Novel Estrogen Receptor-A Binding Sites and Estradiol Target Genes Identified by Chromatin Immunoprecipitation Cloning in Breast Cancer

Estrogen receptor-A (ERA) and its ligand estradiol play critical roles in breast cancer growth and are important therapeutic targets for this disease. Using chromatin immunoprecipitation (ChIP)-on-chip, ligand-bound ERA was recently found to function as a master transcriptional regulator via binding to many cis-acting sites genome-wide. Here, we used an alternative technology (ChIP cloning) and identified 94 ERA target loci in breast cancer cells. The ERA-binding sites contained both classic estrogen response elements and nonclassic binding sequences, showed specific transcriptional activity in reporter gene assay, and interacted with the key transcriptional regulators, including RNA polymerase II and nuclear receptor coactivator-3. The great majority of the binding sites were located in either introns or far distant to coding regions of genes. Forty-three percent of the genes that lie within 50 kb to an ERA-binding site were regulated by estradiol. Most of these genes are novel estradiol targets encoding receptors, signaling messengers, and ion binders/ transporters. mRNA profiling in estradiol-treated breast cancer cell lines and tissues revealed that these genes are highly ERA responsive both in vitro and in vivo. Among estradiol-induced genes, Wnt11 was found to increase cell survival by significantly reducing apoptosis in breast cancer cells. Taken together, we showed novel genomic binding sites of ERA that regulate a novel set of genes in response to estradiol in breast cancer. Our findings suggest that at least a subset of these genes, including Wnt11 , may play important in vivo and in vitro biological roles in breast cancer. [Cancer Res 2007;67(10):5017–24]


Introduction
Estrogen receptor-a (ERa; ref. 1) is a ligand-activated nuclear receptor that regulates transcription of estrogen-responsive genes important for cell growth, differentiation, and malignant transformation in various target cells (2).ERa and its ligand estradiol [17h-estradiol (E 2 )] play critical roles in the growth of breast cancer tissue and are important therapeutic targets (3).Significant progress has been made in understanding the role of ERa as a transcription factor that regulates the expression of target genes by directly binding to an estrogen response element (ERE; ref. 4) or by association with other transcription factors on promoter targets (1,4,5).Until recently, however, little had been known about the distribution of ERa-binding sites within the genome and the identity of genes regulated by these cis-acting elements.
Two recent pioneering publications by Carroll et al. (6,7) revolutionized our understanding of ERa action.Using chromatin immunoprecipitation (ChIP)-on-chip, this group mapped a large number of ERa-binding sites on a chromosome and genome-wide scale, identifying novel cis-regulatory sites and target genes in MCF-7 breast cancer cells (6,7).The majority of these binding sites were distant from the transcription start sites of regulated genes (6,7).
Identification of novel genomic targets and a deeper understanding of their transcriptional regulation by ERa and their physiologic function may lead to the development of more specific and effective treatments for breast cancer.In the current study, we used a technique alternative to ChIP-on-chip (i.e., ChIP-linked target site cloning) for unbiased and potentially genome-wide identification of regulatory targets of estradiol/ERa in breast cancer cells.Our aims were to determine the nature of ERa binding relative to the structure of a gene and increase our understanding of ERa action both in normal tissue and in the malignant state.We characterized these binding sites and regulation of proximal genes by estradiol/ERa both in vitro and in vivo.The majority of the 38 estrogen-regulated genes turned out to be previously unknown ERa targets.We showed in detail previously unknown biological roles of one of these estrogen/ ERa-regulated genes, Wnt11, in breast cancer.

Materials and Methods
Cell lines and tissues.MCF-7, T47D, and MDA-MB-231 cells (American Type Culture Collection) were maintained in MEM (Invitrogen) containing 25 units/mL penicillin, 25 units/mL streptomycin, and 10% fetal bovine serum (FBS) at 37jC and 5% CO 2 .Snap-frozen ER + (n = 25) and ER À (n = 20) breast cancer tissues were obtained from the Northwestern Breast Specialized Programs in Research Excellence Tissue Core Facility.These tissues were collected after obtaining written informed consent approved by the Institutional Review Board of Northwestern University.
Chromatin immunoprecipitation.After MCF-7 cells were grown to 75% to 80% confluence in MEM supplemented with 10% FBS, the cells were serum starved in DMEM/F-12 without phenol red (Invitrogen) and FBS for 24 h.After 3 h of treatment with 10 À9 mol/L E 2 , cells were washed twice with cold PBS and cross-linked with 1% formaldehyde at room temperature for 10 min.The cross-linking reaction was stopped by adding glycine containing a cocktail of protease inhibitors (Sigma) to a final concentration of 125 nmol/L for 5 min at room temperature.Cells were rinsed twice with cold PBS, harvested, and stored at À80jC before use.Cell pellets were lysed and sonicated to shear the DNA into 0.6-to 3.0-kb fragments.Insoluble material was removed by centrifugation, and the extract was precleared by incubation with blocked protein A-agarose/Salmon Sperm DNA (Upstate) for at least 1 h at 4jC to reduce nonspecific interactions.
After centrifugation, the supernatant (50 AL) was collected as input, and the remainder was diluted in buffer [1% Triton X-100, 2 mmol/L EDTA, 50 mmol/L Tris-HCl (pH 8.1)] and subjected to immunoprecipitation overnight at 4jC with a monoclonal antibody against ERa (Upstate).
After immunoprecipitation, 60 AL protein A-agarose/Salmon Sperm DNA beads were added and the incubation was continued for another 1 h.To decrease nonspecific binding, DNAs/protein complexes were washed under high-stringency wash conditions.Precipitates were washed sequentially for 10 min, two times in buffer I [0.1% SDS, 1% Triton X-100, 2 mmol/L EDTA, 20 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl], six times in buffer II [0.1% SDS, 1% Triton X-100, 2 mmol/L EDTA, 20 mmol/L Tris-HCl (pH 8.0), 500 mmol/L NaCl], two times in buffer III [0.25 mol/L LiCl, 1% NP40, 1% deoxycholate, 1 mmol/L EDTA, 10 mmol/L Tris-HCl (pH 8.0)], and twice with 1 mmol/L EDTA and 10 mmol/L Tris-HCl (pH 8.0).Precipitated chromatin complexes were removed from the beads through a 15-min incubation with 50 AL of 1% SDS and 0.1 mol/L NaHCO 3 and vortexing at room temperature.This step was done twice.Eluates were pooled and heated at 65jC overnight to reverse the formaldehyde cross-linking.DNA fragments were purified with a MinElute Reaction Cleanup kit (Qiagen).The immunoprecipitated and input DNA samples were assayed for binding to the MYC promoter region as a positive control before ChIP cloning.For PCR, 1 AL of purified DNA extraction and 40 cycles of amplification were used.
Cloning, sequencing, and analysis of ERA-binding fragments.ChIPderived DNA was subjected to AvaII restriction digestion and PCR linker ligation followed by PCR amplification (40 cycles) generating sufficient material for cloning into the pGEM-T Easy Vector System (Promega).Clones were sequenced and every retrieved ChIP fragment was mapped to the human genome using the BLAT search function of University of California Santa Cruz genome browser or ENSEMBL search.The presence of EREs was analyzed by ERE Finder (8), and activator protein-1 (AP-1) sites were identified using the Transcription Element Search System. 3icroarray-based mRNA profiling.MCF-7 cells at 70% to 80% confluence maintained in MEM as described above were treated with vehicle or E 2 at a concentration of 10 À9 mol/L for 3 and 6 h.Total RNA was extracted as described below.Gene expression profiles of the MCF-7 cells were analyzed on Human Genome U133 Plus 2.0 microarray chips (Affymetrix), which contained 54,613 probe sets.Sample labeling and subsequent hybridization to the array were carried out according to the manufacturer's instructions in the Microarray Core Facility within the Center for Genetic Medicine, Northwestern University.Expression data normalization was done as described previously (9).
Validation of the in vivo ERa-binding sites by real-time PCR.To examine the enrichment of specific ERa-binding fragments in an independent ChIP assay, primers were generated corresponding to the regions examined within each ChIP-derived genomic fragment.Primers were synthesized by Integrated DNA Technologies.For each PCR assay, DNA after ChIP of ERa and IgG as well as input DNA was quantified by PicoGreen dsDNA dye (Invitrogen) from E 2 -treated or vehicle (ethanol) MCF-7 cells.Real-time PCR was done using Applied Biosystems SYBR Green Master kit following manufacturer's instructions.The estrogen-mediated fold enrichment of ERa-binding regions relative to IgG was compared with its vehicle (ethanol) control.For every fragment analyzed, enrichment was measured after three independent immunoprecipitations.To further validate the in vivo binding of ERa to its putative binding regions, ChIP of nuclear receptor coactivator-3 (NCOA3; also known as SRC-1), RNA polymerase II (PolII; Santa Cruz Biotechnologies), and IgG were done.
RNA preparation and validation of ChIP-derived target gene expression by real-time PCR.Total RNA was extracted from the MCF-7, T47D, and MDA-MB-231 breast cancer cell lines and breast cancer tissue samples (n = 45; wet weight % 100 mg) using Tri-Reagent (Sigma) according to the manufacturer's instruction.Total RNA samples were treated with DNase I (Ambion) for 20 min at 37jC according to the product manual.One microgram of total RNA was then reverse transcribed in a final volume of 20 AL using Reverse Transcriptase III (Invitrogen).Real-time PCR was done for measurement of gene expression.A dissociation curve was analyzed for each sample to ensure that a single amplification product was obtained.For real-time PCR, 10 AL of 2Â SYBR Green PCR Master Mix (Applied Biosystems), 5 to 10 Amol/L forward and reverse primers of each gene, and 1 AL cDNA template were added in a 20 AL reaction in triplicate.Forty cycles of PCR amplification (95jC for 30 s and 60jC for 1 min) were done on an Applied Biosystems Prism 7000 or 7900 HT Sequence Detection System.Data are reported as the mean fold change F SD for experiments done in triplicate.
The PCR profile was 3 min at 94jC, followed by 35 cycles of 30 s at 94jC, 30 s at 57jC, and 30 s at 72jC, and a final extension of 10 min at 72jC.The amplified fragments were analyzed on a 1% agarose gel.The PCR fragments were directly cloned into the pGEM-T Easy Vector System as described in the manufacturer's protocol and sequenced to check their fidelity.The inserts were then released from the vector by appropriate restriction enzymes indicated above and subcloned into a pGL4-SV40 vector in sense and antisense orientations.Briefly, SV40 promoter was cloned in the location between a synthetic poly(A) signal/transcriptional pause site and the luc2 gene of the pGL4.10[luc2]vector (Promega).All constructs were reconfirmed by sequencing.
Hormone-depleted MCF-7 cells were transfected with each of the ERabinding domain vectors with Fugene 6 transfection reagent (Roche Applied Science) according to the manufacturer's protocol.Reporter plasmid (0.2 Ag) and pCMVhGal internal control (0.1 Ag) were transfected per well.Twenty-four hours after transfection, DMEM/F-12 was added containing ethanol (vehicle) or E 2 (10 À7 mol/L), and total protein lysate was collected and assayed for luciferase and h-galactosidase activities after 20-to 24-h treatment.
Transfection of small interfering RNA.RNA interference was carried out by using SMARTpool small interfering RNA (siRNA) designed against Wnt11 and siCONTROL nontargeting siRNA as a negative control (Dharmacon).After 3 days of culture in MEM containing 10% charcoalstripped calf serum, siRNA against Wnt11 or control siRNA at a final concentration of 100 nmol/L was transiently transfected into the MCF-7 cells for 48 h.The cells were then stimulated with E 2 (10 À7 mol/L) or vehicle for 20 to 24 h and harvested for analysis.Total RNA and protein were prepared from harvested cells using Tri-Reagent.Knockdown efficiency of target genes was examined by real-time PCR and Western blot.
Cell viability and apoptosis assay.Cell viability was determined by trypan blue exclusion.After various treatments, cells were harvested, washed, and treated with trypan blue at a concentration of 0.4% (w/v).After 10 min, trypan blue uptake (indicating dead cells) was determined by counting on a hemocytometer.Apoptosis in cells was evaluated by a poly(ADP-ribose) polymerase (PARP) cleavage Western blot assay.
Cell cycle distribution analysis.MCF-7 cells were transiently transfected with control siRNA or Wnt11 siRNA (100 nmol/L) for 48 h; the cells were then stimulated for 20 to 24 h with E 2 (10 À7 mol/L) or vehicle and harvested for cell cycle distribution analysis using propidium iodide (PI) staining and flow cytometry as in Keeton and Brown (10) with slight modification.Briefly, 1 Â 10 6 cells were harvested and washed with PBS and then fixed in cold 70% ethanol at À20jC for 2 h.Fixed cells were treated with 1 mL PI solution (50 Ag/mL PI, 0.2 mg/mL RNase A, and 0.1% Triton X-100) for 20 min at 37jC and analyzed for DNA content by flow cytometry by a core facility.
Western blotting.Western blot was done for cleaved PARP analysis and for detection of Wnt11 protein level knockdown.Aliquots of 20 Ag of total protein were electrophoresed on an 8% ( for PARP cleavage) or 10% ( for Wnt11 western) SDS-polyacrylamide gel and transferred to a nitrocellulose membrane.The membrane was blocked overnight at 4jC with 5% milk in TBS followed by hybridization with a rabbit anti-human cleaved PARP antibody at a dilution of 1:1,000 (Cell Signaling) or with rabbit anti-human Wnt11 antibody at a dilution of 1:1,000 (kindly provided by Len Eisenberg, Medical University of South Carolina, Charleston, SC; ref. 11) for confirmation of specificity of Wnt11.The hybridization with antibodies was done for 3 h at room temperature.After washing, the membrane was then incubated for 1 h at room temperature with horseradish peroxidaseconjugated secondary antibody (Sigma) at a dilution of 1:3,000.Immunoreactive bands were stained by a chemiluminescent procedure (Pierce) and visualized by autoradiography.

Results
Identification of genomic ERA-binding sites in human breast cancer cells.To identify the ERa-binding sites, we developed a technique to clone sequences from DNA immunoprecipitated by an anti-ERa antibody.The ChIP-PCR procedure was optimized to achieve amplification using an ERa antibody to detect proteins bound to the promoter of a prototypical ERa target gene, Myc, in the absence of any nonspecific IgG binding after 40 cycles of PCR (Supplementary Fig. S1).Once these conditions were reproducibly achieved in five consecutive experiments, DNA fragments immunoprecipitated by the ERa antibody were extracted, cloned, and sequenced.A total of 130 cloned fragments with insert sizes ranging from f100 to 1,000 bp were then identified by BLAT or ENSEMBL searches for human genome matches.
Ninety-four cloned fragments were mapped to the genome (Supplementary Table S1).The remaining 36 could not be mapped because either the fragment could not be sequenced due to a high GC-content or the cloned fragments were mapped to repetitive sequences across the genome.Of these, as shown in Table 1, f46% were localized within an intron of the open reading frame (ORF) of a gene.Forty percent were located in the 5 ¶-region of a gene (upstream of the transcription start site), and 14% were located in the 3 ¶-region of a gene (downstream of the 3 ¶-untranslated region).Further analysis showed that 23% of the ERa-ChIP fragments were located within the 50 kb 5 ¶-flanking region of a gene, whereas the remaining 17% were located >50 kb 5 ¶ upstream of a gene.Approximately 12% of ERa-ChIP fragments were located within 50 kb 3 ¶-flanking region of a gene.
We next identified the genes that contained an ERa-binding intron or that contained an ERa-binding site within 50 kb of its 5 ¶-or 3 ¶-flanking regions.First, we determined the mRNA expression levels of the most proximal gene after treatment with E 2 (10 À9 mol/L) at 3 or 6 h.Expression levels of the second most proximal gene were also determined after E 2 treatment at similar time points, provided that the coding region of the second gene was closer than 50 kb to the ERa-binding site.If more than two genes were found within 50 kb, we evaluated only the two most proximal genes.Following this strategy, we identified 88 E 2 -regulated genes located proximal to the 94 binding sites identified by ChIP cloning.Expression of 38 of these 88 (43%) genes was significantly regulated after 3 or 6 h of E 2 treatment by z1.5-fold or V0.67-fold compared with baseline (P < 0.05; Table 2).Importantly, 82% (31 of 38) of genes identified were novel E 2 targets, 24% of which were uncharacterized previously.Interestingly, 54% of genes with intronic sequences that bind ERa were regulated by E 2 .
Confirmation of ERA binding to E 2 -regulated genes.Conventional ChIP experiments showed binding of ERa, RNA PolII, NCOA3 (SRC-1), and IgG to 11 randomly selected putative ERa-binding sites (each labeled by the closest gene in Fig. 1).We showed that treatment with E 2 (versus ethanol) significantly enhanced binding of these three transcription factors to these sites.These results further validated these sites as functional ERa-binding sequences (Fig. 1).ERa, RNA PolII, and NCOA3 associations were observed in all 11 binding sites in an estrogendependent manner; the range of E 2 -dependent fold enrichments for ChIP-derived DNA were 2.6 to 7.2 (ERa), 2.9 to 11.0 (RNA poIII), and 2.6 to 6.5 (NCOA3) compared with the vehicle (Supplementary Figs.S2-S4).On the other hand, there were no E 2 -dependent fold enrichment of DNA fragments after ChIP with nonspecific IgG (range of fold change, 0.9-1.1;Supplementary Fig. S5).This independent demonstration of enrichment of these binding sites in response to E 2 suggests that the fragments identified by ChIP cloning were bona fide ERa-binding sites in MCF-7 cells.
Genes identified by ChIP cloning are highly regulated by E 2 .Using the Affymetrix U133 Plus 2.0 chip, which contains probe sets representing a majority of the human genes, we identified E 2regulated genes in MCF-7 cells.We found that 3% of the probe sets (1,618/54,613) showed differential expression (z1.5-fold or V0.67fold regulation; P < 0.05, t test) after 3 or 6 h of E 2 treatment.We compared the E 2 -regulated genes identified by the microarray with the 88 genes that contained proximal ERa targets identified by genome-wide ChIP cloning.We found that genes proximal to ERa target sequences have a much higher chance (43%) of being regulated by E 2 compared with E 2 -regulated probe sets (3%) determined by a genome-wide microarray experiment (P < 0.0001, two sample proportion test).
We next investigated whether the expression pattern of E 2regulated genes identified in MCF-7 cells was similar to that in the T47D cell line, which is also ERa + .We determined the mRNA levels of seven randomly selected E 2 -regulated ERa target genes in T47D cells.As shown in Supplementary Table S2, the results confirmed a very high degree of concordance between MCF-7 and T47D cells with respect to E 2 regulation of novel ERa target genes.On the other hand, none of these seven genes were regulated in the ERa À cell line MDA-MB-231, in which expression of two genes (MYOG and Wnt11) were undetectable at both time points tested.Moreover, the ER antagonist ICI 182780 blocked E 2 -dependent regulation of these genes (data not shown).These observations strongly support the notion that ERa is required for E 2 regulation of the target genes identified by the ChIP cloning method.Association between ERA target gene expression and ERA status of breast cancer tissues.ERa status (the presence or absence of ERa protein determined by immunohistochemistry in a tumor sample) is a prognostic factor in breast cancer and the single most important predictor for response to hormonal treatment (3).To investigate whether expression of ERa target genes identified in E 2 -treated MCF-7 cells are associated with the ERa status in breast cancer tissues, the mRNA levels of three novel ERa target genes were measured in 25 ERa + and 20 ERa À breast cancer tissues.
We selected three prototypical E 2 -regulated genes, each of which represented a distinct ERa-binding pattern: (a) 40 kb 5 ¶ (Wnt11), (b) intron 2 (Adora1), and (c) 2 kb 3 ¶ (SAPS2; Fig. 1; Table 2).We investigated in vivo regulation of these three genes by ERa using regression analyses.The mRNA levels of Wnt11, Adora1, and SAPS2 were plotted against ERa mRNA levels.Regression analysis showed statistically significant correlations between ERa mRNA and Wnt11 mRNA (r = 0.567), ERa mRNA and Adora1 mRNA (r = 0.609), and ERa mRNA and SAPS2 mRNA (r = 0.792; P < 0.00005 for each r value; Fig. 2).These findings indicate that E 2 -responsive genes that lie proximal to ERa-binding sites identified by ChIP cloning in breast cancer cell lines are also regulated by ERa in vivo.
ERA-binding sites exert regulatory activities.To determine whether the ERa-binding sites cloned by ChIP contained sequences with enhanced transcriptional activity, we cloned the ERa-binding fragments proximal to the Wnt11, Adora1, and SAPS2 genes in both the sense and the antisense orientations into an SV40 promoter/ luciferase reporter construct.As described above, the ERa-binding site for each gene represents one of three binding patterns: 5 ¶ proximal for Wnt11, intronic for Adora1, and 3 ¶ proximal for SAPS2.The ERa-binding site located 40 kb 5 ¶ of Wnt11 was not proximal to another gene.On the other hand, the ERa-binding site within intron 2 of Adora1 is also 13 kb 5 ¶ upstream of the Myogenin gene, which is regulated by E 2 in MCF-7 cells.The ERa target site located 2 kb 3 ¶ downstream of the SAPS2 gene is also 1 kb 3 ¶ downstream of the SBF1 gene and is also regulated by E 2 .
Individual ERa-binding site/luciferase reporter constructs were transfected into ERa + MCF-7 cells, which were then treated with vehicle or E 2 for 20 h and assayed for luciferase reporter gene activity.The ERa-binding site 40 kb 5 ¶ of Wnt11 enhanced transcriptional activity by 3.7-fold in the sense direction and 5.0-fold in  the antisense direction on treatment with E 2 (10 À7 mol/L; Fig. 3A).Similar results were obtained with transfection of luciferase reporter constructs containing ERa-binding sites within intron 2 Adora1 and 2 kb 3 ¶ of SAPS2 genes.E 2 treatment enhanced transcription by 4.1-fold and 3.7-fold in cells transfected with ERa-Adora1 sense and antisense constructs, respectively.Cells transfected with ERa-SAPS2 sense construct increased transcription by 6.5-fold on E 2 treatment, whereas transcriptional activity was augmented by 16-fold on E 2 treatment of cells transfected with ERa-SAPS2 antisense construct.
To show that E 2 regulated the transcriptional activity of these constructs in an ERa-dependent fashion, we transfected them into the ERa À MDA-MB-231 cell line.The absence of induction by E 2 supported our conclusion that these sequences confer responsiveness to E 2 in an ERa-dependent manner (Fig. 3B).Overall, these observations confirmed that ERa-binding regions identified for Wnt11, Adora1, and SAPS2 contain E 2 -responsive functional cis-acting elements.
Knockdown of E 2 -induced Wnt11 is associated with increased breast cancer cell death.We showed that Wnt11 is located proximally to an ERa-binding site and its expression is induced by E 2 (Supplementary Table S2).Because Wnt11 belongs to an oncogene family (18), we hypothesized that E 2 -induced Wnt11 expression could be involved in breast cancer pathobiology.MCF-7 cells were cultured in steroid-deprived medium for 3 days, and siRNA against Wnt11 or control siRNA was transiently transfected into the MCF-7 cells for 48 h followed by 20-to 24-h treatment with E 2 or vehicle.To show the efficiency and specificity of depletion of Wnt11, both real-time PCR and Western blot analysis were done (Fig. 4D).
As shown in Fig. 4, knockdown of Wnt11 in MCF-7 cells treated with or without E 2 resulted in a significantly decreased cell survival (12.2% and 13.9%, respectively) compared with cells transfected with control siRNA (5.3% and 6.7%, respectively; P < 0.05 for each case; Fig. 4A).We next determined whether this change in cell survival was due to changes in apoptosis and/or proliferation.Knockdown of Wnt11 increased apoptosis as shown specifically by a striking increase in PARP cleavage (Fig. 4B).The highest level of PARP cleavage was observed in Wnt11-depleted cells incubated with vehicle followed by Wnt11-depleted cells treated with E 2 .The results suggested that Wnt11 depletion increased apoptosis, and E 2 blunted this effect.On the other hand, depletion of Wnt11 by siRNA did not result in significant changes in the fractions of MCF-7 cells in S phases compared with control samples (Fig. 4C).However, significant changes in cell populations in S phases were observed in an E 2 -dependent manner, suggesting that Wnt11 does not have a major role in cell cycle progression in MCF-7 cells.Thus, the observed changes in cell viability seem to be due to an effect of Wnt11 on apoptosis but not on proliferation.

Discussion
Using a ChIP-linked target site cloning strategy, we identified 94 ERa-binding sites within the human genome, many of which represent novel E 2 targets in MCF-7 breast cancer cells.We showed Figure 3. Transcriptional activities of ERa-binding sites proximal to E 2 -induced genes.ERa-binding fragments for Wnt11, Adora1, and SAPS2 were cloned in the sense or antisense orientation into a pGL4-SV40 plasmid.Expression of the empty vector (Empty ) was used as a reference standard.Reporter plasmids (0.2 Ag) and pCMVhGal (0.1 Ag) were cotransfected into ERa + MCF-7 cells (A) and ERa À MDA-MB-231 cells (B) in 24-well plates.After 20 to 24 h, cells were washed with PBS and treated with 10 À7 mol/L E 2 or ethanol for an additional 20 h.Cells were then lysed and assayed for luciferase and h-galactosidase activities.Columns, mean of a representative experiment done from at least three experiments; bars, SE.Constructs oriented in the sense direction for the Wnt11, Adora1, and SAPS2 genes are denoted Wnt11-S, Adora1-S, and SAPS2-S, respectively; whereas constructs that contain the antisense ERa-binding fragments are denoted Wnt11-AS, Adora1-AS, and SAPS2-AS.*, P < 0.05, t test, statistically significant differences compared with empty vector.
both in vivo and in vitro that a representative portion of these genes are highly regulated by E 2 /ERa in breast cancer cells and tissues.We also showed a novel biological role of the E 2 -regulated gene Wnt11 in breast cancer.
A significant volume of work has focused on identifying essential domains within the proximal promoters of E 2 -regulated genes (6,12,13,(19)(20)(21)(22)(23)(24).However, many functionally relevant binding sites for transcription factor likely exist in regions outside of gene promoters, particularly in introns (25,26).These sites would not have been identified using promoter microarray (21) or CpG islandenriched DNA array (20,27), which identify binding sites only within the region proximal to the transcription start site.A recently published pioneering study showed by ChIP-on-chip that ERa binds to thousands of sites genome-wide and interact with transcription factors binding to specific cis-acting elements as a master regulator of many genes (7).
We compared the ERa-binding sites found in this study with those published by Carroll et al. (7).Approximately 95%, 86%, 57%, 39%, 17%, 11%, and 6% of the 94 binding sites that we cloned were located 500, 300, 100, 50, 20, 11, and 5 kb away from the closest binding sequences published by Carroll et al, respectively (7).The distribution of binding sequences across the chromosomes was fairly similar in both studies.We identified one ERa-binding site mapped to the Y chromosome, whereas no binding sites were mapped to this chromosome in the Carroll et al. study (7).We showed the proximity of the 11 sites, which bind ERa, RNA PolII, and NCOA3, from our study to the closest possible binding sites identified in the Carroll et al. study in Supplementary Table S3.The distance ranged from 1.24 to 304 kb.This possibly suggested that the identification of exact binding sites may vary with the technique used.In general, our findings agree with the conclusions published by Carroll et al.For example, both groups found that the majority of the ERa-binding sites fell outside of classically defined promoter regions.
In this study, there were only 18 (of 94) binding sites that did not reside within 50 kb of a gene.Thus, we arbitrarily chose this distance to limit the group of genes to be tested for regulation by estrogen treatment.It is quite possible that ERa-binding sites can also regulate genes that lie more than 50 kb away.In fact, recently published articles from Carroll et al. (7) showed the importance of these far distant sites as regulators of transcription.
The fact that 54% of the genes containing intronic ERa-binding sites were regulated by E 2 suggests that this is an interesting phenomenon with functional significance.ERa-binding fragments located distal to genes or within introns might function through long-range interactions that involve looping of chromatin to bring the elements within proximity of gene promoters (26,28).Indeed, the intronic ERa-binding sites identified in the present study indicate that ERa may regulate target gene transcription by altering local chromatin structure.Recent reports about intronic binding of other transcription factors, such as cAMP-responsive element binding protein and BARX2, provide further support that intronic binding of ERa may be an important mechanism of gene regulation by E 2 (25,29).
E 2 has been shown to regulate transcription through either direct binding of EREs or indirectly by interacting with transcription factor complexes.Analysis of the ERa-binding sequences associated with the 38 identified E 2 -regulated genes revealed that 61% of these sequences contained at least one classic (palindromic) ERE.Eightyseven percent of binding sites missing a palindromic ERE contained one or more AP-1 sites.The comparison of these results with previously published data suggests that there is a higher likelihood of the occurrence of canonical EREs in ERa-binding sites proximal to an E 2 -regulated gene (4,6).Furthermore, in the absence of an ERE, AP-1 or other cis-acting elements may confer E 2 /ERa responsiveness possibly via tethering of ERa on AP-1-binding transcription factors (19).
It has been shown that Wnt11 signaling induces proliferation (30), transformation (31), and prostate cancer progression (32) through a noncanonical pathway (33)(34)(35).However, the role of Wnt11 in the progression of breast cancer remains unknown.Our studies provide an initial insight into the mechanism by which Wnt11 may promote tumor progression in the breast.Upregulation of Wnt11 mRNA on E 2 exposure may activate the Wnt11 signaling pathway and inhibit apoptosis, thus favoring tumor growth.Further investigation of the biological functions of Wnt11 and other identified ERa-regulated genes in breast cancer may lead to the development of specific, targeted treatments.

Figure 1 .
Figure 1.Validation of the binding of a transcription complex to ERa-bound DNA fragments cloned after ChIP.Left, conventional ChIP of ERa and standard PCR of sites adjacent to randomly selected 11 genes; right, ChIP of ERa, RNA PolII, NCOA3, or IgG (control) and real-time PCR of above ERa-binding regions.Data represent estrogen-mediated fold enrichment compared with vehicle (ethanol) control.DNA samples after ChIP of ERa, RNA poIII, NCOA3, and IgG, as well as input DNA (Inp ) from samples treated with or without E 2 were quantified by PicoGreen dsDNA dye.Negative control (N ) was PCR without DNA.The color intensity reflects the fold change as described in the legend.The detailed enrichment graphs of ChIP with ERa, RNA PolII, NCOA3, and IgG are available as Supplementary Figs.S2 to S5.Data are the average of three replicates F SD.

Figure 2 .
Figure 2. Correlation between mRNA levels of ERa and Wnt11, Adora1, and SAPS2 in 45 breast tumor tissues.Total RNA was isolated from immunohistochemically determined 25 ERa + and 20 ERa À breast cancer tissues.mRNA levels of ERa, Wnt11, Adora1, and SAPS2 were measured by real-time PCR and normalized by glyceraldehyde-3-phosphate dehydrogenase.The correlation between mRNA levels of ERa and the three ERa target genes were carried out by regression analysis.

Figure 4 .
Figure 4. siRNA-mediated inhibition of Wnt11 in MCF-7 leads to increased cell death.MCF-7 cells were cultured in hormone-depleted medium for 3 days, and siRNA against Wnt11 or control siRNA was transiently transfected into the MCF-7 cells for 48 h.The cells were then stimulated with E 2 (10 À7 mol/L) or vehicle for 20 to 24 h and harvested for analysis.MCF-7 cells were harvested for determination of percentage of nonviable cells by trypan blue exclusion (A) and for PARP cleavage with rabbit anti-human cleaved PARP antibody (B).Lanes 1 and 2, control siRNA or Wnt11 siRNA-transfected MCF-7 cells treated with vehicle; lanes 3 and 4, control siRNA or Wnt11 siRNA-transfected MCF-7 cells treated with E 2 (10 À7 mol/L).Blots were reprobed with a h-actin antibody to control for loading.Analysis of cell cycle progression after silencing of Wnt11 (C).Knockdown efficiency and specificity of the Wnt11 gene were examined by both real-time PCR and Western blotting using a Wnt11 peptide antibody (D).Columns, mean of three independent experiments; bars, SE. *, P < 0.05, t test, statistically significant differences.

Table 2 .
ERa target genes regulated by estradiol