b-actin, 59-CATCACCATTGGCAATGAGCGGTTCCGCTG-39, was used as probe in hybridization. CAT expression plasmids. A low background promoterless CAT expression vector pUMSVOCAT (26) with modification at its cloning sites (27) was used for preparation of the CAT expression constructs. The 1.2-kb aromatase genomic fragment containing 836 bp of 59 flanking sequences of promoter 1.3, promoter 1.3 region, and entire sequences of exon 1.3 was subcloned into pBluescript vector, resulting in pBS-1.2 kb aro. This plasmid was used as template DNA in the PCR reactions to generate a set of 59 or 39 deletion constructs with sets of 59 and 39 primers flanking the designated regions with a artificial HindIII site at the 59 end and a XbaI site at 39 end. The fragments, 2836/1129 bp, 2716/1129 bp, 2596/1129 bp, 2476/ 1129 bp, 2356/1129 bp, 2304/1129 bp, 2224/1129 bp, 2144/1129 bp, 268/1129 bp, 234/1129 bp, 224/1129 bp, 156/1129 bp, 268/15 bp, and 2144/15 bp, were restricted with HindIII and XbaI, purified on the agarose gel, and subcloned into the HindIII/XbaI site of the modified pUMSVOCAT vector. These constructs are designed as pUMS2836/1129 CAT, pUMS2716/1129CAT, pUMS2596/ 1129CAT, pUMS2476/1129CAT, pUMS2356/1129CAT, pUMS2304/ 1129CAT, pUMS2224/1129CAT, pUMS2144/1129CAT, pUMS2104/ 1129CAT, pUMS268/1129CAT, pUMS234/1129, pUMS224/ 1129CAT, pUMS156/1129, pUMS268/15CAT, and pUMS2144/ 15CAT, respectively. Three mutations, TGAAGTCA (CREaro) 3 TGTTGTCA (MuCREaro 1), TGAAGTCA 3 TGAAGTGT (MuCREaro 2), and TGAAGTCA 3 TGACGTCA (MuCREaro 3, i.e., palindromic CRE), of CREaro element were generated by PCR-directed mutagenesis (28) within 2104/1129 fragment which contains promoter I.3 and the silencer element. The 2104/1129 fragments containing the desired mutations were restricted with HindIII/XbaI, gel purified, and subcloned into HindIII/XbaI sites of pUMSVOCAT. These plasmids were designed as pUMSMuCRE1, pUMSMuCRE2, and pUMSMuCRE3, respectively. The sequences of all PCR amplified fragments were confirmed by direct sequencing. Transfection and CAT assay. SK-BR-3 cells and tumor fibroblasts, WS3TF, were transfected by the use of Lipofectin (Life Technologies, Inc.) according to the manufacture’s instructions. The cotransfection experiments were performed 20 to 24 h after seeding approximately 4 3 10 5 cells per 60-mm tissue culture dish using 10 mg of the test plasmid and 3 mg of the plasmid pSV-b-Gal which was used to normalize the transfection efficiency. After overnight incubation, medium containing lipofectin and DNA were removed, and the cells were cultured in the regular growth medium with or without 20 mM forskolin. After 24 h incubation, the cells were harvested from the plates by scraping, pelleted by centrifugation, resuspended in 0.25 M Tris–HCl, pH8.0, and disrupted by freezing/thawing four times. Aliquots of the lysate were used for assay of b-galactosidase activity (29). CAT activity in the cell extracts containing an equal amount of b-galactosidase activity from each sample was determined by the liquid scintillation counting (LSC) method (30). Briefly, the appropriate amount of cell extracts was incubated in a reaction containing 14C-labeled chloramphenciol and n-Butyl Coenzyme A. The reaction products were extracted with a small volume of xylene. The xylene phase was mixed with scintillant and counted in a scintillation counter. The CAT activity was expressed as -fold increase in activity over that of the pUMSVOCAT vector (1.0) and shown as means 6 SE of three independent transient transfection experiments performed for each construct. Nuclear extract preparations. Nuclear extracts from SK-BR-3 cells and tumor fibroblasts were prepared according to the method of Kingston et al. (31) with a slight modification. Cells were harvested and washed twice with PBS. Pellets were resuspended in 10 vol of buffer A (300 mM sucrose, 10 mM Hepes, pH 7.9, 10 mM KCl, 1.5 mM MgCl 2, 0.1 mM EGTA) and homogenized by 20 strokes of a Kontes all glass Dounce homogenizer (A pestle). Buffer A and all buffers described below contained 0.5 mM DTT, 0.5 mM PMSF, and

181

2 mg/ml each of pepstatin, leupeptin, and antipain. After centrifugation at 4000 rpm for 5 min, the pellets were resuspended in 10 vol of buffer B (10 mM Hepes, pH 7.9, 400 mM NaCl, 1.5 mM MgCl 2, 0.2 mM EGTA, and 5% glycerol) and were dounced 20 times with a B pestle. Homogenates were mixed for 30 min on a rotator-wheel in the cold room and then centrifuged at 15,000g for 30 min. Supernatants were collected and dialyzed for 4 h against 50 vol of buffer C (20 mM Hepes, pH 7.9, 0.1 mM EDTA, 75 mM NaCl, and 20% glycerol). Dialysates were centrifuged at 15,000g for 15 min and supernatants stored at 270°C until used in gel-shift assays and footprinting analyses. DNA mobility shift analysis. The double stranded CREaro-containing oligonucleotide 276/250 bp, 59-AACCTGCTGATGAAGTCACAAAATGAC-39, was end-labeled with [ r-32P]ATP and T4-Kinase and used as a probe in the mobility shift assay. Mobility shift assays were done as described by Singh et al. (32). Briefly, 5 mg of nuclear extract was incubated with 6000 cpm of 32P-labeled probe at room temperature for 30 min in 15 ml of a mixture containing 10 mM Tris–HCl, pH 7.5, 50 mM NaCl, 1 mM MgCl 2, 0.5 mM EDTA, 0.5 mM DTT, 4% (v/v) glycerol, and 0.1 mg/ml poly(dI– dC). The reaction mixture was electrophoresed on 6% acrylamide:bis acrylamide (74:1) gel with 0.5 3 TBE at 10 v/cm in a cold room. Gels were dried and autoradiographed. For competition experiments, the conditions used for binding of the nuclear factors to each probe were the same as those described above except that the appropriate amounts of the unlabeled DNA fragments and poly(dI– dC) were supplemented in the binding reaction mixture as a specific and a nonspecific competitor DNA, respectively. In order to evaluate the interaction of nuclear proteins with CREaro, the double-stranded oligonucleotides for CREB consensus oligonucleotide (59-AGAGATTGCCTGACGTCAGAGAGCTAG-39, Santa Cruz Biotechnology), CREB mutant oligonucleotide (59-AGAGATTGCTGTGGTCAGAGAGCTAG-39, Santa Cruz Biotechnology), and the mutated CREaro were used as competitors in the gel mobility shift analysis. The sequences of these mutations-containing oligonucleotides and the wide type CREaro-containing oligonucleotide are as follows (only sense strands are shown): Wild type CREaro: 59-AACCTGCTGATGAAGTCACAAAATGAC-39 Mu CREaro 1: 59-AACCTGCTGATGTTGTCACAAAATGAC-39 Mu CREaro 2: 59-AACCTGCTGATGAAGTGTCAAAATGAC-39 Mu CREaro 3 (CREpal): 59-AACCTGCTGATGACGTCACAAAATGAC-39.

The CRE-like sequences are underlined and the mutated bases are in bold. For supershift assays, 4 ml of rabbit polyclonal IgG against human CREB1 was incubated with 5 mg of nuclear extracts on ice for 1 h before incubating with 32P-labeled CREaro fragment. Western blot analysis. Nuclear proteins from SK-BR-3 cells, MCF-7 cells, and tumor fibroblasts were prepared as described under Nuclear Extract Preparations. Nuclear proteins, 50 mg of each sample, were electorphoresed on 10% SDS–PAGE. The fractionated proteins were then electrophoretically transferred to nitrocellulose membrane (Trans-Blot, transfer medium, Bio-Rad) in 25 mM Trizma base/190 mM glycine. The electrophoretic transfer was performed in a Trans-Blot SD semidry electrophoretic transfer cell (Bio-Rad) at 200 mA for 1 h. The membrane was pretreated with 5% nonfat milk in 1X PBS buffer, pH 7.5, for 1 h at room temperature. The nitrocellulose was then washed three times with PBS for 5 min each. After washing, the blocked nitrocellulose membrane was incubated with the 1:1000 diluted polyclonal anti-CREB1 IgG (Santa Cruz) for 1 h at RT with shaking. The filter was washed three times with PBS containing 0.05% Tween 20, 5 min each, and incubated with 1:2500 diluted secondary antibody, goat anti-rabbit HRP conjugate (Pierce). A sensitivity enhanced chemiluminescent substrate, SuperSignal Substrate (Pierce), was used for the coloring reaction according to the manufacturer’s instructions. UV crosslinking studies. The UV crosslinking experiments were performed as described by Chodosh et al. (33) with some modifica-

182

ZHOU AND CHEN

tions. The BrdU-substituted oligonucleotides for CREaro were synthesized by the DNA synthesis laboratory at the City of Hope. Double-stranded, BrdU-substituted CREaro fragment was end-labeled with [ r-32P]ATP with T4 Kinase and used as a probe in UV crosslinking experiments. In these experiments, 10 mg of nuclear extract were incubated with 1 3 10 5 cpm of 32P-labeled, BrdU-substituted CREaro and 2.5 mg of poly(dI– dC) for 30 min at room temperature in 20 ml of the same binding buffer as described for the DNA mobility shift experiments. The mixtures were irradiated for 6.5 min under a UV transillumintor at 254 nm at an intensity of 120,000 mW/cm 2 (UVC.515 ultraviolet multilinker, Ultra-Lum, Inc., Carson, CA) at a distance of 5 cm from the UV source. The mixtures were electrophoresed on a 10 –20% gradient SDS-polyacrylamide gel, and the gels were then dried and autoradiographed.

RESULTS

Induction of Exon 1.3-Specific Aromatase mRNA in SK-BR-3 Cells and in Tumor Fibroblasts WS3TF by Forskolin To determine if cAMP affects aromatase expression in breast cancer cells and stromal cells, we investigated the change in aromatase mRNA levels and in alternative promoter usage in response to forskolin in breast cancer SK-BR-3 cells and in tumor fibroblasts WS3TF. After 24 h incubation in the presence of 20 mM forskolin (an inducer of cAMP synthesis), total RNA was extracted from both treated and untreated cells and subjected to RT-PCR-Southern analysis as described under Materials and Methods. The results showed that the addition of forskolin increased the total aromatase mRNA (as estimated by the level of exon II-specific RT-PCR product) by four- and five-fold in SK-BR-3 cells and WS3TF cells, respectively (data not shown). As shown in Fig. 1, the levels of exon 1.3-specific aromatase transcripts in forskolin-treated SK-BR-3 cells and tumor fibroblasts were about two- and five-fold higher than that in the untreated cells, respectively. On the other hand, as an internal control, the levels of b-actin mRNA in forskolin treated and untreated cells remain about the same (Fig. 1). Identification of a CRE-Like (CREaro) Element within the Promoter 1.3 Region of the Human Aromatase Gene To determine whether the cAMP-induced increase of the transcriptional function of promoter 1.3 was mediated through specific DNA sequences, a series of 59 deletion constructs was created in the 59-flanking region of promoter 1.3. DNA fragments which contain deletions of the 59-flanking sequences of promoter 1.3 and 129-bp exon 1.3 sequences were prepared by PCR and fused to the coding sequences of the bacterial reporter gene, CAT, in pUMSVOCAT vector. The promoter activities and transcriptional regulation by cAMP of these deleted DNA fragments were first examined by transient transfection of SK-BR-3 cells. Fibroblasts would be our first choice as host cells for this

FIG. 1. Effects of forskolin on the exon 1.3-specific aromatase mRNA level. 1.5 mg total RNA from control (forskolin 2) or forskolintreated (20 mM for 24 h) SK-BR-3 cells and tumor fibroblasts was subjected to RT-PCR as described under Materials and Methods. One-fifth of the amount of the RT-PCR product was analyzed by Southern hybridization with oligonucleotide cDNA probe derived from the middle of exon II of aromatase gene, or from b-actin gene. The membrane was exposed to a PhosphorImager, and the radioactivity of PCR products was determined. The intensities of exon 1.3-specific and b-actin-specific PCR products in forskolin treated SK-BR-3 cells and tumor fibroblasts were shown by taking those of the untreated samples as 100%.

study because a much higher level of cAMP induction was detected in tumor fibroblasts, as shown in Fig. 1. However, a significant quantity of cells would be needed for large-scale transient transfection experiments. Culturing such a large amount of fibroblasts presents technical difficulties due to the fact that fibroblasts are a primary culture and grow slowly. Since the usage of exon I.3/promoter I.3 in SK-BR-3 cells can be induced by forskolin, and the transfection efficiency of this cell line is reasonable, we decided to use SK-BR-3 cells as host cells for the deletion mutation analysis to first map the cAMP-responsive region. We then selected several of the deletion CAT constructs to transfect tumor fibroblasts to determine if the results from SK-BR-3 cells hold true for tumor fibroblasts. As shown in Fig. 2A, after treatment with 20 mM of forskolin for 24 h, a four- to five-fold increase over the basal levels was detected in SK-BR-3 cells transfected with pUMS2104/1129CAT and pUMS268/1129CAT plas-

CHARACTERIZATION OF cAMP-RESPONSIVE ELEMENT

183

FIG. 2. Mapping of the cAMP-responsive element within the promoter 1.3 region of the human aromatase gene by 59-deletion analysis. (A) A map of promoters 1.3 and II region of human aromatase gene is shown at the top. The relative positions of TATA boxes, CRE, and silencer elements are indicated. Schematic diagrams of various deleted promoter 1.3-CAT fusion constructs are shown at the left. The number are counted from the transcription start site of promoter 1.3 (11). UMS, upstream mouse sequence; CAT, chloramphenicol acetyltransferase. Deletion constructs containing different lengths of 59-flanking region of promoter 1.3 were transfected into SK-BR-3 cells. The transfected cells were cultured in the medium with or without 20 mM forskolin for 24 h. The basal and induced CAT activities of the transfected SK-BR-3 cells are shown at the right, and expressed as the mean 6 SD of three independent experiments. (B) CAT activities in transfected tumor fibroblasts. The experimental conditions were the same as described in A. The CAT activities in forskolin-treated transfected cells were expressed as the fold of that in untreated cells transfected with respective constructs.

184

ZHOU AND CHEN

FIG. 3. The nucleotide sequence of the 2104/1129-bp region containing promoter 1.3 of the human aromatase gene. The important regulatory segments, CREaro (identified in this study), promoter I.3 (B1, reference 24), CLS (CRE-like sequence, reference 23), silencer (reference 25), and ERRa-1/SF-1 binding site (Refs. 22 and 36), are underlined.

mids. However, forskolin induction of the CAT activity was much lower in the cells transfected with pUMS234/1129, suggesting that the sequences between 268 and 235 bp are required for the full response of promoter 1.3 to cAMP. In addition, the basal activity of the pUMS268/1129CAT transfected cells was significantly higher than that of the pUMS234/ 1129CAT transfected cells, suggesting an action of endogenous cAMP. As the control, no induction could be detected for SK-BR-3 cells transfected with the pUMSVOCAT vector without the aromatase genomic segments. Similar results were obtained from the transient transfection experiments with tumor fibroblasts as shown in Fig. 2B. As showed in Fig. 2B, the forskolin treatment caused a seven- to eleven-fold induction of CAT activity in fibroblasts transfected with several CAT constructs carrying the promoter 1.3 region from 2304/1129 to 268/1129. Further deletion at 59 termini of promoter 1.3 region, from 268/1129 to 234/ 1129, resulted in the reduction in cAMP induction of CAT activity, from seven-fold induction to three-fold induction. This result supports the hypothesis that the cAMP-responsive element is located within the 268/ 234 region. In the presence of forskolin, the CAT activity of fibroblasts transfected with pUMS234/ 1129CAT was found to be three times higher than the basal level. This result suggests that other sequences such as Ad4BP/SF1 element (22) and a CRE-like sequence (23) in the 234/1129 region may also be needed for promoter 1.3 to fully respond to cAMP in tumor fibroblasts.

Mutational Analysis of the cAMP-Responsive Element CREaro The nucleotide sequence of 2104/1129 fragment is shown in Fig. 3. The CRE consensus sequence homology analysis of 268/235 fragment reveals the presence of a CRE-like sequence between 266 and 259 bp, TGAAGTCA. This eight-base segment was named as CREaro that is one base different from that of the CRE consensus sequence, TGACGTCA, here named CREpal. To determine whether CREaro is a true cAMP responsive element, we generated three mutations in CREaro as described under Materials and Methods and shown in Fig. 4. The mutant plasmids were transfected into both SK-BR-3 cells and WS3TF cells to determine whether CREaro is functional in both breast cancer cells and stromal fibroblasts. The relative increases of CAT activities of transfected tumor fibroblasts and SK-BR-3 cells, cultured in the medium with or without forskolin, were shown in Fig. 4. All three mutations reduced the responses to forskolin. The pattern of relative decreases was different in two cell lines. In pUMSMuCRE1, the AA bases of CREaro were mutated to TT (see Fig. 4). This mutation resulted in a marked decrease in cAMP induction in WS3TF cells, while the mutation with the last two bases changed from CA to GT has less effect on cAMP induction. Surprisingly, cAMP induction of CAT expression in tumor fibroblasts transfected with pUMSMuCRE3, which contains the CRE consensus sequence, CREpal, was also significantly reduced. In SK-BR-3 cells, the

CHARACTERIZATION OF cAMP-RESPONSIVE ELEMENT

185

FIG. 4. Mutational analysis of cAMP-responsive element upstream from promoter 1.3 of the human aromatase gene. The DNA fragments, 2104/1129 bp, with original CRE (CREaro) or mutated CREs were generated and fused with CAT reporter gene as described under Materials and Methods. Tumor fibroblasts or SK-BR-3 cells were transfected with these CAT plasmids and cultured in the presence or absence of forskolin. Results of the CAT assays are presented as fold of induction, and expressed as mean 6 SD of three experiments.

largest decrease in CAT expression was found in cells transfected with pUMSMuCRE2. These results indicate that cAMP stimulation of promoter 1.3 transcription is mediated through CREaro. Since CREpal has a lower activity than CREaro, trans-acting factors other than CREB1 may interact with CREaro. In addition, considering the finding that the activity was reduced to different degrees with three mutants in breast cancer SK-BR-3 cells and tumor fibroblasts WS3TF, cAMP induction may involve different CRE-binding proteins or distinct mechanisms in these two types of cells. Gel Mobility Analysis of Proteins Binding to the CREaro Sequence Mobility shift analysis was used to determine whether specific complexes could form between CREaro and proteins in nuclear extracts from forskolin-treated fibroblasts and SK-BR-3 cells. We first used as a probe, the end-labeled oligonucleotide with sequences corresponding to the region 276 to 250 bp with respect to the transcription start site of exon 1.3. As shown in Fig. 5A, lane 2, and in Fig. 5C, lanes 3– 6, a major DNA/protein complex was formed between the probe and the nuclear protein(s) in forskolin-treated fibroblasts. However, fewer complexes were formed when the nuclear extracts from untreated fibroblasts were used (Fig. 5C, lanes 10 –14). The DNA/protein complex was totally competed by the addition of a 500-fold molar excess of unlabeled CREaro-containing DNA (Fig. 5A, lane 3, and Fig. 5C, lane 7). A 500-fold molar excess of CREpal was also able to compete for the protein binding to the CREaro probe (Fig. 5A, lane 4). Since it has been shown that CREpal containing

CAT construct (i.e., MuCRE3) has a reduced response to forskolin treatment, the CREpal–protein complex may not be as effective as the CREaro–protein complex. The competitive gel shift assays, titrating CREaro and CREpal competitor oligos, were performed to compare the respective affinities of these two oligos. As showed in Fig. 5D, under the experimental conditions used, both CREaro-containing oligo and CREpalcontaining oligo have similar binding affinities to the nuclear proteins present in forskolin-treated fibroblasts. To determine if CREB1 protein is present in the complex formed between CREaro probe and the nuclear proteins in forskolin-treated tumor fibroblasts, a polyclonal antibody against human CREB1 was added to the mobility shift assay reaction. No supershift band could be detected (Fig. 5A, lane 5), suggesting that the nuclear protein(s) bound to CREaro in response to the treatment of forskolin is different from CREB1. This finding was confirmed by Western blot analysis (Fig. 6) which did not detect CREB1 in the nuclear extract from the tumor fibroblasts. However, the CREB1 protein is thought to be able to bind to CREaro since a clear complex was formed between the purified CREB1 DNA binding domain peptide and CREaro probe (Fig. 5A, lane 6), and the complex was supershifted by antiCREB1 antibody (Fig. 5A, lane 7). These results suggest that CREB1 protein is able to bind to CREaro, but it is not expressed or is expressed at a very low level in our tumor fibroblast line. The mobility pattern with nuclear extracts from SKBR-3 cells, as shown in Fig. 5B, was different from that observed with nuclear extracts of WS3TF. At least two specific complexes, complexes A and B, were formed

186

ZHOU AND CHEN

FIG. 5. DNA mobility shift assays of CREaro containing oligonucleotide in the presence of nuclear protein extracts from SK-BR-3 cells and tumor fibroblasts. (A) The 276/250-bp oligonucleotide from promoter 1.3 region was end-labeled, incubated with nuclear protein extracts from fibroblasts (lanes 2–5) or CREB DNA binding domain (lanes 6 and 7) or without any protein (lane 1), and analyzed by gel electrophoresis. Gel shift assays were performed in the absence (lane 2) or presence of 500-fold molar excess of various unlabeled competitor oligonucleotides as indicated above each lane. Some reactions contained the CREB antibody (lanes 5 and 7). (B) Gel shift assays were the same as described for A, except that the nuclear protein extracts from SK-BR-3 cells were used. Specific DNA/protein complexes, A and B, are indicated by arrows. The supershifted band C in lane 5 is also indicated by an arrow. (C) The conditions for gel shift assay were the same as described for A. The nuclear extracts from both forskolin-treated and untreated tumor fibroblasts were used as indicated. (D) Competitive gel shift assays by titrating CREaro-containing oligo and CREpal-containing oligo. For dI– dC, 500-fold molar excess was used. The nuclear extracts from forskolin-treated tumor fibroblasts were used in this experiment.

CHARACTERIZATION OF cAMP-RESPONSIVE ELEMENT

187

FIG. 5—Continued

between the proteins in SK-BR-3 cells and the CREaro probe (Fig. 5B, lane 2), and these complexes could be competed by a 500-fold molar excess of unlabeled CREaro (Fig. 5B, lane 3). A 500-fold excess of CREpal containing oligonucleotide was able to compete for complex A, but not complex B (Fig. 5B, lane 4). In contrast to the DNA/protein complex formed by nuclear protein in fibroblasts, complex A formed between CREaro probe and nuclear protein(s) in SK-BR-3 cells was su-

pershifted by the addition of anti-CREB1 antibody (Fig. 5D, lane 5, supershifted band C), suggesting that the nuclear proteins in SK-BR-3 cells contain CREB1 which binds to CREaro. This conclusion is further supported by the fact that CREB1 protein was detected in the nuclear extract of SK-BR-3 cells by Western blot analysis (Fig. 6). In addition, a mutant form of CREpal (muCREpal) was found to be unable to compete for either complex (results not shown). A 500-fold excess of two CREaro mutants, MuCRE1 and MuCRE2, did compete partially for the specific complexes formed between CREaro containing oligonucleotide and nuclear proteins in SK-BR-3 cells. These results agree with the findings shown in Fig. 3 that these mutations partially reduced, but not completely, the forskolin-induced responses in SK-BR-3 cells. UV CrossLinking Analysis of Tumor Fibroblast Nuclear Proteins Bound to CREaro Containing Oligonucleotide

FIG. 6. Western blot analysis of the nuclear proteins in SK-BR-3 cells, MCF-7 cells, and in tumor fibroblasts WS3TF. Nuclear extracts, 50 mg of each sample, were electrophoresed on 10% SDS– PAGE and transferred to nitrocellulose membrane. After being blocked by 5% nonfat milk, the membrane was probed with 1:1000 dilution of polyclonal rabbit antibody against CREB1 (Santa Cruz Biotechnology Co). CREB and anti-CREB1 complexes were detected by a goat anti-rabbit IgG HRP conjugate (1:2500 dilution, Pierce) followed by the SuperSignal Substrate (Pierce).

Since the DNA/protein complex pattern formed between CREaro containing oligonucleotide and nuclear proteins in tumor fibroblasts was found to be less complex and the binding protein(s) is probably not CREB1, we decided to further characterize the CREaro binding protein by UV crosslinking experiments. For the UV crosslinking analysis, radiolabeled CREaro containing oligonucleotide was incubated with nuclear extracts from forskolin-treated tumor fibroblasts and exposed to UV irradiation. A major protein, with molecular

188

ZHOU AND CHEN

FIG. 7. Characterization of CREaro-binding proteins in tumor fibroblasts by UV crosslinking experiments. UV crosslinking experiments were performed as described under Materials and Methods using 32P-labeled, BrdU-substituted CREaro containing oligonucleotide, 276/250 bp as probe. Ten micrograms of fibroblasts nuclear extract was incubated with 1 3 10 5 cpm of probe in the absence (lane 3) or presence of 70-, 300-, or 3000-fold molar excess of unlabeled CREaro-containing oligonucleotide (lanes 4 – 6). Seventyfold molar excess of poly(dI– dC) was also included in one of the reactions as nonspecific competitor (lane 7). Reaction mixtures were either exposed to UV light (protein 1 UV), or not exposed to UV light (No UV).

weight of approximately 60 kDa was found to bind to CREaro containing oligonucleotide (see Fig. 7). This DNA/protein complex was competed by a 70-fold molar excess of unlabeled CREaro containing DNA, but not by a 70-fold excess of dI– dC. Two additional minor complexes, migrating corresponding to proteins with molecular weight of approximately 70 and 43 kDa, were also observed. DISCUSSION

By performing primer-specific RT-PCR analyses, we and two other laboratories (9, 20, 21) have found that exons 1.3 and PII are the two major exon 1s present in aromatase mRNAs isolated from breast tumors, suggesting that promoters 1.3 and II are the major promoters directing aromatase expression in breast cancer and surrounding stromal cells and fibroblasts. In the present study, we have shown that exons 1.3 and PII are the two major exon 1s in aromatase mRNAs isolated from a fibroblast line derived from a breast tumor tissue (data not shown). Since it is known that adipose stromal cells and fibroblasts isolated from noncancerous tissue contain mainly exon 1.4 containing aromatase mRNA (16, 17), a switch of the promoter usage from 1.4 to 1.3 and II occurs in stromal cells and fibroblasts in breast cancer tissue. Harada and his

colleagues reported that the aromatase mRNA level in breast cancer tissue is 2.6 times that in nonmalignant breast tissue (34), and forskolin treatment leads to an increase of the exon 1.3-containing message in cultured breast adipose stromal cells (9). In addition, studies from Simpson’s laboratory (22, 23) have revealed that promoter II is a cAMP-responsive promoter. In the present study, we have demonstrated that promoter 1.3 is a cAMP-responsive promoter by the identification of a cAMP responsive element, CREaro, between 266 and 259 bp relative to the transcription start site of promoter 1.3. Based on these findings, it is proposed that an increase of the usage of promoters 1.3 and II in tumor stromal tissue results from an enhanced cAMP production. There are three considerations associated with the finding of CREaro upstream from promoter I.3. First, while it is generally accepted that aromatase is mainly expressed in adipose stromal cells and fibroblasts in normal breast tissue, both stromal cells (including fibroblasts) and cancer cells in tumor tissue express aromatase. The expression of aromatase in cancer cells has been demonstrated by immunocytochemistry (4, 7), in situ hybridization (7), RT-PCR analysis on microdissected cancerous tissue (20) as well as on breast cancer cell lines (20), and activity measurement in

CHARACTERIZATION OF cAMP-RESPONSIVE ELEMENT

breast cancer cells (12). In the present study, we have shown that cAMP is capable of inducing aromatase expression in breast cancer cells such as SK-BR-3 cells. Second, there may be promoter-specific CREs for promoters 1.3 and II or different cAMP induction mechanisms in different tissues. Investigation from Simpson’s laboratory (22, 23) suggests that an Ad4BP/ SF1 element and a CRE-like sequence (CLS) (see Fig. 3) upstream from promoter II are critical for the cAMP induction of promoter II in ovary tissue. The interaction of these elements and the mechanism of cAMP induction of promoter II are similar to those described for the transcriptional control of aromatase expression in rat granulosa cells (35). We have focused our studies on promoter 1.3 activity in breast cancer tissue (in both cancer and stromal cells). The Ad4BP/SF1 element reported by Michael et al. (22) is positioned within the silencer element identified by us (25). As published previously from our laboratory (25), the silencer downregulates the transcriptional activity of promoter 1.3 in breast. Recent RT-PCR analysis has revealed that Ad4BP/SF1 is not expressed in breast cancer tissue (36). The latter finding may explain why cells transfected with pUms224/1129 do not respond to forskolin treatment. This genomic region contains promoter 1.3, the Ad4BP/SF1 element and the CRE-like sequence (116/124 bp) reported by Michael et al. (23) (see Fig. 3). However, in the presence of forskolin, the CAT activity in cells transfected with pUms 234/1129 CAT was found to be two to three times higher than the basal level. The results suggest the 234/224 region may also be needed for promoter 1.3 to respond to cAMP. While we have shown that cAMP upregulates promoter 1.3 in breast tissue mediated through CREaro, which is discussed in this paper, we do not yet know the cAMP induction mechanism on promoter II in breast cancer tissue. It may be that cAMP activation of promoter II in breast cancer tissue involves the interaction of the CRE-like region (116/124 bp) and a different transcriptional factor that is expressed in breast cancer tissue and binds to the site for Ad4BP/ SF1. Our recent yeast one-hybrid screen studies have identified that several nuclear receptors can bind to this regulatory region (36). In addition, we have not determined whether CREaro can upregulate promoter II, although we have shown that the silencer element can suppress promoter II function (25). Third, the genomic region studied here contains the silencer element, 194/1123 bp (see Fig. 3). Our results indicate that the positive action of CREaro can overcome the negative action of silencer (by comparing the CAT activity between pUms268/1129CAT- and pUms234/1129CAT-transfected cells (see Fig. 2)). Experiments are being performed to study how these regulatory elements affect each other and modulate the function of promoter I.3.

189

Transcription factors have been implicated in the modulation of gene transcription by binding to specific DNA control element located in the promoter regions of target genes and to thereby activate or repress transcription. The ATF/CREB family consists of a series of such transcription factors that function through binding to the cAMP-responsive element (CRE) palindromic octanucleotide, TGACGTCA. The best characterized members of this gene family include CREB-1, CREB-2, ATF-1, ATF-2, ATF-3, and ATF-4 (for review, see Ref. 37). We thus investigated if such a factor was present in forskolin-treated tumor fibroblasts and SKBR-3 cells using mobility shift assay and supershift assay with a CREB polyclonal antibody. The binding pattern by the nuclear proteins from tumor fibroblasts to CREaro containing DNA was different from that observed with nuclear extract from SK-BR-3 cells. Using nuclear extracts from tumor fibroblasts, one major DNA/protein complex was formed, and this complex could not be shifted by the addition of anti-CREB1 antibody, suggesting that the transactivation process in tumor fibroblasts may involve a CRE binding protein which is different from CREB1. This conclusion is further supported by the facts that (1) the molecular weight of CREaro binding protein, 60 kDa (as shown by UV crosslinking experiment), is different from that of most of the known CRE-binding proteins, from 35 to 45 kDa, (2) the cAMP induction of CAT expression was also markedly reduced in the cells transfected with the CAT construct containing MuCREaro 3, i.e., CRE consensus sequence (Fig. 4), and (3) CREB1 is not present in tumor fibroblasts, as demonstrated by Western blot analysis (Fig. 6). In contrast to the binding pattern observed with nuclear extracts from tumor fibroblasts, at least two specific complexes were detected using nuclear extracts from SK-BR-3 cells. Furthermore, the complex A was able to be shifted by the addition of polyclonal antibody against CREB1 while complex B was not shifted by anti-CREB antibody (Fig. 5B, lane 5), suggesting that the transcriptional stimulation of promoter 1.3 in response to cAMP in SK-BR-3 cells involves CREB1 as well as other CRE-binding proteins interacting with CREaro element. CREB1 is present in the nuclear extract of SK-BR-3 cells, as demonstrated by Western blot analysis (Fig. 6). In summary, we present and discuss a mechanism for the cAMP induction of the promoter I.3 activity. Results from our and other laboratories reveal that cAMP plays a critical role in upregulating the expression of aromatase/increasing estrogen biosynthesis in breast cancer tissue. Several factors have been suggested to induce the level of cAMP in breast cancer tissue. For example, Zhao et al. (38) suggested that prostaglandin PGE2 synthesized in breast cancer cells induces cAMP response. Furthermore, estrogen is capable of increasing cAMP production in breast cancer

190

ZHOU AND CHEN

cells by stimulating adenylate cyclase (39). We have found that estradiol at 100 nM is capable of inducing CAT expression in tumor fibroblasts transfected with pUms268/1129 (unpublished results). These observations suggest a paracrine loop between estrogen production (by aromatase) and cAMP synthesis in breast cancer tissue. ACKNOWLEDGMENTS This research was supported by the National Institutes of Health Grant CA44735 and a University of California BCRP grant 1RB0118. These authors are members of the City of Hope Breast Cancer Program (CA 65767). D. Zhou was also supported by China National Nature Science grant 39870836. We thank Ms. Chun Yang for her help in construction of CAT plasmids used in Fig. 4, and Mr. Keith M. Quach for his help in the Western blot analysis on CREB1 expression in two types of cells.

REFERENCES 1. Davidson, N. E., and Lippman, M. E. (1989) Crit. Rev. Oncog. 1, 89 –111. 2. Edman, C. D., Aiman, E. J., Porter, J. C., and MacDonald, P. C. (1978) Am. J. Obstet. Gynocol. 130, 439 – 447. 3. James, V. H. T., McNeill, J. M., Lai, L. C., Newton, C. L., Ghilchik, M. W., and Reed, M. J. (1987) Steroids 50, 269 –279. 4. Esteban, J., Warsi, Z., Haniu, M., Hall, P., Shiverly, J. E., and Chen, S. (1992) Am. J. Pathol. 140, 337–343. 5. Sasano, H., Nagura, H., Harada, N., Goukon, Y., and Kimura, M. (1994) Hum. Pathol. 25, 530 –535. 6. Santen, R. J., Martel, H., Hoagland, M., Naftolin, F., Roa, L., Harada, N., Hafer, L., Zaino, R., and Santner, S. J. (1994) J. Clin. Endocrinol. Metab. 79, 627– 632. 7. Lu, Q., Nakmura, J., Savinov, A., Yue, W., Weisz, J., Dabbs, D. J., Wolz, G., and Brodie, A. (1996) Endocrinology 137, 3061–3068. 8. Bulum, S. E., Price, T. M., Mahendroo, M. S., Aitken, J., and Simpson, E. R. (1993) J. Clin. Endocrinol. Metab. 77, 1622–1628. 9. Harada, N. (1997) J. Steroid Biochem. Mol. Biol. 61, 175–184. 10. Santner, S. J., Chen, S., Zhou, D., Korsunsky, Z., Martel, J., and Santen, R. J. (1993) J. Steriod Biochem. Mol. Biol. 44, 611– 616. 11. Yue, W., Zhou, D., Chen, S., and Brodie, A. (1994) Cancer Res. 54, 5092–5095. 12. Sun, X.-Z., Zhou, D., and Chen, S. (1997) J. Steroid Biochem. Mol. Biol. 63, 29 –36. 13. Tekmal, R. R., Ramachandra, N., Gubba, S., Durgam, V. R., Mantione, J., Toda, K., Shizuta, Y., and Dillehay, D. L. (1996) Cancer Res. 56, 3180 –3185. 14. Mahendroo, M. S., Means, G. D., Mendelson, C. R., and Simpson, E. R. (1991) J. Biol. Chem. 266, 11276 –11278. 15. Means, G. D., Kilgore, M. W., Mahendroo, M. S., Mendelson, C. R., and Simpson, E. R. (1991) Mol. Endocrinol. 5, 2005–2013.

16. Harada, N. (1993) Biochem. Biophys. Res. Commun. 189, 1001– 1007. 17. Mahendroo, M. S., Mendelson, C. R., and Simpson, E. R. (1993) J. Biol. Chem. 268, 19463–19470. 18. Harada, N., Utsumi, T., and Takagi, Y. (1993) Proc. Natl. Acad. Sci. USA 90, 11312–11316. 19. Zhao, Y., Mendelson, C. R., and Simpson, E. R. (1995) Mol. Endocrinol. 9, 340 –349. 20. Zhou, C., Zhou, D., Esteban, J., Murai, J., Sitteri, P. I., Wilczynski, S., and Chen, S. (1996) J. Steroid Biochem. Mol. Biol. 59, 163–171. 21. Agarwal, V. R., Bulun, S. E., Leitch, M., Rohrich, R., and Simpson, E. R. (1996) J. Clin. Endocrinol. Metab. 81, 3843– 3849. 22. Michael, M. D., Kilgore, M. W., Morohashi, K-I., and Simpson, E. R. (1995) J. Biol. Chem. 270, 13561–13566. 23. Michael, M. D., Michael, L. F., and Simpson, E. R. (1997) Mol. Cell. Endocrinol. 134, 147–156. 24. Zhou, D., Clarke, P., Wang, J., and Chen, S. (1996) J. Biol. Chem. 271, 15194 –15202. 25. Zhou, D., and Chen, S. (1998) Arch. Biochem. Biophys. 353, 213–220. 26. Salier, J-P., Hirosawa, S., and Kurachi, K. (1990) J. Biol. Chem. 265, 7062–7068. 27. Yanagawa, Y., Chen, J. C., Hsu, L. C., aand Yoskida, A. (1995) J. Biol. Chem. 270, 17521–17527. 28. Nelson, R. M., and Long, G. L. (1989) Anal. Biochem. 180, 147– 151. 29. Rosenthal, N. (1987) Methods Enzymol. 152, 705–720. 30. Seed, B., and Sheen, J. Y. (1988) Gene 67, 271–277. 31. Kingston, R. E. (1991) In Current Protocols in Molecular Biology (Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seldman, J. G., Smith, J. A., and Struhl, K., Eds.), pp. 12.1.1–12.1.9. Wiley, New York. 32. Singh, H., Sen, R., Baltimore, D., and Sharp, P. A. (1986) Nature 319, 154 –158. 33. Chodosh, L. A., Carthew, R. W., and Sharp, P. A. (1986) Mol. Cell Biol. 6, 4723– 4733. 34. Utsumi, T., Harada, N., Maruta, M., and Takagi, Y. (1996) J. Clin. Endocrinol. Metab. 81, 2344 –2349. 35. Carlone, D. L., and Richards, J. S. (1997) Mol. Endocrinol. 11, 292–304. 36. Yang, C., Zhou, D., and Chen, S. (1998) Cancer Res. 58, 5695– 5700. 37. Meyer, T. E., and Habener, J. F. (1993) Endocrne Reviews, 14, 269 –290. 38. Zhao, Y., Agarwal, V. R., Mendelson, C. R., and Simpson, E. R. (1997) J. Steroid Biochem. Mol. Biol. 61, 203–210. 39. Aronica, S. M., Kraus, W. L., and Katzenellenbogen, B. S. (1994) Proc. Natl. Acad. Sci. USA 91, 8517– 8521.