Implications for Breast Cancer Development

ORIGINAL ARTICLE RASSF1A inhibits estrogen receptor alpha expression and estrogen-independent signalling: implications for breast cancer development

S Thaler1, M Schmidt2, A Schad3 and JP Sleeman1,4

The Ras association domain family 1 isoform A (RASSF1A) is a tumor suppressor whose inactivation is implicated in the development of many human cancers, including breast carcinomas. Little is known about the tumor-suppressive function of RASSF1A in breast tissue and whether its inactivation is mechanistically involved in the initiation and progression of breast tumors. Here, we show that RASSF1A inhibits breast cancer growth in vivo, and suppresses estrogen receptor (ERa) expression and function. Reconstitution of RASSF1A in MCF7 cells led to decreased ERa levels and reduced sensitivity to estrogen (E2). Concomitantly, we observed decreased expression of Id1 as well as the E2-responsive genes Bcl-2 and c-Myc that cooperatively contribute to the immortalization and transformation of breast epithelial cells. This downregulation was associated with induction of cell-cycle arrest and senescence that constitute early barriers to cancer initiation and progression. Knockdown of ERa showed that downregulation of ERa sufﬁces to increase senescence and inhibit expression of Bcl-2, c-Myc and Id1. However, enforced expression of ERa only partially rescued RASSF1A-mediated growth inhibition and senescence, suggesting that suppression of ERa expression and activity is not the only mechanism by which RASSF1A inhibits growth and survival of breast cancer cells. Ectopic expression of Bcl-2, c-Myc and Id1 had little or no effect on RASSF1A-mediated growth arrest, indicating that RASSF1A acts dominantly over these oncogenes. Mechanistically, RASSF1A was found to suppress ERa expression through Akt1. It also transiently inhibited ERa-induced Ras-MAPK activity after exposure of cells to E2. Together, our data show that RASSF1A acts as a tumor suppressor in ERa þ mammary epithelial cells, in part through inhibiting ERa expression and activity. These ﬁndings suggest that RASSF1A has a key role in suppressing the transformation of human breast epithelial cells and ERa þ breast cancer initiation.

Oncogene (2012) 31, 4912--4922; doi:10.1038/onc.2011.658; published online 23 January 2012 Keywords: RASSF1A; estrogen receptor a; senescence; breast cancer inhibition

INTRODUCTION and phosphatidylinositol-3-OH kinase (PI3K).12,13 Thus, ERa is a key Ras association domain family 1 isoform A (RASSF1A) is frequently player in E2-mediated proliferation, survival and differentiation inactivated in breast carcinomas,1 and is associated with estrogen through regulating the transcription of E2-target genes as well as receptor (ERa þ ) status, as aberrant DNA methylation is thought through activation of signal transduction pathways. to inactivate RASSF1A more frequently in ERa þ breast carcino- Many observations suggest an important role for ERa in breast mas.2--4 Furthermore, RASSF1A expression is lost in the ERa þ cancer tumorigenesis.14 In contrast to normal breast tissue where 5 mammary carcinoma cell lines MCF7, T47-D and ZR75-1. ERa þ cells are in the minority, overexpression of ERa is found in Moreover, inactivation of RASSF1A occurs early during the around 70% of human breast carcinomas.14 Furthermore, ERa development of ER þ breast cancer and persists during cancer expression in benign breast epithelial lesions is a risk factor for progression.2--4 Thus, RASSF1A inactivation might be an important breast malignancies,15 whereas low ERa expression in normal and early event in the genesis of ERa þ breast cancer. breast tissue is associated with reduced breast cancer incidence.16 ERa is required for ductal elongation during mammary gland In normal breast tissue, ERa expression and proliferation are 6 development. It belongs to the steroid hormone receptor inversely associated, indicating that ERa þ breast cells are non- superfamily and functions as a ligand-activated transcription dividing, or that ERa is downregulated as cells enter the cell factor. In the nucleus, ERs undergo conformational changes upon cycle.17 This relationship is completely or partially lost in benign interaction with estrogen (E2), allowing them to bind DNA and ERa þ breast lesions, where proliferating ERa þ cells are promote transcription of target genes.7,8 In addition to direct found.18,19 Thus, upregulation of ERa expression associated with transcriptional effects, ERs localized to the plasma membrane can enhanced proliferation might be an important and early event in inﬂuence the activity of signalling pathways through coupling the progression of precancerous lesions to breast cancer. 7 directly or indirectly to G-proteins by non-genomic mechanisms. The proliferation of ERa þ cells in benign breast lesions and This leads to activation of downstream effectors such as protein cancers may reﬂect an escape from apoptosis, cell-cycle arrest and kinase C-d,9 mitogen activated protein kinase ERK (MAPK)9,10,11 senescence, barriers that suppress transformation and transition

1Centre for Biomedicine and Medical Technology Mannheim (CBTM), Medical Faculty Mannheim, University of Heidelberg, Mannheim, Germany; 2Department of Obestrics and Gynecology, Johannes Gutenberg University, Mainz, Germany; 3Department of Pathology, Johannes Gutenberg University, Mainz, Germany and 4KIT Campus Nord, Institut fu¨r Toxikologie und Genetik, Karlsruhe, Germany. Correspondence: Dr S Thaler, Centre for Biomedicine and Medical Technology Mannheim (CBTM), Universita¨tsmedizin Mannheim, University of Heidelberg, TRIDOMUS-Gebaüde Haus C, Ludolf-Krehl-Strasse 13 --17, 68167 Mannheim, Germany. E-mail: [email protected] Received 13 July 2011; revised 5 December 2011; accepted 21 December 2011; published online 23 January 2012 RASSF1A inhibits ERa S Thaler et al 4913 from precancerous hyperplasias into invasive cancer.20,21 At the RESULTS molecular level, ERa mediates E2-dependent transcription of RASSF1A inhibits E2-dependent outgrowth of orthotopic proto-oncogenes that counteract cell-cycle checkpoints, senes- xenografts cence and apoptosis. For example, the anti-apoptotic protein Bcl-2 MCF7 is an ERa þ breast cancer cell line that is deficient for 22 is an E2-responsive gene that is widely expressed in ERa þ endogenous RASSF1A expression.5 To determine whether RASS- 23,24 tumors and correlates with recurrence following anti-hormo- F1A has a tumor-suppressive role in these cells, we developed 25 nal therapies. Bcl-2 also cooperates with c-Myc, ras and certain MCF7 cells in which RASSF1A expression is inducibly restored in viral oncogenes during cell immortalization and transforma- response to doxycycline (Figure 1A), then examined the effect of 26,27 tion. c-Myc itself is an E2-responsive gene that induces cell- restored RASSF1A expression on E2-dependent growth of MCF7 28 cycle progression in breast cancer cells through transcriptional tumors in vivo. To this end, 5 106 MCF7 cells that conditionally Cip1/Waf1 29 Â repression of p21 , and can transform immortalized express RASSF1A in response to doxycycline were implanted into breast epithelial cells in combination with either constitutive the mammary fat pad of E2-substituted female NOD/SCID mice. 30 PI3K signalling or with constitutive Akt1 and Rac1 activity. Mice fed with doxycycline displayed a significant inhibition of Expression of the helix-turn helix protein Id1 has been proposed orthotopic xenograft outgrowth in comparison with controls 31 to be regulated by E2, although evidence for direct regulation is (Figure 1B). To verify RASSF1A expression upon doxycycline lacking. Id1 negatively regulates differentiation, stimulates G1-S administration, sections of xenografts were stained for RASSF1A Cip1/Waf1(refs 32,33) phase progression through inhibition of p21 and expression (Figure 1C). Doxycycline itself had no effect on the Ink4A 34 35 p16 , and suppresses senescence. In addition, Id1 immorta- outgrowth of parental MCF7 tumors (Supplementary Figure 1). 36 lizes embryonic fibroblasts in cooperation with Bcl-2. Thus, a Together, these data demonstrate that RASSF1A has a tumor- complex interplay of signalling pathways and oncogenes is suppressive effect in this ERa þ breast cancer model. Accordingly, required to transform breast epithelial cells, and E2-mediated ductal and lobular epithelial cells in normal human breast tissue gene expression might counteract mechanisms that normally stained positively for RASSF1A (Figure 1D). suppress the initiation of breast cancer. In this study, we hypothesized that RASSF1A may act as a tumor suppressor by regulating the expression and/or function of ERa. RASSF1A expression leads to growth inhibition, senescence and We found that reconstitution of RASSF1A in MCF7 cells led to downregulation of ERa decreased ERa levels, reduced expression of Id1 and the E2- To determine whether RASSF1A has an effect on ERa expression, responsive genes Bcl-2 and c-Myc, upregulation of p21Cip1/Waf1 we treated the inducible RASSF1A-MCF7 cells with and without induction of cell-cycle arrest and senescence, and inhibition of doxycycline. In response to RASSF1A expression, we observed signalling pathways involved in breast epithelial cell transforma- downregulation of ERa and inhibition of growth, as evidenced by tion. These data provide important mechanistic insights into how a reduced colony-forming ability (Figure 2a). Similar results were RASSF1A suppresses human breast cancer tumorigenesis. obtained with doxycycline-induced expression of RASSF1A

1400 ] –+doxycycline 3 1200 – Doxycycline + Doxycycline p <0.05 RASSF1A 1000 actin 800 600 400 200 Tumor volume [mm 0 101321 24 27 31 35 38 – + doxycycline days – Doxycycline + Doxycycline

anti- anti- RASSF1A RASSF1A

H & E anti- RASSF1A

– Doxycycline + Doxycycline Figure 1. RASSF1A inhibits E2-dependent outgrowth of orthotopic xenografts. (A) Conditional MCF7 cells were plated at equal densities, and RASSF1A was induced by doxycycline treatment. Induction was conﬁrmed by immunoblotting (upper panel). Induction of RASSF1A expression strongly inhibited clonogenic survival (lower panel). (B) Conditional MCF7 cells were implanted into the mammary fat pad of female NOD.SCID mice treated weekly with E2. Groups of mice either received doxycycline to induce RASSF1A expression or were left untreated as a control. Left panel: Tumor growth was monitored and plotted against time post implantation of tumor cells (days). Mean values ±s.d. and the P-value for day 38 are indicated. Right panel: Representative pictures of tumor-bearing mice. (C) Conditional MCF7 cells were implanted into the mammary fat pad of E2 substituted female NOD.CB17 SCID mice to establish tumors. After 30 days, one group of mice received doxycycline whereas the other group was left untreated. Four days later, the animals were killed and sections of tumors from both groups were analyzed for RASSF1A expression using immunohistochemical staining. (D) Immunohistochemical analysis of RASSF1A in sections of four independent normal human breast tissues (a--d).

& 2012 Macmillan Publishers Limited Oncogene (2012) 4912 --4922 RASSF1A inhibits ERa S Thaler et al 4914 p < 0.05 180 80 * 12 * 160 dox 70 *p < 0.05 140 + dox 60 10 – dox 120 8 – + doxycycline 50 + dox 100 40 6 ER alpha 80 30 4 60 * RASSF1A 20 * *

number of cells [%] * 2

number of colonies 40

10 number of cells [%] 20 actin 0 0 SubG 0 SubG1 G1 SG2-M D 1 MCF7

80 * p < 0.05 70 *p < 0.05 5 * – dox 60 400 + dox 4 350 50 dox – + doxycycline + dox 300 40 3 250 ER alpha 30 200 * 2 150 20 RASSF1A number of cells [%] * 100 n.s. 1 10 * number of cells [%]

number of colonies 50 actin 0 0 0 T47 D SubG1 G1 SG2-M D SubG1

40 p < 0.05 35 30 25 – Dox 20 + Dox 15 SAβ-gal -gal+ cells [%]

β 10 5 SA 0

60 p < 0.05 50 anti- 40 – Dox ER alpha 30 + Dox + cells [%]

α 20 10 ER MCF7 - doxycycline MCF7 + doxycycline 0 MCF7

25 p < 0.05 20

15 RASSF1A 10 + RASSF1A -gal+ cells [%] β 5 SA 0 T47 D Figure 2. RASSF1A mediates downregulation of ERa and promotes growth inhibition and senescence. MCF7 cells (a) or T47D cells (b) were induced or non-induced with doxycycline to conditionally express RASSF1A as indicated. Cell extracts obtained 48 h after doxycycline administration and 24 h after E2 exposure were analyzed by immunoblotting with the indicated antibodies. Cells were plated at equal densities. After induction of conditional RASSF1A expression for 10--14 days, clonogenic survival (number of colonies) was compared with non-transduced cells. Mean values±s.d. from four independent experiments are presented. Right panels: Cells in the subG1,G1,S,G2-M phases of the cell cycle as well as doublets (D) were quantiﬁed. Mean values ±s.d. from four independent experiments are presented. P-value o0.05 are indicated by asterisks. (c) Conditional MCF7 cells were grown for 6 days in the absence or presence of doxycycline as indicated. RASSF1A expression induced a senescence morphology accompanied by SAb-gal activity, and a reduction in ERa expression as evidenced by immunostaining (right panels). In the quantiﬁcation of these data (left panels), mean values ±s.d. of six independent experiments are presented. (d) T47D cells were transfected with either pcDNA3.1-RASSF1A or empty vector. RASSF1A expression induced a senescence morphology accompanied by SAb-gal activity. Mean values ±s.d. of four independent experiments are presented. (e) Conditional MCF7 cells grown under E2-free conditions and in the absence or presence of doxycycline as indicated were double immunostained with anti-RASSF1A (red) and ERa (green) antibodies.

in T47D (Figure 2b), another ERa þ breast cancer cell line that is Conditional RASSF1A expression also led to an increase in G1 deﬁcient for endogenous RASSF1A expression.5 These data phase for both cell lines (Figures 2a and b, right panels) and to indicate that RASSF1A-mediated downregulation of ERa is not induction of senescence, as evidenced by increased senescence- cell line speciﬁc. associated b-galactosidase (SAb-gal) activity (Figures 2c and d).

Oncogene (2012) 4912 --4922 & 2012 Macmillan Publishers Limited RASSF1A inhibits ERa S Thaler et al 4915 Similar findings were made when single cell clones instead of suggesting that RASSF1A may affect ERa protein levels in a pooled transduced cells were used (Supplementary Figure 2). proteasome-dependent manner. Immunofluorescence double stainings revealed a reciprocal RASSF1A-expressing MCF7 cells grown in the absence of E2 did staining pattern for RASSF1A and ERa, consistent with negative not express Bcl-2 or activated c-Myc (Figure 3a). However, Id1 was regulation of ERa by RASSF1A (Figure 2e). Collectively, these still present, indicating that its expression is E2 independent. We findings suggest that RASSF1A expression in ERa þ breast cancer observed reduced protein expression of c-Myc and Bcl-2 in cells causes reduced ERa expression, induction of cell-cycle arrest RASSF1A-expressing cells at all E2 concentrations used (Figure 3a; and senescence and reduced clonogenic survival. Supplementary Table 1). Transcription of Bcl-2 and c-Myc but not Id1 was also significantly reduced upon expression of RASSF1A RASSF1A expression leads to downregulation of E2-responsive (Figure 3d). RASSF1A expression in the inducible MCF7 cells in the genes that are positive regulators of G1-S phase progression and presence and absence of E2 had no effect on expression of the counteract senescence ERa cofactor SRC3 (Supplementary Figure 3), ruling this out as a ERa mediates E2-dependent transcription of c-Myc and Bcl-2, potential mechanism for the observed effects on Bcl-2 and c-Myc genes that act together to protect cells from cell-cycle arrest, transcription. oncogene-induced senescence and apoptosis. E2 also has been In addition to being downregulated by RASSF1A, we also found reported to regulate Id1 expression.31 We therefore hypothesized that E2 itself modestly downregulated ERa, and that this was that loss of RASSF1A might cause inappropriate upregulation of associated with a concomitant increase in Bcl-2 expression ERa, in turn enhancing expression of these genes and thereby (Supplementary Figure 4), demonstrating an E2-dependent overriding transformation barriers. To investigate this hypothesis, negative feedback loop on ERa expression. Nevertheless, RASSF1A we examined whether RASSF1A affects expression of ERa, Id1, expression substantially reduced levels of ERa over and above that c-Myc and Bcl-2. These experiments were performed in the induced by the presence of E2 (Figure 3a). presence of different E2 concentrations to allow effects on E2- To determine whether RASSF1A-mediated downregulation of dependent expression of these genes to be observed. ERa is sufficient to induce senescence and repress the expression Using the doxycycline-inducible MCF7 cells, we found that of Id1, c-Myc and Bcl-2, we employed stable knockdown of ERa in RASSF1A-expressing cells display lower ERa protein and transcript MCF7 cells. Knockdown of ERa reduced expression of Id1, levels in comparison with controls in the presence of all E2 c-Myc and Bcl-2 (Figure 4a). Furthermore, E2 reduced the concentrations used (Figures 3a and b; Supplementary Table 1). relative expression of p21Cip1/Waf1 in the presence and absence The proteasome inhibitor Bortezomib strongly augmented RASS- of RASSF1A (Figure 3a), while efficient knockdown of ERa F1A protein expression, yet levels of the ERa protein were upregulated p21Cip1/Waf1 expression (Figure 4a). The extent of unaffected by RASSF1A in the presence of Bortezomib (Figure 3c), ERa ablation corresponded to SAb-gal activity and increased

010-7 10-8 10-10 E2 (M) -+-+-- + + doxycycline RASSF1A 1.0 * * ER alpha 0.8 * * = p <0.05 * 0.6 Id1 – Dox + Dox Bcl-2 0.4

c-myc 0.2 relative expression

Cip1/Waf1 0 p21 010-7 10-8 10-10 E2 (M) ER alpha actin

––+ + – – + + Bortezomib [10 nM] * =p <0.05 * =p <0.05 – + – + – + – + doxycycline 2.5 * 3 * – Dox – Dox ER alpha 2 + Dox 2.5 + Dox (short exp.) 2 1.5 n.s. n.s. ER alpha 1.5 (long exp.) 1 * * 1 RASSF1A * 0.5 0.5 * relative expression actin relative expression 0 0

Id1 Id1 Bcl-2 c-myc Bcl-2 c-myc 0 15 0 30 time [h] Cip1/Waf1 Cip1/Waf1 1 p21 p2 10-7 E2 (M) 10-10 E2 (M) Figure 3. Conditional RASSF1A expression leads to downregulation of E2-response genes that are positive regulators of G1-S phase progression and counteract senescence. (a) Conditional MCF7 cells were grown in the presence or absence of doxycycline, and in the presence or absence of different E2 concentrations as indicated. Cell extracts obtained 48 h after doxycycline administration and 24 h after E2 exposure were analyzed by immunoblotting with the indicated antibodies. (b) RASSF1A downregulates transcription of ERa. Conditional MCF7 cells were grown in the presence or absence of doxycycline, and in the presence or absence of different E2 concentrations as indicated. RNA was harvested 48 h after doxycycline administration and 24 h after E2 exposure, and ERa transcript levels were analyzed by quantitative PCR. (c) Conditional MCF7 cells were grown in the presence or absence of doxycycline, and in the presence or absence of the indicated concentration of Bortezomib. Cell extracts obtained 0, 15 and 30 h after doxycycline and Bortezomib administration were analyzed by immunoblotting with the indicated antibodies. Two different exposure times for the ERa blot are presented. (d) RASSF1A downregulates transcription of Bcl2, c-Myc but not Id1. Conditional MCF7 cells were grown in the presence or absence of doxycycline, and in the presence or absence of different E2 concentrations as indicated. RNA was harvested 48 h after doxycycline administration and 24 h after E2 exposure, and transcript levels of the indicated genes were analyzed by quantitative PCR.

& 2012 Macmillan Publishers Limited Oncogene (2012) 4912 --4922 RASSF1A inhibits ERa S Thaler et al 4916 30 * =p <0.05 70 25 * 60 * 50 20 * neg. controlESR1 k.o.1ESR1 k.o.2 40 15 30 10 SAβ-gal ER alpha -gal+ cells [%] 20 β *

Apoptotic cells [%] 5

SA 10

Id1 0 0

80 * 70 Bcl-2 100 60 80 50 anti- c-myc 40 ER alpha 60 -gal+ cells [%] * β 30 * 40 + cells [%]

p21Cip1/Waf1 α 20 20 10 ER 0 * 0

actin Apopt. + SA MCF7 neg. control MCF7 ESR1 k.o.1

neg. controlESR1ESR1 k.o.1 k.o.2 neg. controlESR1ESR1 k.o.1 k.o.2 Figure 4. Knockdown of ERa expression leads to downregulation of E2-response genes, and induction of senescence and apoptosis. (a) Western blots of lysates from two independent knockdowns of ERa expression in MCF7 cells (ESR1k.o.1 and ESR1k.o.2) were probed with the indicated antibodies. Cell extracts were obtained 48 h after lentiviral transduction. (b) Knockdown of ERa using shRNA in MCF7 cells (ESR1k.o.1) induced growth inhibition, senescence morphology accompanied with SAb-gal activity and a reduction in ERa expression as evidenced by immunostaining 48 h after lentiviral transduction (right panels). Left panels: Quantiﬁcation of the percentage of SAb-gal-positive, ERa-positive and apoptotic cells in two independent knockdowns of ERa expression in MCF7 cells (ESR1k.o.1 and ESR1k.o.2) is presented as mean values ±s.d. of four independent experiments.

p21Cip1/Waf1 levels (Figure 4), characteristics of senescent cells. RASSF1A-induced downregulation of Bcl-2 was rescued when Accordingly, SAb-gal activity was found in 20--60% of the cells ERa expression levels were maintained ectopically at physiological following ERa knockdown, whereas apoptosis was detected in levels (Figure 6a). However, c-Myc and Id1 expression was reduced only 15--25% of the cells (Figure 4b). Similarly, chemical ablation and p21Cip1/Waf1 levels were increased compared with those in of ERa protein using fulvestrant reduced Bcl-2, c-Myc and Id1 RASSF1A-expressing MCF7 cells without ectopic ERa expression expression, stimulated expression of p21Cip1/Waf1 increased the (Figure 6a). number of cells in G1, and reduced colony formation concomi- Ectopic expression of ERa in MCF7 cells in the absence of tantly with induction of senescence and apoptosis (Figure 5). RASSF1A induction caused a slight decrease in colony-forming The above findings are consistent with the notion that a ability (Figure 6b). In contrast, when RASSF1A was induced by consequence of RASSF1A expression in ERa þ breast cancer cells doxycycline, MCF7 cells expressing ectopic ERa exhibited an is induction of senescence through negative regulation of ERa, increase in clonogenic survival, although colony formation was not decreased E2-dependent oncogene expression and reduced cell rescued to the levels observed in the absence of RASSF1A survival. However, not all E2-responsive genes (for example, (Figure 6b). Similarly, RASSF1A-induced senescence was only cyclin D1) were differentially regulated upon RASSF1A expression moderately rescued by ectopic ERa expression (Figure 6c). (Supplementary Figure 5). Furthermore, Id1 protein levels were Furthermore, even though knockdown of ERa resulted in increased modulated by RASSF1A independently of E2 (Figure 3a), but no p21Cip1/Waf1 levels (Figure 4a), MCF7 cells conditionally expressing effect of RASSF1A on Id1 transcription was observed (Figure 3d). In RASSF1A displayed increased p21Cip1/Waf1 levels in both the control experiments, doxycycline treatment of parental MCF7 presence and absence of enforced ERa expression (Figure 6a), cells in the presence and absence of E2 had no effect on Id1 again indicating that ectopic ERa expression does not completely expression (Supplementary Figure 6). These data suggest that rescue the effect of RASS1A on p21Cip1/Waf1 levels. Collectively, RASSF1A can post-translationally regulate Id1 expression inde- these findings suggest that RASSF1A exerts its tumor suppressor pendently of E2. function only partially through downregulation of ERa expression. We also tested whether high E2 concentrations counteracts RASSF1A-mediated growth inhibition and induction of senes- RASSF1A inhibits cell growth and survival through ERa-dependent cence. However, this was not the case (Figure 6d). and independent mechanisms Finally, we determined whether Id1, Bcl-2 and c-Myc have the To determine whether ERa downregulation is the only mechanism potential to combat RASSF1A-mediated growth inhibition. Neither by which RASSF1A mediates cell-cycle arrest and senescence, we ectopic overexpression of Id1 alone or in combination with Bcl-2, established RASSF1A conditional MCF7 cells that ectopically nor overexpression of c-Myc was able to circumvent RASSF1A- express physiologically relevant levels of ERa constitutively. Upon mediated inhibition of clonogenic survival (Supplementary induction of RASSF1A in these cells, levels of ERa remained Figures 7a and b). Although enforced expression of Bcl-2 and equivalent to those in RASSF1A conditional MCF7 cells without c-Myc together increased cellular survival in the presence and doxycycline treatment (Figure 6a), indicating that the cells absence of RASSF1A expression, the relative difference between maintain physiologically relevant levels of ERa in the presence RASSF1A-expressing and non-expressing cells was not signifi- of RASSF1A. Furthermore, western blot analysis showed increased cantly different to control cells (Supplementary Figure 7b, left Bcl-2 and c-Myc expression in the non-doxycycline-treated cells panel). Furthermore, RASSF1A expression increased p21Cip1/Waf1 transfected with the ERa expression construct compared with levels even in MCF7 cells overexpressing c-Myc (Supplementary controls, indicating functionality of the ectopically expressed Figure 7b, right panel). Thus, RASS1A acts dominantly over Bcl-2, ERa (Figure 6a). Id1 and c-Myc.

Oncogene (2012) 4912 --4922 & 2012 Macmillan Publishers Limited RASSF1A inhibits ERa S Thaler et al 4917 μ – + Fulvestrant [1 M] 90 * + + E2 [10-9 M] 80 * p < 0.05 ER alpha 70 3 * – Fulvestrant 60 Bcl-2 + Fulvestrant 50 2 c-myc 40

Id1 30

number of cells [%] * 1 20 p21Cip1/Waf1 * number of cells [%] 10 n.s. * actin 0 0 SubG SubG1 G1 S G2-M D 1

70 * 60 * 60 50 50 40 40 30 30

-gal+ cells [%] 20

20 β 10 10 SA Number of colonies 0 0 + E2 [10-9 M] + E2 [10-9 M] 30 80 * * + Fulvestrant [1μM] 70 25 60 *p < 0.05 20 50

15 -gal+ cells [%] 40 – Fulvestrant β 30 + Fulvestrant 10 20 5

Apoptotic cells [%] 10 0 0 Apopt. + SA Figure 5. Fulvestrant leads to degradation of ERa, downregulation of E2-response genes, and induction of senescence and apoptosis. (a) Conditional MCF7 cells were grown in the absence of doxycycline, and in the presence or absence of the indicated fulvestrant concentrations. Cell extracts obtained 48 h after doxycycline administration and 24 h after E2 exposure were analyzed by immunoblotting with the indicated antibodies. Right panel: Cells in the subG1,G1,S,G2-M phases of the cell cycle as well as doublets (D) were quantiﬁed 48 h after fulvestrant exposure. Mean values±s.d. from four independent experiments are presented. P-values o0.05 are indicated by asterisks. (b) Degradation of ERa using the anti-estrogen fulvestrant in MCF7 cells induced growth inhibition, senescence morphology accompanied by SAb-gal activity (right panel) and induction of apoptosis after 5 days exposure. Left panels: Quantiﬁcation of the percentage of SAb-gal-positive and apoptotic cells, as well as the clonogenic survival of MCF7 cells in the presence and absence of fulvestrant is presented as mean values±s.d. of four independent experiments.

RASSF1A inhibits ERa expression via Akt1, and counteracts between Akt1 and either ERa expression or protein stabilization. mechanisms that lead to E2-independent growth Accordingly, conditional MCF7 cells grown in the absence of A variety of mechanisms that confer E2-independent growth and doxycycline exhibited decreased expression of endogenous ERa in turn resistance against drugs that target ERa have been when incubated with the PI3K inhibitor Ly294.002 (Figure 7c). identified, including constitutive activation of Akt1 and 3(refs 37,38) Furthermore, while reduced colony formation consequent to and hyperactivation of the Ras-MAPK pathway.39 As RASSF1A is chemical ablation of ERa protein by fulvestrant was substantially able to modulate the MAPK pathway and inhibit Akt,40 we next rescued by MyrAkt1 (Figure 7d), this rescue was abrogated by investigated whether RASSF1A might regulate signalling pathways RASSF1A (Figure 7e). Collectively, these findings indicate that associated with E2 independence. RASSF1A downregulates ERa through inhibition of Akt1 activity, First, we established RASSF1A conditional MCF7 cells that and that RASSF1A inhibits E2-independent growth. express MyrAkt1 or MyrAkt3,37,38 myristylated forms of Akt1 and Next, we investigated whether RASSF1A-mediated downregula- Akt3 that localize to the plasma membrane and are therefore tion of ERa also requires modulation of the Ras-MAPK pathway. activated. In agreement with our previous findings,40 conditional First, control experiments were performed to demonstrate that expression of RASSF1A inhibited MyrAkt1 and MyrAkt3 through doxycycline in parental MCF7 cells had no effect on E2-induced phosphorylation at Ser473 (Figure 7a, right blot). Accordingly, while ERK phosphorylation (Supplementary Figure 8). Next, when expression of either MyrAkt1 or MyrAkt3 supported outgrowth of conditional MCF7 cells were grown in the absence of doxycycline the cells under E2-free conditions, neither MyrAkt1 nor MyrAkt3 and treated with the MEK inhibitor UO126, decreased expression was able to relieve RASSF1A-mediated growth inhibition as of endogenous ERa was observed (Figure 7c). Furthermore, measured by colony formation (Figure 7b). RASSF1A suppressed the enhanced ERK phosphorylation observed Molecularly, we found that MyrAkt1 inhibited RASSF1A- 35 min after treatment with E2 (Figure 7f), consistent with the mediated ERa downregulation, whereas MyrAkt3 had no such notion that RASSF1A-dependent suppression of E2-mediated effect (Figure 7a, right blot). Moreover, MyrAkt1-expressing MCF7 activation of the Ras-MAPK pathway might be involved in the cells displayed a higher clonogenic survival under E2-free regulation of ERa. However, when conditional MCF7 cells were conditions in comparison with their MyrAkt3-expressing counter- cultured in the presence of E2 for 24 h, RASSF1A expression parts (Figure 7b). These findings point to a potential interplay promoted sustained ERK phosphorylation (Figure 7a, right blot),

& 2012 Macmillan Publishers Limited Oncogene (2012) 4912 --4922 RASSF1A inhibits ERa S Thaler et al 4918 ––++ doxycycline 120 – dox, neg. vector RASSF1A 100 + dox, neg. vector – dox, ER alpha 80 ER alpha + dox, ER alpha 60 Bcl-2

c-Myc 20

Id1 number of colonies 15 10 Cip1/Waf1 p21 5 actin 0 neg. vectorER alpha neg. vectorER alpha neg. vectorER alpha neg. ERvector alphaneg.ER vector alphaneg.ER vector alpha

pQCXIN - pQCXIN ER alpha

10-7 M 10-9 M 10-10 M (E2) 10-7 M1010-9 M -10 M (E2)

* n.s. * 180 100 100 * 90 160 * * * 90 * * – dox * 80 80 + dox 70 140 60 120 70 * 50 60 * = p <0.05 40 100 * 50 -gal+ cells [%] 30 *

β 80 20 40 * = p <0.05 -gal+ cells [%] β SA 10 60 30 * * * *

0 SA

number of colonies 40 neg. vectorER alpha neg. vectorER alpha 20 20 * 10 0 0 010-7 10-8 10-9 10-10 10-11 10-12 10-13 E2 (M) 010-7 10-8 10-9 10-10 10-11 10-12 10-13 E2 (M) 10-7 M10-10 M (E2) Figure 6. Enforced ERa expression partially rescues RASSF1A-mediated senescence. (a) Conditional MCF7 cells were retrovirally transduced to express human ERa (pQCXIN-ERa), or were transduced with empty vector (pQCXINÀ). After selection, surviving cells were cultured for 48 h either in the presence or in the absence of doxycycline, then cell extracts were analyzed by immunoblotting with the indicated antibodies. (b) RASSF1A conditional MCF7 cells transduced either to express ERa or with empty vector were cultivated in the presence or absence of doxycycline together with the indicated E2 concentrations for 3 weeks. Clonogenic survival (number of colonies) in the presence of the indicated E2 concentrations is presented as mean values ±s.d. from four independent experiments. (c) The induction of senescence in MCF7 cells as assessed by the proportion of SAb-gal þ cells was analyzed with and without conditional RASSF1A expression and with and without ectopic expression of ERa in the presence of the indicated concentrations of E2. The color of the bars corresponds to the conditions used in Figure 6b. Cells were treated for 6--7 days with doxycycline together with the indicated E2 concentrations as indicated. Mean values ±s.d. from four independent experiments are indicated. (d) Clonogenic survival (number of colonies) and the induction of senescence (% SAb-gal þ ) with and without induction of RASSF1A expression ( þ dox, Àdox) was analyzed in MCF7 cells cultured in the presence of the indicated E2 concentrations. Mean values ±s.d. from three independent experiments are presented. P-value o0.05 are indicated by asterisks.

indicating that RASSF1A can also negatively regulate ERa levels A variety of mechanisms enhance ERa levels, including independently of ERK phosphorylation. Accordingly, inhibition of amplification of the ERa gene, although almost half of breast PI3K activation led to downregulation of ERa, despite concomitant tumors with high ERa do not show such an amplification.42 activation of the Ras-MAPK pathway (Figure 7c). Mutation of caveolin-1(ref. 43) and loss of expression of miR206(ref. 44) have also been implicated in enhancing ERa expression in breast cancer cells. Here, we present a novel mechanism namely that loss DISCUSSION of RASSF1A can also contribute to enhanced ERa levels. We Here we show that RASSF1A can exert a tumor-suppressive suggest that RASSF1A expression in non-transformed cells serves activity in breast cancer cells in two ways. First, it can negatively to restrict the levels of ERa through inhibiting Akt1. Furthermore, regulate ERa levels, leading to inhibition of signal transduction RASSF1A also blocked early E2-mediated activation of the pathways that are activated by E2 and thus decreased Ras-MAPK pathway, and chemical inhibition of ERK phosphoryla- E2-dependent expression of oncogenes that can override trans- tion caused ERa downregulation, identifying a further possible formation barriers such as senescence and apoptosis. Second, mechanism by which RASSF1A can suppress ERa expression. RASSF1A inhibits signalling pathways involved in E2-independent However, this latter mechanism appears to operate only growth that are involved in the transformation of breast epithelial transiently, as RASSF1A caused sustained ERK phosphorylation cells. These results therefore identify RASSF1A expression as a following prolonged exposure of cells to E2. Inhibition of PI3K barrier that suppresses transformation of breast epithelial cells. activation was sufficient to downregulate ERa even though the A major finding of this paper is that RASSF1A expression Ras-MAPK pathway was activated. downregulates ERa, implying that loss of RASSF1A during tumor Our findings indicate that RASSF1A-mediated inhibition of Akt1 development serves to increase ERa levels. Increased ERa serves to reduce either ERa expression or protein stability. These expression and function can make cells more susceptible to observations are consistent with studies using transgenic animals E2-induced stimulation of proliferation and are thought to be an that express MyrAkt1 in the mammary epithelium, in which Akt1 initial, important event during breast carcinogenesis.15,17--19 activity results in increased expression of ERa during carcinogen- Enhanced ERa levels may act by endowing tumor cells with a induced mammary tumorigenesis.45 MyrAkt1 was shown to more robust response to E2. Furthermore, the response to E2 may induce E2-independent outgrowth and tamoxifen resistance of be qualitatively different, as the mechanism of transactivation and breast cancer cells.37 Our data suggest that by inhibiting Akt1, target gene regulation differ when ERa levels are enhanced.41 RASSF1A serves as a transformation barrier in breast epithelial

Oncogene (2012) 4912 --4922 & 2012 Macmillan Publishers Limited RASSF1A inhibits ERa S Thaler et al 4919 -9 – – + ++– – +++ – – E2 (10 M, 24h) * - + - UO126 – + – ++– +++– – – doxycycline – Doxycyclin --+ Ly294.002 160 * + Doxycyclin RASSF1A - - - doxycycline 140 ER alpha p-AKTS473 120 100 * ER alpha p-ERK1/2 80 * p-ERK1/2 60 p-AKTS473 40 ERK number of colonies 20 * = p <0.05 ERK actin 0 neg. vector MyrAkt1 MyrAkt3 AKT

–MyrAKT1 MyrAKT3 actin

* * * * 100 100 ––+ + + + + + E2 (10-9 M) * = p <0.05 * p < 0.05 – + –––+ + + doxycycline 80 * 80 – dox RASSF1A 60 60 – Fulvestrant + dox + Fulvestrant – dox, + Fulvestrant ER alpha 40 40 + dox, + Fulvestrant p-ERK 42/44 20 20 number of colonies [%]

number of colonies [%] ERK 0 0 neg. vector MyrAkt1 MyrAkt1 MyrAkt1 actin

0601535 time [min] Figure 7. RASSF1A counteracts mechanisms that lead to E2-independent growth. (a) Conditional MCF7 cells transduced to express MyrAkt1 or MyrAkt3 or with empty vector (neg. vector) were cultured in the presence or absence of doxycycline for 48 h to induce RASSF1A expression. Before harvesting, cells were cultured for 24 h in the presence or absence of E2 (10À9M). Western blots of lysates of these cells were probed with the indicated antibodies. (b) Equal numbers of conditional MCF7 cells transduced to express MyrAkt1 or MyrAkt3 or with empty vector (neg. vector) were cultured under E2-free conditions either in the presence or absence of doxycycline for 14 days to induce RASSF1A expression. Clonogenic survival (number of colonies) was analyzed (mean values ±s.d. from four independent experiments). (c) RASSF1A conditional MCF7 cells grown in the absence of doxycycline were either treated with the PI3K inhibitor LY294.002 (30 mmol/l for 16 h) or with the MEK inhibitor UO126 (20 mmol/l for 16 h). Cell extracts were analyzed by immunoblotting using the indicated antibodies. (d) Equal numbers of RASSF1A conditional MCF7 cells transduced to express MyrAkt1 were cultured either in the presence or absence of fulvestrant for 9 days in the absence of doxycycline. Clonogenic survival (number of colonies) was analyzed (mean values ±s.d. from four independent experiments). (e) RASSF1A conditional MCF7 cells transduced to express MyrAkt1 were cultured in the presence or absence of doxycycline either with or without fulvestrant. Clonogenic survival (number of colonies) was analyzed (mean values ±s.d. from four independent experiments). (f) Conditional MCF7 cells were grown in the presence or absence of doxycycline to induce RASSF1A expression for 48 h. Before harvesting, cells were cultured for 15, 35 and 60 min in the presence or absence of E2 (10À9 M). Western blots of lysates of these cells were probed with the indicated antibodies.

cells that acts in part at the level of ERa, in part through HB-EGF, in turn inhibiting the rapid and transient activation of the E2-independent mechanisms. Ras-MAPK pathway (Figure 8). On the other hand, Akt activity A consequence of RASSF1A-mediated suppression of ERa is inhibits Raf signalling in MCF7 cells.49 As expression of RASSF1A reduced expression of E2-dependent genes such as Bcl-2 and inhibits Akt activity, Akt-mediated inhibition of Raf would thereby c-Myc. Although Id1 has been reported to be E2 regulated,31 we be relieved, resulting in enhanced ERK phosphorylation. This effect observed no effect of E2 on Id1 expression at either the protein or would be more pronounced in the sustained presence of E2 as the transcriptional levels. Nevertheless, both knockdown and ERa is downregulated, further reducing Akt activity. Future work chemical ablation of ERa reduced expression of Id1. How this is will aim to provide experimental substantiation for this notion. achieved remains unclear, but we note that ERb has been reported In previous studies, we have demonstrated that RASSF1A to directly bind to Id1.46 If ERa also complexes with Id1, then it is executes cell-cycle arrest and senescence in lung cancer cells conceivable that the stability of Id1 may thereby be enhanced in through p21Cip1/Waf1 upregulation in a p53-independent manner.40 an E2-independent manner. In further studies, we will investigate In this report, we demonstrate that RASSF1A has the ability to if this is the case. upregulate p21Cip1/Waf1 in ERa þ breast cancer cells through We show that RASSF1A inhibits initial E2-mediated activation of downregulation of ERa. Id1 and c-Myc are negative regulators of the Ras-MAPK pathway (Figure 7f), whereas after 24 h exposure to p21Cip1/Waf1(refs 29,33) and are also downregulated by RASSF1A. E2, RASSF1A enhances ERK phosphorylation (Figure 7a, left panel, However, RASSF1A was able to upregulate p21Cip1/Waf1 even in the lanes 2 and 4). This latter effect is also observed in the absence of context of enforced expression of ERa or c-Myc. Thus, additional E2, albeit to a lesser extent (Figure 7a, left panel, lanes 1 and 2). mechanisms in addition to downregulation of ERa and E2-target E2 treatment has been reported to lead to rapid and transient genes contribute to RASS1A-mediated upregulation of activation of the Ras-MAPK pathway that is thought to proceed p21Cip1/Waf1. In lung cancer cells, RASSF1A-mediated p21Cip1/Waf1 through a multistep process involving E2-bound ERa, Src and the upregulation is dependent on Akt inhibition,40 and thus the p85 regulatory subunit of PI3K.11,13,47,48 In turn, Src mediates inhibition of Akt activity we observed in MCF7 cells in response to activation of matrix metalloproteinases that liberate HB-EGF RASSF1A expression is likely to account at least in part for the bound to the surface of the MCF7 cells, leading to epidermal ERa-independent upregulation of p21Cip1/Waf1. growth factor receptor transactivation and activation of Ras-MAPK Our data suggest that RASSF1A suppresses transformation of pathway.11 Our results are consistent with the notion that breast cancer cells through the regulation of several signalling RASSF1A suppresses E2-mediated liberation of membrane-bound pathways (Figure 8), contributing to our understanding of the

& 2012 Macmillan Publishers Limited Oncogene (2012) 4912 --4922 RASSF1A inhibits ERa S Thaler et al 4920 EGF

EGF EGF

MMP’s MMP’s Ras Ras Src α Src ER ERα p85 E2 p85 E2 PI3K RASSF1A

Raf Akt1 Raf Akt1

MEK MEK

ERK α α ER ERK ER suceptibility to E2 suceptibility to E2 Cip1/Waf1 Cip1/Waf1 p21 Id1 proliferation survival p21 Id1 survival growth factors c-Myc c-Myc E2 independence senescence ERα Bcl-2 ERα Bcl-2 E2 E2

transcription of E2 target genes transcription of E2 target genes Figure 8. RASSF1A inhibits E2-mediated activation of the MAPK pathway. Schematic model of a possible molecular network regulated by RASSF1A, based on the ﬁndings in this paper and on the published literature. In the absence of RASSF1A (left panel), E2 treatment triggers rapid and transient activation of the Ras-MAPK and PI3-K/Akt pathways via a multistep process encompassing E2-bound ERa, Src and the p85 regulatory subunit of PI3-K. Src-mediated activation of matrix metalloproteinases results in liberation of the growth factor HB-EGF, leading to epidermal growth factor receptor transactivation and activation of Ras-MAPK. RASSF1A interferes with this network by inhibiting signalling via Akt1. It also transiently suppresses the Ras-MAPK pathway upon E2 stimulation, but later enhances activation of this pathway as a consequence of Akt1 inhibition. An important consequence of the inhibitory activity of RASS1A is reduced ERa levels, which in turn affects the expression and activity of E2n-target genes. For more details, see text.

signalling pathways that act in conjunction with oncogenes to purchased from Santa Cruz Biotechnology (Heidelberg, Germany). transform human breast epithelial cells. In the absence of RASSF1A, Antibodies against p-Akt Ser473 and p-44/42 Erk1/2 were from CST augmented PI3K activity leads to activation of Akt1 and thereby (Danvers, MA, USA), and the anti-RASSF1A antibody was from Abcam increased expression of ERa. In turn, in the presence of E2, ERa (Cambridge, UK). induces expression of genes such as Bcl-2 and c-Myc that increase survival and suppress expression of p21Cip1/Waf1. Consistently, Cell culture and viral transduction c-Myc can transform immortalized early passage human breast epithelial cells either together with constitutive PI3K signalling or in MCF7 and T47D were purchased from ATCC. Unless otherwise indicated, combination with constitutive Akt1 and Rac1.30 When RASSF1A is parental cell lines were maintained in RPMI supplemented with 10% fetal expressed, Akt1 activation is suppressed, leading to inhibition of bovine serum (tetracycline-free fetal bovine serum (Tet-approved FBS, ERa expression, and reduced transcription of E2-dependent Clontech) for conditional cells), 1% L-glutamine and 1% penicillin/ oncogenes. This leads to upregulation of p21Cip1/Waf1 and streptomycin. The packaging cell lines Phoenix GP and HEK293T were induction of senescence. used for generation of retro- or lentiviruses following standard calcium Collectively, these findings suggest that RASSF1A has a central phosphate protocols. MCF7 and T47D cells were transduced using the retroviral vectors pRevTRE (containing the human RASSF1A cDNA) and role in suppressing transformation of human breast epithelial cells 53 40 through ERa-dependent and independent mechanisms that serve pQCXIP (containing rtTA-M2) as previously described. Cells were to activate early barriers to transformation such as senescence. Our selected with puromycin and hygromycin. For all experiments, pooled data therefore shed important new light onto how RASSF1A acts as transduced selected cells rather than single cell clones were used unless a tumor suppressor during the genesis of ERa þ breast tumors. otherwise indicated. Conditional RASSF1A expression was achieved by addition of doxycycline (1 mg/ml). Transduction efficiencies were monitored by flow cytometric detection of eGFP expression. For transient expression of RASSF1A in T47D cells, transfections with pcDNA3.1-RASSF1A MATERIALS AND METHODS were performed with Lipofectamine 2000 (Invitrogen, Karlsruhe, Germany) Plasmids and reagents according to manufacturer’s instructions. A FLAG-tagged human RASSF1A cDNA was cloned into the retroviral vector pRevTRE (Clontech, Mountain View, CA, USA) to allow conditional Other methods 40 transgene expression controlled by doxycycline in the presence of rt TA. Western blot analysis, fluorescence-activated cell sorting analysis, detec- Lentiviral shRNA constructs pLKO.1 CMV tGFP ESR1K.O 1 and 2 and tion of SAb-gal-positive cells and clonogenic assays were performed as scramble were purchased from Sigma-Aldrich (Taufkirchen, Germany). A previously described.40 Quantitative RT--PCR was performed with Sybr cDNA encoding human ERa was cloned into the retroviral vector pQCXIN (ref. 50) Green (Invitrogen) according to manufacturer’s instructions. The primers (Clontech). pBabe-neo, pBabe-Neo-Myr-Flag-AKT1 and pWZL Neo used are listed in Supplementary Information. Myr Flag AKT3,50 pQCXIN-ERa, pBabe-neo-Id1 and pMIG Bcl-2(ref. 51) and pWZL Blast myc52 were used for transduction of RASSF1A-MCF7. E2 (17- estradiol), fulvestrant and doxycycline were purchased from Sigma-Aldrich. Immunohisto- and immunofluorescence staining UO126 and LY294.002 were purchased from Calbiochem (Nottingham, UK). Paraffin-embedded samples of human breast tissue were obtained from Antibodies against ERa, Bcl2, c-Myc, Id1, p21, ERK1, Akt and b-actin were the Department of Obestrics and Gynecology, Johannes Gutenberg

Oncogene (2012) 4912 --4922 & 2012 Macmillan Publishers Limited RASSF1A inhibits ERa S Thaler et al 4921 University, Mainz. Their use approved by the local ethical review board. 10 Migliaccio A, Di Domenico M, Castoria G, de Falco A, Bontempo P, Nola E. Tyrosine 40 Immunohistochemistry was performed as previously described, and kinase/p21ras/MAP-kinase pathway activation by estradiol-receptor complex in quantified using the Automated Cellular Imaging System ACIS III (Dako, MCF-7 cells. EMBO J 1996; 15: 1292 --1300. Glostrup, Denmark). Apopotic cells were detected using the TUNEL 11 Razandi M, Pedram A, Park ST, Levin ER. Proximal events in signaling in plasma Apoptosis Detection Kit (Millipore, Schwalbach, Germany). Double- membrane estrogen receptors. J Biol Chem 2003; 278: 2701 --2712. immunofluorescence staining of ERa and RASSF1A was performed using 12 Simoncini T, Hafezi-Moghadam A, Brizil DP, Ley K, Chin WW, Liao JK. Interaction of Alexa-596-coupled anti-mouse (RASSF1A) and Alexa-488-coupled anti- estrogen receptor with the regulatory subunit of phosphatidylinositol-3-OH kinase. Nature 2000; 407: 538 --541. rabbit (ERa) secondary antibodies. 13 Castoria G, Migliaccio A, Bilancio A, Di Domenico M, de Falco A, Lombardi M. PI3-kinase in concert with Src promotes the S-phase entry of estradiol-stimulated Animal experiments MCF-7 cells. EMBO J 2001; 20: 6050 --6059. Female NOD.SCID CB-17 mice were housed under specific pathogen-free 14 Shoker BS, Jarvis C, Sibson DR, Walker C, Sloane JP. Oestrogen receptor expression in the normal and precancerous breast. J Pathol 1999; 188: 237 --244. conditions. Mice were injected subcutaneously 1 day before implantation 15 Khan SA, Rogers MA, Khurana KK, Meguid MM, Numann PJ. Estrogen receptor of MCF7 cells with E2 (0.1 mg per mouse) dissolved in sesame oil, then expression in benign breast epithelium and breast cancer risk. J Natl Cancer Inst at weekly intervals. Mice were inoculated in the mammary fat pad 1998; 90:37--42. 6 with 5 Â 10 cells. For induction of conditional RASSF1A expression, 16 Lawson JS, Field AS, Champion S, Tran D, Ishikura H, Trichopoulos D. Low drinking water was supplemented with doxycycline (1 mg/ml). Tumor oestrogen receptor alpha expression in normal breast tissue underlies low breast growth was monitored using a caliper. All in-vivo experiments were cancer incidence in Japan. Lancet 1999; 354: 1787 --1788. performed in accordance with institutional guidelines and the German 17 Clarke RB, Howell A, Potten CS, Anderson E. Dissociation between steroid receptor Animal Protection Law. expression and cell proliferation in the human breast. Cancer Res 1997; 57: 4987 --4991. 18 Shaaban AM, Sloane JP, West CR, Foster CS. Breast cancer risk in usual ductal Statistical analysis hyperplasia is defined by estrogen receptor-a and Ki-67 expression. Am J Pathol Differences between experimental groups were assessed using Student’s 2002; 160: 597 --604. t-test (BMDP Statistical Software, Saugus, MA, USA). P-values 19 Shoker BS, Jarvis C, Clarke RB, Anderson E, Munro C, Davis MPA. Abnormal regulations of the oestrogen receptor in benign breast lesions. J Clin Pathol 2000; of 0.05 were considered significant. o 53: 778 --783. 20 Bartkova J, Rezaei N, Liontos M, Karakeidos P, Kletsas D, Issaeya N et al. Oncogene- induced senescence is part of the tumorigenesis barrier imposed by DNA damage CONFLICT OF INTEREST checkpoints. Nature 2006; 444: 633 --637. The authors declare no conflict of interest. 21 Di Micco R, Fumagalli M, Cicalese A, Piccinin S, Gasparini P, Luise C et al. Oncogene-induced senescence is a DNA damage response triggered by DNA hyper-replication. Nature 2006; 444: 638 --642. ACKNOWLEDGEMENTS 22 Perillo B, Sasso A, Abbondanza C, Palumbo G. 17b-Estradiol inhibits apoptosis in MCF-7 cells, inducing bcl-2 expression via two estrogen-responsive elements We are grateful to Selma Huber and workers in the KIT rodent facility and Gitta Flaig present in the coding sequence. Mol Cell Biol 2000; 20: 2890 --2901. for excellent technical assistance. We thank Robert Weinberg for kindly providing 23 Magro G, Bisceglia M, Michal M. Expression of steroid hormone receptors, their pBabe-neo; William Hahn for providing pBabe-Neo-Myr-Flag-AKT1 and pWZL Blast regulated proteins, and bcl-2 protein in myofibroblastoma of the breast. myc and Jean Zhao for pWZL Neo Myr Flag AKT3. pMIG Bcl-2 was created in the Histopathology 2000; 36: 515 --521. laboratory of Stanley Korsmeyer and was kindly provided from Dana Farber Cancer 24 Yang Q, Sakurai T, Jing X, Utsunomiya H, Shan L, Nakamura Y et al. Expression of Institute. This study was supported by Deutsche Forschungsgemeinschaft grant Bcl-2, but not Bax, correlates with estrogen receptor status and tumor TH1523/1-2 to ST and MAIFOR (ST and MS). proliferation in invasive breast carcinoma. Pathol Int 1999; 49: 775 --780. Author contributions: ST designed research; ST, AS and MS performed research, 25 Planas-Silva MD, Bruggeman RD, Grenko RT, Smith JS. Overexpression of c-Myc MS and AS contributed new reagents/analytic tools, ST analyzed data; ST and JPS and Bcl-2 during progression and distant metastasis of hormone-treated breast wrote the paper. cancer. Exp Mol Pathol 2007; 82:85--90. 26 Reed JC, Haldar S, Croce CM, Cuddy MP. Complementation by BCL-2 and C-HA-RAS oncogenes in malignant transformation of rat embryo fibroblast. Mol REFERENCES Cell Biol 1990; 10: 4370 --4374. 1 Dammann R, Schagdarsurengin U, Seidel C, Strunnikova M, Rastetter M, Baier K 27 Strasser A, Harris AW, Bath ML, Cory S. Novel primitive lymphoid tumors induced et al. The tumor suppressor RASSF1A in human carcinogenesis: an update. Histol in transgenic mice by co-operation between myc and bcl-2. Nature 1990; Histopathol 2005; 20: 645 --663. 348: 331 --333. 2 Sunami E, Shinozaki M, Sim MS, Nguyen SL, Vu AT, Giuliano AE et al. Estrogen 28 Dubik D, Shiu RP. Mechanism of estrogen activation of c-myc oncogene receptor and HER2/neu status affect epigenetic differences of tumor-related expression. Oncogene 1992; 7: 1587 --1594. genes in primary breast tumors. Breast Cancer Res 2008; 10: R46. 29 Gartel AL, Ye X, Goufman E, Shianova P, Hay N, Najmabandi F. Myc represses the 3 Feng W, Shen L, Wen S, Rosen DG, Jelinek J, Hu X et al. Correlation between CpG p21(WAF1/CIP1) promoter and interacts with Sp1/Sp3. Proc Natl Acad Sci USA methylation profiles and hormone receptor status in breast cancers. Breast Cancer 2001; 98: 4510 --4515. Res 2007; 9: R57. 30 Zhao JJ, Gjoerup OV, Subramanian RR, Cheng Y, Chen W, Roberts TM et al. Human 4 Shinozaki M, Hoon DS, Giuliano AE, Hansen NM, Wang HJ, Turner R. Distinct mammary epithelial cell transformation through the activation of phosphatidy- hypermethylation profile of primary breast cancer is associated with sentinel linositol 3-kinase. Cancer Cell 2003; 3: 483 --495. lymph node metastasis. Clin Cancer Res 2005; 11: 2156 --2162. 31 Lin CQ, Singh J, Murata K, Itahana Y, Parrinello S, Liang SH et al. A role for Id-1 in 5 Strunnikova M, Schagdarsurengin U, Kehlen A, Garbe JC, Stampfer MR, Dammann the aggressive phenotype and steroid hormone response of human breast cancer R. Chromatin inactivation precedes de novo DNA methylation during the cells. Cancer Res 2000; 60: 1332 --1340. progressive epigenetic silencing of the RASSF1A promoter. Mol Cell Biol 2005; 32 Peverali FA, Ramqvist T, Saffrich R, Barone MV, Philipson L. Regulation of G1 10: 3923 --3933. progression by E2A and Id helix-loop-helix proteins. EMBO J 1994; 13: 4291 --4301. 6 Mueller SO, Clark JA, Myers PH, Korach KS. Mammary gland development in adult 33 Prabhu S, Ignatova A, Park ST, Sun XH. Regulation of the expression of cyclin- mice requires epithelial and stromal estrogen receptor {alpha}. Endocrinology dependent kinase inhibitor p21 by E2A and Id. Mol Cell Biol 1997; 17: 5888 --5896. 2002; 143: 2357 --2365. 34 Ohtani N, Zebedee Z, Hout TJG, Stinson JA, Sugimoto M, Ohashi Y et al. Opposing 7 Levin ER. Integration of the extranuclear and nuclear actions of estrogen. Mol effects of ETS and Id proteins on p16INK4a expression during cellular senescence. Endocrinol 2005; 19: 1951 --1959. Nature 2001; 409: 1067 --1070. 8 Dahlman-Wright K, Cavailles V, Fuqua SA, Jordan VC, Katzenellenbogen JA, Korach 35 Swabrick A, Roy E, Allen T, Bishop JM. Id1 cooperates with oncogenic Ras to KS. International Union of Pharmacology. LXIV. Estrogen receptors. Pharmacol Rev induce metastatic mammary carcinoma by subversion of the cellular senescence 2006; 58: 773 --781. response. Proc Natl Acad Sci USA 2008; 105: 5402 --5407. 9 Keshamouni VG, Mattingly RR, Reddy KB. Mechanism of 17-b-estradiol-induced 36 Norton JD, Atherton GT. Coupling of cell growth control and apoptosis functions Erk1/2 activation in breast cancer cells. J Biol Chem 2002; 277: 22558 --22565. of Id proteins. Mol Cell Biol 1998; 18: 2371 --2381.

& 2012 Macmillan Publishers Limited Oncogene (2012) 4912 --4922 RASSF1A inhibits ERa S Thaler et al 4922 37 Campell RA, Bhat-Nakshatri P, Patel NM, Constantinidou D, Ali S, Nakshatri H. ER-positive mammary tumors that mimic human breast cancer. Carcinogenesis Phosphatidylinositol 3-kinase/AKT-mediated activation of estrogen receptor 2007; 28: 584 --594. alpha: a new model for anti-estrogen resistance. J Biol Chem 2001; 46 Chen L, Qui J, Yang C, Yang X, Chen X, Jiang J et al. Identification of a 276: 9817 --9824. novel estrogen receptor beta1 binding partner, inhibitor of differentiation-1, 38 Faridi J, Wang L, Endemann G, Roth RA. Expression of constitutively active Akt-3 in and role of ERbeta1 in human breast cancer cells. Cancer Lett 2009; 278: MCF-7 breast cancer cells reverses the estrogen and tamoxifen responsivity of 210 --219. these cells in vivo. Clin Cancer Res 2003; 9: 2933 --2939. 47 Migliaccio A, Castoria G, Di Domenico M, De Falco A, Bilancio A, Auricchio F. 39 Kasid A, Lippman ME, Papageorge AG, Lowy DR, Gelmann EP. Transfection of Src is an initial target of sex steroid hormone action. Ann NY Acad Sci 2002; v-rasH DNA into MCF-7 human breast cancer cells bypasses dependence on 963: 185 --190. estrogen for tumorigenicity. Science 1985; 228: 725 --728. 48 Razandi M, Oh P, Pedram A, Schnitzer J, Levin ER. ERs associate with and 40 Thaler S, Ha¨hnel PS, Schad A, Dammann R, Schuler M. RASSF1A mediates recapitulate the production of caveolin: implications for signalling and cellular p21Cip1/Waf1-dependent cell cycle arrest and senescence through modulation of actions. Mol Endocrinol 2002; 16: 100 --115. the Raf-MEK-ERK pathway and inhibition of Akt. Cancer Res 2009; 69: 1748 --1757. 49 Moelling K, Schad K, Bosse M, Zimmermann S, Schwenker M. Regulation of Raf- 41 Fowler AM, Solodin NM, Valley CC, Alarid ET. Altered target gene regulation Akt cross-talk. J Biol Chem 2002; 277: 31099 --31106. controlled by estrogen receptor-a concentrations. Mol Endocrinol 2006; 50 Boehm JS, Zhao JJ, Yao J, Kim SY, Firestein R, Dunn IF et al. Integrative 20: 291 --301. genomic approaches identify IKBKE as a breast cancer oncogene. Cell 2007; 42 Holst F, Stahl PR, Ruiz C, Hellwinkel O, Jehan Z, Wendland M et al. Estrogen 129: 1065 --1079. receptor alpha (ESR1) amplification is frequent in breast cancer. Nat Genet 2007; 51 Cheng EH, Wei MC, Weiler S, Flavell RA, Mak TW, Lindsten T et al. BCL-2, BCL-X(L) 39: 655 --660. sequester BH3 domain-only molecules preventing BAX- and BAK-mediated 43 Li T, Sotgia F, Vuolo MA, Li M, Yang WC, Pestell RG et al. Caveolin-1 mutations in mitochondrial apoptosis. Mol Cell 2001; 8: 705 --711. human breast cancer: functional association with estrogen-receptor alpha- 52 Boehm JS, Hession MT, Bulmer SE, Hahn WC. Transformation of human positive status. Am J Pathol 2006; 168: 1998 --2013. and murine fibroblasts without viral oncoproteins. Mol Cell Biol 2005; 25: 44 Kondo N, Toyama T, Sugiura H, Fujii Y, Yamashita H. miR-206 expression is down- 6468 --6474. regulated in estrogen receptor alpha-positive human breast cancer. Cancer Res 53 Urlinger S, Baron U, Thellmann M, Hasan MT, Bujard H, Hillen W. Exploring the 2008; 68: 5004 --5008. sequence space for tetracycline-dependent transcriptional activators: novel 45 Blanco-Aparicio C, Pe´rez-Gallego L, Pequenõ B, Leal JF, Renner O, Carnero A. mutations yield expanded range and sensitivity. Proc Natl Acad Sci USA 2000; Mice expressing myrAKT in the mammary gland develop carcinogen-induced 97: 7963 --7968.

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

Oncogene (2012) 4912 --4922 & 2012 Macmillan Publishers Limited