Eliminating Epigenetic Barriers Induces Transient Hormone-Regulated Gene Expression in Estrogen Receptor Negative Breast Cancer Cells

Oncogene (2008) 27, 4075–4085 & 2008 Nature Publishing Group All rights reserved 0950-9232/08 $30.00 www.nature.com/onc ORIGINAL ARTICLE Eliminating epigenetic barriers induces transient hormone-regulated gene expression in estrogen receptor negative breast cancer cells

L Fleury1,2, M Gerus1,2,3, AC Lavigne1,2,3, H Richard-Foy1,2 and K Bystricky1,2

1Laboratoire de Biologie Mole´culaire Eucaryote, Universite´ de Toulouse, Toulouse, France and 2UMR 5099, CNRS, Toulouse, France

In breast cancer, approximatelyone-third of tumors 2004). Estrogens exert their function by binding to the express neither the estrogen receptor (ERa) nor estro- estrogen receptor-a (ERa), a transcription factor con- gen-regulated genes such as the progesterone receptor trolling the expression of genes involved in cell gene (PR). Our studyprovides new insights into the proliferation and differentiation. While two-thirds of mechanism allowing hormone-activated expression of breast cancers are ERa-positive, nearly one-third of ERa target genes silenced in ERa-negative mammary breast tumor cells do not express ERa.ERa-negative tumor cells. In cell lines derived from ERa-negative tumors are of higher grade and highly invasive (Sorlie MDA-MB231 cells, stable expression of different levels of et al., 2001) and are not responsive to endocrine ERa from a transgene did not result in transcription of therapies designed to blockER a function. PR. A quantitative comparative analysis demonstrates The absence of ERa does not appear to be a result of that inhibiting DNA methyltransferases using 5-aza-20- mutations that are found in less than 1% of ER-negative deoxycytidine or speciﬁc disruption of DNMT1 by small tumors (Roodi et al., 1995; Lacroix et al., 2004). interfering RNAs and treatment with the histone-deace- Moreover, in ERa-negative breast cancer cells, silencing tylase inhibitor trichostatin A enabled ERa-mediated of the ERa gene is accompanied by alteration of hormone-dependent expression of endogenous PR.We estrogen-regulated gene expression. Indeed, some genes show that demethylation of a CpG island located in the such as the trefoil factor 1 gene (pS2/TFF1) or the ﬁrst exon of PR was a prerequisite for ERa binding progesterone receptor gene (PR) are silenced while to these regulatorysequences. Although not a general other genes become constitutively expressed (such as requirement, DNA demethylation is also necessary for Cathepsin D (Cat-D) (Touitou et al., 1991)). Whether or derepression of a subset of ERa target genes involved in not the silencing of a subset of estrogen-regulated genes tumorigenesis. PR transcription did not subsist 4 days is a direct consequence of the absence of ERa remains after removal of the DNA methyltransferase blocking debated. agents, suggesting that hormone-induced expression of Furthermore, the consequences of de novo ERa ERa target genes in ERa-negative tumor cells is transient. expression for the expression of hormone-regulated Our observations support a model where an epigenetic genes that are normally silenced in these cells remain mark confers stable silencing byprecluding ER a access to controversial. Indeed, transient vector-based expression promoters. of ERa under the control of a SV40 promoter in ERa- Oncogene (2008) 27, 4075–4085; doi:10.1038/onc.2008.41; negative mammary tumor cell lines such as MDA- published online 3 March 2008 MB231 did not lead to stimulated transcription of either pS2/TFF1 or PR. Yet, hormone-induced expression of Keywords: breast cancer; chromatin; ERa; PR; methy- pS2/TFF1 was observed after infection of MDA-MB231 lation with an adenovirus carrying the ERa gene (Lazennec et al., 1999) or stable expression of ERa under the control of a CMV promoter (Metivier et al., 2003). Remarkably, hormone-induced expression of PR has never been observed in any of these systems. Introduction Complete loss of ERa expression in ERa-negative mammary tumor cell lines has been linked to aberrant Estradiol is a key regulator for normal growth and hypermethylation of a CpG island contained in its differentiation of mammary glands as well as for the promoter (Ottaviano et al., 1994; Roodi et al., 1995). In malignant progression of breast cancer (Platet et al., recent years, epigenetic mechanisms of gene regulation have been increasingly associated with establishment Correspondence: Dr K Bystricky, Laboratoire de Biologie Mole´ culaire and progression of cancer (Jones and Baylin, 2002; Eucaryote, Universite´ de Toulouse, IBCG, 118 route de Narbonne, Momparler, 2003; Ting et al., 2006). DNA cytosine Toulouse, Midi Pyrenees 31062, France. methyltransferase 1 (DNMT1) is the most abundant E-mail: [email protected] 3These authors contributed equally to this work. and catalytically active DNA methyltransferase. It Received 28 August 2007; revised 27 December 2007; accepted 31 induces the covalent addition of a methyl group to the January 2008; published online 3 March 2008 50 position of cytosine, predominantly within CpG PR promoter accessibility and ERa levels L Fleury et al 4076 dinucleotides whose occurrence in the human genome is generally increased in promoter regions (Lande-Diner et al., 2007). DNMTs can be inhibited by the nucleoside MCF-7 MDA-MB231 HE-5 MDA-66 ERα antimetabolite 5-aza-20-deoxycytidine (AZA) (Jones GAPDH and Taylor, 1980). In addition, chromatin at inactive promoters is generally hypoacetylated by histone deacetylases (HDAC) that can be inhibited by HDAC 100 inhibitors such as Trichostatin A (TSA). Because 80 DNMT1 has been found to interact physically with 60 [%]

HDACs, DNA-methylation and histone deacetylation α 40 may function through a common mechanistic pathway ER

to repress transcription (Rountree et al., 2000; Dobosy 20 and Selker, 2001; Robertson, 2002). The use of both Relative expression of 0 TSA and AZA has synergistic effects in activating MCF-7 MDA-MB231 HE-5 MDA-66 epigenetically silenced genes (Cameron et al., 1999). In

particular, a number of tumor suppressor genes that are 10 β-actin speciﬁcally silenced in various cancers have been PR reactivated using a combination of both drugs (Bachman 8 et al., 2003; Lopez-Serra et al., 2006). PS2/TFF1 The goal of the study described here was to investigate 6 the role of ERa and chromatin structure in regulating expression of the progesterone receptor gene. Here, we 4

used two MDA-MB231-derived cell lines that stably 2 express different levels of ERa to quantify PR expres- Relative mRNA expression sion in response to general DNA methyltransferase 0 inhibition, to speciﬁc disruption of DNMT1 by small MCF-7 MDA-MB231 HE-5 MDA-66 interfering RNA, and to inhibitors of HDACs. We 2.5 β-actin combined chromatin immunoprecipitation (ChIP) ana- Cat-D lysis and bisulﬁte sequencing to determine whether the 2 extent of ERa binding correlates with the methylation status of regulatory sequences of the PR gene. Finally, 1.5

we demonstrate that the epigenetic mechanism involved 1 in repression of PR can be extended to a subset of ERa target genes silenced in ERa-negative cells. 0.5 Relative mRNA expression

0 MCF-7 MDA-MB231 HE-5 MDA-66 Results Figure 1 Absence of progesterone receptor gene (PR) and pS2/ TFF1 gene expression but not Cat-D in MDA-MB231-derived breast cancer cell lines. (a) Analysis of ERa expression. Top panel: Stable ERa expression is not sufﬁcient to promote western blot analysis. Cells stimulated with 10À8 M E2 were hormone-induced PR expression in ERa-negative breast immunoblotted. Bottom panel: western blot quantiﬁcation. ERa/ cancer cell lines GAPDH ratio in MCF-7 cells set to 100%. (b) quantitative real- We analysed the expression levels of ERa target genes time (qRT)-PCR using primer sets for progesterone receptor gene (PR), pS2/TFF1 or b-actin (as an internal control). (c) qRT-PCR that are silenced in ERa-negative cell lines after ectopic, using primer sets for Cat-D or b-actin. Expression levels from E2- stable expression of ERa to different levels. This treated relative to mock-treated cell lines as indicated (see Materials approach allowed us to investigate the effect of ERa and methods) are represented. Data shown are the average of three expression in a homogenous and controlled back- independent experiments, error bars represent ±s.e. mean. ground. Steady state ERa protein levels in HE-5 (Touitou et al., 1991) and MDA-66 cell lines (Metivier et al., 2004) treated with estradiol (E2) were 10 and increase (otwofold) in pS2/TFF1 transcription in 40%, respectively, of the one found in MCF-7 cells MDA-66 cells, this level of expression is smaller than (Figure 1a). Hormone-induced expression of three the one reported previously (Metivier et al., 2003) and estradiol-regulated genes, PR, pS2/TFF1 and Cat-D, negligible relative to the one observed in MCF-7 cells. was analysed by quantitative real-time (qRT)-PCR in Here, we provided a complete analysis using qRT-PCR these cell lines (Figures 1b and c). In MCF-7 cells, and ChIP followed by qPCR. Because all studies were estrogens stimulated PR expression 2.5-fold, pS2/TFF1 performed in parallel, the results for different genes and expression nearly 10-fold and Cat-D twofold. In ERa- under different conditions can be compared quantita- negative MDA-MB231 cells these genes are not induced tively. Although we cannot exclude that the ERa by E2 (Figures 1b and c). Surprisingly, we did not transgene bearing cell lines provided by F Gannon observe hormone regulation of PR expression in HE-5 (MDA-66) may have changed phenotype over time and or MDA-66 (Figure 1b). While we observed a small cultures, we did not observe any differences in mRNA

Oncogene PR promoter accessibility and ERa levels L Fleury et al 4077 levels from cells before and after dozens of passages in DNA demethylation and histone deacetylation trigger our laboratory (data not shown). hormone-dependent expression of PR and pS2/TFF1 We confirmed that the presence of ERa in the cell is in MDA-MB231-derived cell lines, but do not affect sufficient to confer hormone-regulated Cat-D expres- hormone-regulation of Cat-D expression sion, since both HE-5 and MDA-66 exhibit a 1.8-fold Limited access of ERa to PR and pS2/TFF1 regulatory increase in Cat-D mRNA levels similar to the twofold sequences may result from epigenetic mechanisms, stimulation observed in MCF-7 cells (Figure 1c). These involving DNA methylation, as has been demonstrated results taken together with previous observations that for the ERa gene itself (Yang et al., 2001). Thus, we the chromatin structure of the PR and pS2/TFF1 quantified expression levels of PR, pS2/TFF1 and Cat-D promoters remains insensitive to DNAseI cleavage in genes following AZA, TSA or combination treatments the HE-5 cell line (Giamarchi et al., 1999) led us to (Figure 2a). Treatment of MDA-66 cells with TSA, led postulate that promoter accessibility may hamper ERa to a 2- and a 10-fold induction of PR and pS2/TFF1 binding to PR and, to some extent, to pS2/TFF1 in mRNAs respectively. Cat-D gene expression was not MDA-MB231-derived cell lines. affected by these treatments. Exposure to AZA resulted

TSA AZA E2 Stop 50 MDA-66 -72h0 +16h β-actin 40 Cat-D PR 30 PS2/TFF1

10 Relative mRNA expression

0 E2E2/TSA E2/AZA E2/TSA-AZA

TSA AZA E2+/-ICI Stop 50 MDA-66 -72h0 +16h PR 40 PS2/TFF1

Relative mRNA expression 10

0 - E2 E2/ICI - E2 E2/ICI - E2 E2/ICI - E2 E2/ICI - TSA AZA TSA-AZA Figure 2 DNA demethylation and histone deacetylation trigger hormone-dependent expression of progesterone receptor gene (PR) and pS2/TFF1 in MDA-66 cells, but do not affect Cat-D expression. (a) quantitative real-time (qRT)-PCR using primer sets for PR, pS2/TFF1, Cat-D or b-actin in MDA-66 cells. Cells were treated with 5 mM of 5-aza-20-deoxycytidine) (AZA) before addition of 50 mgmlÀ1 of Trichostatin A (TSA) and E2 10À8 M or EtOH. Expression levels from E2-treated relative to mock-treated cells are represented. (b) Expression of PR and pS2/TFF1 in MDA-66 cells is mediated by ERa. Cells were incubated with AZA and TSA as in (a) before treatment with E2 and ICI 10À7 M. qRT-PCR analysis using primer sets for PR and pS2/TFF1.

Oncogene PR promoter accessibility and ERa levels L Fleury et al 4078 in a dramatic increase in hormone-stimulated expres- Figure 2b). These results suggest that DNA demethyla- sion of PR (10-fold) and pS2/TFF1 (35-fold) that was tion directly affects target gene promoter accessibility by significantly higher than their expression levels observed increasing basal transcription through an ERa-indepen- after TSA treatment. Combining AZA and TSA dent mechanism. treatments resulted in even greater hormone-stimulated PR and pS2/TFF1 mRNA synthesis. Simultaneous DNA demethylation is pivotal in PR transcription treatment of the cells that had been exposed to TSA, regulation AZA or a combination treatment, with estradiol and the Methylation of CpG dinucleotides is maintained by the pure anti-estrogen ICI 182 780 abolished the hormone- action of DNMT1 (Pradhan et al., 1999). We investi- induced increase of pS2/TFF1 or PR transcription gated the role of DNMT1 in PR and PS2/TFF1 gene (Figure 2b). As ICI binds to ERa and induces its silencing in MDA-66 cells using a siRNA based degradation (Wijayaratne and McDonnell, 2001), this approach to deplete DNMT1 protein. Western blot inhibition demonstrates that hormone-regulated activa- analysis (Figure 3a) demonstrates significant reduction tion of ERa target gene expression is dependent on of DNMT1 protein levels present in MDA-66 trans- ERa. We noted, however, that in the absence of fected with two distinct siRNAs directed against estradiol, expression levels of both PR and pS2/TFF1 DNMT1. Electroporation with a combination of the were significantly higher in AZA-treated cells compared two siRNAs had the same effect as each siRNA alone. to untreated cells and identical to the levels observed for We observed a significant increase in hormone-stimu- cells treated with E2/ICI (a 5- or 15-fold induction; lated accumulation of PR (fivefold) and pS2/TFF1

si RNA electroporation E2 Stop

10 β-actin -32h0 +16h 9 Cat-D 8 PR 7 PS2/TFF1 6

no transfection mock transfection siDNMT1.1 siDNMT1.2 siDNMT1.1 and 1.2 5 DNMT1 4 GAPDH 3 MDA-66 2

Relative mRNA expression 1 0 no transfection si mock si DNMT1.1 si DNMT1.2 si DNMT1.1 si DNMT1.2

9 40 β-actin β-actin 8 PR 35 PR 7 PS2/TFF1 30 PS2/TFF1 6 25 5 20 4 15 3 2 10 Relative mRNA expression

1 Relative mRNA expression 5 0 0 - E2 E2/ICI - si DNMT1 AZAsi DNMT1 AZA si RNA E2 4 days after electroporation +/- ICI Stop

-32h0 +16h Figure 3 DNA demethylation using 5-aza-20-deoxycytidine (AZA) or siDNMT1 triggers hormone-dependent expression of progesterone receptor gene (PR) and pS2/TFF1 in MDA-66 cells, but do not affect Cat-D expression. (a) Western blot analysis. MDA-66 cells were transfected mockor two distinct siRNA against DNMT1. Cellular proteins from siRNA-treated and control cells were immunoblotted. (b) quantitative real-time (qRT)-PCR analysis using primer sets for PR, pS2/TFF1, Cat-D, b-actin. (c) Expression of PR and pS2/TFF1 in MDA-66 cells is mediated by ERa. Cells were transfected with both siDNMT1.1 and/or siDNMT1.2 as in (a) before addition of E2 10À8 M±ICI 10À7 M. RT-PCR analysis using primer sets for PR, pS2/TFF1 and b-actin. Expression levels are represented relative to mock-transfected, mock-treated cells. (d) Expression of PR and pS2/TFF1 in MDA-66 cells treated with AZA for 72 h or transfected with siDNMT1 is transient. qRT-PCR analysis using primers for PR, PS2/TFF1, Cat-D and b-actin. Expression levels were analysed immediately following treatment or 4 days after removal of treatments.

Oncogene PR promoter accessibility and ERa levels L Fleury et al 4079 (sevenfold) mRNAs 48 h post siDNMT1 transfection ChIP to regulatory sequences of PR known to lead to (Figure 3b). In contrast, upon siDNMT1 transfection estrogen activation using the B promoter (Figure 5a). Cat-D expression remained inducible and unchanged This region located in the first exon contains a half ERE, compared to expression levels found in mock-trans- is associated with an AP1 site and contains a CpG island fected cells. Furthermore, in MDA-66 cells transfected (Petz et al., 2004). We observed hormone-induced with siDNMT1 treated with E2 in combination with ICI binding of ERa to pS2/TFF1 promoter (14-fold the estradiol-induced increase in PR and pS2/TFF1 increase) and PR regulatory regions (12-fold increase) expression was abolished (Figure 3c). Thus, the ob- in MCF-7 cells, but ERa was never recruited to these served transcription after DNMT1 depletion is mediated regions in MDA-MB231-derived cells (Figure 5b). In by ERa. Steady state levels of mRNA after AZA contrast to the lackof ER a binding to these promoters treatment compared to levels after siDNMT1 transfec- observed in HE-5 and MDA-66 cells, siDNMT1 or tion were roughly twofold for PR and fourfold for pS2/ AZA treatments enabled estrogen-stimulated ERa TFF1 (Figure 3d). This could stem from the conse- binding to PR and pS2/TFF1 regulatory regions in quence of the more general AZA treatment that affects MDA-66 cells (Figure 5c). Simultaneous treatment with all DNA-methyltransferases, and thus indirectly from E2 and ICI abolished hormone-dependent recruitment transcription of number of other genes. When quantify- of ERa. To ascertain that the methylation status of the ing PR and pS2/TFF1 mRNAs 4 days after complete regulatory regions of PR was the cause for the lackof removal of AZA or TSA treatments we found that hormone-induced ERa recruitment, we determined the neither PR nor pS2/TFF1 was expressed. This indicates methylation status of a CpG island located within the that DNA demethylation causes transient re-expression PR regulatory region (Figure 5d) using bisulfite sequen- of these two genes. The reversibility of the repressive cing in MDA-66 cells before and after exposure to AZA chromatin structure suggests that an underlying signal (Figures 5a and d). Amplified DNA sequences from independent of the DNA methylation status may render AZA-treated cells exhibited a significantly lower percen- these promoters prone to silencing in MDA-MB231- tage of methylated CpGs than sequences from untreated derived cell lines. cells. Taken together, ChIP and bisulfite sequencing results suggest that ERa access to the PR regulatory regions is blocked by CpG methylation. DNA demethylation and histone deacetylation trigger In addition, ERa fixation to PR regulatory sequences hormone-dependent expression of ERa in in MDA-MB231, HE-5 or MDA-66 cells treated with MDA-MB231-derived cells AZA increases with intracellular ERa protein levels We observed an important increase in hormone- (compare Figures 6 and 4). We conclude that DNA regulated PR and pS2/TFF1 gene expression after demethylation allows quantitative binding of ERa to AZA and AZA/TSA treatments. Knowing that such promoter sequences of ERa target genes leading to treatments also affect endogenous ERa gene expression, hormone-stimulated gene expression. we investigated how intracellular ERa levels influence the extent of estrogen-regulated target gene expression. PR and pS2/TFF1 expression strongly increased reach- The epigenetic mechanism of PR silencing is common to ing respectively 8- and 12-fold stimulations in HE-5 cells some but not all ER target genes that are repressed and a 20-fold stimulation in MDA-66 cells (Figure 4a). in ER-negative cell lines Concomitantly, an increase in ERa protein levels was Finally, we analysed the expression levels of several, also noted (Figure 4b). Quantification of western blots newly identified estrogen receptor target genes to test from three independent experiments revealed a max- whether our observations could be extended to genes imum fivefold increase in ERa protein levels present in distinct from PR and PS2/TFF1. We crossed several HE-5 cells and a maximum 20-fold increase in MDA-66 studies from the recent literature based on microarray cells (Figure 4c). Thus, stimulation of ERa target gene transcriptome analyses and ChIP-chip, ChIP-DSL or transcription appears to tightly correlate with the ChIP-paired end di-tag data obtained from mammary quantity of ERa protein available in the cell. However, tumor cell lines (Lin et al., 2004, 2007; Laganiere et al., expression of PR and pS2/TFF1 is not dictated by ERa 2005; Kininis et al., 2007; Kwon et al., 2007) to select a protein thresholds in the absence of treatments modify- group of relevant genes based on the following criteria: ing chromatin structure. Indeed, transient transfection hormone-stimulated gene expression in ERa-positive of MDA-66 cells with increasing quantities of a vector mammary tumor cell lines, complete lackof expression bearing the ERa gene resulting in ERa protein levels up in ERa-negative breast cancer cell lines (Nagaraja et al., to 20 times the one found in MCF-7 cells did not allow 2006) and ERa target identified by ChIP. Thus, we target gene transcription (data not shown). investigated gene regulation of Annexin A9 (ANXA9), a gene coding for a protein of the annexin superfamily CpG methylation of the PR promoter impedes ERa (Raynal and Pollard, 1994; Gerke and Moss, 2002), the fixation in MDA-MB231-derived breast cancer cell lines RET proto-oncogene, which encodes a protein receptor expressing ERa tyrosine kinase with a zinc finger domain associated We next wanted to know whether epigenetic modifica- with dominantly inherited cancer syndromes tions of regulatory sequences of PR and pS2/TFF1 (Goodfellow and Wells, 1995), the tumor protein D52-like impede ERa binding. We analysed ERa binding using 1(TPD52L1 (D53)) upregulated in human breast and

Oncogene PR promoter accessibility and ERa levels L Fleury et al 4080 TSA PS2/TFF1 mRNA 50 AZA E2 Stop PR mRNA 45 -72h0 +16h 40 35 30 25 21 x 20 15

Relative mRNA expression 10 20 x

5 12 x 8 x 0 - TSA AZA TSA - TSA AZA TSA - TSA AZA TSA AZA AZA AZA MDA-MB231 HE-5 MDA-66

ERα ERα ERα GAPDH GAPDH GAPDH - TSA AZA TSA - TSA AZA TSA - TSA AZA TSA EtOH AZA AZA AZA MDA-MB231 HE-5 MDA-66 MCF-7

TSA AZA E2 Stop 10 ERα protein 9 -72h0 +16h 8 7 6 expression α 5 4 20 x 3

Relative ER 2 1 5 x 0 - TSA AZA TSA - TSA AZA TSA - TSA AZA TSA AZA AZA AZA MDA-MB231 HE-5 MDA-66 Figure 4 The extent of progesterone receptor gene (PR) and PS2/TFF1 gene activation correlates with ERa protein levels MDA- MB231, HE-5 and MDA-66 cells stimulated as in Figure 2. (a) quantitative real-time (qRT)-PCR analysis using primers for PR and PS2/TFF1. Expression levels are represented relative to mock-treated cells. (b) Top panel: western blot analysis of ERa and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) after total protein extraction of cells treated as in (a). Bottom panel: ImageQuant analysis of three independent western blot analyses. Protein levels are represented relative to ERa protein concentration measured in untreated MCF7 cells.

prostate cancers (Balleine et al., 2000; Ahram et al., (Figure 7a) and assessed changes in relative expression 2002; Pollack et al., 2002; Rubin et al., 2004), bone levels in the MDA-66 cells in the absence and in the morphogenetic protein-6 (BMP6), a multifunctional presence of AZA. molecule of the transforming growth factor-b super- Similar to PR, the expression of ERa is not sufﬁcient family overexpressed in breast, prostate and salivary to induce estrogen regulated expression of ANXA9, gland cancers (Hamdy et al., 1997; Heikinheimo et al., and, to a lesser extent, of the RET proto-oncogene in 1999) and the gene regulated in breast cancer 1 MDA-66 cells (Figure 7b). Hormone-dependent stimu- (GREB1), whose product contributes to the growth- lation of transcription of these genes required DNA promoting effects of estrogens in MCF-7 cells (Rae demethylation (Figure 7b). In contrast, expression of et al., 2005). We conﬁrmed that expression of these TPD52L1 was weakly hormone regulated in the genes was stimulated by estradiol in MCF7 cells presence of ectopic ERa in MDA-66 cells (Figure 7b).

Oncogene PR promoter accessibility and ERa levels L Fleury et al 4081 18 β-actin promoter Promoter B Promoter A 16 PR promoter PS2/TFF1 promoter 1/2 ERE 1/2 ERE 14 (+83) SP1 (+571) SP1 SP1 (-49) (+580) 12 (-61) SP1 AP1 10 CCAAT (+590) (+90) (-70) 8 +1 +1

Fold induction 6 4 -711+31 +464+751 +1105 BSP 2 ChIP 0 MCF-7 MDA-MB231 HE-5 MDA-66

14 MDA-66 β-actin promoter MDA-66 mock 5mCpG PR promoter CpG 12 PS2/TFF1 promoter

6 ERE/AP1

Fold induction +58 +254

4 MDA-66 AZA treatment

0 - E2 E2/ICI - E2 E2/ICI AZA si DNMT1

E2 si RNA E2 ERE/AP1 AZA +/- ICI Stop electroporation +/- ICI Stop +58 +254

-87h0 +1h -47h0 +1h Figure 5 DNA demethylation triggers hormone-dependent estrogen receptor (ERa) recruitment to progesterone receptor gene (PR) and pS2/TFF1 promoters in MDA-66 cells. (a) Organization of the PR gene regulatory region. The gray area is CpG rich. Fragments amplified for Chromatin immunoprecipitation (ChIP) (b) and BSP (d) analyses are indicated. (b) ChIP followed by qPCR analysis of ERa recruitment to PR, PS2/TFF1 and b-actin promoters. MCF-7, MDA-MB231, HE-5 and MDA-66 cells were stimulated with E2 10À8 M.(c) ChIP followed by qPCR analysis of ERa recruitment to PR, PS2/TFF1 and b-actin promoters. MDA-66 cells were stimulated with E2 ± ICI following AZA treatment or siDNMT1.1 post transfection. Results are shown as fold induction of ERa binding upon demethylation. (d) Bisulfite Sequencing (BSP) analysis of the CpG methylation status of the CpG island containing PR regulatory region. Genomic DNA from MDA-66 cells treated with AZA or mock-treated was exposed to bisulfite. PCR amplified fragments of the region of interest were cloned and sequenced. Methylation patterns are represented by a succession of circles symbolizing the methylation status of cytosines part of CpG di-nucleotides; filled: methylated, open: unmethylated.

Transcription of TPD52L1 further increased in the Discussion presence of AZA. For BMP6 and for GREB1 the presence of ectopic ERa is also sufﬁcient for hormone- Our study provides new insights into the mechanisms regulated expression (Figure 7b). Exposure to AZA allowing hormone-activated expression of ERa target did not increase BMP6 or GREB1 expression levels genes that are silenced in ERa-negative mammary suggesting that DNA methylation is not directly tumor cell lines. We demonstrated that the ﬁrst step involved in downregulation of these genes in MDA- needed to stimulate estrogen regulated gene expression MB231 cells. Previously demonstrated upregulation of of PR and a subset of estrogen target genes silenced in BMP6 gene expression in AZA-treated MDA-MB231 ERa-negative mammary tumor cell lines is to eliminate (Zhang et al., 2007) was thus directly due to re- epigenetic barriers to allow ERa access to regulatory expression of ERa. sequences. In particular, we demonstrated that ERa

Oncogene PR promoter accessibility and ERa levels L Fleury et al 4082 12 MDA-66 β-actin promoter ERa to activate gene expression as shown recently for PR promoter the pS2/TFF1 promoter (Baron et al., 2007) could be 10 altered in ERa-negative breast cancer cells. We postulate that a distinct chromatin environment that sur- rounds ERa target genes in their silent conformation 8 may translate into distinct requirements for cofactors that are necessary to overcome the repressive chromatin 6 structure during gene activation. Finally, our observation that hormone-induced ex-

Fold induction 4 pression of ERa target genes in ERa-negative tumor cells is transient supports a model where an epigenetic markconfers stable silencing by precluding ER a access 2 to promoters. Silent ERa target gene promoters have also been shown to accumulate DNA methylation 0 following induced depletion of ERa (Leu et al., 2004). MDA-MB231 HE-5 MDA-66 Nevertheless the opposite is not true. In fact, we showed AZA that the presence of ERa in cells expressing PR as a AZA E2 Stop result of AZA treatment is not sufficient to maintain an open chromatin conformation and to sustain PR -87h0 +1h expression (Figure 3). In a different approach, DNA Figure 6 Quantitative estrogen receptor (ERa) recruitment to methylation was erased through mutating DNMT1 and demethylated progesterone receptor gene (PR) regulatory DNMT3, which allowed activation of the tumor sequences. ChIP followed by quantitative real-time (qRT)-PCR suppressor gene p16INK4 normally found to be silenced analysis of ERa recruitment to PR regulatory sequences in in colorectal cancer cells (Bachman et al., 2003). MDA-MB231, HE-5 and MDA-66 cells. Cells were stimulated following 5-aza-20-deoxycytidine (AZA) treatment. Results are However, a weeklater this gene was again completely shown as fold induction of ERa binding upon E2 stimulation. silenced indicating that re-expression was transient. Here, silencing was possible through epigenetic mod- ifications in the absence of DNA methylation, thus binding to PR was specifically impaired by DNA suggesting that a distinct markspecifically identifies hypermethylation of CpGs downstream of the PR B genes targeted for long-term repression. The identifica- promoter. In addition, we showed that the level of ERa tion of the signal that imposes selective pressure to target gene expression correlates positively with ERa silence a subset of genes in breast cancer cells may open protein levels and binding present in breast cancer cells new avenues for the development of new therapeutic (Figures 4 and 6). agents. Some reports claim that genes such as PR or pS2/ TFF1 could be expressed simply by ectopically introducing ERa into these cell lines (Lazennec et al., 1999; Materials and methods Metivier et al., 2003), while other reports indicate that expression of ERa is not sufficient to transcribe these Reagents silenced genes (Touitou et al., 1991; Giamarchi et al., Estradiol, AZA and TSA were purchased from Sigma-Aldrich 1999). However, it was previously shown that treatment (Saint-Quentin Fallavier, France). ICI 182780 (ICI) was of ERa-negative breast cancer cell lines such as MDA- purchased from TOCRIS (MO, USA). Antibodies, rabbit MB231 or MDA-MB435 with DNA methylation polyclonal anti-ERa antibodies (HC20) were purchased from inhibitors can lead to expression of ERa protein (Yang SantaCruz (Le Perray en Yvelynes, France), mouse monoclonal anti-Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) et al., 2001). Our data demonstrate that although some antibodies from Chemicon International (Hants, UK) and ERa target genes that are silenced in MDA-MB231 mouse monoclonal anti-DNMT1 antibodies from Imgenex cells, including TPD52L1, BMP6 and GREB1,canbe (San Diego, CA, USA). expressed by re-introducing ERa in ERa-negative cell lines, a subset of ERa target genes, including PR, Cell lines and tissue culture ANXA9 and RET, require DNA demethylation of their MCF-7 cells were maintained in Dulbecco’s modified Eagle’s regulatory regions in addition to the presence of ERa medium (DMEM) F-12 with Glutamax containing 50 mgmlÀ1 for transcription in MDA-MB231-derived cells. In gentamicin, 1 mM sodium pyruvate and 10% heat-inactivated contrast to PR, AZA induced overexpression of ERa fetal calf serum (FCS) (Invitrogen, Cergy Pontoise, France). did not lead to increased transcription of BMP6 and MDA-MB231 were maintained in DMEM containing 4.5 g lÀ1 GREB1. To induce transcription of these genes, it is not glucose, 50 mgmlÀ1 gentamicin, 1 mM sodium pyruvate, and necessary to eliminate epigenetic barriers for ERa 10% heat-inactivated FCS. HE-5 and MDA-66 cells are stable binding even at low local ERa protein concentrations. cell lines derived from MDA-MB 231 cell lines: the HE-5 cell line carries an ERa transgene under the control of a SV40 The observed heterogeneity of repression mechanisms promoter (Touitou et al., 1991) and the MDA-66 cell line may be related to the structure of regulatory sequences carries an ERa transgene under the control of a CMV of these genes. The interplay between ERa and other promoter (Metivier et al., 2004). HE-5 and MDA-66 cells transcription activators that modulate the ability of were maintained under the same conditions as MDA-MB231

Oncogene PR promoter accessibility and ERa levels L Fleury et al 4083 7 MCF-7

2 estrogen induction

PR RET BMP6 β-actin ANXA9 GREB1 TPD52L1

12 MDA-66

AZA E2+/- Stop - 10 -72h0 +16h E2 - /AZA 8 E2/AZA

4 Relative mRNA expression

PR RET BMP6 β-actin ANXA9 GREB1 TPD52L1 Figure 7 The epigenetic mechanism of progesterone receptor gene (PR) silencing is common to some but not all estrogen receptor (ERa) target genes that are repressed in ERa-negative cell lines (a) quantitative real-time (qRT)-PCR using primer sets for PR, ANXA9, RET, TPD52L1, BMP6, GREB1, b-actin in MCF-7 cells. Cells were incubated in steroid-free medium, treated with E2 10À8 M or EtOH for 16 h. Expression levels from E2-treated relative to mock-treated cells are represented. (b) (qRT)-PCR using same primer as (a) in MDA-66 cells. Cells were incubated in steroid-free medium, treated with 5 mM of 5-aza-20-deoxycytidine (AZA) for 72 h before addition of E2 10À8 M or EtOH for 16 h. Expression levels from treated cells relative to mock-treated cells are represented. cells with 100 mgmlÀ1 and 600 mgmlÀ1 hygromicin, respectively. alkaline phosphatase-conjugated anti-rabbit or anti-mouse All cells were grown at 37 1C in a humidiﬁed atmosphere immunoglobulin G (Sigma-Aldrich). The membrane was containing 10% CO2. washed again and ERa or DNMT1 were detected with ECF To study the effects of estrogens and anti-estrogens, cells substrate (Amersham Biosciences) on a Fluorimager 595 were grown for 2 days in media without phenol red, without (Molecular Dynamics, Sunnyvale, CA, USA). Quantiﬁcation gentamicin and 10% of serum stripped of endogenous steroids. analysis was performed using the Imagequant software. The Cells were treated or not with 10 nM E2, 5 mM of AZA, 100 nM quantity of extract loaded on gels was normalized to total ICI, 50 mgmlÀ1 TSA for the indicated times. protein content, assayed by the AmidoSchwartz technique.

Western blot Chromatin immunoprecipitation assay Samples were separated by SDS–polyacrylamide gel electro- Four million cells were seeded in 14 cm diameter plates. After phoresis and then blotted with the indicated antibody onto AZA treatment or siDNMT1 electroporation, cells were Hybond-P nitrocellulose membrane (Amersham Biosciences, treated for 1 h with EtOH (as mock), E2 or ICI. ChIP assays Piscataway, NJ, USA). Anti-DNMT1 and HC20 antibodies were performed essentially as described in (Tyteca et al., 2006) were used at 1/200 dilution; anti-GAPDH was used at 1/1000 using HC20 (4 mg/30 mg of chromatin) or no antibody as a dilution. The membrane was washed and incubated with an control. qPCR analysis was performed on an i-Cycler device

Oncogene PR promoter accessibility and ERa levels L Fleury et al 4084 (Bio-Rad Laboratories, Hercules, CA, USA) using the Cell electroporation and RNA interference platinum SYBR Green qPCR SuperMix (Invitrogen), accord- The transfection of small interfering RNA was performed ing to the manufacturer’s instructions. The following primer by electroporation using a Gene Pulser Xcell apparatus pairs were used to amplify genomic DNA from ChIPs: for the (Bio-Rad Laboratories) set to 250 V, 950 mF, in 4 mm cuvettes pS2/TFF1 promoter fragment, 50-GGCCATCTCTCACTAT (Eurogentec, Seraing, Belgique). The final concentration of GAATCACTTCTGC-30 and 50-GGCAGGCTCTGTTTGC siRNA was 20 mM for 8 Â 106 cells. After electroporation, TTAAAGAGCG-30; for PR regulatory sequences DNA cells were plated and grown for 2 days in phenol-red free. fragment, 50-ATCTACAACCCGAGG CG-30 and 50-CCCAG Then, cells were treated or not with E2 or ICI for 16 h for GAAGGGTCGGACTT-30; for the b-actin promoter, 50-TGG qPCR and for 1 h for ChIP experiments. siRNA directed ACTTCGAGCAAGAGATG-30 and 50-GAAGGAAGGCTG against DNMT1 were chosen as in (Suzuki et al., 2004) GAAGAGTG-3. (Eurogentec). Amplification conditions were 3 min at 95 1C followed by 50 cycles (20 s at 95 1C, 30 s at 58 1C (for PR amplification) or 62.5 1C (for pS2/TFF1 and b-actin amplifications), 20 s at 72 1C). Bisulfite genomic sequencing 340 mg of genomic DNA were boiled for 1 min at 98 1C and qRT-PCR experiments denaturated by 3 M NaOH for 15 min at 37 1C. Bisulphite qRT-PCR experiments were performed as described in Baron conversion was performed as indicated in MethylDetector kit et al. (2007). The following primer pairs were used to amplify manufacturer’s instruction (Active Motif, Rixensart, Belgi- cDNAs after reverse transcription experiments: for GAPDH, que). The primer sequences used for bisulfite genomic 50-ACAGCAACAGGGTGGTGGAC-30 and 50-GACCATT sequencing were 50-TGTGGGTGGTATTTTTAATGA GCTGGGGCTGGTG-30; for pS2/TFF1,50-GTACACG GAAT-30 and 50-CCCCTCACTAAAACCCTAAAACTA-30. GAGGCCCAGACAGA-30 and 50-AGGGCGTGACACCAG PCR products were diluted 1/500 fold and a second PCR was GAAA-30; for PR,50-GGCCATACCTATCTCCCTGGA-30 performed using identical conditions. Resulting amplicons and 50-CTCCACGTCCGACAGCGACT-30; for b-actin, 50- were subcloned into the plasmid pGEM-T (Promega, Char- GGTGACAGCAGTCGGTTGGA-30 and 50-CACAATAGT bonnie` res, France), and sequenced. CCTCGGCCACATT-30; for ANXA9,50-CCGCTGTACT TTGCTGACAA-30 and 50-GTTCAGCCAAACACGGAA AT-30; for BMP6 50-CCCTCTTCATGCTGGATCTG-30 and Acknowledgements 50-AGGGGAGAACTCCTTGTCGT-30; for TPD52L1 50-AA CCGTTGCAAGGAACAGAC-30 and 50-ATGCCAGCTTT We thankF Gannon for cell line MDA-66, F Vignon for TGCTGAAGT-30; for ret 50-TCCTGGGAGAAGCTCAGT cell line HE-5, D Trouche and members of his group GT-30 and 50-GATGTTGGGGCACAAGAACT-30;forGREB1 for insightful discussions and assistance and D Trouche for 50-GTGGTAGCCGAGTGGACAAT-30 and 50-AAACCCGTC critical reading of the manuscript. This workwas partially TGTGGTACAGC-30. supported by the following grants: ANR JC08_42115, La Amplification conditions were 30 min at 45 1C and 3 min Ligue contre le Cancer, comite´ du Tarn and comite´ de Haute at 95 1C followed by 50 cycles (20 s at 95 1C, 30 s at 58 1C, Garonne, INCa, contract no. PL 06-045 and Region Midi- 20 s 72 1C). Pyre´ ne´ es.

References

Ahram M, Best CJ, Flaig MJ, Gillespie JW, Leiva IM, Chuaqui RF Hamdy FC, Autzen P, Robinson MC, Horne CH, Neal DE, Robson et al. (2002). Proteomic analysis of human prostate cancer. Mol CN. (1997). Immunolocalization and messenger RNA expression of Carcinog 33: 9–15. bone morphogenetic protein-6 in human benign and malignant Bachman KE, ParkBH, Rhee I, Rajagopalan H, Herman JG, Baylin prostatic tissue. Cancer Res 57: 4427–4431. SB et al. (2003). Histone modiﬁcations and silencing prior to DNA Heikinheimo KA, Laine MA, Ritvos OV, Voutilainen RJ, Hogan BL, methylation of a tumor suppressor gene. Cancer Cell 3: 89–95. Leivo IV. (1999). Bone morphogenetic protein-6 is a marker of Balleine RL, Fejzo MS, Sathasivam P, Basset P, Clarke CL, Byrne JA. serous acinar cell differentiation in normal and neoplastic human (2000). The hD52 (TPD52) gene is a candidate target gene for events salivary gland. Cancer Res 59: 5815–5821. resulting in increased 8q21 copy number in human breast Jones PA, Baylin SB. (2002). The fundamental role of epigenetic events carcinoma. Genes Chromosomes Cancer 29: 48–57. in cancer. Nat Rev Genet 3: 415–428. Baron S, Escande A, Alberola G, Bystricky K, Balaguer P, Richard- Jones PA, Taylor SM. (1980). Cellular differentiation, cytidine analogs Foy H. (2007). Estrogen receptor alpha and the AP1 complex and DNA methylation. Cell 20: 85–93. cooperate during IGF-I-induced transcriptional activation of the Kininis M, Chen BS, Diehl AG, Isaacs GD, Zhang T, Siepel AC et al. pS2/TFF1 gene. J Biol Chem 282: 11732–11741. (2007). Genomic analyses of transcription factor binding, histone Cameron EE, Bachman KE, Myohanen S, Herman JG, Baylin SB. acetylation, and gene expression reveal mechanistically distinct (1999). Synergy of demethylation and histone deacetylase inhibition classes of estrogen-regulated promoters. Mol Cell Biol 27: 5090– in the re-expression of genes silenced in cancer. Nat Genet 21: 5104. 103–107. Kwon YS, Garcia-Bassets I, Hutt KR, Cheng CS, Jin M, Liu D et al. Dobosy JR, Selker EU. (2001). Emerging connections between DNA (2007). Sensitive ChIP-DSL technology reveals an extensive estrogen methylation and histone acetylation. Cell Mol Life Sci 58: 721–727. receptor alpha-binding program on human gene promoters. Proc Gerke V, Moss SE. (2002). Annexins: from structure to function. Natl Acad Sci USA 104: 4852–4857. Physiol Rev 82: 331–371. Lacroix M, Toillon RA, Leclercq G. (2004). Stable ‘portrait’ of breast Giamarchi C, Solanas M, Chailleux C, Augereau P, Vignon F, tumors during progression: data from biology, pathology and Rochefort H et al. (1999). Chromatin structure of the regulatory genetics. Endocr Relat Cancer 11: 497–522. regions of pS2 and cathepsin D genes in hormone-dependent and Laganiere J, Deblois G, Lefebvre C, Bataille AR, Robert F, Giguere V. -independent breast cancer cell lines. Oncogene 18: 533–541. (2005). From the cover: location analysis of estrogen receptor alpha Goodfellow PJ, Wells Jr SA. (1995). RET gene and its implications for target promoters reveals that FOXA1 deﬁnes a domain of the cancer. J Natl Cancer Inst 87: 1515–1523. estrogen response. Proc Natl Acad Sci USA 102: 11651–11656.

Oncogene PR promoter accessibility and ERa levels L Fleury et al 4085 Lande-Diner L, Zhang J, Ben-Porath I, Amariglio N, Keshet I, Hecht tion, and comparison of de novo and maintenance methylation. M et al. (2007). Role of DNA methylation in stable gene repression. J Biol Chem 274: 33002–33010. J Biol Chem 282: 12194–12200. Rae JM, Johnson MD, Scheys JO, Cordero KE, Larios JM, Lippman Lazennec G, Alcorn JL, Katzenellenbogen BS. (1999). Adenovirus- ME. (2005). GREB 1 is a critical regulator of hormone dependent mediated delivery of a dominant negative estrogen receptor gene breast cancer growth. Breast Cancer Res Treat 92: 141–149. abrogates estrogen-stimulated gene expression and breast cancer cell Raynal P, Pollard HB. (1994). Annexins: the problem of assessing the proliferation. Mol Endocrinol 13: 969–980. biological role for a gene family of multifunctional calcium- and Leu YW, Yan PS, Fan M, Jin VX, Liu JC, Curran EM et al. phospholipid-binding proteins. Biochim Biophys Acta 1197: 63–93. (2004). Loss of estrogen receptor signaling triggers epigenetic Robertson KD. (2002). DNA methylation and chromatin—unraveling silencing of downstream targets in breast cancer. Cancer Res 64: the tangled web. Oncogene 21: 5361–5379. 8184–8192. Roodi N, Bailey LR, Kao WY, Verrier CS, Yee CJ, Dupont WD et al. Lin CY, Strom A, Vega VB, Kong SL, Yeo AL, Thomsen JS et al. (1995). Estrogen receptor gene analysis in estrogen receptor-positive (2004). Discovery of estrogen receptor alpha target genes and and receptor-negative primary breast cancer. J Natl Cancer Inst 87: response elements in breast tumor cells. Genome Biol 5: R66. 446–451. Lin CY, Vega VB, Thomsen JS, Zhang T, Kong SL, Xie M et al. Rountree MR, Bachman KE, Baylin SB. (2000). DNMT1 binds (2007). Whole-genome cartography of estrogen receptor alpha HDAC2 and a new co-repressor, DMAP1, to form a complex at binding sites. PLoS Genet 3: e87. replication foci. Nat Genet 25: 269–277. Lopez-Serra L, Ballestar E, Fraga MF, Alaminos M, Setien F, Esteller Rubin MA, Varambally S, Beroukhim R, Tomlins SA, Rhodes DR, M. (2006). A Profile of methyl-CpG binding domain protein Paris PL et al. (2004). Overexpression, amplification, and androgen occupancy of hypermethylated promoter CpG islands of tumor regulation of TPD52 in prostate cancer. Cancer Res 64: 3814–3822. suppressor genes in human cancer. Cancer Res 66: 8342–8346. Sorlie T, Perou CM, Tibshirani R, Aas T, Geisler S, Johnsen H et al. Metivier R, Penot G, Carmouche RP, Hubner MR, Reid G, Denger S (2001). Gene expression patterns of breast carcinomas distinguish et al. (2004). Transcriptional complexes engaged by apo-estrogen tumor subclasses with clinical implications. Proc Natl Acad Sci USA receptor-alpha isoforms have divergent outcomes. EMBO J 23: 98: 10869–10874. 3653–3666. Suzuki M, Sunaga N, Shames DS, Toyooka S, Gazdar AF, Minna JD. Metivier R, Penot G, Hubner MR, Reid G, Brand H, Kos M et al. (2004). RNA interference-mediated knockdown of DNA methyl- (2003). Estrogen receptor-alpha directs ordered, cyclical, and transferase 1 leads to promoter demethylation and gene re- combinatorial recruitment of cofactors on a natural target expression in human lung and breast cancer cells. Cancer Res 64: promoter. Cell 115: 751–763. 3137–3143. Momparler RL. (2003). Cancer epigenetics. Oncogene 22: 6479–6483. Ting AH, McGarvey KM, Baylin SB. (2006). The cancer epigenome— Nagaraja GM, Othman M, Fox BP, Alsaber R, Pellegrino CM, Zeng components and functional correlates. Genes Dev 20: 3215–3231. Y et al. (2006). Gene expression signatures and biomarkers of Touitou I, Vignon F, Cavailles V, Rochefort H. (1991). Hormonal noninvasive and invasive breast cancer cells: comprehensive profiles regulation of cathepsin D following transfection of the estrogen or by representational difference analysis, microarrays and proteomics. progesterone receptor into three sex steroid hormone resistant Oncogene 25: 2328–2338. cancer cell lines. J Steroid Biochem Mol Biol 40: 231–237. Ottaviano YL, Issa JP, Parl FF, Smith HS, Baylin SB, Davidson NE. Tyteca S, Vandromme M, Legube G, Chevillard-Briet M, Trouche D. (1994). Methylation of the estrogen receptor gene CpG island marks (2006). Tip60 and p400 are both required for UV-induced apoptosis loss of estrogen receptor expression in human breast cancer cells. but play antagonistic roles in cell cycle progression. EMBO J 25: Cancer Res 54: 2552–2555. 1680–1689. Petz LN, Ziegler YS, Schultz JR, Kim H, Kemper JK, Nardulli AM. Wijayaratne AL, McDonnell DP. (2001). The human estrogen (2004). Differential regulation of the human progesterone receptor receptor-alpha is a ubiquitinated protein whose stability is affected gene through an estrogen response element half site and Sp1 sites. differentially by agonists, antagonists, and selective estrogen J Steroid Biochem Mol Biol 88: 113–122. receptor modulators. J Biol Chem 276: 35684–35692. Platet N, Cathiard AM, Gleizes M, Garcia M. (2004). Estrogens and Yang X, Phillips DL, Ferguson AT, Nelson WG, Herman JG, their receptors in breast cancer progression: a dual role in cancer Davidson NE. (2001). Synergistic activation of functional estrogen proliferation and invasion. Crit Rev Oncol Hematol 51: 55–67. receptor (ER)-alpha by DNA methyltransferase and histone PollackJR, Sorlie T, Perou CM, Rees CA, Jeffrey SS, Lonning PE deacetylase inhibition in human ER-alpha-negative breast cancer et al. (2002). Microarray analysis reveals a major direct role of DNA cells. Cancer Res 61: 7025–7029. copy number alteration in the transcriptional program of human Zhang M, Wang Q, Yuan W, Yang S, Wang X, Yan JD et al. (2007). breast tumors. Proc Natl Acad Sci USA 99: 12963–12968. Epigenetic regulation of bone morphogenetic protein-6 gene Pradhan S, Bacolla A, Wells RD, Roberts RJ. (1999). Recombinant expression in breast cancer cells. J Steroid Biochem Mol Biol 105: human DNA (cytosine-5) methyltransferase. I. Expression, purifica- 91–97.

Oncogene