Oncogene (2008) 27, 1019–1032 & 2008 Nature Publishing Group All rights reserved 0950-9232/08 $30.00 www.nature.com/onc ONCOGENOMICS A genome-wide study of the repressive effects of on estrogen receptor alpha signaling in breast cancer cells

C Williams, K Edvardsson, SA Lewandowski, A Stro¨ m and J-A˚ Gustafsson

Department of Biosciences and Nutrition at Novum, Karolinska Institutet, Huddinge, Sweden

Transcriptional effects of estrogen result from its activa- mediated via two estrogen receptor (ER) isoforms, that tion of two estrogen receptor (ER) isoforms; ERa that is, ER alpha (ERa) and ER beta (ERb) (Hanstein et al., drives proliferation and ERb that is antiproliferative. 2004). The second ER isoform, ERb, was discovered Expression of ERb in xenograft tumors from the T47D some 10years ago (Kuiper et al., 1996). ERa and ERb breast cancer cell line reduces tumor growth and are homologous, especially in their DNA-binding angiogenesis. If ERb can halt tumor growth, its introduc- domains (97%), but they differ in their ligand-binding tion into cancers may be a novel therapeutic approach to domains (59% identity) and in their transcriptional the treatment of estrogen-responsive cancers. To assess activating function-1 (AF-1) domains. Estrogen is non- the complete impact of ERb on transcription, we have selective for the two receptors and the binding of made a full transcriptome analysis of ERa- and ERb- estrogen results in a receptor–ligand complex that binds mediated regulation in T47D cell line with Tet-Off with high affinity to estrogen responsive elements regulated ERb expression. Of the 35 000 and (EREs) on DNA. Transcriptional regulation by the transcripts analysed, 4.1% (1434) were altered by ERa ERs may occur through a direct interaction of ERs with activation. Tet withdrawal and subsequent ERb expres- EREs or through an interaction of ER with other sion inhibited the ERa regulation of 998 genes and, in transcription factors (Sp1, AP-1 and NF-kB; McDon- addition, altered expression of 152 non-ERa-regulated nell and Norris, 2002). The two receptors can hetero- genes. ERa-induced and ERb-repressed genes were dimerize, and they may modulate each other’s effects involved in proliferation, steroid/xenobiotic metabolism (Gustafsson, 2006). ERb often behaves as an antagonist and ion transport. The ERb repressive effect was further to ERa (Nilsson et al., 2001). The basis of this confirmed by proliferation assays, where ERb was shown antagonism has been shown to be due to the reduction to completely oppose the ERa–E2 induced proliferation. in ERa level and reduced recruitment of the Additional analysis of ERb with a mutated DNA-binding activating protein-1 complex (Matthews et al., 2006). domain revealed that this mutant, at least for a quantity of Detection of ERa protein in a breast tumor helps to genes, antagonizes ERa even more strongly than ERb wt. identify those breast cancer patients who may respond From an examination of the genes regulated by ERa and to hormonal intervention, and measurement of ERa has ERb, we suggest that introduction of ERb may be an become standard in the clinical management of breast alternative therapeutic approach to the treatment of cancer. However, only about 50% of ERa-expressing certain cancers. tumors respond well to hormonal therapy. ERb is Oncogene (2008) 27, 1019–1032; doi:10.1038/sj.onc.1210712; expressed at high concentration in normal human published online 13 August 2007 mammary gland tissue and expression is lost or decreased in breast cancer (Palmieri et al., 2002; Keywords: estrogen receptor; microarray; breast cancer; Esslimani-Sahla et al., 2005). When it is reintroduced gene regulation into breast cancer cells, ERb is antiproliferative (Paruthiyil et al., 2004; Strom et al., 2004; Murphy et al., 2005; Hartman et al., 2006). Several investigators have used the breast cancer cell ONCOGENOMICS Introduction line MCF-7 to evaluate the transcriptional effect of ERa in response to estrogen (Frasor et al., 2003, 2004; Estrogen is involved in the regulation of the reproduc- Buterin et al., 2006), and shown that many genes tive, immune, cardiovascular, musculo/skeletal and associated with the control of , proliferation central nervous systems. The effects of estrogens are and apoptosis are regulated by ERa. Few reports have focused on ERb, in terms of global transcriptional effects, and no direct transcriptional targets of ERb Correspondence: C Williams, Department of Biosciences and Nutrition have been convincingly described. In an effort to at Novum, Karolinska Institutet, Ha¨ lsova¨ gen 7-9, 14157 Huddinge, elucidate the gene regulatory function of ERb and the Sweden. E-mail: [email protected] mechanisms behind its suggested anti-tumorigenic role, Received 16 January 2007; revised 19 June 2007; accepted 5 July 2007; we performed a full transcriptome analysis of the cell published online 13 August 2007 line used in our recently reported experimental model of Role of ERb in breast cancer cells C Williams et al 1020 ERb-dependent reduction of breast tumors (Strom changes caused by ERa and ERb et al., 2004; Hartman et al., 2006). We identify clear We found that the activation of ERa alone by E2 differences between gene regulation by ERa and ERb, (Comparison I, referred to as the ‘ERa profile’) resulted respectively, and report previously unknown targets of in the differential expression of 1 434 out of the analysed estrogen regulation. Our data provide further insight 35 000 transcripts. Correlations of M-values (2 log of into the interplay between the two receptors and the FC) between the replicated arrays are shown in antiproliferative actions of ERb. Figure 1d, where the arrow in Comparison (I) indicates the maximally ERa-regulated gene TFF1/pS2, FC 34.8 (confirmed with real-time PCR as upregulated 301 times). Of the 50strongest regulated genes, 40(80%) Results and discussions were upregulated. Overall the proportion of upregulated genes was 63% (897 out of 1 434 genes). When the cells We have recently reported on the use of the breast expressed both ERa and ERb (TetÀ culture), E2 cancer cell line T47D with a tetracyclin (Tet) responsive treatment affected 588 genes (Comparison IV). The element regulating ERb expression to study the role of pS2 gene was now upregulated only 4.4-fold (confirmed ERb in tumors formed from T47D cells. In the absence with real-time PCR as upregulated 53 times). The ERb- of Tet, ERb was induced in these tumors resulting in the mediated negative effect on this gene is in line with reduction of tumor size and inhibition of angiogenesis previously published data (Matthews et al., 2006). Venn (Hartman et al., 2006). To investigate the mechanisms diagram in Figure 2a further shows that Tet withdrawal by which ERb opposes the growth of ERa-positive and the subsequent ERb expression more or less tumors, we analysed the genome-wide transcriptional inhibited ERa regulation of 998 genes and, in addition, effects of ERb induction in T47D breast cancer cells. We altered expression of 152 genes not regulated when ERa examined the response of the cells to estrogen in the was the only ER expressed. The remaining 436 ERa- presence of ERa alone or ERa and ERb together, and regulated genes were still differentially expressed when the effects of induction of ERb. In the presence of Tet, ERb was present at high levels. Selective induction of ERa is the predominant receptor expressed and follow- ERb (Comparison II), excluding gene regulation possi- ing Tet withdrawal, ERb is induced to ERa at a ratio of bly inherent to the Tet-Off system as determined by approximately 4:1 (Strom et al., 2004). A mock T47D control analysis, strongly affected the expression of 196 Tet-Off PBI control was analysed to define non-ERb- genes as a result of ERb expression. Among the 50most related gene expression inherent in the Tet-Off model affected genes, 37 (74%) were downregulated following used. A 240-fold increase in ERb transcript levels was ERb expression. Thus, ERa induces and ERb negatively observed after Tet withdrawal in 17b-estradiol (E2) modulates a majority of their regulated genes. Tables 1 treated cultures; the relative ERb mRNA levels as and 2 show the top regulated genes of the ‘ERa profile’ analysed by real-time PCR are shown in Figure 1a. and ‘ERb profile’, respectively. Observation of ERb protein co-expressed with green We used the classification and the fluorescent protein (GFP) confirms that there was a EASE package for a two-step functional analysis to corresponding increase from undetectable to clearly identify biological themes that were overrepresented visible levels of induced protein in the cells (Figure 1b). among the differentially expressed genes (a Gene To estimate ERa-induced gene expression, E2-treated Ontology annotation was assigned to approximately cells without the expression of ERb were compared to an 50% of the regulated genes). The most overrepresented equivalent cell line treated with the antiestrogen ICI gene group of the ‘ERa profile’ was the upregulated 182780(ICI) (Comparison I: E2 T47D Tet þ compared to genes within the ‘cell cycle’ (EASE score 4.2 e–034). ICI T47D Tet þ ). To obtain ERb-induced alterations, Here, 101 genes were significantly upregulated upon E2 three separate studies were performed comparing genes stimulation; of these 53 belonged to the subgroup altered upon the introduction of ‘active’ ERb (Compar- ‘regulation of cell cycle’ and 11 to the subgroup of ‘cell ison II: E2 T47D TetÀ compared to E2 T47D Tet þ ); cycle checkpoint’ (BRCA2, TP53, BUB1, BUB1B, genes altered upon the introduction of ‘inactive’ ERb BUB3, CCNA2, CHEK1, MAD2L1, RBBP8, TTK (Comparison III: ICI T47D TetÀ compared to ICI T47D and GTSE1). Only one ‘cell cycle checkpoint’ gene Tet þ ); and genes altered upon E2 treatment in cultures (CCNG2) was downregulated by ERa. Other over- expressing both ERa and ERb (Comparison IV: E2 T47D represented upregulated gene groups were ‘cytokinesis’ TetÀ compared to ICI T47D TetÀ). The design of the and ‘sterol biosynthesis’. Among the genes downregu- microarray experiment is shown in Figure 1c. Agreement lated by ERa, the groups of ‘morphogenesis’, ‘cell withrespecttofoldchange(FC)andP-values was communication’, ‘protein amino acid dephosphoryla- observed in several cases where independent probes tion’, ‘cell adhesion’, ‘oncogenesis’ and ‘apoptosis’ were represented different regions of the same gene, validating overrepresented. For the ERb-regulated genes, the the accuracy of the method. Differential expression of ‘regulation of cell proliferation’ gene group was the regulated genes was confirmed with real-time PCR of 37 most downregulated one (EASE score 0.006) followed randomly selected genes. A selection of the results is by ‘transition metal ion homeostasis’ and ‘xenobiotic presented in Tables 1–4, and Figures 1–3; complete results metabolism’. Overrepresented gene groups that were are available as Supplementary Infomation and deposited upregulated following the ERb induction belonged to in ArrayExpress (E-MEXP-969). ‘negative regulation of ’, ‘energy pathways’,

Oncogene Role of ERb in breast cancer cells C Williams et al 1021

Figure 1 Estrogen receptor beta (ERb) expression and microarray analysis of the T47D ERb Tet-Off cell line. (a) Real-time PCR analysis of ERb expression under four different culture conditions; Tet þ and ICI, TetÀ and ICI, Tet þ and E2, TetÀ and E2. Relative expression is compared to expression in Tet þ and ICI culture. (b) Picture of cells (upper panel) and green fluorescent protein (GFP) detection (lower panel) of Tet þ and TetÀ cultured cells treated with E2. (c) Design of microarray experiment of the four conditions analysed. Direct comparisons were performed in a loop-wise design. Comparison (I) measures ERaÀE2 gene regulation; (II) measures regulation due to the induced expression of ERb in active state (E2); (III) measures regulation due to the induced expression of ‘inactive’ ERb (ICI); (IV) measures ERa/ERb–E2 regulation. (d) Correlation of M-values (2 log of fold change) between replicates of corresponding analysis. Gray arrow (in I) indicates the most prominent ERa-induced gene pS2 (TFF1). Black arrows (in Comparisons II and III) indicate the probes corresponding to induced ERb (3–4 probes).

‘lipid metabolism’, ‘ion homeostasis’ and ‘apoptosis’. induction of ERb. Overrepresentation analysis of this Details of overrepresented groups are shown in Table 4. group shows that the biological context where the opposing effect of ERb was predominant was in ‘cell cycle’, ‘xenobiotic metabolism’ and ‘ion transport’. The ERb compared to ERa gene regulation reverse pattern, where genes strongly reduced by ERa Comparison I in relation to Comparison IV (Venn (upon E2 stimulation) were considerably increased as a diagram in Figure 2a) indicated that 998 less genes were consequence of ERb expression, was seen in a smaller estrogen regulated when Tet was withdrawn and ERb set of 12 transcripts (for example, TP53INP1, NSE2, and ERa were co-expressed, compared to when ERa RIN2, CPEB4, PSAP, LIPH, PDCD6IP and SEP1). was expressed alone. However, a full transcriptome ERb was synergized with ERa in the regulation of comparison has a relatively large degree of uncertainty transcription of only 10genes; ER a was one such gene. and, although it clearly indicates that ERb opposes ERa All genes are listed in Supplementary Information. regulation of many genes, it is not sufficient for a complete comparative evaluation. To investigate the action of ERb in relation to that of ERa in detail, we Genes uniquely regulated by ERb have studied how the different receptor isoforms Totally 104 genes or transcript variants were scored as regulate one and the same gene. Of the genes in the being regulated by ERb only (Figure 2b), and after ERb profile, 92 out of the 196 genes (Comparison II) eliminating border-line ERa-regulated genes (genes were also regulated by ERa (Figure 2b). Of these, 76% possibly regulated by ERa but below cut-off), 49 genes were upregulated by ERa and reduced upon the listed in Table 3 remained as exclusively ERb-regulated.

Oncogene Role of ERb in breast cancer cells C Williams et al 1022 Table 1 The 35 genes most regulated by E2 activation of ERa Gene symbol Gene name GO biological process FC comparison

I II III IV ERa ERb–E2 ERb–ICI ERa/b

Upregulated by ERa TFF1 Trefoil factor 1 (pS2) Cell growth and/or maintenance; defense 301I 0.2I 0.5I 53I response; digestion GREB1II GREB1 protein Biological role unknown 8.3 0.6 1.0 4.7 PKIB Protein kinase (cAMP-dependent, Negative regulation of protein kinase 6.3 1.01.0 4.4 catalytic) inhibitor beta activity CXCL12 Chemokine (C-X-C motif) ligand 12 Calcium ion homeostasis; cell adhesion; 5.6 0.5III 1.0 1.8 (SDF-1) cell–cell signaling IGFBP4 Insulin-like growth factor binding Regulation of cell growth; signal 5.5 0.7 1.0 2.8 protein 4 transduction; skeletal development SIAH2 Seven in absentia homolog 2 Development 17I 0.8I 0.5I 28I HSPB8 Heat-shock 22 kDa protein 8 Biological role unknown 4.1 0.8 0.8 3.1 KCNK5II Potassium channel, subfamily K, Excretion; potassium ion transport 4.0 0.4 n/a 1.9 member 5 IL20Interleukin 20 Proinflammatory cytokine 248I 0.1I 0.3I 117I SLC26A2II Solute carrier family 26 (sulfate Sulfate transport 5.1I 0.6I 1.0I 3I transporter), member 2 UGT2B17 UDP glycosyltransferase 2 family, Steroid metabolism; xenobiotic metabolism 3.3 0.4 n/a 1.3 polypeptide B17 CCNA2 Cyclin A2 Mitotic G2 checkpoint 7.9I 0.9I 0.2I,III 33I TUBA1 Tubulin, alpha, ubiquitous Microtubule-based movement 3.2 0.8 0.9 3.0 ITPK1 Inositol 1,3,4-triphosphate 5/6 kinase Signal transduction 3.2 0.7 1.2 1.9 HMGB2 High-mobility group box 2 DNA packaging; regulation of transcription 3.2 0.6III 0.6III 3.2

OLFM1II Olfactomedin 1 Neurogenesis 3.1 0.4III 1.0 1.6 BUB1 BUB1 budding uninhibited by Mitotic spindle checkpoint; protein amino 3.1 0.7 0.6III 4.1 benzimidazoles 1 homolog (yeast) acid phosphorylation Q8TF51 KIAA1950protein Proteolysis and peptidolysis 3.0 1.1 0.9 2.8 C19orf14 19 open reading frame 14 3.0 0.7 0.7 3.3 CDC20CDC20celldivision cycle 20homologRegulation of cell cycle; - 3.0 0.6 0.7 2.6 dependent protein catabolism SLC7A5 Solute carrier family 7 (cationic amino Amino acid metabolism; amino acid 3.0 n/a 1.0 2.1 acid transporter, y+ system), member 5 transport RNASEH2A Ribonuclease H2, large subunit DNA replication; RNA catabolism 2.9 0.8 0.7 3.2 – Similar to ubiquitin-conjugating 2.9 0.8 0.9 2.3 E2S UBE2S Ubiquitin-conjugating enzyme E2S Ubiquitin cycle 2.9 0.6 1.0 1.9 THBS1 Thrombospondin 1 Blood coagulation; cell adhesion; 2.9 0.6 1.0 1.9 neurogenesis C10orf3 Chromosome 10 open reading frame 3 Cytoskeleton organization and biogenesis 2.9 0.6III 0.5III 3.4 TPD52L1 Tumor protein D52-like 1 Oncogenesis 2.8 0.8 0.9 2.8 PPM1E Protein phosphatase 1E (PP2C domain Protein amino acid dephosphorylation 2.8 0.7 1.1 1.9 containing) C20orf129 Chromosome 20 open reading frame 129 2.8 0.8 0.7 2.8

Downregulated by ERa TP53INP1 Tumor protein inducible nuclear 0.1I 2.9I 0.8I 0.5I protein 1 EDG3 Endothelial differentiation, sphingolipid Cytosolic calcium ion concentration 0.3 0.8 0.5III 0.7 G-protein-coupled receptor, 3 elevation; positive regulation of cell proliferation ERBB2 v-erb-b2 erythroblastic leukemia viral Cell proliferation; oncogenesis; 0.3 1.4 1.0 0.5 oncogene homolog 2, neuro/glioblastoma transmembrane receptor protein tyrosine derived oncogene homolog (avian) kinase signaling pathway CYP1A1 Cytochrome P450, family 1, subfamily A, Electron transport 0.3 0.7 0.5 0.5 polypeptide 1 ZFP36L2 Zinc-finger protein 36, C3H type-like 2 Cell proliferation 0.4 0.8 0.5 0.6 EFNB2 Ephrin-B2 Cell–cell signaling; neurogenesis 0.2I 2.5I 0.8I 0.6I

Abbreviation: n/a, not applicable. Genes are listed according to the ranking in array. Columns I, II, III and IV show fold-change values for the four different comparisons, that is, expression changes due to (I) E2 activation of ERa, (II) ERb induction in the presence of E2, (III) ERb induction in the presence of ICI and (IV) E2 activation of ERa/b. Fold change indicates results by array, except when real-time PCR has been performed (indicated by I). Two or more probes corresponding to the same gene (same or different splice variants) showing similar results are indicated by II. Differential expression above cut-off is indicated by bold FC values; an array determined FC between B0.7 and 1.3 indicates essentially unchanged genes. III indicates that changes may be partly or fully related to the Tet-Off system and not due to ERb expression (indications from control analysis).

Oncogene Role of ERb in breast cancer cells C Williams et al 1023 Table 2 The 35 genes most regulated by induction of ERb Gene symbol Gene name GO biological process FC comparison

I II III IV IERa ERb–E2 ERb–ICI ERa/b

Upregulated by the introduction of ERb expression ESR2II Estrogen receptor 2 (ERb) Negative regulation of cell growth; regulation 0.8I 323I 120I 2.0I of transcription LRRC15 Leucine-rich repeat-containing 15 — 1.4I 51I 0.4I 162I UGT1A9 UDP glycosyltransferase 1 family, Metabolism 1.1 3.8 n/a n/a polypeptide A9 APOD Apolipoprotein D Lipid metabolism; transport 0.8I 14I 1.5I 16.6I RIN2 Ras and Rab interactor 2 Endocytosis; neuropeptide signaling pathway 0.6 2.2 1.2 0.9 HMGCL 3-Hydroxymethyl-3-methylglutaryl- Amino acid metabolism; energy pathways 0.9 1.9 1.4 1.1 Coenzyme A lyase (hydroxymethylglutaricaciduria) KIF21A Kinesin family member 21A — 1.2 1.9 0.9 2.2 PLOD2 Procollagen-lysine, 2-oxoglutarate Protein modification 1.5 1.9 0.9 2.6 5-dioxygenase (lysine hydroxylase) 2 P2RY2 Purinergic receptor P2Y, G-protein G-protein signaling; cell ion homeostasis 1.2I 2.0I 1.5I 1.6I coupled, 2 NSE2 Breast cancer membrane protein 101 — 0.7 1.8 1.2 0.8 DKFZP434C212 DKFZP434C212 protein Phosphoenolpyruvate-dependent 1.8 1.2 0.9 sugar phosphotransferase system 1.0 QSCN6 Quiescin Q6 Negative regulation of cell proliferation 0.7I 3.0I 1.0I 2.1I Downregulated by introduction of ERb expression TFF1 Trefoil factor 1 (pS2) Cell growth and/or maintenance; defense 301I 0.2I 0.5I 53I response; digestion IL20Interleukin 20 Proinflammatory cytokine 248I 0.1I 0.3I 117I TMEM46 Transmembrane protein 46 — 0.7 0.3 0.5 0.8 SLC26A2 II Solute carrier family 26 (sulfate Sulfate transport 5.1I 0.6I 1.0I 3I transporter), member 2 BCL2II B-cell CLL/lymphoma 2 Antiapoptosis; humoral immune response; 6.3I 0.3I 0.4I 4.5I regulation of cell cycle KCNK5II Potassium channel, subfamily K, Excretion; potassium ion transport 4.0 0.4 n/a 1.9 member 5 UGT2B17 UDP glycosyltransferase 2 family, Steroid metabolism; xenobiotic metabolism 3.3 0.4 n/a 1.3 polypeptide B17 CXCR7 Chemokine (C-X-C motif) receptor 7 G-protein-coupled receptor protein signaling 0.8 0.4 0.7 0.7 pathway; HSPD1II Heat shock 60kDa protein 1 Regulation of apoptosis; response to unfolded 2.2 0.5 0.6 1.7 (chaperonin) protein CLDN1 Claudin 1 Cell adhesion 0.6 0.5 n/a 0.5 DHRS2 II Dehydrogenase/reductase (SDR Carbohydrate metabolism 2.0 0.5 1.01.4 family) member 2 UGT2B28 UDP glucuronosyltransferase 2 Estrogen metabolism; xenobiotic metabolism 2.5 0.5 1.1 1.4 family, polypeptide B28 S100A6 S100 calcium-binding protein A6 Axonogenesis; cell–cell signaling; positive n/a 0.5I 0.8 0.6 (calcyclin) regulation of fibroblast proliferation; regulation of cell cycle SSRP1 Structure-specific recognition protein 1 Regulation of transcription 1.5 0.5 0.9 1.3 GFRA1 GDNF family receptor alpha 1 Cell surface receptor linked signal transduction 1.6 0.5 0.8 1.4 FABP5 Fatty acid binding protein 5 Epidermal differentiation; lipid metabolism; 2.2 0.5 1.0 1.7 (psoriasis associated) transport SLC16A1 Solute carrier family 16, member 1 Mevalonate transport 2.2 0.5 1.3 1.5 TFF3 (intestinal) Defense response; digestion 2.6 0.5 1.2 1.4 MT1F Metallothionein 1F (functional) Copper ion homeostasis 1.4 0.6 0.9 n/a GREB1II GREB1 protein Biological process unknown 5.5 0.6 1.0 3.9 TMEM107 Transmembrane protein 107 — 1.6 0.6 0.8 1.0 EDN1 Endothelin 1 Cell–cell signaling; positive regulation of cell 0.9I 0.6I 0.7I 0.7I proliferation; regulation of transcription MYC v-myc myelocytomatosis viral Cell cycle arrest; iron ion homeostasis; 22I 0.7I 1.5I 10I oncogene homolog (avian) regulation of transcription from Pol II promoter

Abbreviation: n/a, not applicable. Genes are listed according to the ranking in array. Columns I, II, III and IV show fold-change values for the four different comparisons, that is, expression changes due to (I) E2 activation of ERa, (II) ERb induction in the presence of E2, (III) ERb induction in the presence of ICI and (IV) E2 activation of ERa/b. Fold change indicates results by array, except when real-time PCR has been performed (indicated by I). Two or more probes corresponding to the same gene (same or different splice variants) showing similar results is indicated by II. Differential expression above statistical cut-off indicated by bold FC values; an array determined FC between B0.7 and 1.3 may be considered as indicating essentially unchanged genes.

Oncogene Role of ERb in breast cancer cells C Williams et al 1024 Table 3 The genes regulated only by the induction of ERb Gene symbol Gene name GO biological process FC

Upregulated by ERb exclusively LRRC15 Leucine-rich repeat-containing 15 52I UGT1A9 UDP glucuronosyltransferase 1 Xenobiotic metabolism 3.8 family, polypeptide A9 APOD Apolipoprotein D Lipid metabolism; transport 14I HMGCL 3-Hydroxymethyl-3-methylglutaryl- Amino acid metabolism; energy pathways 1.9 Coenzyme A lyase KIF21A Kinesin family member 21A Microtubule-based movement 1.9 P2RY2 Purinergic receptor P2Y, G-protein G-protein signaling; cell ion homeostasis 2.0I coupled, 2 DKFZP761D0211 Hypothetical protein Cell growth; cell–matrix adhesion 1.8 DKFZp761D0211 DKFZP434C212 DKFZP434C212 protein Phosphoenolpyruvate-dependent sugar phosphotransferase system 1.8 QSCN6 Quiescin Q6 Negative regulation of cell proliferation 3.0I PTK9L PTK9L protein tyrosine kinase 9-likeIntracellular signaling cascade 1.7 OXR1 oxidation resistance 1 Cell wall catabolism 1.7 ANXA9 A9 Skeletal development 1.7 J00191 1.7 SEPT9 9 Negative regulation of progression through cell cycle; protein 1.7 heterooligomerization LOC114990Hypothetical protein BC013767 1.7 UQCRFS1 Ubiquinol-cytochrome c reductase, Electron transport 1.6 Rieske iron-sulfur polypeptide 1 ALG14 Asparagine-linked 14 1.6 homolog GSTT2 Glutathione S-transferase theta 2 Response to stress 1.6 KCTD11 Potassium channel tetramerization Potassium ion transport 1.6 domain containing 11 NDRG3 NDRG family member 3 Cell differentiation; negative regulation of cell growth 1.6 CYB5R3 cytochrome b-5 reductase Cholesterol biosynthesis; energy pathways; iron ion transport 1.6 STAU Staufen, RNA-binding protein Development 1.6 RG9MTD2 RNA (guanine-9-) methyltransferase 1.6 domain containing 2 STK3 Serine/threonine kinase 3 Apoptosis; protein amino acid phosphorylation; signal transduction 1.5 PIGK Phosphatidylinositol glycan, class K Proteolysis and peptidolysis 1.5 HIG1 Likely ortholog of mouse hypoxia Phosphoenolpyruvate-dependent sugar phosphotransferase system 1.5 induced gene 1 NRIP3 Nuclear receptor interacting protein 3 1.5 LOC283932 Hypothetical protein LOC283932 1.5 JJAZ1 Joined to JAZF1 Chromatin modification; regulation of transcription 1.5 LDHC Lactate dehydrogenase C Glycolysis 1.5

Downregulated regulated by ERb exclusively CBX5 Chromobox homolog 5 (HP1 alpha Chromatin assembly/disassembly 0.5 homolog, Drosophila) S100A6 S100 calcium binding protein A6 Axonogenesis; cell–cell signaling; regulation of cell cycle 0.5I (calcyclin) EDN1 Endothelin 1 Cell–cell signaling; positive regulation of cell proliferation; regulation of 0.6I transcription IRA1 Likely ortholog of mouse IRA1 0.6 protein SIPA1L2 Signal-induced proliferation- 0.6 associated 1 like 2 ADM Adrenomedullin cAMP biosynthesis; cell–cell signaling; excretion; progesterone biosynthesis; 0.4I regulation of transcription; response to wounding PANK2 Pantothenate kinase 2 Coenzyme A biosynthesis 0.6 RPL21 Ribosomal protein L21 Protein biosynthesis; ribosome biogenesis 0.6 MGC5306 Hypothetical protein MGC5306 Development 0.6 FIBL-6 Hemicentin Protein amino acid phosphorylation 0.6 MRPS16 Mitochondrial ribosomal protein S16Protein biosynthesis 0.7

ERb-regulated genes not showing any tendency to be regulated by E2 activation of ERa (and not affected by Tet withdrawal in the mock T47D PBI cell line). Fold change indicates results by array, except when real-time PCR has been performed (indicated by I; FC denote comparison TetÀ E2 versus Tet+ E2). Only differentially expressed genes above cut-off are listed. Seventy-one percent of the genes within this subgroup upregulated and the cell–cell signaling genes (ADM (35/49 genes) were upregulated as a consequence of ERb and EDN1) as downregulated. Two of the strongest expression. Here, we especially note the genes involved ERb specifically upregulated genes were apolipoprotein in ‘energy pathways’ of cholesterol biosynthesis and D (APOD) and leucine-rich repeat-containing 15 lipid metabolism (LDHC, HMGCL and CYB5R3) as (LRRC15), confirmed by real-time PCR as 14- and

Oncogene Role of ERb in breast cancer cells C Williams et al 1025 Table 4 The most regulated biological gene groups Differential expression Biological processes Genes EASE score

ERaupregulated Mitotic cell cycle 802.6 e–040 Cell proliferation 116 1.3 e–024 Cell cycle checkpoint 11 0.00000071 Cytokinesis 17 0.0000015 Ribosome biogenesis and assembly 14 0.000002 Response to endogenous stimulus 23 0.000032 Sterol biosynthesis 7 0.00072 ERadownregulated Morphogenesis 34 0.0076 Cell communication 80 0.0092 Protein amino acid dephosphorylation 8 0.018 Intracellular signaling cascade 27 0.024 Organogenesis 28 0.038 Muscle development 7 0.058 Cell motility 12 0.059 Development 46 0.077 Cell adhesion 17 0.12 Oncogenesis 4 0.13 ERb upregulated Negative regulation of cell growth 2 0.048 Energy pathways 3 0.10 Lipid metabolism 4 0.16 Ion homeostasis 2 0.20 Apoptosis 3 0.27 Signal transduction 9 0.3 ERb downregulated Regulation of cell proliferation 6 0.006 Transition metal ion homeostasis 3 0.007 Regulation of cell cycle 7 0.009 Xenobiotic metabolism 3 0.03 Metal ion homeostasis 3 0.03 Serine family amino acid metabolism 2 0.09 Positive regulation of cell proliferation 3 0.10

Enriched biological processes of the differentially expressed genes, as defined by Gene Ontology. ERa -regulated genes denotes genes affected by E2 in T47D Tet+ culture (predominantly expressing ERa), Comparison I; ERb-regulated genes denote genes affected by ERb induction in T47D TetÀ culture (compared to T47D Tet+ culture, both treated with E2 24 h), Comparison II.

70-fold increased, respectively (Figure 3b). These genes by ERa and decrease of S100A6/calcyclin by ERb were further analysed in Tet þ and TetÀ cultures at correlates to the earlier findings of in vivo gene several time intervals after E2 treatment, where these expression in wt and ER knockout mouse bone tissue genes were shown to be strongly induced 3 h after E2 (Lindberg et al., 2003). treatment, with a maximum response at 18–48 h Since other array analyses of ERa action have been (Figure 3b, right). published previously, we compared our results with another study (Frasor et al., 2003). It should be noted that the experiments compared were performed using Comparisons with literature different cell lines (MCF-7 versus our T47D, both breast Several previously known ERa-regulated genes were cancer cell lines predominantly expressing ERa but with detected in this study, for example, TFF1/pS2 (FC 34.8), large differences in chromosomal abnormalities and GREB1 (FC B8 for three transcript variants), SIAH 2 mutations). In our study, E2 treatment was compared to (FC 5.3), SDF1/CXCL12 (FC 5.6), IGFBP-4 (FC 5.5), treatment with ICI, which was different from the CCNA2 (FC 4.2), STC2 (FC 2.3), PR (FC 1.9), RBBP-8 protocol employed by Frasor et al. (2003), and different (FC 1.7), NRIP1/RIP140(FC 1.7), CTSD (FC 1.6) and technology platforms were utilized (Affymetrix Gene- NR5A2/LRH1 (FC 1.7), all confirming the data Chip versus Operon’s oligomer spotted array). Compari- previously published (Roberts et al., 1988; Inoue et al., sons can only be made between annotated genes present 2002; Hall and Korach, 2003; Annicotte et al., 2005; on both arrays and with comparable annotations (7597 Frasor et al., 2005). Also downregulated genes such as genes). Venn diagrams of the comparisons are shown in ERa itself (FC 0.6), EFNB2 (FC 0.4), CYPIA1 (FC 0.3), Figures 2c and d. Of 45 genes upregulated by ERa (at IL1R1 (FC 0.4), ERBB2 (FC 0.5) and CCNG2 (FC 0.4) 24 h) in (Frasor et al., 2003), we confirm that 19 genes are have previously been shown or indicated to be down- significantly upregulated also in T47D (24 h E2 treat- regulated by estrogen in breast or other tissues ment), and for most of the remaining 25 genes we still (Nikolova et al., 1998; Ricci et al., 1999; Inoue et al., observe an upregulation, but below our cut-off. Further- 2002; Frasor et al., 2004; Schaefer et al., 2005; Matthews more, we describe an additional 585 upregulated genes, et al., 2006; Stossi et al., 2006). Furthermore, the including known genes (for example, pS2, SIAH2, estrogen stimulated the increase of IGFBP4 and LIG1 GREB1 and BCL2) as well as a large number of

Oncogene Role of ERb in breast cancer cells C Williams et al 1026

Figure 2 Comparisons of estrogen receptor (ER) profiles. (a) Our study. Overlap between ‘ERa profile’ (differentially expressed genes from Comparison I) and ‘ERa/ERb profile’ (differentially expressed genes from Comparison IV) after 24 h E2 treatment. (b) Overlap between ‘ERa profile’ and ‘ERb profile’ (differentially expressed genes from Comparison II) after 24 h E2 treatment. Overlapping 92 genes are differently regulated, and bar below shows the type of regulation by ERa and ERb (up- or downregulated). (c) Comparison between ERa profiles of our T47D data set and previously published of MCF-7 cells (Frasor et al., 2003), both 24 h E2 treatment, upregulated genes (left) and downregulated genes (right), respectively.

previously unreported genes. For only one gene we show in the transcription of many genes, especially those a contradictory downregulation in our study (CCBP2). active in ‘cell proliferation’ and ‘ion homeostasis’. Of 93 downregulated genes in MCF-7 cells, 19 are also Similar to our findings, these authors observed that downregulated in our T47D cells; again we see indica- the induction of ERb itself, without E2 treatment, tions for many of the remaining genes to be down- elicited changes in gene expression. In the study by regulated but below the cut-off. There are two genes Chang et al. (2006) no complete lists of regulated genes where our results differ (IER3/IEX1, which we have were published, but for several of the genes described in confirmed by real-time PCR, and LDLR) and we show the article we confirm the regulation by ERb (for 324 more genes to be downregulated. example, THBS1 and CLDND1). We also observed a Taken together, correlation with literature and con- similar downregulation of other genes (for example, firmation by real-time PCR support the high consistency CXCL12/SDF1), but where the regulation could not be of our data, and we present a large number of novel differentiated from the Tet-Off system itself and, ERa-regulated genes. consequently, these genes were excluded from our ERb Previously, few, if any, studies of the transcriptional profile. For other genes, we could not confirm the results profile of ERb in breast cancer cells have been of Chang et al. (2006) (for example, BIRC3, IL17RB, published, but recently Chang et al. (2006) published a CSRP2, BMP7, ERBB2, CDC25A, FOXM1, E2F1, first analysis of transcriptional changes of genes induced CXCL1, 2, 10and 20).Other differences between the by ERb in breast cancer cell line MCF-7. This study is two studies are as follows: (1) expression of 16 genes similar to ours, the major differences being the use of involved in the TGFb pathway was affected by E2 different cell lines (MCF-7 versus T47D), the mode of treatment but we confirmed the downregulation of only induction of ERb expression (adenoviral gene delivery two of these genes (TGFb3 and SMAD6) by ERa and versus Tet-Off construct), the ratio of ERb to ERa an opposing regulation of THBS1. Furthermore, in our (approximately 1:3 in Chang et al. (2006) versus our study TGFb2 was upregulated by ERa; (2) semaphorins ratio of 4:1) as well as the use of ICI (not used by Chang and two of their receptors were regulated in the Chang et al., 2006) and the type of technology (Affymetrix et al. (2006) study, but no semaphorins detected in GeneChip versus Operon’s oligomer spotted array). our study (sema3B, 3C, 3F, 4F, 4G and 6A1) were Despite these differences, several major conclusions are changed by ERb (sema3C was upregulated by ERa); (3) identical. In both studies ERb was found to oppose ERa Chang et al. (2006) found 44 genes to be specifically

Oncogene Role of ERb in breast cancer cells C Williams et al 1027

Figure 3 Real-time PCR analysis confirms the microarray data. (a) Expression analysis of 10genes with upregulated expression after 24 h E2 treatment, when estrogen receptor alpha (ERa) is the predominant ER expressed (Tet þ culture). Relative expression of each gene is set to 1 when cells are cultured in ICI. (b) Left: Expression analysis of two genes (APOD and LRRC15) found to be exclusively regulated by ERb (TetÀ culture) and not affected by ERa (Tet þ culture), and corresponding expression in mock cell line (Tet þ and TetÀ culture, striped bars). Relative expression of each gene is set to 1 when cells are cultured in Tet þ and ICI. Right: Their respective mRNA induction followed over time, 1–48 h after E2 addition in separate experiment.

upregulated by ERb and we detect 49 genes; however, development and genomic instability (Lo et al., 2005) there is no similarity in this gene set in the two studies. and PTTG1 is overexpressed in breast cancer, induces Some of the differences may be related to the cell lines, aneuploidy and is associated with tumor metastasis mode of ERb delivery, levels of ERb, effects induced by (Vlotides et al., 2007). ICI and to the differences in the presence of genes on the Further evidence of the proliferative actions of ERa is array. that the most strongly ERa-downregulated gene was the cell cycle repressor TP53INP1, a downregulation that was opposed by ERb. Increase of TP53INP1 leads to Antiproliferative and antitumorigenic actions of ERb cycle arrest in G1 and enhanced p53-mediated apoptosis The majority of the genes whose expression we found to (Tomasini et al., 2005). In addition, several antiproli- be increased following the introduction of ERb oppose ferative genes (QSCN6, NDRG3, SEPT9, KCTD11 and proliferation or invasiveness of breast cancer. Examples STK3) are affected by ERb alone, strongly supporting are the three strongly ERa upregulated genes, which are the notion that ERb is antiproliferative and capable of all breast cancer related: pS2, BCL2 and GREB1 (Real inhibiting or reducing the growth of tumors. Subse- et al., 2002; Kang et al., 2005; Rae et al., 2005). The quently, several cyclins are changed by the ERs; strong ERb negative influence on the ERa upregulation CCNA2, CCNB1, CCNB2, CCND1 and CCNF are of these genes would most likely help in the reduction of all upregulated by ERa and opposed by Tet withdrawal tumor proliferation. In addition, the cell cycle–related induced ERb expression. genes MYC, NME2, SAP30, MATK, NOL1 and The effect of ERb expression on the proliferation was NOLC1 are upregulated by ERa and opposed by further demonstrated by proliferation assay. Here, ERb. MYC is a direct target of ERa transcription T47D breast cancer cells, Tet-Off ERb expressing clone (Park et al., 2005) and NME2 is a downstream target of as well as Tet-Off PBI expressing mock control were MYC. Decreasing MYC protein levels in MCF-7 cells treated with ICI and E2 and proliferation measured significantly inhibits tumor growth (Wang et al., 2005). after 5 days. E2 addition in both Tet þ cultures Two other oncogenes (AURUKA and PTTG1) were increases the proliferation significantly compared to found to be increased by ERa and opposed by ERb. ICI and non-treated cultures. Also E2 addition to TetÀ AURUKA has been associated with breast cancer PBI culture increases the proliferation, whereas E2

Oncogene Role of ERb in breast cancer cells C Williams et al 1028 IL-20was one of the strongest ER a-induced genes detected in this study (Figure 3a), and has not earlier been described as estrogen induced or breast cancer related, but its promoter has been determined as a target of ERa in MCF-7 breast cancer cells via genome-wide immunoprecipitation (Laganiere et al., 2005). IL-20 is a proinflammatory cytokine and a key player in the pathology of rheumatoid arthritis (Hsu et al., 2006), atherosclerosis (Chen et al., 2006) and psoriasis (Wang et al., 2006).

The mechanism by which ERb opposes ERa regulation We note a strong opposing effect on many ERa- regulated genes upon the introduction of ERb. This effect could be mediated in several ways; ERb down- regulation of ERa at the mRNA level, ERb may oppose Figure 4 Proliferation assay confirms estrogen receptor beta ERa at the protein level, for example, via heteroduplex (ERb) reduces proliferation. ERb Tet-Off T47D and mock PBI formation, ERb may compete with ERa for binding to Tet-Off T47D cells were synchronized and cultured in serum-free DNA sites (for example, EREs) or oppose ERa by media with and without Tet. After 5 days of incubation with occupying necessary cofactors. The basis for ERb’s respective treatment (ICI, E2 or no treatment) the cell viability was antagonism against ERa has previously been shown to assayed. Cells treated with ICI or no treatment did not increase in proliferation, E2 induced viability with approximately 50%, except be due to the reduction in ERa protein level and reduced when ERb was expressed and, apparently, abolished E2-induced recruitment of the activating protein-1 complex upon proliferation. ERb expression (Matthews et al., 2006). We confirm that ERb downregulates ERa at the mRNA level and, thus, some of the opposing effect on ERa signaling by ERb may be directly related to less expression of ERa. addition to TetÀ ERb-expressing culture induces no Also, ERa downregulates its own transcript upon E2 proliferation (Figure 4). This was replicated twice, in treatment and in addition, in our material, ERa strongly culture conditions with either 5 or 2% dextran-coated upregulates a splice variant of ERa lacking exon 7. This charcoal-treated FBS (DCC). Thus, the expression of variant is commonly increased in breast cancer tissue ERb at the levels used here appears to oppose ERa–E2 and acts in a dominant-negative way (Poola and Speirs, induced proliferation completely. 2001; Marshburn et al., 2004). This implies a double Furthermore, E2 regulates many genes involved in cell negative feedback loop for ERa at both the transcript adhesion, which plays a significant role for the (downregulation of wt variant) and protein level capability of tumors to metastasize. Our results show (upregulation of dominant-negative variant), whereas that ERa appears to reduce while ERb increases cell ERb opposes ERa by downregulating its wt variant but adhesion. ERa downregulates adhesion by decreasing does not induce the dominant negative ERa variant. In the levels of 17 transcripts (NCAM2, ALCAM, addition, we observe that LRH-1 (NR5A2), which L1CAM, LAMB2, SCARB2, THBS3, COL5A1, controls the expression of aromatase, is upregulated by CYR61, ITGB4, MLLT4, PTPRF, CCL2, ZYX, ERa and opposed by ERb, thus affecting local estrogen LRRN1, NTN4, ANTXR1 and CLDN1). CLDN1 is production, which for breast tumors in vivo is thought to one of the few genes downregulated by both ERa and be the most important estrogen source (Russo and ERb, as was also discussed by Chang et al. (2006). Loss Russo, 2006). of expression of this gene has been suggested to play a In an effort to elucidate the mechanism behind the role in invasion and metastasis of breast cancer (Tokes ERb-mediated opposing effect on ERa signaling, we et al., 2005). ERb upregulates two adhesion genes expressed an ERb protein with a mutated DNA-binding (COL6A1 and ANNEXIN A9). Furthermore, ERa domain (DBDm) in T47D cells using lentivirus delivery, upregulates adhesion genes linked to metastasis or to compare the effect to that with ERa and ERb wt. oncogenic potential (CD44, RET, CXCL12/SDF-1 Here, we see that genes strongly upregulated by ERa and THBS1) and an additional five genes involved in and opposed by ERb wt expression (such as pS2 with cell adhesion (ITGA6, MICB, BYSL, TROAP and known ERE-binding and SPINK4 with several half-site TPBG) of which ERb opposes THBS1. In addition, EREs within the 5 kb promoter area) are even more PDCD6IP reported to have roles in regulating adhesion opposed by ERb DBDm (Figure 5). Also for genes was increased by ERb and reduced by ERa. Possibly, strongly downregulated by ERa (TP53INP1), this effect ERb may change adhesion properties, as well as is modulated by ERb wt but completely inhibited by metastasis potential, of breast cancer cells. ERb DBDm (Figure 5). In addition, the ERa upregulated IL-20, shown to From this experiment, we conclude that the opposing promote angiogenesis in endothelial cells (Hsieh et al., effect of ERb can be accentuated when ERb lacks the 2006), was not upregulated when ERb was expressed. ability to bind to DNA. The ERb DBDm may act as a

Oncogene Role of ERb in breast cancer cells C Williams et al 1029

Figure 5 Effect of estrogen receptor beta (ERb) wt and ERb DNA-binding domain (DBDm) in T47D breast cancer cells via lentivirus delivered expression. Genes detected as oppositely regulated in the Tet-Off ERb T47D system were further analysed in T47D parental cells infected with ERb wt, ERb DBDm and mock, respectively. Opposing effects by ERb wt are clear, both for the ERa upregulated genes pS2 and SPINK4 and for the ERa downregulated gene TP53INP1. When ERb DBDm is introduced, the ERa regulation is even further antagonized. dominant negative on ERa, possibly by heterodimeriz- novel. Furthermore, results from the expression of a ing with ERa and preventing the heterodimer from DBD-mutated ERb indicates that the ERa opposing binding to EREs. When ERb wt, on the other hand, action of ERb can be accentuated if ERb cannot bind to forms a heterodimer with ERa the heterodimer can still DNA. It should be noted that this report describes bind to EREs and regulate the genes, although not with changes related to 24 h of E2 activation. It is possible the same efficiency as the ERa homodimer (explaining that some immediate regulatory effects, by this time, why ERb wt negatively modulates, but does not inhibit have disappeared. ERa transcription). The lower efficiency may depend on In conclusion, ERb opposes ERa at the transcriptome reduced transactivating functions, in line with the results level and in so doing exhibits antiproliferative actions as by Gougelet et al. (2007), where the transactivation well as suppression of potent proinflammatory cyto- function-1 is implied as the keystone of ERb-mediated kines. Since ERb is expressed in small amounts along transcriptional repression of ERa. For other genes, with ERa in approximately 70% of breast cancers however, other mechanisms may be important, for (Kurebayashi et al., 2000), breast cancer therapy example, for MYC (upregulated by ERa) where we involving specific ERb agonists may prove to be an saw a similar reduction by both ERb wt and ERb interesting complement in future therapies. DBDm and for PR (upregulated by ERa) where we did not see any strong opposing effect by ERb (neither in the Tet-Off expression model nor using lentivirus Materials and methods delivery) nor by ERb DBDm at this time-point (24 h). Cell culture Transfections of T47D cell line and cell culture have previously been described by Strom et al. (2004) and Hartman et al. Conclusions (2006). Cells were cultured in DMEM/F12 mixed (1:1) medium supplemented with 5% FBS. For synchronization the medium Uncovering the contribution of ERb in estrogenic was changed to phenol red-free DMEM/F12 mixed (1:1) regulation of breast tumor growth is important for medium supplemented with 5% DCC for 24 h; the serum was understanding and treatment of this disease. In the then reduced to 0.5% DCC and 10 nM ICI 182780was added. present study, we describe the effect at the transcriptome Tet was withdrawn in half of the plates 12 h before the start of level of introducing ERb into ERa-expressing breast treatment. Tet þ and TetÀ cultures were treated with either 10n M of the pure ER antagonist ICI or 10n M of ER ligand E2 cancer cells, using a model shown earlier to reduce (17b-estradiol) for 24 h after which all cells were collected xenograft tumor growth (Hartman et al., 2006). We simultaneously for RNA extraction. RNA was extracted using show that the induced ERb expression opposes the ERa TRIzol precipitation. This was subsequently repeated with a driven upregulation of cell proliferation genes, which we mock-control T47D stably transfected with an empty Tet-Off functionally confirm by proliferation assays. ERb also PBI construct. affects several other processes and genes; not all of these actions of ERb oppose ERa and some transcripts Lentivirus vectors and infection of T47D cells appear to be uniquely regulated by ERb. In addition, Follow-up experiments to study the effect of DBD-mutated we present many novel genes not previously reported as ERb were performed using lentivirus vectors and infection regulated by ERa. For ERb regulation there are very of T47D cells. The plasmid pcDNA3-FLAG ERb was used as few published reports, thus most data presented here are a template for TOPO cloning into pLenti6/V5-D-TOPO

Oncogene Role of ERb in breast cancer cells C Williams et al 1030 according to the instructions (Invitrogen, Carlsbad, CA, Technology, Sweden), hybridized for 35–40h at 42 1C, washed USA). The pLenti6/V5-D-FLAG ERb DBDm was con- and scanned at 10-mm resolution using the G2565BA DNA structed from pLenti6/V5-D-FLAG ERb using Quick-Change microarray scanner (Agilent Technologies, Palo Alto, CA, (Stratagene, La Jolla, CA, USA) according to the manufac- USA) photo multiplier tube set to 100. The obtained Tiff turer’s instructions. Forward oligo ATCACTATGGAG images were analysed using the GenePix Pro 6.0software TCTGGTCGTTG CA GCATGTAAGGCCTTTTTTAAAA (Axon Instruments, MDS Analytical Technologies, Sunnyvale, GA changing E167 and G168 to A. CA, USA). All data analysis steps were performed in the R Lentivirus was produced with the ViraPower Lentivirus environment for statistical computing and programming Expression system. The titer of the virus was estimated according essentially as previously described (Richter et al., 2006). to the instructions (Invitrogen). T47D cells were spread on six- Differentially expressed genes were identified using an well plates at a density of 200 000 cells/well. The next day empirical Bayes moderated t-test (Smyth, 2004) ranking the lentivirus at 2 MOI was added in 1 ml of growth media genes according to significance; cut-off for differential expres- supplemented with 6ı`g of polybrene. After 24 h at 371C the cells sion was set to B-value >0.0 (an FC of at least 1.4, and were washed and 2 ml of normal growth media was added. P-value o0.005), and confirmed by real-time PCR analysis Blasticidine to 5 mg/ml was added after another 24 h and the cells of genes close to this cut-off. In addition, genes regulated by were then incubated for another 5 days at 371C. Non-infected the Tet-Off system, as determined from the mock control, were dead cells were washed off and the remaining cells were removed from the ERb profile. The raw data and detailed trypsinized and spread onto 100 mm plates. One well/100 mm protocols are available from the ArrayExpress data repository plate in 10ml of phenol red-free media was supplemented with using the accession number A-MEXP-571 and E-MEXP-969. 5% DCC, after 24 h the media was changed to 0.5% DCC cells Classification into Gene Ontology functional groups (Harris incubated at 371Cfor24hafterwhich10nM 17b-estradiol was et al., 2004) and analysis of overrepresented themes were added to half of the plates and the cells were incubated at 371C performed using the Expression Analysis Systematic Explorer for 24 h. A parallel mock infection was performed, where (EASE) package (Hosack et al., 2003). The complete human lentivirus without ERb construct was used. Cells were harvested transcriptome was used for calculation of the expected in 1 ml TRIzol/100 mm plate; RNA was extracted with chloro- frequencies in the overrepresentation analysis, and a Gene form and cleaned using Qiagen RNeasy spin columns with on- Ontology theme was considered overrepresented if the column DNase I digestion. Infection procedure (mock, ERb wt calculated EASE score was below 0.3. and ERb DBDm) was repeated twice on different occasions, Gene expression changes caused by the Tet-Off system once with subsequent ICI versus E2 treatment and once with (T47D transfected with a Tet-Off PBI empty vector) were non-treatment control versus E2. characterized under the same conditions as above, but to exclude all possibly regulated genes, the cut-off was lowered to B À4. All these genes were excluded from the comparisons Cell proliferation assay o and analyses, unless real-time PCR proved them to be Cells were plated at 2500 cells/well in 96-well plates and cultured significantly changed only in the ERb-expressing cell line. in 2% DCC and 1 mg/ml Tet. The next day cells were synchronized using 10n M ICI 182780for 24 h, the following day cells were washed once with PBS and the medium was Real-time PCR changed to include respective treatments (Tet 1 mg/ml, E2 10n M RNA was isolated as described above and DNA degraded using and ICI 10n M). After 5 days of incubation the cell viability was DNase I, 1 mg of the total RNA was used for each cDNA assayed using MTS kit (Promega, Madison, WI, USA) synthesis. First strand cDNA was synthesized using Superscript (‘CellTiter 96 Aqueous Non-Radioactive Cell Proliferation III and random hexamers (Invitrogen). Linear phase of Assay’). Assay was replicated with cells cultured in 5% DCC. logarithmic amplification was used for quantification, and cycle The absorbance was measured at 490nm. Results are presented number was compared between triplicate samples using cDNA in Figure 4 as percent of control (control is Tet þ ,ICI). template corresponding to 5 ng RNA per sample, 1 pmol of each forward and reverse primer and SYBR green PCR master mix in a total volume of 10 ml. Runs were performed using ABI 7500 Microarray experiment and analysis instrument (Applied Biosystems, Foster City, CA, USA) We used two-color comparative microarray technique, cover- according to the manufacturer’s instructions. The expression of ing the fully known transcriptome of 35 000 genes and variants genes was normalized to the expression of 18S, ARHGDIA and using Operon’s long-oligonucleotide spotted array. In addi- GAPDH. Relative expression and standard deviation were tion, the array was complemented with 133 oligos specifically calculated using DDCt formula. Primer pairs for 37 genes are synthesized for a detailed and robust analysis of nuclear detailed in Supplementary Data 2. receptors, splice variants and co-regulators. Each comparison was replicated and dye-swapped. In addition, we performed a Comparisons between studies using different microarray formats duplicated dye-swap comparison and real-time PCR controls To compare our results to published studies where Affymetrix of the mock control (Tet-Off PBI T47D) for comparison and GeneChips were utilized, we used the EASE software to match for identification of differential expression caused by the features across platforms. By providing Affy-ID to GeneID, a system by itself. direct comparison was possible to perform. Only features with Microarray analysis was performed essentially as described annotation present in GeneID database and present on both by Richter et al. (2006). For each cDNA synthesis, 10 mgof arrays are directly comparable (7597 genes). total RNA was used. Samples to be co-hybridized on a slide were pooled and direct comparisons were performed using replicated samples, cDNA synthesis and hybridizations with Acknowledgements the Cy5/Cy3 dye assignments reversed. Eight hybridizations (each comparison replicated) were performed (and an addi- We thank Dr Chunyan Zhao and Associate Professor Karin tional two for the mock control); design of the experiment is Dahlman-Wright for valuable discussions. This study was shown in Figure 1C. Operon’s human 35 K arrays were used supported by grants from the Swedish Cancer Fund and from (printed by the microarray facility at KTH, Royal Institute of KaroBio AB.

Oncogene Role of ERb in breast cancer cells C Williams et al 1031 References

Annicotte JS, Chavey C, Servant N, Teyssier J, Bardin A, Hsu Y-H, Li H-H, Hsieh M-Y, Liu M-F, Huang K-Y, Chin Licznar A et al. (2005). The nuclear receptor liver receptor L-S et al. (2006). Function of interleukin-20 as a proin- homolog-1 is an estrogen receptor target gene. Oncogene 24: flammatory molecule in rheumatoid and experimental 8167–8175. arthritis. Arthritis Rheum 54: 2722–2733. Buterin T, Koch C, Naegeli H. (2006). Convergent transcrip- Inoue A, Yoshida N, Omoto Y, Oguchi S, Yamori T, Kiyama tional profiles induced by endogenous estrogen and distinct R et al. (2002). Development of cDNA microarray for xenoestrogens in breast cancer cells. Carcinogenesis 27: expression profiling of estrogen-responsive genes. J Mol 1567–1578. Endocrinol 29: 175–192. Chang EC, Frasor J, Komm B, Katzenellenbogen BS. (2006). Kang H, Watkins G, Parr C, Douglas-Jones A, Mansel R, Impact of estrogen receptor beta on gene networks regulated Jiang W. (2005). Stromal cell derived factor-1: its influence by estrogen receptor alpha in breast cancer cells. Endocri- on invasiveness and migration of breast cancer cells in vitro, nology 147: 4831–4842. and its association with prognosis and survival in human Chen W-Y, Cheng B-C, Jiang M-J, Hsieh M-Y, Chang M-S. breast cancer. Breast Cancer Res 7: R402–R410. (2006). IL-20 is expressed in atherosclerosis plaques and Kuiper GGJM, Enmark E, Pelto-Huikko M, Nilsson S, promotes atherosclerosis in apolipoprotein E-deficient mice. Gustafsson J-A. (1996). Cloning of a novel estrogen receptor Arterioscler Thromb Vasc Biol 26: 2090–2095. expressed in rat prostate and ovary. Proc Natl Acad Sci USA Esslimani-Sahla M, Kramar A, Simony-Lafontaine J, Warner 93: 5925–5930. M, Gustafsson J-A, Rochefort H. (2005). Increased estrogen Kurebayashi J, Otsuki T, Kunisue H, Tanaka K, Yamamoto receptor {beta}cx expression during mammary carcinogen- S, Sonoo H. (2000). Expression levels of estrogen receptor- esis. Clin Cancer Res 11: 3170–3174. {alpha}, estrogen receptor-{beta}, coactivators, and core- Frasor J, Danes JM, Funk CC, Katzenellenbogen BS. (2005). pressors in breast cancer. Clin Cancer Res 6: 512–518. Estrogen down-regulation of the corepressor N-CoR: Laganiere J, Deblois G, Lefebvre C, Bataille AR, Robert F, mechanism and implications for estrogen derepression of Giguere V. (2005). From the cover: location analysis of N-CoR-regulated genes. Proc Natl Acad Sci USA 102: estrogen receptor {alpha} target promoters reveals that 13153–13157. FOXA1 defines a domain of the estrogen response. Proc Frasor J, Danes JM, Komm B, Chang KCN, Lyttle CR, Natl Acad Sci USA 102: 11651–11656. Katzenellenbogen BS. (2003). Profiling of estrogen up- and Lindberg MK, Moverare S, Skrtic S, Gao H, Dahlman-Wright down-regulated gene expression in human breast cancer K, Gustafsson J-A et al. (2003). Estrogen receptor (ER)- cells: insights into gene networks and pathways underlying {beta} reduces ER{alpha}-regulated gene transcription, estrogenic control of proliferation and cell phenotype. supporting a ‘Ying Yang’ relationship between ER{alpha} Endocrinology 144: 4562–4574. and ER{beta} in mice. Mol Endocrinol 17: 203–208. Frasor J, Stossi F, Danes JM, Komm B, Lyttle CR, Lo Y-L, Yu J-C, Chen S-T, Yang H-C, Fann CSJ, Mau Y-C Katzenellenbogen BS. (2004). Selective estrogen receptor et al. (2005). Breast cancer risk associated with genotypic modulators: discrimination of agonistic versus antagonistic polymorphism of the mitosis-regulating gene Aurora-A/ activities by gene expression profiling in breast cancer cells. STK15/BTAK. Int J Cancer 115: 276–283. Cancer Res 64: 1522–1533. Marshburn PB, Zhang J, Bahrani-Mostafavi Z, Mostafavi BZ, Gougelet A, Mueller SO, Korach KS, Renoir J-M. (2007). Marroum M-C, Mougeot J-LC et al. (2004). Estrogen Oestrogen receptors pathways to oestrogen responsive receptor-[alpha] messenger RNA variants that lack exon 5 or elements: the transactivation function-1 acts as the keystone exon 7 are coexpressed with wild-type form in human of oestrogen receptor (ER)[beta]-mediated transcriptional endometrium during all phases of the menstrual cycle. Am J repression of ER[alpha]. J Steroid Biochem Mol Biol 104: Obstet Gynecol 191: 626–633. 110–122. Matthews J, Wihlen B, Tujague M, Wan J, Strom A, Gustafsson J. (2006). ERb scientific visions translate to clinical Gustafsson J-A. (2006). Estrogen receptor (ER) {beta} uses. Climacteric 9: 156–160. modulates ER{alpha}-mediated transcriptional activation Hall JM, Korach KS. (2003). Stromal cell-derived factor 1, a by altering the recruitment of c-Fos and c-Jun to estrogen- novel target of estrogen receptor action, mediates the responsive promoters. Mol Endocrinol 20: 534–543. mitogenic effects of estradiol in ovarian and breast cancer McDonnell DP, Norris JD. (2002). Connections and cells. Mol Endocrinol 17: 792–803. regulation of the human estrogen receptor. Science 296: Hanstein B, Djahansouzi S, Dall P, Beckmann M, Bender H. 1642–1644. (2004). Insights into the molecular biology of the estrogen Murphy LC, Peng B, Lewis A, Davie JR, Leygue E, Kemp A receptor define novel therapeutic targets for breast cancer. et al. (2005). Inducible upregulation of oestrogen receptor- Eur J Endocrinol 150: 243–255. {beta}1 affects oestrogen and tamoxifen responsiveness Harris MA, Clark J, Ireland A, Lomax J, Ashburner M, in MCF7 human breast cancer cells. J Mol Endocrinol 34: Foulger R et al. (2004). The Gene Ontology (GO) database 553–566. and informatics resource. Nucleic Acids Res 32: D258–D261. Nikolova Z, Djonov V, Zuercher G, Andres A, Ziemiecki A. Hartman J, Lindberg K, Morani A, Inzunza J, Strom A, (1998). Cell-type specific and estrogen dependent expression Gustafsson J-A. (2006). Estrogen receptor {beta} inhibits of the receptor tyrosine kinase EphB4 and its ligand ephrin- angiogenesis and growth of T47D breast cancer xenografts. B2 during mammary gland morphogenesis. J Cell Sci 111: Cancer Res 66: 11207–11213. 2741–2751. Hosack DA, Dennis Jr G, Sherman BT, Lane HC, Lempicki Nilsson S, Makela S, Treuter E, Tujague M, Thomsen J, RA. (2003). Identifying biological themes within lists of Andersson G et al. (2001). Mechanisms of estrogen action. genes with EASE. Genome Biol 4: R70. Physiol Rev 81: 1535–1565. Hsieh M-Y, Chen W-Y, Jiang M-J, Cheng B-C, Huang T-Y, Palmieri C, Cheng G, Saji S, Zelada-Hedman M, Warri A, Chang M-S. (2006). Interleukin-20 promotes angiogenesis in Weihua Z et al. (2002). Estrogen receptor beta in breast a direct and indirect manner. Genes Immun 7: 234–242. cancer. Endocr Relat Cancer 9: 1–13.

Oncogene Role of ERb in breast cancer cells C Williams et al 1032 Park K-J, Krishnan V, O’Malley BW, Yamamoto Y, Gaynor by estradiol via down-regulation of IL-1 receptor type I in RB. (2005). Formation of an IKK[alpha]-dependent tran- uterine epithelial cells. J Immunol 175: 6509–6516. scription complex is required for estrogen receptor-mediated Smyth G. (2004). Linear models and empirical Bayes methods gene activation. Mol Cell 18: 71–82. for assessing differential expression in microarray experi- Paruthiyil S, Parmar H, Kerekatte V, Cunha GR, Firestone ments. Stat Appl Genet Mol Biol 3: Article 3. GL, Leitman DC. (2004). Estrogen receptor {beta} inhibits Stossi F, Likhite VS, Katzenellenbogen JA, Katzenellenbogen human breast cancer cell proliferation and tumor formation BS. (2006). Estrogen-occupied estrogen receptor represses by causing a G2 cell cycle arrest. Cancer Res 64: 423–428. cyclin G2 gene expression and recruits a repressor Poola I, Speirs V. (2001). Expression of alternatively spliced complex at the cyclin G2 promoter. J Biol Chem 281: estrogen receptor alpha mRNAs is increased in breast cancer 16272–16278. tissues. J Steroid Biochem Mol Biol 78: 459–469. Strom A, Hartman J, Foster JS, Kietz S, Wimalasena J, Rae JM, Johnson MD, Scheys JO, Cordero KE, Larios JM, Gustafsson J-A. (2004). Estrogen receptor {beta} inhibits Lippman ME. (2005). GREB1 is a critical regulator of 17{beta}-estradiol-stimulated proliferation of the breast hormone dependent breast cancer growth. Breast Cancer Res cancer cell line T47D. Proc Natl Acad Sci USA 101: Treat 92: 141–149. 1566–1571. Real PJ, Sierra A, De Juan A, J.C S, Lopez-Vega JM, Tokes A-M, Kulka J, Paku S, Szik A, Paska C, Novak P et al. Fernandez-Luna JL. (2002). Resistance to chemotherapy via (2005). Claudin-1, -3 and -4 and mRNA expression Stat3-dependent overexpression of Bcl-2 in metastatic breast in benign and malignant breast lesions: a research study. cancer cells. Oncogene 21: 7611–7618. Breast Cancer Res 7: R296–R305. Ricci MS, Toscano DG, Mattingly CJ, Toscano Jr WA. Tomasini R, Seux M, Nowak J, Bontemps C, Carrier A, (1999). Estrogen receptor reduces CYP1A1 induction Dagorn J-C et al. (2005). TP53INP1 is a novel p73 target in cultured human endometrial cells. J Biol Chem 274: gene that induces cell cycle arrest and cell death by 3430–3438. modulating p73 transcriptional activity. Oncogene 24: Richter K, Wirta V, Dahl L, Bruce S, Lundeberg J, Carlsson L 8093–8104. et al. (2006). Global gene expression analyses of hemato- Vlotides G, Eigler T, Melmed S. (2007). Pituitary tumor poietic stem cell-like cell lines with inducible Lhx2 expres- transforming gene: physiology and implications for tumor- sion. BMC Genomics 7: [E-pub ahead of print]. igenesis. Endocr Rev 28: 165–186. Roberts M, Wallace J, Jeltsch J-M, Berry M. (1988). The 50 Wang F, Lee E, Lowes MA, Haider AS, Fuentes-Duculan J, flanking region of the human pS2 gene mediates its Abello MV et al. (2006). Prominent production of IL-20 transcriptional activation by estrogen in MCF-7 cells. by CD68+/CD11c+ myeloid-derived cells in psoriasis: Biochem Biophys Res Commun 151: 306–313. gene regulation and cellular effects. J Invest Dermatol 126: Russo J, Russo IH. (2006). The role of estrogen in the 1590–1599. initiation of breast cancer. J Steroid Biochem Mol Biol 102: Wang Y-h, Liu S, Zhang G, Zhou C-q, Zhu H-x, Zhou X-b 89–96. et al. (2005). Knockdown of c-Myc expression by RNAi Schaefer TM, Wright JA, Pioli PA, Wira CR. (2005). IL- inhibits MCF-7 breast tumor cells growth in vitro and in 1{beta}-mediated proinflammatory responses are inhibited vivo. Breast Cancer Res 7: R220–R228.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene