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

Research Article

Zhihong Lin,1 Scott Reierstad,1 Chiang-Ching Huang,2 and Serdar E. Bulun1

Departments of 1Obstetrics and Gynecology and 2Preventive Medicine, Feinberg School of Medicine, Northwestern University, Chicago, Illinois

Abstract by association with other transcription factors on promoter targets A A (1, 4, 5). Until recently, however, little had been known about Estrogen receptor- (ER ) and its ligand estradiol play a critical roles in breast cancer growth and are important the distribution of ER -binding sites within the genome and the therapeutic targets for this disease. Using chromatin immu- identity of genes regulated by these cis-acting elements. A Two recent pioneering publications by Carroll et al. (6, 7) noprecipitation (ChIP)-on-chip, ligand-bound ER was re- a cently found to function as a master transcriptional regulator revolutionized our understanding of ER action. Using chromatin cis immunoprecipitation (ChIP)-on-chip, this group mapped a large via binding to many -acting sites genome-wide. Here, we a used an alternative technology (ChIP cloning) and identified number of ER -binding sites on a chromosome and genome-wide scale, identifying novel cis-regulatory sites and target genes in 94 ERA target loci in breast cancer cells. The ERA-binding sites contained both classic estrogen response elements and MCF-7 breast cancer cells (6, 7). The majority of these binding sites nonclassic binding sequences, showed specific transcriptional were distant from the transcription start sites of regulated genes activity in reporter gene assay, and interacted with the key (6, 7). Identification of novel genomic targets and a deeper understand- transcriptional regulators, including RNA polymerase II and a nuclear receptor coactivator-3. The great majority of the ing of their transcriptional regulation by ER and their physiologic binding sites were located in either introns or far distant to function may lead to the development of more specific and effective coding regions of genes. Forty-three percent of the genes that treatments for breast cancer. In the current study, we used a lie within 50 kb to an ERA-binding site were regulated by technique alternative to ChIP-on-chip (i.e., ChIP-linked target site estradiol. Most of these genes are novel estradiol targets cloning) for unbiased and potentially genome-wide identification of regulatory targets of estradiol/ERa in breast cancer cells. Our aims encoding receptors, signaling messengers, and ion binders/ were to determine the nature of ERa binding relative to the structure transporters. mRNA profiling in estradiol-treated breast a cancer cell lines and tissues revealed that these genes are of a gene and increase our understanding of ER action both in A in vitro in vivo normal tissue and in the malignant state. We characterized these highly ER responsive both and . Among a estradiol-induced genes, Wnt11 was found to increase cell binding sites and regulation of proximal genes by estradiol/ER both in vitro and in vivo. The majority of the 38 estrogen-regulated genes survival by significantly reducing apoptosis in breast cancer a cells. Taken together, we showed novel genomic binding sites turned out to be previously unknown ER targets. We showed in A detail previously unknown biological roles of one of these estrogen/ of ER that regulate a novel set of genes in response to a estradiol in breast cancer. Our findings suggest that at least a ER –regulated genes, Wnt11, in breast cancer. subset of these genes, including Wnt11, may play important in vivo in vitro and biological roles in breast cancer. [Cancer Materials and Methods Res 2007;67(10):5017–24] Cell lines and tissues. MCF-7, T47D, and MDA-MB-231 cells (American Introduction Type Culture Collection) were maintained in MEM (Invitrogen) containing 25 units/mL penicillin, 25 units/mL streptomycin, and 10% fetal bovine a a j + À Estrogen receptor- (ER ; ref. 1) is a ligand-activated nuclear serum (FBS) at 37 C and 5% CO2. Snap-frozen ER (n = 25) and ER receptor that regulates transcription of estrogen-responsive genes (n = 20) breast cancer tissues were obtained from the Northwestern Breast important for cell growth, differentiation, and malignant transfor- Specialized Programs in Research Excellence Tissue Core Facility. These mation in various target cells (2). ERa and its ligand estradiol tissues were collected after obtaining written informed consent approved by the Institutional Review Board of Northwestern University. [17h-estradiol (E2)] play critical roles in the growth of breast cancer tissue and are important therapeutic targets (3). Significant Chromatin immunoprecipitation. After MCF-7 cells were grown to 75% to 80% confluence in MEM supplemented with 10% FBS, the cells were progress has been made in understanding the role of ERa as a serum starved in DMEM/F-12 without phenol red (Invitrogen) and FBS for transcription factor that regulates the expression of target genes À9 24 h. After 3 h of treatment with 10 mol/L E2, cells were washed twice by directly binding to an estrogen response element (ERE; ref. 4) or 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 Note: Supplementary data for this article are available at Cancer Research Online of 125 nmol/L for 5 min at room temperature. Cells were rinsed twice (http://cancerres.aacrjournals.org/). with cold PBS, harvested, and stored at À80jC before use. Cell pellets were Requests for reprints: Serdar E. Bulun, Department of Obstetrics and Gynecology, lysed and sonicated to shear the DNA into 0.6- to 3.0-kb fragments. Feinberg School of Medicine, Northwestern University, Chicago, IL 60611. Phone: 312- Insoluble material was removed by centrifugation, and the extract was 503-1600; Fax: 312-503-0095; E-mail: [email protected]. I2007 American Association for Cancer Research. precleared by incubation with blocked protein A-agarose/Salmon Sperm doi:10.1158/0008-5472.CAN-06-3696 DNA (Upstate) for at least 1 h at 4jC to reduce nonspecific interactions. www.aacrjournals.org 5017 Cancer Res 2007; 67: (10). May 15, 2007

After centrifugation, the supernatant (50 AL) was collected as input, and microgram of total RNA was then reverse transcribed in a final volume of the remainder was diluted in buffer [1% Triton X-100, 2 mmol/L EDTA, 20 AL using Reverse Transcriptase III (Invitrogen). Real-time PCR was done 50 mmol/L Tris-HCl (pH 8.1)] and subjected to immunoprecipitation for measurement of gene expression. A dissociation curve was analyzed for overnight at 4jC with a monoclonal antibody against ERa (Upstate). each sample to ensure that a single amplification product was obtained. After immunoprecipitation, 60 AL protein A-agarose/Salmon Sperm DNA For real-time PCR, 10 ALof2Â SYBR Green PCR Master Mix (Applied beads were added and the incubation was continued for another 1 h. To Biosystems), 5 to 10 Amol/L forward and reverse primers of each gene, and decrease nonspecific binding, DNAs/protein complexes were washed under 1 AL cDNA template were added in a 20 AL reaction in triplicate. Forty high-stringency wash conditions. Precipitates were washed sequentially for cycles of PCR amplification (95jC for 30 s and 60jC for 1 min) were done on 10 min, two times in buffer I [0.1% SDS, 1% Triton X-100, 2 mmol/L EDTA, an Applied Biosystems Prism 7000 or 7900 HT Sequence Detection System. 20 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl], six times in buffer II [0.1% Data are reported as the mean fold change F SD for experiments done in SDS, 1% Triton X-100, 2 mmol/L EDTA, 20 mmol/L Tris-HCl (pH 8.0), 500 triplicate. mmol/L NaCl], two times in buffer III [0.25 mol/L LiCl, 1% NP40, 1% Plasmids, transfections, and luciferase assays. ERa-binding regions deoxycholate, 1 mmol/L EDTA, 10 mmol/L Tris-HCl (pH 8.0)], and twice identified from ChIP cloning for Wnt11, Adora1, and SAPS2 were amplified by with 1 mmol/L EDTA and 10 mmol/L Tris-HCl (pH 8.0). Precipitated PCR. To investigate the effect of orientation of ERa-binding fragments on chromatin complexes were removed from the beads through a 15-min transcriptional activities, ERa-binding fragments were cloned in both the incubation with 50 AL of 1% SDS and 0.1 mol/L NaHCO3 and vortexing at sense and the antisense orientations in reporter gene expression vectors. room temperature. This step was done twice. Eluates were pooled and For orientation cloning, primers were used to introduce KpnI and XhoI sites heated at 65jC overnight to reverse the formaldehyde cross-linking. DNA into amplified fragments of Wnt11 and Adora1 and KpnI and SacI sites fragments were purified with a MinElute Reaction Cleanup kit (Qiagen). The into the fragment of SAPS2. The primers for the sense fragments were the immunoprecipitated and input DNA samples were assayed for binding to following: Wnt11,5¶-GGTACCAGTTGTTAGGAATGGCTATGTTCC-3¶ (for- the MYC promoter region as a positive control before ChIP cloning. For ward) and 5¶-CTCGAGTAGGCTACCAAGTCTATGGTATTCTG-3¶ (reverse); PCR, 1 AL of purified DNA extraction and 40 cycles of amplification were Adora1,5¶-GGTACCTCTGTCTATGCACACCATGCA-3¶ ( forward) and 5¶- used. CTCGAGCCAGTCTGAAAACTGGTAATAATACCT-3¶ (reverse); and SAPS2, Cloning, sequencing, and analysis of ERA-binding fragments. ChIP- 5¶-GGTACCGTCCAACTGGTTTCAACT-3¶ ( forward) and 5¶-GAGCTCCG- derived DNA was subjected to AvaII restriction digestion and PCR linker TAACCGGAGTCCATGTTT-3¶ (reverse). The primers for the antisense ligation followed by PCR amplification (40 cycles) generating sufficient fragments were the following: Wnt11,5¶-CTCGAGAGTTGTTAGGAATGGC- material for cloning into the pGEM-T Easy Vector System (Promega). Clones TATGTTCC-3¶ (forward) and 5¶-GGTACCTAGGCTACCAAGTCTATGG- were sequenced and every retrieved ChIP fragment was mapped to the TATTCTG-3¶ (reverse); Adora1,5¶-CTCGAGTCTGTCTATGCACACCATGCA- human genome using the BLAT search function of University of California 3¶ (forward) and 5¶-GGTACCAGTCTGAAAACTGGTAATAATACCT-3¶ Santa Cruz genome browser or ENSEMBL search. The presence of EREs was (reverse); and SAPS2,5¶-GAGCTCGTCCAACTGGTTTCAACT-3¶ (forward) analyzed by ERE Finder (8), and activator protein-1 (AP-1) sites were and 5¶-GGTACCGTAACCGGAGTCCATGTTT-3¶ (reverse). identified using the Transcription Element Search System.3 The PCR profile was 3 min at 94jC, followed by 35 cycles of 30 s at 94jC, Microarray-based mRNA profiling. MCF-7 cells at 70% to 80% 30 s at 57jC, and 30 s at 72jC, and a final extension of 10 min at 72jC. The confluence maintained in MEM as described above were treated with amplified fragments were analyzed on a 1% agarose gel. The PCR fragments À9 vehicle or E2 at a concentration of 10 mol/L for 3 and 6 h. Total RNA was were directly cloned into the pGEM-T Easy Vector System as described in extracted as described below. Gene expression profiles of the MCF-7 cells the manufacturer’s protocol and sequenced to check their fidelity. The were analyzed on Human Genome U133 Plus 2.0 microarray chips inserts were then released from the vector by appropriate restriction (Affymetrix), which contained 54,613 probe sets. Sample labeling and enzymes indicated above and subcloned into a pGL4-SV40 vector in sense subsequent hybridization to the array were carried out according to the and antisense orientations. Briefly, SV40 promoter was cloned in the manufacturer’s instructions in the Microarray Core Facility within the location between a synthetic poly(A) signal/transcriptional pause site and Center for Genetic Medicine, Northwestern University. Expression data the luc2 gene of the pGL4.10[luc2] vector (Promega). All constructs were normalization was done as described previously (9). reconfirmed by sequencing. Validation of the in vivo ERa-binding sites by real-time PCR. To Hormone-depleted MCF-7 cells were transfected with each of the ERa- examine the enrichment of specific ERa-binding fragments in an binding domain vectors with Fugene 6 transfection reagent (Roche Applied independent ChIP assay, primers were generated corresponding to the Science) according to the manufacturer’s protocol. Reporter plasmid (0.2 Ag) regions examined within each ChIP-derived genomic fragment. Primers and pCMVhGal internal control (0.1 Ag) were transfected per well. Twenty-four were synthesized by Integrated DNA Technologies. For each PCR assay, DNA hours after transfection, DMEM/F-12 was added containing ethanol (vehicle) or a À7 after ChIP of ER and IgG as well as input DNA was quantified by E2 (10 mol/L), and total protein lysate was collected and assayed for luciferase PicoGreen dsDNA dye (Invitrogen) from E2-treated or vehicle (ethanol) and h-galactosidase activities after 20- to 24-h treatment. MCF-7 cells. Real-time PCR was done using Applied Biosystems SYBR Green Transfection of small interfering RNA. RNA interference was carried Master kit following manufacturer’s instructions. The estrogen-mediated out by using SMARTpool small interfering RNA (siRNA) designed against fold enrichment of ERa-binding regions relative to IgG was compared with Wnt11 and siCONTROL nontargeting siRNA as a negative control its vehicle (ethanol) control. For every fragment analyzed, enrichment was (Dharmacon). After 3 days of culture in MEM containing 10% charcoal- measured after three independent immunoprecipitations. To further stripped calf serum, siRNA against Wnt11 or control siRNA at a final validate the in vivo binding of ERa to its putative binding regions, ChIP concentration of 100 nmol/L was transiently transfected into the MCF-7 cells À7 of nuclear receptor coactivator-3 (NCOA3; also known as SRC-1), RNA for 48 h. The cells were then stimulated with E2 (10 mol/L) or vehicle for polymerase II (PolII; Santa Cruz Biotechnologies), and IgG were done. 20 to 24 h and harvested for analysis. Total RNA and protein were prepared RNA preparation and validation of ChIP-derived target gene from harvested cells using Tri-Reagent. Knockdown efficiency of target expression by real-time PCR. Total RNA was extracted from the MCF-7, genes was examined by real-time PCR and Western blot. T47D, and MDA-MB-231 breast cancer cell lines and breast cancer tissue Cell viability and apoptosis assay. Cell viability was determined by samples (n = 45; wet weight 100 mg) using Tri-Reagent (Sigma) according trypan blue exclusion. After various treatments, cells were harvested, to the manufacturer’s instruction. Total RNA samples were treated with washed, and treated with trypan blue at a concentration of 0.4% (w/v). DNase I (Ambion) for 20 min at 37jC according to the product manual. One 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 trans- 3 http://www.cbil.upenn.edu/cgi-bin/tess/tess fected with control siRNA or Wnt11 siRNA (100 nmol/L) for 48 h; the cells

Cancer Res 2007; 67: (10). May 15, 2007 5018 www.aacrjournals.org

À7 were then stimulated for 20 to 24 h with E2 (10 mol/L) or vehicle and 94 binding sites identified by ChIP cloning. Expression of 38 of these harvested for cell cycle distribution analysis using propidium iodide (PI) 88 (43%) genes was significantly regulated after 3 or 6 h of E2 staining and flow cytometry as in Keeton and Brown (10) with slight treatment by z1.5-fold or V0.67-fold compared with baseline Â 6 modification. Briefly, 1 10 cells were harvested and washed with PBS and (P < 0.05; Table 2). Importantly, 82% (31 of 38) of genes identified then fixed in cold 70% ethanol at À20jC for 2 h. Fixed cells were treated A were novel E2 targets, 24% of which were uncharacterized previously. with 1 mL PI solution (50 g/mL PI, 0.2 mg/mL RNase A, and 0.1% Triton X- a 100) for 20 min at 37jC and analyzed for DNA content by flow cytometry by Interestingly, 54% of genes with intronic sequences that bind ER a core facility. were regulated by E2. A Western blotting. Western blot was done for cleaved PARP analysis and Confirmation of ER binding to E2-regulated genes. for detection of Wnt11 protein level knockdown. Aliquots of 20 Ag of total Conventional ChIP experiments showed binding of ERa, RNA protein were electrophoresed on an 8% (for PARP cleavage) or 10% (for PolII, NCOA3 (SRC-1), and IgG to 11 randomly selected putative Wnt11 western) SDS-polyacrylamide gel and transferred to a nitrocellulose ERa-binding sites (each labeled by the closest gene in Fig. 1). j membrane. The membrane was blocked overnight at 4 C with 5% milk in We showed that treatment with E2 (versus ethanol) significantly TBS followed by hybridization with a rabbit anti-human cleaved PARP enhanced binding of these three transcription factors to these antibody at a dilution of 1:1,000 (Cell Signaling) or with rabbit anti-human sites. These results further validated these sites as functional Wnt11 antibody at a dilution of 1:1,000 (kindly provided by Len Eisenberg, ERa-binding sequences (Fig. 1). ERa, RNA PolII, and NCOA3 Medical University of South Carolina, Charleston, SC; ref. 11) for confirmation of specificity of Wnt11. The hybridization with antibodies associations were observed in all 11 binding sites in an estrogen- was done for 3 h at room temperature. After washing, the membrane was dependent manner; the range of E2-dependent fold enrichments then incubated for 1 h at room temperature with horseradish peroxidase– for ChIP-derived DNA were 2.6 to 7.2 (ERa), 2.9 to 11.0 (RNA poIII), conjugated secondary antibody (Sigma) at a dilution of 1:3,000. Immuno- and 2.6 to 6.5 (NCOA3) compared with the vehicle (Supplementary reactive bands were stained by a chemiluminescent procedure (Pierce) and Figs. S2–S4). On the other hand, there were no E2-dependent visualized by autoradiography. 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 Results sites in response to E2 suggests that the fragments identified Identification of genomic ERA-binding sites in human by ChIP cloning were bona fide ERa-binding sites in MCF-7 breast cancer cells. To identify the ERa-binding sites, we cells. developed a technique to clone sequences from DNA immunopre- Genes identified by ChIP cloning are highly regulated by E2. cipitated by an anti-ERa antibody. The ChIP-PCR procedure was Using the Affymetrix U133 Plus 2.0 chip, which contains probe sets optimized to achieve amplification using an ERa antibody to detect representing a majority of the human genes, we identified E2- proteins bound to the promoter of a prototypical ERa target gene, regulated genes in MCF-7 cells. We found that 3% of the probe sets Myc, in the absence of any nonspecific IgG binding after 40 cycles of (1,618/54,613) showed differential expression (z1.5-fold or V0.67- PCR (Supplementary Fig. S1). Once these conditions were repro- fold regulation; P < 0.05, t test) after 3 or 6 h of E2 treatment. We ducibly achieved in five consecutive experiments, DNA fragments compared the E2-regulated genes identified by the microarray with immunoprecipitated by the ERa antibody were extracted, cloned, the 88 genes that contained proximal ERa targets identified by and sequenced. A total of 130 cloned fragments with insert sizes genome-wide ChIP cloning. We found that genes proximal to ERa ranging from f100 to 1,000 bp were then identified by BLAT or target sequences have a much higher chance (43%) of being ENSEMBL searches for human genome matches. regulated by E2 compared with E2-regulated probe sets (3%) Ninety-four cloned fragments were mapped to the genome determined by a genome-wide microarray experiment (P < 0.0001, (Supplementary Table S1). The remaining 36 could not be mapped two sample proportion test). because either the fragment could not be sequenced due to a high Characterization of ERA target genes in breast cancer cell GC-content or the cloned fragments were mapped to repetitive lines. As described above, we confirmed that 38 of the genes 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 Table 1. Genome-wide location of ERa-binding sites the 3¶-region of a gene (downstream of the 3¶-untranslated region). relative to a gene Further analysis showed that 23% of the ERa-ChIP fragments were located within the 50 kb 5¶-flanking region of a gene, whereas Proximity to the ERa-binding sites, the remaining 17% were located >50 kb 5¶ upstream of a gene. closest gene ChIP clones (%) Approximately 12% of ERa-ChIP fragments were located within 50 kb 3¶-flanking region of a gene. Intronic 43 (45.7) ¶ We next identified the genes that contained an ERa-binding 5 -Flanking region (kb) 38 (40.4) <10 12 (12.8) intron or that contained an ERa-binding site within 50 kb of its 5¶-or ¶ 11–50 10 (10.6) 3 -flanking regions. First, we determined the mRNA expression levels >50 16 (17.0) À9 of the most proximal gene after treatment with E2 (10 mol/L) at 3 3¶-Flanking region (kb) 13 (13.8) or 6 h. Expression levels of the second most proximal gene were also <4 9 (9.57) determined after E2 treatment at similar time points, provided that 4–11 1 (1.06) the coding region of the second gene was closer than 50 kb to the 11–50 1 (1.06) ERa-binding site. If more than two genes were found within 50 kb, >50 2 (2.13) we evaluated only the two most proximal genes. Following this Total 94 (100) strategy, we identified 88 E2-regulated genes located proximal to the www.aacrjournals.org 5019 Cancer Res 2007; 67: (10). May 15, 2007

Table 2. ERa target genes regulated by estradiol

Gene Gene ID Genomic Induction Function

location by E2*

c Cathepsin D (CTSD) 1509 1.0 kb 5¶ 2.27 F 0.376 Protease c V-myc myelocytomatosis viral 4609 1.0 kb 5¶ 2.71 F 0.04 Transcription factor, proliferation oncogene homologue (MYC) c Hypothetical LOC402592 402592 2.0 kb 5¶ 2.00 F 0.17 Unknown c FLJ40194 protein 124871 3.5 kb 5¶ 7.78 F 1.27 Unknown Hypothetical LOC339583 339583 4.0 kb 5¶ 0.42 F 0.01 Unknown b Olfactory receptor, family 7, 26333 5.0 kb 5¶ 9.82 F 2.25 GPCR, signal transduction subfamily A, member 17 (OR7A17) b Chromosome 9 ORF 84 158401 8.0 kb 5¶ 1.94 F 0.08 Unknown c Myogenin (MYOG) 4656 13.0 kb 5¶ 2.55 F 0.19 Transcription factor b Methyl-CpG binding domain protein 4 (MBD4) 8930 13.0 kb 5¶ 1.52 F 0.07 DNA repair c Hypothetical gene supported by BC009385 400043 18.0 kb 5¶ 16.46 F 1.56 Unknown c PDZ domain containing (PDZK1) 5174 28.0 kb 5¶ 1.87 F 0.09 Ion transport c Plexin domain containing 1(PLXDC1) 57125 29.0 kb 5¶ 2.55 F 0.54 Receptor activity, oncogenesis c Zinc finger and BTB domain containing 9 (ZBTB9) 221504 36.0 kb 5¶ 1.50 F 0.05 Transcription factor b Heat shock transcription factor 1(HSF1) 3297 40.0 kb 5¶ 1.52 F 0.04 Transcription factor b Wingless-type MMTV integration site family, member 11 (Wnt11) 7481 40.0 kb 5¶ 7.31 F 0.51 Signal transduction, oncogene b G-protein coupled receptor 173 (GPR173) 54328 49.0 kb 5¶ 0.38 F 0.07 GPCR, signal transduction b Protein phosphatase 2, regulatory subunit B, c isoform (PPP2R2C) 5522 Intron 1 1.85 F 0.11 Signal transduction c Solute carrier family 9, isoform 1 (SLC9A1) 6548 Intron 1 0.58 F 0.07 Ion transport b Nitric oxide synthase 1 (NOS1) 4842 Intron 1 0.12 F 0.05 Enzyme, signal transduction b Glutamate receptor, ionotropic, kainate 4 (GRIK4) 2900 Intron 1 2.34 F 0.10 Ion transport, signal transduction c Hypothetical gene supported by AK022887; AK056417 (FLJ12825) 440101 Intron 1 1.73 F 0.26 Unknown c Adenosine A1 receptor (ADORA1) 134 Intron 2 6.24 F 0.70 GPCR, signal transduction, apoptosis c Sparc/osteonectin, cwcv and kazal-like domains proteoglycan 1 (SPOCK) 6695 Intron 2 1.53 F 0.06 Cell motility, proliferation c Sideroflexin 5 (SFXN5) 94097 Intron 2 2.55 F 0.19 Ion transport c Q9H5L9 (no description) N/A Intron 2 1.63 F 0.23 Unknown WD repeat domain 10 (WDR10) 55764 Intron 3 1.59 F 0.11 Signal transduction, apoptosis b Block of proliferation 1(BOP1) 23246 Intron 3 2.66 F 0.18 rRNA processing b Hypothetical LOC389167 389167 Intron 3 2.34 F 0.32 Unknown b p21(CDKN1A)-activated kinase 7 (PAK7) 57144 Intron 3 1.71 F 0.22 Phosphatase b Serine/threonine kinase 32B (STK32B) 55351 Intron 3 0.57 F 0.05 Phosphatase, signal transduction c Neutrophil cytosolic factor 2 (NCF2) 4688 Intron 4 2.27 F 0.08 Immune response c Protein phosphatase 1K (PP2C domain containing; PPM1K) 152926 Intron 5 2.50 F 0.19 Phosphatase, signal transduction c Glutamate receptor interacting protein 1 (GRIP1) 23426 Intron 8 0.53 F 0.03 Transcription coactivator c Adaptor-related protein complex 1, b1 subunit (AP1B1) 162 Intron 11 1.57 F 0.15 Endocytosis, vesicle transport c Oxysterol binding protein-like 5 (OSBPL5) 114879 Intron 15 2.06 F 0.28 Receptor, lipid transport c SET binding factor 1(SBF1) 6305 1.0 kb 3¶ 1.60 F 0.08 Phosphatase, signal transduction c SAPS domain family, member 2 (SAPS2) 9701 2.0 kb 3¶ 3.39 F 0.71 Unknown c Calcium channel, voltage-dependent, b1 subunit (CACNB1) 782 3.8 kb 3¶ 1.61 F 0.13 Ion transport

*mRNA expression induction by E2 at 3 or 6h. cRepresents ERE. bRepresents AP-1 binding sites. containing ERa-binding sites identified by ChIP cloning were the T47D cell line, which is also ERa+.Wedeterminedthe a significantly up-regulated or down-regulated by E2 (P < 0.05). These mRNA levels of seven randomly selected E2-regulated ER target genes encoded novel E2 targets with various functions, such as genes in T47D cells. As shown in Supplementary Table S2, the proliferation (Myc and SPOCK), apoptosis (Adora1), cell motility results confirmed a very high degree of concordance between (SPOCK), Wnt signaling (Wnt11), transcription factors or coregu- MCF-7 and T47D cells with respect to E2 regulation of novel lators (Myc, ZBTB9, HSF1, MYOG, PLXDC1,andGRIP1), ion ERa target genes. On the other hand, none of these seven genes À transport (SLC9A1, AP1B1, and CACBN1), and signal transduction were regulated in the ERa cell line MDA-MB-231, in which (OR7A17, Wnt11, GPR173, STK32B, and PPM1K; Table 2). Of note, expression of two genes (MYOG and Wnt11) were undetectable that several genes known previously to be regulated by E2, including at both time points tested. Moreover, the ER antagonist ICI Myc (12–14), CTSD (13, 15), ADORA1 (16), and GRIP1 (17), were 182780 blocked E2-dependent regulation of these genes (data identified in our experiments validates the ChIP cloning technique. not shown). These observations strongly support the notion that a We next investigated whether the expression pattern of E2- ER is required for E2 regulation of the target genes identified regulated genes identified in MCF-7 cells was similar to that in by the ChIP cloning method.

Cancer Res 2007; 67: (10). May 15, 2007 5020 www.aacrjournals.org

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 E2 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.

Association between ERA target gene expression and ERA sites identified by ChIP cloning in breast cancer cell lines are also status of breast cancer tissues. ERa status (the presence or regulated by ERa in vivo. absence of ERa protein determined by immunohistochemistry in a ERA-binding sites exert regulatory activities. To determine tumor sample) is a prognostic factor in breast cancer and the whether the ERa-binding sites cloned by ChIP contained sequences single most important predictor for response to hormonal with enhanced transcriptional activity, we cloned the ERa-binding treatment (3). To investigate whether expression of ERa target fragments proximal to the Wnt11, Adora1, and SAPS2 genes in both genes identified in E2-treated MCF-7 cells are associated with the the sense and the antisense orientations into an SV40 promoter/ ERa status in breast cancer tissues, the mRNA levels of three novel luciferase reporter construct. As described above, the ERa-binding À ERa target genes were measured in 25 ERa+ and 20 ERa breast site for each gene represents one of three binding patterns: cancer tissues. 5¶ proximal for Wnt11, intronic for Adora1, and 3¶ proximal for We selected three prototypical E2-regulated genes, each of SAPS2. The ERa-binding site located 40 kb 5¶ of Wnt11 was not which represented a distinct ERa-binding pattern: (a)40kb5¶ proximal to another gene. On the other hand, the ERa-binding site (Wnt11), (b) intron 2 (Adora1), and (c)2kb3¶ (SAPS2; Fig. 1; within intron 2 of Adora1 is also 13 kb 5¶ upstream of the Myogenin Table 2). We investigated in vivo regulation of these three genes by gene, which is regulated by E2 in MCF-7 cells. The ERa target site ERa using regression analyses. The mRNA levels of Wnt11, located 2 kb 3¶ downstream of the SAPS2 gene is also 1 kb 3¶ a Adora1, and SAPS2 were plotted against ER mRNA levels. downstream of the SBF1 gene and is also regulated by E2. Regression analysis showed statistically significant correlations Individual ERa-binding site/luciferase reporter constructs were between ERa mRNA and Wnt11 mRNA (r = 0.567), ERa mRNA transfected into ERa+ MCF-7 cells, which were then treated with a and Adora1 mRNA (r = 0.609), and ER mRNA and SAPS2 mRNA vehicle or E2 for 20 h and assayed for luciferase reporter gene (r = 0.792; P < 0.00005 for each r value; Fig. 2). These findings activity. The ERa-binding site 40 kb 5¶ of Wnt11 enhanced trans- indicate that E2-responsive genes that lie proximal to ERa-binding criptional activity by 3.7-fold in the sense direction and 5.0-fold in

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 ERaÀ 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.

www.aacrjournals.org 5021 Cancer Res 2007; 67: (10). May 15, 2007

Figure 3. Transcriptional activities of ERa-binding sites proximal to E2-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 ERaÀ MDA-MB-231 cells (B) in 24-well plates. After 20 to 24 h, cells were washed with PBS and treated with À7 10 mol/L E2 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.

À7 the antisense direction on treatment with E2 (10 mol/L; Fig. 3A). 24-h treatment with E2 or vehicle. To show the efficiency and Similar results were obtained with transfection of luciferase reporter specificity of depletion of Wnt11, both real-time PCR and Western constructs containing ERa-binding sites within intron 2 Adora1 blot analysis were done (Fig. 4D). and 2 kb 3¶ of SAPS2 genes. E2 treatment enhanced transcription As shown in Fig. 4, knockdown of Wnt11 in MCF-7 cells treated by 4.1-fold and 3.7-fold in cells transfected with ERa-Adora1 sense with or without E2 resulted in a significantly decreased cell survival and antisense constructs, respectively. Cells transfected with ERa- (12.2% and 13.9%, respectively) compared with cells transfected SAPS2 sense construct increased transcription by 6.5-fold on with control siRNA (5.3% and 6.7%, respectively; P < 0.05 for each E2 treatment, whereas transcriptional activity was augmented by case; Fig. 4A). We next determined whether this change in cell 16-fold on E2 treatment of cells transfected with ERa-SAPS2 survival was due to changes in apoptosis and/or proliferation. antisense construct. Knockdown of Wnt11 increased apoptosis as shown specifically by To show that E2 regulated the transcriptional activity of these a striking increase in PARP cleavage (Fig. 4B). The highest level of constructs in an ERa-dependent fashion, we transfected them into PARP cleavage was observed in Wnt11-depleted cells incubated aÀ the ER MDA-MB-231 cell line. The absence of induction by E2 with vehicle followed by Wnt11-depleted cells treated with E2. The supported our conclusion that these sequences confer responsive- results suggested that Wnt11 depletion increased apoptosis, and E2 ness to E2 in an ERa-dependent manner (Fig. 3B). Overall, these blunted this effect. On the other hand, depletion of Wnt11 by observations confirmed that ERa-binding regions identified for siRNA did not result in significant changes in the fractions of Wnt11, Adora1, and SAPS2 contain E2-responsive functional MCF-7 cells in S phases compared with control samples (Fig. 4C). cis-acting elements. However, significant changes in cell populations in S phases were Knockdown of E2-induced Wnt11 is associated with in- observed in an E2-dependent manner, suggesting that Wnt11 does creased breast cancer cell death. We showed that Wnt11 is not have a major role in cell cycle progression in MCF-7 cells. Thus, located proximally to an ERa-binding site and its expression is the observed changes in cell viability seem to be due to an effect of induced by E2 (Supplementary Table S2). Because Wnt11 belongs Wnt11 on apoptosis but not on proliferation. to an oncogene family (18), we hypothesized that E2-induced Wnt11 expression could be involved in breast cancer pathobiology. Discussion MCF-7 cells were cultured in steroid-deprived medium for 3 days, Using a ChIP-linked target site cloning strategy, we identified 94 and siRNA against Wnt11 or control siRNA was transiently ERa-binding sites within the human genome, many of which transfected into the MCF-7 cells for 48 h followed by 20- to represent novel E2 targets in MCF-7 breast cancer cells. We showed

Cancer Res 2007; 67: (10). May 15, 2007 5022 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2007 American Association for Cancer Research. ERa-Binding Sites and Estradiol Target Genes both in vivo and in vitro that a representative portion of these genes A significant volume of work has focused on identifying essential are highly regulated by E2/ERa in breast cancer cells and tissues. We domains within the proximal promoters of E2-regulated genes also showed a novel biological role of the E2-regulated gene Wnt11 in (6, 12, 13, 19–24). However, many functionally relevant binding sites breast cancer. 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 island- enriched 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 E2 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 E2 (25, 29). E2 has been shown to regulate transcription through either direct Figure 4. siRNA-mediated inhibition of Wnt11 in MCF-7 leads to increased cell binding of EREs or indirectly by interacting with transcription death. MCF-7 cells were cultured in hormone-depleted medium for 3 days, and a siRNA against Wnt11 or control siRNA was transiently transfected into the MCF-7 factor complexes. Analysis of the ER -binding sequences associated À7 cells for 48 h. The cells were then stimulated with E2 (10 mol/L) or vehicle for with the 38 identified E2-regulated genes revealed that 61% of these 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 sequences contained at least one classic (palindromic) ERE. Eighty- cleavage with rabbit anti-human cleaved PARP antibody (B). Lanes 1 and 2, control seven percent of binding sites missing a palindromic ERE contained siRNA or Wnt11 siRNA-transfected MCF-7 cells treated with vehicle; lanes 3 and 4, À7 one or more AP-1 sites. The comparison of these results with control siRNA or Wnt11 siRNA-transfected MCF-7 cells treated with E2 (10 mol/L). Blotswerereprobedwithah-actin antibody to control for loading. Analysis previously published data suggests that there is a higher likelihood of cell cycle progression after silencing of Wnt11 (C). Knockdown efficiency and of the occurrence of canonical EREs in ERa-binding sites proximal 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 to an E2-regulated gene (4, 6). Furthermore, in the absence of an experiments; bars, SE. *, P <0.05,t test, statistically significant differences. ERE, AP-1 or other cis-acting elements may confer E2/ERa www.aacrjournals.org 5023 Cancer Res 2007; 67: (10). May 15, 2007

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2007 American Association for Cancer Research. Cancer Research responsiveness possibly via tethering of ERa on AP-1–binding tumor growth. Further investigation of the biological functions of transcription factors (19). Wnt11 and other identified ERa-regulated genes in breast cancer It has been shown that Wnt11 signaling induces proliferation may lead to the development of specific, targeted treatments. (30), transformation (31), and prostate cancer progression (32) through a noncanonical pathway (33–35). However, the role of Wnt11 in the progression of breast cancer remains unknown. Our Acknowledgments studies provide an initial insight into the mechanism by which Received 10/6/2006; revised 1/31/2007; accepted 3/15/2007. Wnt11 may promote tumor progression in the breast. Up- Grant support: NIH grant RO1-CA67167. The costs of publication of this article were defrayed in part by the payment of page regulation of Wnt11 mRNA on E2 exposure may activate the charges. This article must therefore be hereby marked advertisement in accordance Wnt11 signaling pathway and inhibit apoptosis, thus favoring with 18 U.S.C. Section 1734 solely to indicate this fact.

References 13. Shang Y, Hu X, DiRenzo J, Lazar MA, Brown M. 24. Metivier R, Penot G, Hubner MR, et al. Estrogen Cofactor dynamics and sufficiency in estrogen receptor- receptor-a directs ordered, cyclical, and combinatorial 1. David G, DeNardo H-TK, Susan Hilsenbeck, Valerie regulated transcription. Cell 2000;103:843–52. recruitment of cofactors on a natural target promoter. Cuba, Anna Tsimelzon, Powel H. Brown. Global gene 14. Dubik D, Shiu RP. Transcriptional regulation of c- Cell 2003;115:751–63. expression analysis of estrogen receptor transcription myc oncogene expression by estrogen in hormone- 25. Stevens TA, Iacovoni JS, Edelman DB, Meech R. factor cross talk in breast cancer: identification of responsive human breast cancer cells. J Biol Chem 1988; Identification of novel binding elements and gene estrogen-induced/activator protein-1-dependent genes. 263:12705–8. targets for the homeodomain protein BARX2. J Biol Mol Endocrinol 2006;19:362–4. 15. Giamarchi C, Solanas M, Chailleux C, et al. Chromatin Chem 2004;279:14520–30. 2. Pettersson K, Gustafsson JA. Role of estrogen receptor structure of the regulatory regions of pS2 and cathepsin D 26. Wang Q, Carroll JS, Brown M. Spatial and temporal h in estrogen action. Annu Rev Physiol 2001;63:165–92. genes in hormone-dependent and -independent breast recruitment of androgen receptor and its coactivators 3. Kun Y, How LC, Hoon TP, et al. Classifying the cancer cell lines. Oncogene 1999;18:533–41. involves chromosomal looping and polymerase tracking. estrogen receptor status of breast cancers by expression 16. Soulez M, Parker MG. Identification of novel Mol Cell 2005;19:631–42. profiles reveals a poor prognosis subpopulation exhibit- oestrogen receptor target genes in human ZR75-1 breast 27. Yan PS, Chen CM, Shi H, et al. Dissecting complex ing high expression of the ERBB2 receptor. Hum Mol cancer cells by expression profiling. J Mol Endocrinol epigenetic alterations in breast cancer using CpG island Genet 2003;12:3245–58. 2001;27:259–74. microarrays. Cancer Res 2001;61:8375–80. 4. O’Lone R, Frith MC, Karlsson EK,Hansen U. Genomic 17. Klinge CM, Jernigan SC, Mattingly KA, Risinger KE, 28. Wells J, Farnham PJ. Characterizing transcription targets of nuclear estrogen receptors. Mol Endocrinol Zhang J. Estrogen response element-dependent regula- factor binding sites using formaldehyde crosslinking 2004;18:1859–75. tion of transcriptional activation of estrogen receptors a and immunoprecipitation. Methods 2002;26:48–56. 5. Gottlicher M, Heck S, Herrlich P. Transcriptional and h by coactivators and corepressors. J Mol Endocrinol 29. Impey S, McCorkle SR, Cha-Molstad H, et al. Defining cross-talk, the second mode of steroid hormone 2004;33:387–410. the CREB regulon: a genome-wide analysis of transcrip- receptor action. J Mol Med 1998;76:480–9. 18. Clement G, Braunschweig R, Pasquier N, Bosman FT, tion factor regulatory regions. Cell 2004;119:1041–54. 6. Carroll JS, Liu XS, Brodsky AS, et al. Chromosome- Benhattar J. Alterations of the Wnt signaling pathway 30. Ouko L, Ziegler TR, Gu LH, Eisenberg LM, Yang VW. wide mapping of estrogen receptor binding reveals long- during the neoplastic progression of Barrett’s esopha- Wnt11 signaling promotes proliferation, transformation, range regulation requiring the forkhead protein FoxA1. gus. Oncogenel 2006;25:3084–92. and migration of IEC6 intestinal epithelial cells. J Biol Cell 2005;122:33–43. 19. DeNardo DG, Kim HT, Hilsenbeck S, et al. Global Chem 2004;279:26707–15. 7. Carroll JS, Meyer CA, Song J, et al. Genome-wide gene expression analysis of estrogen receptor transcrip- 31. Christiansen JH, Monkley SJ, Wainwright BJ. Murine analysis of estrogen receptor binding sites. Nat Genet tion factor cross talk in breast cancer: identification of WNT11 is a secreted glycoprotein that morphologically 2006;38:1289–97. estrogen-induced/activator protein-1-dependent genes. transforms mammary epithelial cells. Oncogene 1996;12: 8. Bajic VB, Tan SL, Chong A, et al. Dragon ERE Finder Mol Endocrinol 2005;19:362–78. 2705–11. version 2: a tool for accurate detection and analysis of 20. Watanabe T, Inoue S, Hiroi H, et al. Isolation of 32. Zhu H, Mazor M, Kawano Y, et al. Analysis of Wnt estrogen response elements in vertebrate genomes. estrogen-responsive genes with a CpG island library. gene expression in prostate cancer: mutual inhibition by Nucleic Acids Res 2003;31:3605–7. Mol Cell Biol 1998;18:442–9. WNT11 and the androgen receptor. Cancer Res 2004;64: 9. Irizarry RA, Hobbs B, Collin F, et al. Exploration, 21. Laganiere J, Deblois G, Lefebvre C, et al. From the 7918–26. normalization, and summaries of high density oligonucle- cover: location analysis of estrogen receptor a target 33. Mulholland DJ, Dedhar S, Coetzee GA, Nelson CC. otide array probe level data. Biostatistics 2003;4:249–64. promoters reveals that FOXA1 defines a domain of the Interaction of nuclear receptors with the Wnt/h- 10. Keeton EK, Brown M. Cell cycle progression stimu- estrogen response. Proc Natl Acad Sci U S A 2005;102: catenin/Tcf signaling axis: Wnt you like to know? lated by tamoxifen-bound estrogen receptor-a and 11651–6. Endocr Rev 2005;26:898–915. promoter-specific effects in breast cancer cells deficient 22. Mao DY, Watson JD, Yan PS, et al. Analysis of Myc 34. Pandur P, Lasche M, Eisenberg LM, Kuhl M. in N-CoR and SMRT. Mol Endocrinol 2005;19:1543–54. bound loci identified by CpG island arrays shows that Wnt-11 activation of a non-canonical Wnt signal- 11. Eisenberg CA, Gourdie RG, Eisenberg LM. Wnt-11 is Max is essential for Myc-dependent repression. Curr ling pathway is required for cardiogenesis. Nature expressed in early avian mesoderm and required for the Biol 2003;13:882–6. 2002;418:636–41. differentiation of the quail mesoderm cell line QCE-6. 23. Hall JM, McDonnell DP, Korach KS. Allosteric 35. Koyanagi M, Haendeler J, Badorff C, et al. Non- Development 1997;124:525–36. regulation of estrogen receptor structure, function, and canonical Wnt signaling enhances differentiation of 12. Dubik D, Shiu RP. Mechanism of estrogen activation coactivator recruitment by different estrogen response human circulating progenitor cells to cardiomyogenic of c-myc oncogene expression. Oncogene 1992;7:1587–94. elements. Mol Endocrinol 2002;16:469–86. cells. J Biol Chem 2005;280:16838–4.

Cancer Res 2007; 67: (10). May 15, 2007 5024 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2007 American Association for Cancer Research. Novel Estrogen Receptor-α Binding Sites and Estradiol Target Genes Identified by Chromatin Immunoprecipitation Cloning in Breast Cancer

Zhihong Lin, Scott Reierstad, Chiang-Ching Huang, et al.

Cancer Res 2007;67:5017-5024.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/67/10/5017

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2008/05/19/67.10.5017.DC1

Cited articles This article cites 35 articles, 10 of which you can access for free at: http://cancerres.aacrjournals.org/content/67/10/5017.full#ref-list-1

Citing articles This article has been cited by 7 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/67/10/5017.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/67/10/5017. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.