Genome-wide activity of unliganded estrogen receptor-α in breast cancer cells Livia Caizzia,b,c, Giulio Ferreroa,d, Santina Cutrupia,d, Francesca Corderoa,e, Cecilia Ballaréc,f, Valentina Mianoa,d, Stefania Reinerib, Laura Riccib, Olivier Friarda, Alessandro Testoria, Davide Coràa,g, Michele Casellea,h, Luciano Di Crocec,f,i, and Michele De Bortolia,d,1 aCenter for Molecular Systems Biology, University of Turin, 10043 Orbassano, Turin, Italy; bBioindustry Park Silvano Fumero, 10010 Colleretto Giacosa, Turin, Italy; cDepartment of Gene Regulation, Stem Cells and Cancer, CRG Center for Genomic Regulation, 08003 Barcelona, Spain; dDepartment of Clinical and Biological Sciences, University of Turin, 10043 Orbassano, Turin, Italy; eDepartment of Computer Science, University of Turin, 10149 Turin, Italy; fUniversitat Pompeu Fabra, 08018 Barcelona, Spain; gDepartment of Oncology, Institute for Cancer Research and Treatment (IRCC), University of Turin, 10060 Candiolo, Turin, Italy; hDepartment of Physics, University of Turin, 10125 Turin, Italy; and iInstitució Catalana de Recerca i Estudis Avançats (ICREA), 08010 Barcelona, Spain Edited* by Michael G. Rosenfeld, University of California at San Diego, La Jolla, CA, and approved February 20, 2014 (received for review August 19, 2013) Estrogen receptor-α (ERα) has central role in hormone-dependent rearrangements (14), made it difficult to evaluate ERα binding in breast cancer and its ligand-induced functions have been extensively cells treated with vehicle alone. As a consequence, most authors characterized. However, evidence exists that ERα has functions that have dismissed the question of hormone-independent binding as are independent of ligands. In the present work, we investigated the compromised peak calling or as unspecific background (15, 16). binding of ERα to chromatin in the absence of ligands and its func- However, especially for the clinical problem concerning the re- tions on gene regulation. We demonstrated that in MCF7 breast can- sponse to aromatase inhibitors, identifying possible ERα geno- cer cells unliganded ERα binds to more than 4,000 chromatin sites. mic actions in the absence of ligands would be very relevant. In Unexpectedly, although almost entirely comprised in the larger group this work, we first identified bona fide genomic ERα binding sites α of estrogen-induced binding sites, we found that unliganded-ER in the absence of estrogen in breast cancer cells. We then eval- binding is specifically linked to genes with developmental functions, uated the effect of ERα silencing on gene transcription and compared with estrogen-induced binding. Moreover, we found that α CELL BIOLOGY α binding of pioneer factors, demonstrating that ER controls ge- siRNA-mediated down-regulation of ER in absence of estrogen is nomic activity by binding to several chromatin sites independently accompanied by changes in the expression levels of hundreds of of estrogen exposure. Thus, unliganded ERα may participate, coding and noncoding RNAs. Down-regulated mRNAs showed enrich- together with other factors, in the definition of the chromatin ment in genes related to epithelial cell growth and development. Stable ERα down-regulation using shRNA, which caused cell growth landscape of hormone-dependent breast cancer cells. α arrest, was accompanied by increased H3K27me3 at ER binding Results sites. Finally, we found that FOXA1 and AP2γ binding to several sites Unliganded ERα Cistrome in MCF7 Cells. α is decreased upon ERα silencing, suggesting that unliganded ERα To identify ER binding in participates, together with other factors, in the maintenance of the the absence of estrogen, MCF7 cells were maintained in hor- luminal-specific cistrome in breast cancer cells. mone-depleted (HD) medium, transfected with control (siCTR) or ERα-specific (siERα) siRNA and subjected to chromatin chromatin binding | transcriptome | enhancer | pioneer factors | immunoprecipitation followed by high-throughput sequencing epigenetics Significance strogen receptor-α (ERα) expression in breast cancer defines Ethe Luminal A phenotype, which represents the subset of Estrogen receptor-α (ERα) is a key protein in breast cancer and tumors that are responsive to endocrine treatments. Spontane- treatments targeting ERα are among the most widely used and ous or experimentally induced (1) loss of ERα elicits growth effective in clinics. Although the role of estrogen-stimulated arrest or epithelial to mesenchymal transition in vitro, whereas ERα in breast cancer has been exhaustively described, the estrogen withdrawal from culture media, albeit reducing pro- functions of ERα in the absence of estrogen is hill-defined. In liferation rate, has no such effect. These data suggest that loss of this work, we show that ERα binds extensively to the genome ERα does not equal depletion of estrogen. ERα is a DNA- of breast cancer cells in the absence of estrogen, where it binding, ligand-activated transcription factor, but it can be acti- regulates the expression of hundreds of genes endowed vated in absence of ligands by diverse mechanisms, especially by with developmental functions. Our data suggest that ERα phosphorylation through different pathways, including protein has a fundamental role in the homeostasis of luminal epi- kinase A, mitogen-activated protein kinases, and others (ref. 2 thelial cells also when estrogen is ablated physiologically or and references therein). Ligand-independent activity of ERα was pharmacologically. reported by several groups on individual genes (3–5). Genome- wide ERα binding in the absence of estrogen was also described Author contributions: L.D.C. and M.D.B. designed research; L.C., G.F., S.C., F.C., and V.M. performed research; C.B., S.R., L.R., and M.C. contributed new reagents/analytic tools; L.C., in breast cancer cells acquainted with growing in hormone- G.F., S.C., F.C., C.B., O.F., A.T., and D.C. analyzed data; and L.C., G.F., L.D.C., and M.D.B. depleted media (6–8) and in mouse uterus (9). These data suggest wrote the paper. that ERα may have a wide genomic function in breast cancer cells The authors declare no conflict of interest. independent of its ligands. Estrogen response in breast cancer *This Direct Submission article had a prearranged editor. cells was extensively characterized in terms of chromatin Freely available online through the PNAS open access option. binding and gene-expression regulation using both cell lines and Data deposition: The data reported in this paper have been deposited in the Gene Ex- human tumor biopsies. In vitro models were especially useful pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE53533). α because they allowed correlating ER -binding events, which are 1To whom correspondence should be addressed. E-mail: [email protected]. Man- rarely located at gene promoters, with gene-expression data (10– uscript number: 201315445. 13). In these studies, the experimental setting, together with the This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. fact that breast cancer cell lines show a high grade of genomic 1073/pnas.1315445111/-/DCSupplemental. www.pnas.org/cgi/doi/10.1073/pnas.1315445111 PNAS Early Edition | 1of6 Downloaded by guest on September 26, 2021 (ChIP-seq) using antibodies against ERα or IgG as control. may represent “residual binding” after estrogen deprival. Using Analysis of ERα enrichment over IgG in siCTR conditions evi- ChIP-quantitative PCR (qPCR), we verified that apo-ERα binding denced 4,232 unliganded ERα binding sites (apo-ERα binding to several sites was stable up to 12 d in HD medium (Fig. S1E), thus sites, aERBS) (P < 1e-05). These sites were almost entirely excluding simple estrogen carryover when cells were switched to contained in the ERα cistrome reported in MCF7 cells cultured HD medium. Furthermore, using GREAT analysis (18), we found in full medium (FM-ERBS) or after 17β-estradiol (E2) treatment that aERBS lie close to genes associated with development, cell (E2-ERBS) (Fig. 1A) (15, 17). Accordingly, aERBS showed ge- differentiation, and morphogenesis, whereas E2-ERBS and FM- nomic distribution similar to estrogen-induced events, with in- ERBS,notincommonwithaERBS,showedenrichmentinme- creased prevalence of intergenic location (Fig. S1A). tabolism, lipid metabolism and biosynthesis terms (Datasets S1 and To verify the specificity of the signal, we examined how siERα, S2). This difference was clearly shown by semantic analysis of the which reduced ERα protein level by 80% (Fig. S1B), affected associated Gene Ontology (GO) terms (19), as shown in Fig. 2A. these binding events. ChIP signal was strongly reduced upon Thus, this result suggests that ERα chromatin binding in absence of ERα knockdown (Fig. 1B and Fig. S1C), confirming that these hormone has different functions than estrogen-induced binding. are bona fide ERBS in the absence of hormone. Comparison of Transcription factor binding sites (TFBS) analysis confirmed ERα binding enrichment in siCTR over siERα allowed ranking that apo-ERα binding is most likely facilitated by cooperating aERBS by significance (Fig. 1B) and this unraveled diversity factors, as previously shown for liganded ERα (11, 12, 20). aERBS among aERBS. Analysis of top 25% aERBS revealed a higher are frequently accompanied by forkhead box protein A1 (FOXA1/ average number of reads and a full estrogen-response element HNF3A), activating enhancer binding protein 2 gamma (AP2γ/ (fERE) as the most represented motif at peak center