Mechanisms of Transcriptional Activation of bcl-2Gene Expression by 17β-Estradiol in Breast Cancer Cells*

bcl-2 gene expression is induced by 17β-estradiol (E2) in T47D and MCF-7 human breast cancer cells, and the mechanism of E2 responsiveness was further investigated by analysis of the bcl-2 gene promoter. The −1602 to −1534 distal region (bcl-2j) of the promoter was E2-responsive; however, in gel mobility shift assays, the estrogen receptor α (ERα) did not bind [32P]bcl-2j, whereas Sp1 protein formed a retarded band complex. Further analysis demonstrated that the upstream region (−1603 to −1579) of the bcl-2 gene promoter contained two GC/GA-rich sites at −1601 (5′-GGGCTGG-3′) and −1588 (3′-GGAGGG-5′) that bound Sp1 protein. Subsequent studies confirmed that transactivation by E2 was dependent on ERα/Sp1 interactions with both GC-rich sites, and this was confirmed byin vitro footprinting. In contrast, a 21-base pair E2-responsive downstream region (−1578 to −1534) did not bind Sp1 or ERα protein; however, analysis of a complex binding pattern with nuclear extracts showed that ATF-1 and CREB-1 bound to this motif. These data coupled with results of transient transfection studies demonstrated that transcriptional activation by E2 of the −1578 to −1534 region of the bcl-2 gene promoter was dependent on induction of cAMP and subsequent activation through a cAMP response element. Thus, hormone regulation of bcl-2 gene expression in breast cancer cells involves multiple enhancer elements and E2-mediated transactivation does not require direct binding of the estrogen receptor with promoter DNA.

Human breast cancer cell lines have been extensively used as models for understanding the role of bcl-2 in development and growth of breast tumors and their response to chemotherapeutic drugs. Several lines of evidence suggest that resistance of breast cancer cells to treatment with chemotherapeutic drugs is linked to bcl-2 expression in these cells (23)(24)(25)(26)(27)(28). For example, intracellular expression of single chain antibodies to bcl-2 in MCF-7 human breast cancer cells decreased bcl-2 levels and increased the sensitivity of these cells to drug-induced cytotoxicity (24). It was also shown that stable overexpression of bcl-x s , a dominant negative inhibitor of bcl-2, also increased the sensitivity of MCF-7 cells to growth inhibitory effects of the chemotherapeutic agents VP-16 and taxol (7). Expression of bcl-2 and drug resistance of breast cancer cells is also hormonedependent (23, 29 -31). Estrogens induce bcl-2 gene and/or protein expression in T47D, ZR-75, and MCF-7 cells, and these responses are inhibited by androgens (31) and progestins (30). Both androgens and progestins alone decrease bcl-2.
This study probes the molecular mechanism of E2 1 -induced bcl-2 gene expression in T47D and MCF-7 breast cancer cell lines by analysis of constructs containing bcl-2 gene promoter inserts. A 70-bp distal promoter fragment (Ϫ1603 to Ϫ1534) was identified as estrogen-responsive, and this region of the promoter did not contain perfect or imperfect estrogen-responsive elements (EREs). Further analysis identified fragments of 25 (Ϫ1603 to Ϫ1579) and 21 (Ϫ1554 to Ϫ1534) bp that were estrogen-responsive. The more distal G-rich sequence bound Sp1 protein at two sites, and transcriptional activation by E2 was associated with ER ␣ /Sp1 interactions with cis-genomic Sp1 binding sites. In contrast, the E2-responsive 21-bp sequence contained a cAMP response element (CRE) that bound ATF-1 and CREB-1, and transactivation was associated with induction of cAMP by E2. Thus, transcriptional activation of bcl-2 by E2 in MCF-7 and T47D cells involves at least three different distal cis-genomic elements and does not require direct binding of ER ␣ to the bcl-2 gene promoter.

MATERIALS AND METHODS
Chemicals, Cells, and Antibodies-MCF-7 and T47D cells were obtained from the American Type Culture Collection (ATCC, Rockville, MD). Cells were maintained in minimum essential medium (MEM) with 1 mM sodium pyruvate, 1 g of glucose, and 6 g of insulin per liter (for MCF-7 cells) or ␣-MEM for T47D cells. Media for these cells were supplemented with 5% fetal bovine serum plus 10 ml/liter antibioticantimycotic solution. Cells were grown in 150-cm 2 culture flasks in an * This work was supported by National Institutes of Health Grants CA76636 and ES09106 and grants from the Welch Foundation and the Texas Agricultural Experiment Station. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ Poly(dI)-(dC), restriction enzymes (HindIII, KpnI, and BamHI), and T4-polynucleotide kinase were purchased from Roche Molecular Biochemicals. Recombinant Sp1 and ER ␣ proteins and the ␤-galactosidase enzyme assay system were purchased from Promega (Madison, WI). The human estrogen receptor (hER) expression plasmid was kindly provided by Dr. Ming-jer Tsai (Baylor College of Medicine, Houston, TX). All other chemicals and biochemicals were the highest quality available from commercial sources. Oligonucleotides were synthesized by the Gene Technology Laboratory, Texas A&M University (College Station, TX) (Table I).
Fragments were amplified with Vent DNA polymerase and the plasmid p18 -21 as a template. After gel purification, the PCR products were digested with HindIII and KphI and then subcloned into pucSV9CAT. The oligonucleotides from the human bcl-2 promoter listed above were cloned into the pBLCAT2 at the HindIII and BamHI sites to give pbcl-2j, bcl-2k, bcl-2km1, bcl-2km2, bcl-2km3, bcl-2l, pbcl-2m, mt1pbcl-2m, mt2-pbcl-2m, mt3-pbcl-2m, and mt4-pbcl-2m plasmids, respectively. pPac/Sp1 (provided by Dr. R. Tjian, University of California, Berkeley) was digested with XhoI, and the phosphate group was removed by treatment with calf intestinal alkaline phosphatase. After being treated with 10 nM EDTA and heated at 80°C for 15 min, the reaction was extracted with phenol/chloroform (two times) and the vector was separated from the Sp1 insert by electrophoresis and gel extraction. hER was released from the hER expression plasmid by digestion with EcoRI. The hER fragment was filled by dATP and dTTP, and an XhoI linker was ligated to the blunted hER fragment. After digestion with XhoI, the hER fragment was ligated to the pPac vector, which was treated with XhoI and calf intestinal alkaline phosphatase as described above. All ligation products were transformed into DH5␣competent Escherichia coli cells, plasmids were isolated, and correct clonings and orientation were confirmed by restriction enzyme mapping and DNA sequencing using Sequitherm cycle sequencing kit from Epicentre Technologies (Madison, WI). Plasmid preparations for transfections were carried out using a Qiagen Plasmid Mega Kit.
Transient Transfection Assays-Cells were plated in 100-mm culture dishes, grown to 60% confluence, and transfected by the calcium phosphate method (33) with various amounts of the appropriate plasmids. For each 100-mm dish, 5-10 g of plasmid was mixed with 500 l of 2ϫ HBC (NaCl 1.636%, Hepes 1.188%, Na 2 HPO 4 0.02%, pH 7.05-7.12) and 62 l of CaCl 2 (2 M) to give a final volume of 1 ml with distilled water. The mixture of DNA and CaCl 2 was added into 2ϫ HBS dropwise with gentle vortexing and allowed to settle at 20°C for 30 min. DNA was added to the medium dropwise, and formation of fine particles was determined microscopically. After adding DNA, cells were grown for 6 h and then shocked with 20% glycerol for 90 s. Medium was then changed, and cells were treated with appropriate chemicals for 24 -48 h. Cells were then washed with phosphate-buffered saline and scraped from the plates. Cell lysates were prepared in 0.3 ml of 0.25 M Tris-HCl (pH 7.5) by three freeze-thaw-sonication cycles (3 min/each). After centrifugation at 4°C for 15 min, the supernatant was incubated at 56°C for 7 min to remove endogenous deacetylase activity and protein (100 -150 g) was incubated in a reaction mixture containing 0.25 M Tris, pH 7.5, 1 mM acetyl-CoA, 0.2 mCi of [ 14 C]chloramphenicol at 37°C for various times. After incubation, acetylated metabolites were isolated by extraction with ethyl acetate and separated by thin layer chromatography in chloroform:methanol (95:5). The plate was air-dried, and radioactive metabolites were quantified using a Betascope 603 Blot analyzer (In-telliGenetics, Mountain View, CA). Cells were cotransfected with 2.0 g of a ␤-galactosidase-LacZ plasmid (InVitrogen, Carlsbad, CA), and the results were used to normalize CAT activities in various treatment groups.

5Ј-AGC
a The GC-rich sites and CRE sites are shown in bold and base mutations are underlined.
E2 or solvent carrier (ethanol) for about 48 h and harvested by scraping. Data are combined from two separate experiments using the same experimental protocols, and transfection efficiency was normalized with ␤-galactosidase activity as described above.
Reverse Transcription-PCR for Determination of mRNA Level-Total RNA was isolated by the guanidinium thiocyanate/acid phenol extraction method (34); 200 ng of RNA was reverse transcribed using murine leukemia virus reverse transcriptase obtained from Perkin Elmer. The first strand cDNA was directly amplified by 28 PCR cycles using a Taq polymerase; 10 mCi of [␥-32 P]ATP was used to label the primer (forward) for each quantitation. The PCR cycle procedure was 95°C for 45 s, 68°C for 45 s, and 72°C for 90 s for 28 cycles, followed by 72°C for 10 min. After PCR, products were separated by 6% polyacrylamide gel and visualized by autoradiography. ␤-Actin mRNA also was amplified to normalize bcl-2 mRNA levels.
Primer size of bcl-2 mRNA was 385 bp; sequence was as follows.

SEQUENCE 7
Nuclear Extract Preparation and Gel Mobility Shift Assay-Nuclear extracts were prepared from MCF-7 cells treated with Me 2 SO (0.1% v/v) or 10 nM E2 for 4 h utilizing cells maintained in serum-free medium for 3 days. Oligonucleotides were annealed and labeled at the 5Ј end using T4-polynucleotide kinase and [␥-32 P]ATP. Gel electrophoretic mobility shift assays were performed by incubating 5-40 ng of pure Sp1 protein in 25 l of 1ϫ binding buffer (6% glycerol, 1 mM MgCl 2 , 0.5 mM EDTA,0.5 mM dithiothreitol, 50 mM NaCl, 10 mM Tris-HCl, pH 8.0), 0.1 mg/ml BSA. After incubation for 10 min at 4°C, 32 P-labeled oligonucleotide (50,000 cpm) was added to the reaction mixture in the presence of 1 g of poly[d(I-C)] and incubated for an additional 15 min at 25°C. Excess unlabeled DNA was added 5 min before adding 32 P-labeled oligonucleotides. The following procedure was used for ER ␣ enhanced Sp1 binding studies. (a) 200 -800 fmol of pure hER protein in 1ϫ binding buffer containing 40 mM E2 and BSA were added and incubated for 15 min at 4°C; (b) 5-20 ng of Sp1 protein was added to the mixture and incubated on ice for 5 min; (c) 32 P-labeled oligonucleotides (50,000 cpm) were added to the reaction mixture in the presence of 1 g of poly(dI)-(dC), and the mixture was incubated for an additional 15 min at 20°C. Nuclear extracts from control (Me 2 SO) or E2-treated cells were incubated for 15 min at 0°C in HEGD (2 mM Hepes, 1.5 mM EDTA, 1.0 mM dithiothreitol, 10% glycerol (v/v), pH 7.6) buffer with 1 mg of poly(dI)-(dC) to bind nonspecific DNA-binding proteins, and 200 -500fold excess of unlabeled wild-type or mutant oligonucleotide competitors for the competition experiments. Following addition of 32 P-labeled DNA, the mixture (final volume: 20 ml) was incubated for an additional 20 min at 20°C. For experiments using saturation concentrations of Sp1 protein (10 -30 ng) alone or in combination with 200 fmol of ER ␣ protein, the ER ␣ and Sp1 proteins were added simultaneously (early addition) or ER ␣ was added 5 min after the radiolabeled probe (late addition). For gel supershift experiments, antibodies were added after standard gel mobility shift assay procedure and reactions were further incubated for 20 -30 min at 20°C. Samples were loaded onto a 5% polyacrylamide gel (acrylamide-bisacrylamide ratio, 30 mRNA was isolated and quantitated as described under "Materials and Methods." E2 significantly induced blc-2 mRNA levels in both cell lines (p Ͻ 0.05), and results are expressed as means Ϯ S.E. for at least three separate determinations. B, CAT activity in T47D cells. The constructs were transiently transfected into T47D cells, treated with Me 2 SO or E2, and CAT activity was determined as described under "Materials and Methods." Significant induction (p Ͻ 0.05) by E2 was observed for all constructs, and results are expressed as means Ϯ S.E. for at least three separate determinations for each treatment group. C, CAT activity in T47D cells. The constructs were transiently transfected into T47D cells, treated with Me 2 SO or E2, and CAT activity was determined as described under "Materials and Methods." Significant induction (p Ͻ 0.05) by E2 was observed only for pbcl-2d and pbcl-2e, and results are expressed as means Ϯ S.D. for at least three separate determinations for each treatment group.

FIG. 2. Deletion analysis of the bcl-2 gene promoter in MCF-7 (A and B) and T47D (C) cells.
The constructs were transiently transfected into cells, treated with Me 2 SO or E2, and CAT activity was determined as described under "Materials and Methods." Significant induction (p Ͻ 0.05) by E2 is indicated (*), and results are expressed as means Ϯ S.E. for at least three separate determinations for each treatment group.
Kodak X-Omat film for appropriate time at Ϫ80°C.
In Vitro SssI Footprinting-SssI is a CpG methylase that converts the 5Ј cytosine to a 5-methylcytosine. Because the DNA is not sheared or digested, this method is highly sensitive for footprinting weak interactions that require a longer or more complex binding sequence (35,36). Since 5-methylcytosines are resistant to deamination, they are readily detected after PCR amplification with deamination specific primers. Higher resolution is attained by sequencing with the deamination primer that contains the G to A transition and using ddGTP to detect residual guanine residues paired to the 5-methylcytosines. Fifty g of pbcl-2b containing a 1.6-kb promoter fragment from the human bcl-2 gene was restricted with XhoI and diluted to a concentration of 10 ng/l. One l of diluted plasmid was then incubated with increasing concentrations of recombinant human Sp1 protein alone, human ER protein alone, or Sp1 protein plus increasing concentrations of ER ␣ protein.
Reactions were carried out in 1ϫ NS binding buffer (0.02 M HEPES, 0.1 M KCl, 0.005 M MgCl 2 , 0.004 mM EDTA, 5% glycerol, 50 mM S-adenosylmethionine) in a volume of 25 l. The protein-DNA binding reactions were incubated on ice for 5 min, and then equilibrated to room temperature for 10 min; 2 l of 1:2 dilution of purified SssI (New England Biolabs) was then added to the equilibrated reactions, which were then incubated at 30°C for 5 min. After 15 min at 75°C, 10 l of freshly made deamination denaturation buffer (0.9 N NaOH, 25 mM EDTA, 0.2 mg/ml sheared salmon sperm DNA) was added. Following 5 min at 98°C, 200 l of a saturated solution of sodium metabisulfite was added and the samples were processed as described (35,36). The primers used to amplify from the purified deaminated plasmid DNA were bcl-2 B1 (5Ј-TCCACAAACCTAAACAAAAAACC-3Ј) and bcl-2 B2 (5Ј-GGTGTTT-GTTTTTTTATTTTATTTTTTG-3Ј). PCR products were purified and cleaned up using the Wizard PCR prep kit from Promega Corp. Purified PCR products were sequenced with radiolabeled bcl-2B1 primer in the presence of a 5 M solution of dATP, dCTP, and dTTP using 50 M ddGTP as the stop nucleotide. Sequitherm 10ϫ buffer and Sequitherm thermostable DNA polymerase (Epicentre Technologies, Madison, WI) were used for the sequencing reactions. Sequencing reactions were run on 6% polyacrylamide-urea sequencing gels. The dried gels were exposed to a phosphor screen for 12 h and analyzed on a Molecular Dynamics Storm instrument.
Statistics-Results are expressed as means Ϯ S.E. for at least three independent (replicate) experiments for each treatment group. Statistical significance was determined by analysis of variance and Student's t test, and the levels of probability are noted for each experiment. Fig. 1A summarizes the effects of E2 on bcl-2 mRNA levels in MCF-7 and T47D cells, and these data are comparable to those previously reported (23, 29 -31). Deletion analysis of the 7.0-kb fragment from the bcl-2 gene promoter (Fig. 1B) focused on the Ϫ3.0 kb region utilizing transient transfection studies in T47D cells. These plasmids contained bcl-2 gene promoter inserts in pUCSVO-CAT containing an SV40 early-region promoter 5Ј of the HindIII site (32), and the construct was readily transfected into T47D but not MCF-7 cells due to poor conversion in these cells. Results obtained for pbcl-2a-pbcl-2d showed that treatment with E2 resulted in a 3.4 -6.9-fold increase in CAT activity and E2 responsiveness was retained in a promoter region from Ϫ1647 to Ϫ1293. Further analysis of a series of constructs containing bcl-2 gene promoter inserts from within the Ϫ1647/Ϫ1289 region (pbcl-2d-pbcl2i; Fig. 1C) indicated that hormone responsiveness was localized within a minimal 70-bp sequence from Ϫ1603 to Ϫ1534.

Transcriptional Activation of bcl-2 mRNA and Promoter-Reporter Constructs by E2-
Synthetic oligonucleotides within Ϫ1603/Ϫ1534 region of the bcl-2 gene promoter were cloned into pBLCAT3, transiently transfected into MCF-7 cells, and treated with 10 nM E2 or Me 2 SO. The results ( Fig. 2A) indicate that E2 responsiveness of the 71-bp construct (pcbl-2j) is associated with a 25-bp up- FIG. 3. Sp1 binding to bcl-2j and bcl-2k. A, binding to Sp1 protein. Recombinant Sp1 protein was incubated with 32 P-labeled oligonucleotide as indicated and protein-DNA binding was analyzed by gel mobility shift assays as indicated under "Materials and Methods." A retarded band was observed using only radiolabeled bcl-2j and bcl-2k but not bcl-2l; competition with 100-fold excess of a consensus Sp1 oligonucleotide eliminated retarded band formation (data not shown). B, Sp1/Sp3 antibody supershifts. Nuclear extract from MCF-7 cells treated with E2 was incubated with 32 P-labeled oligonucleotides as indicated in the presence of Sp1 (lanes 3 and 4) or Sp3 (lanes 5 and 6) antibodies, and protein-DNA binding was analyzed by gel mobility shift assays as indicated under "Materials and Methods." Sp1 antibody supershifted the major Sp1-DNA complex (lanes 3 and 4), and a weak supershifted Sp3-DNA complex could also be detected. stream (pbcl-2k) and a 21-bp downstream (pbcl-2m) region of the bcl-2 gene promoter. The Ϫ1603 to Ϫ1579 region of the bcl-2 gene promoter contains two G-rich sequences at Ϫ1601 (5Ј-GGGCTGG-3Ј) and Ϫ1588 (GGAGGG) and the role of these sites in E2 responsiveness was further investigated using pbcl-2k and constructs containing mutations at one or both G-rich sites (pbcl-2km1, pbcl-2km2 and pbcl-2km3). The resulting plasmids were transiently transfected into MCF-7 and T47D cells (Fig. 2, A and B), and the results showed that E2 significantly induced CAT activity with wild-type or single-site mutant constructs (pbcl-2k, pbcl-2km1 and pbcl-2km2), whereas no significant induction was observed with the double mutant (pbcl-2km3).
Binding of human recombinant Sp1 protein to [ 32 P]bcl-2k oligonucleotide was concentration-dependent over a range of Sp1 protein concentrations from 5-20 ng/incubation (lanes 2-4) and intensity of the retarded band complex was decreased after coincubation with unlabeled wild-type (lane 5) but not mutant (lane 6) Sp1 oligonucleotides. Binding of [ 32 P]bcl-2km1 (lanes 7-9) and [ 32 P]bcl-2km2 (lanes 10 -12) to recombinant Sp1 protein was also observed; it was apparent that the former oligonucleotide exhibited lower binding affinity for comparable amounts of Sp1 protein as higher amounts (20 -40 ng) were required to observe retarded complexes (bound DNA). The effects of ER ␣ on Sp1-DNA complex formation were also investigated (Fig. 4B) using 32 P-labeled wild-type (lanes 1-4) and mutant bcl-2k oligonucleotides (lanes 5-10). Although absolute band intensities were variable, the intensity of the Sp1-DNA complex (bound DNA) after incubation of [ 32 P]bcl-2k, [ 32 P]bcl-2km1, and [ 32 P]bcl-2km2 with Sp1 protein and increasing amounts of ER ␣ protein (0.2-0.8 pmol) resulted in a relative 2-4-fold increase in the intensity of the Sp1-DNA retarded band. All of these enhanced binding studies utilized subsaturating amounts of Sp1 protein. Therefore, the gel mobility shift FIG. 4. Sp1 binding to G-rich sites and enhancement by ER ␣ protein. A, binding of Sp1 protein to wild-type and mutant bcl-2k oligonucleotides. Recombinant Sp1 protein was incubated with 32 Plabeled oligonucleotide as indicated, and protein-DNA binding was analyzed by gel mobility shift assays as indicated under "Materials and Methods." A retarded band was observed using wild-type and mutant [ 32 P]bcl-2k oligonucleotides; however, higher amounts of Sp1 protein were required using [ 32 P]bcl-2km1 due to weaker binding. B, enhanced Sp1-DNA binding by ER ␣ protein. Recombinant Sp1 protein was incubated with 32 P-labeled oligonucleotides and different amounts of ER ␣ as indicated, and protein-DNA binding was analyzed by gel mobility shift assays as indicated under "Materials and Methods." A retarded band was observed using wild-type and variant bcl-2k oligonucleotides and 0.2-0.8 pmol of ER ␣ caused a dose-dependent increase (2-6-fold) in Sp1-DNA complex formation. Competition with 100-fold excess of a consensus Sp1 oligonucleotide eliminated retarded band formation (data not shown). assays were then repeated with a consensus Sp1 oligonucleotide and saturating concentrations of Sp1 protein followed by early or late addition of ER ␣ protein (Fig. 5A). Supershifted bands were not observed using these conditions; moreover, similar results were obtained using nuclear extracts from Schneider Drosophila SL2 cells transfected with ER ␣ or Sp1 expression plasmid (Fig. 5B). The failure to observe supershifted ER ␣ /Sp1-DNA ternary complexes in these studies was similar to previous reports using GC-rich oligonucleotides from other E2-responsive gene promoters (41)(42)(43)(44)(45).
SssI is a CpG viral methylase that we have utilized to further probe ER ␣ /Sp1 interactions. The bcl-2k region of the bcl-2 gene promoter contains a CpG site and in SssI in vitro footprinting studies (Fig. 5C), pure Sp1 protein alone (lanes 2-4) only slightly protected the CpG site (q) in bcl-2k. Increasing concentrations of pure human recombinant ER ␣ only slightly increased protection at the CpG site ( lanes 5 and 6); however, incubation with ER ␣ (100 and 200 fmol) plus the lowest amount of Sp1 protein (20 ng) completely protected this site and confirmed interactions between ER ␣ and Sp1 proteins in this region of the bcl-2 gene promoter. However, the extent of protection is likely to span all of this region, and sites proximal to it since the Sp1 enhancement is also seen 36 bases downstream of bcl-2k (E). A ghost CpG band, which does not correspond to any known CpG in this sequence and is a likely result of a PCR artifact, appears at random in lanes 1, 5, and 7 (*).
Interactions of ER ␣ and Sp1 protein in the bcl-2k region of the promoter were further investigated in Schneider SL-2 cells that do not express these transcription factors. In cells transfected with pcbl-2k and different amount of Sp1 expression plasmid alone (10 -2000 ng), there was a 30-fold increase in CAT activity. In cells cotransfected with Sp1 (100 ng) and ER ␣ (2-50 ng) expression plasmids, there was a maximal 3.6-fold increase in CAT activity (ER ␣ ϭ 40 ng) compared with cells transfected with Sp1 plasmid alone. ER ␣ expression plasmid alone (50 ng) did not affect basal CAT activity (data not shown). Thus, the functional interactions of ER ␣ and Sp1 observed at G-rich sites in the bcl-2 gene promoter in human breast cancer cells (Fig. 2) are also observed in Schenider SL2 cells. Ϫ1534): Interactions with ATF-1 and CREB-1-In addition to binding of transcription factors Sp1 and ER ␣ to upstream Grich sequences, interactions of [ 32 P]bcl-2m (Ϫ1554 to Ϫ1534) with other nuclear proteins was also investigated. Preliminary studies showed that ER ␣ , Sp1 or their combination did not form a complex with this oligonucleotide (data not shown). The results in Fig. 6A summarize binding of nuclear extracts from E2-treated MCF-7 cells with [ 32 P]bcl-2m (the 21-bp downstream sequence). At least five retarded bands were detected (B1-B5) (lane 1). Competition with 100-fold excess unlabeled bcl-2m resulted in decreased intensities of all retarded bands (lane 2). Competition with unlabeled oligonucleotides mutated within the bcl-2m (Ϫ1550/Ϫ1545) sequence only slightly reduced the intensity of retarded band complexes (lanes 3 and 4).

Binding of Nuclear Extracts to [ 32 P]bcl-2m (Ϫ1554 to
In contrast, mt3-bcl-2m, containing a mutation of the core DRE sequence, competitively decreased all bands (B1-B5; similar to wild-type bcl-2m) (lane 5). Unlabeled CRE oligonucleotide decreased intensity or eliminated B1, B3, B4, and B5 (but not B2) (lane 7), whereas mutant CRE, Sp1, and ERE oligonucleotides were relatively inactive as competitors (lanes 8 -10, respectively). Direct binding studies using 32 P-labeled bcl-2m, mt1-bcl-2m, mt2-2bcl-2m, and mt3-bcl-2m and MCF-7 nuclear extracts (lanes [11][12][13][14] corroborated the results obtained by competitive binding studies. The five bands formed with wild-type bcl-2m were decreased in intensity after incubation with mt1-bcl-2m and mt2-bcl-2m, whereas all five bands were observed with [ 32 P]mt3-bcl-2m (lane 14). In addition, a new band (B1Ј) was Cyclic AMP Responsiveness of pbcl-2l (Ϫ1578 to Ϫ1534)-The finding that CREB-1 and ATF-1 bound to [ 32 P]bcl-2m oligonucleotide prompted us to investigate the cAMP responsiveness of the 21-bp downstream sequence using transient transfection experiments and the pbcl-2l reporter plasmid that contains the Ϫ1578 to Ϫ1534 region of the promoter (Fig. 7). The results showed that 8-bromo cAMP significantly induced CAT activity, and similar results were observed after cotransfection with an expression plasmid for protein kinase A (pPKA). Thus, both cAMP and E2 (Fig. 2) induced pbcl-2l in MCF-7 cells. In contrast, cotransfection with a dominant negative form of CREB (KCREB expression plasmid) significantly decreased CAT activity. In addition, the cAMP inhibitor, H8, blocked transcriptional activation by E2 in cells transfected with pbcl-2l. These results demonstrate that E2 responsiveness of the Ϫ1578 to Ϫ1534 region of the bcl-2 gene promoter is due to up-regulation of the cAMP/protein kinase A pathway. DISCUSSION bcl-2 gene or protein expression in breast cancer patients correlated positively with ER ␣ -positive tumors and is a prognostic factor for increased disease-free survival for breast cancer patients (17)(18)(19)(20)(21). In contrast, bcl-2 plays a role in drug resistance of ER ␣ -positive breast cancer cells due, in part, to inhibition of drug-induced apoptosis (23)(24)(25)(26)(27)(28). Other studies have shown that the tumor suppressor gene p53 can inhibit expression of bcl-2 (32,46), and this also correlates with an inverse relationship observed between bcl-2 and p53 proteins in breast tumors determined by immunohistochemical analysis (19). Thus, the role of bcl-2 in breast cancer may be complex and dependent on tumor stage, chemotherapeutic regimens and coexpression of other factors critical for cancer cell growth, differentiation and apoptosis.
E2 induces bcl-2 gene expression in ER ␣ -positive breast cancer cells (23, 29 -31) (Fig. 1). In this study, we have investigated the molecular determinants associated with this hormone-induced response of the bcl-2 gene. Analysis of the bcl-2 gene promoter in MCF-7 and T47D breast cancer cells (Fig. 1) resulted in identification of an E2-responsive distal region at Ϫ1647 to Ϫ1289 and results of further 5Ј and 3Ј deletions identified a 70-bp sequence (Ϫ1603/Ϫ1532) that retained hormone-inducibility in transient transfection assays in both T47D and MCF-7 cells (Figs. 1 and 2). The 70-bp oligonucleotide (bcl-2j) did not contain a perfect (or imperfect) palindromic ERE and gel mobility shift assays showed that [ 32 P]bcl-2j did not bind ER ␣ to form a retarded band (data not shown). Subsequent analysis of the 70-bp region indicates that E2 responsiveness was complex and requires both upstream (Ϫ1603 to Ϫ1579) and downstream (Ϫ1554 to Ϫ1534) sequences ( Fig. 2A), suggesting possible regulation by multiple transcription factors and cis-elements.
Typically, ER ␣ enhances Sp1-DNA binding in gel mobility shift assays but does not form a supershifted ternary complex (41)(42)(43)(44)(45), and the results in Figs. 3 and 4 show enhanced Sp1 binding to [ 32 P]bcl-2k after coincubation with ER ␣ as previously reported for other GC-rich sequences. In previous studies, enhanced Sp1-DNA binding by ER ␣ was observed using nonsaturating concentrations of Sp1 protein, and we hypothesized that failure to observe a supershifted complex may be due to low (nonsaturating) Sp1 protein concentrations. However, coincubation of ER ␣ with saturating amounts of Sp1 protein or coincubation of Sp1 and ER ␣ expressed in Schneider SL-2 did not result in formation of a supershifted complex (Fig. 5).
Protein-enhanced binding of other transcription factors to their cognate cis-genomic elements has been reported for several nuclear proteins and usually involves enhanced rate of retarded band formation or a decreased rate of protein-DNA dissociation (43, 48 -54). Methylation of CpG sites by the viral methylase has been utilized as a highly sensitive technique for footprinting weak protein-DNA interactions and, therefore, we used this method to examine ER ␣ -Sp1 interactions in the E2responsive G-rich sequence of the bcl-2 gene promoter ( Fig.   FIG. 7. Transcriptional activation of pbcl-2l (؊1578/؊1534) by E2 in MCF-7 cells. MCF-7 cells were transiently transfected with pbcl-2l (8 g), ER ␣ expression plasmid (4 g), and 4.0 g of ␤-galactosidase-LacZ plasmid, treated with various chemicals/constructs, and CAT activity (corrected for transfection efficiency) was determined as described under "Materials and Methods." CAT activity in the control (DMSO) group was significantly increased by E2, cAMP and pPKA (expression plasmid) and decreased with increasing expression of a dominant negative form of CREB (pKCREB). E2-induced CAT activity was significantly inhibited by the cAMP inhibitor H8. All results are expressed as means Ϯ S.E. for at least three separate determinations. 5C). The results show that ER ␣ or Sp1 protein alone exhibit minimal binding at the CpG site (q); however, after coincubation with both proteins, this site was completely protected and methylation was not observed at this site. These results provide important new confirmation of ER ␣ -Sp1 interactions at G/GC-rich sites and complement results of previous studies on this transcription factor complex (41)(42)(43)(44)(45).
The downstream E2-responsive region at Ϫ1578 to Ϫ1534 was further localized to a 21-bp sequence at Ϫ1554 to Ϫ1534 and results of gel mobility shift assays indicated that neither Sp1 (Fig. 3) or ER ␣ (data not shown) proteins bound to this region of the bcl-2 gene promoter. The 21-bp sequence contains a CRE motif and estrogen action via induction of cAMP has previously been reported and confirmed by transcriptional activation of constructs containing a CRE motif linked to a CAT reporter gene (55)(56)(57). Nuclear extracts from MCF-7 cells treated with E2 bind to [ 32 P]bcl-2m (Ϫ1554 to Ϫ1534) to give at least 5 retarded bands (Fig. 6A); however, in competition and antibody supershift assays, it was shown that ATF-1 and CREB-1 were bound to this oligonucleotide (Fig. 6). These data, coupled with results of transient transfection assays (Fig. 7) demonstrate that E2-induced transactivation of bcl-2 is dependent on a CRE motif (Ϫ1554 to Ϫ1534) in the bcl-2 gene promoter. Thus, transcriptional activation of bcl-2 by E2 does not involve direct ER ␣ -DNA binding but is associated with ER ␣ /Sp1 interactions at 2 GC-rich sites and hormone-induced activation of cAMP through a CRE motif. Multiple E2-responsive enhancer sequences have now been identified in promoter regions of several E2-regulated genes (42-44, 58 -63), and this promoter complexity may be important for cell-specific differences in hormone-activated gene expression (63). Current studies in this laboratory are now utilizing bcl-2 gene promoter constructs to investigate the role of different motifs in mitogenand E2-mediated activation of bcl-2 in various cell lines.