Biochimica et Biophysica Acta 1829 (2013) 480–486

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta

j ourna l homepage: www.e lsev ie r .com/ locate /bbagrm

Retinoic acid impairs estrogen signaling in breast cancer cells byinterfering with activation of LSD1 via PKA

Maria Neve Ombra a, Annalisa Di Santi b, Ciro Abbondanza b, Antimo Migliaccio b,Enrico Vittorio Avvedimento c, Bruno Perillo a,⁎a Istituto di Scienze dell’Alimentazione, C.N.R., 83100 Avellino, Italyb Dipartimento di Patologia Generale, Seconda Università degli Studi di Napoli, 80138 Naples, Italyc Dipartimento di Biologia e Patologia Cellulare e Molecolare “L. Califano”, Università degli Studi “Federico II”, 80131 Naples, Italy

⁎ Corresponding author at: I.S.A. via Roma, 52, 83100,299210.

E-mail address: perillo@unina.it (B. Perillo).

1874-9399/$ – see front matter © 2013 Elsevier B.V. Alhttp://dx.doi.org/10.1016/j.bbagrm.2013.03.003

a b s t r a c t

a r t i c l e i n f o

Article history:Received 13 December 2012Received in revised form 7 February 2013Accepted 7 March 2013Available online 16 March 2013

Keywords:Nuclear receptorTranscription regulationEstrogenRetinoic acidGene loopingHistone demethylation

More than 70% of breast cancers in women require estrogens for cell proliferation and survival. 17β-estradiol(E2) effect on mammary target cells is almost exclusively mediated by its binding to the estrogen receptor-α(ERα) that joins chromatin where it assembles active transcription complexes. The proliferative andpro-survival action of estrogens is antagonized in most cases by retinoic acid (RA), even though the cognateretinoic acid receptor-α (RARα) cooperates with ERα on promoters of estrogen-responsive genes.We have examined at the molecular level the crosstalk between these nuclear receptors from the pointof view of their control of cell growth and show here that RA reverts estrogen-stimulated transcriptionof the pivotal anti-apoptotic bcl-2 gene by preventing demethylation of dimethyl lysine 9 in histone H3(HeK9me2). As we previously reported, this is obtained by means of E2-triggered activation of thelysine-specific demethylase 1 (LSD1), an enzyme that manages chromatin plasticity in order to allow specificmovements of chromosomal regions within the nucleus. We find that E2 fuels LSD1 by inducing migration ofthe catalytic subunit of protein kinase A (PKA) into the nucleus, where it targets estrogen-responsive loci. RArescues LSD1-dependent disappearance of H3K9me2 at bcl-2 regulatory regions upon the prevention of PKAassembly to the same sites.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

The great majority of humanmammary tumors, the most commonform of cancer in women, are positive for expression of the nuclearreceptor estrogen receptor-α and are responsive to 17β-estradiol(E2) challenge that stimulates proliferation and promotes survivalof breast cancer cells [1,2]. Liganded estrogen receptor-α recognizesestrogen responsive elements (EREs) on DNA and controls gene expres-sion, as its associated transcription factors generatewell-documented co-valent modifications at the N-terminal tails of nucleosomal histones thatconfer chromatin the flexibility essential for access to buried DNA se-quences [3,4]. E2 protects human breast cancer cells from programmedcell death mainly through the transcriptional stimulation of the anti-apoptotic bcl-2 gene [5] that responds to hormone challenge via twoEREs locatedwithin the coding region [6]. In agreementwith the ‘histonecode’ theory [7] stating thatmethylation of lysine 9 in histone H3 usuallymarks silenced genes [8], we have already shown that E2-induced bcl-2transcription is triggered by demethylation of this lysine driven by theLSD1 demethylase. The process produces reactive oxygen species (ROS)

Avellino, Italy. Tel.: +39 0825

l rights reserved.

with oxidation of adjacent Gs and consequent nicks on DNA strands [9],necessary to loop the enhancers either with the upstream promoter aswell as the downstream polyadenylation sites [10].

Conversely, retinoic acid inhibits proliferation and promotes apo-ptosis in breast cancer cells where it has been previously shown todown-regulate the expression of genes crucial for proliferation andsurvival [11–13]. The effect of RA on responsive cells is mediated bytwo subfamilies of nuclear receptors, the retinoic acid receptor-αand the retinoid X receptors (RXRs) [14–16]. RARs dimerize withRXRs and join responsive sequences on DNA (RARE), functioning astranscription factors that govern the expression of target genesthrough molecular mechanisms in part still undiscovered. In fact, itis unclear whether RA-repressed genes are regulated directly byunliganded heterodimers residing on chromatin, or are controlled in-directly through RA-dependent mediators [17]. Similarly, little isknown on the extent of crosstalk between RA signaling and other nu-clear receptor subfamilies among which ERα plays a pivotal role. In-terestingly, gene expression analysis has revealed that transcriptionof RARα is induced by E2 [18], while, in a study that investigates ona genome-wide scale the relationship between ERα and RARα, Huaet al. report a symmetrical control by RA on the gene coding for theestrogen receptor-α [19]. Hua et al. also show that the two receptorsshare a subset of binding regions on chromatin and postulate that

481M.N. Ombra et al. / Biochimica et Biophysica Acta 1829 (2013) 480–486

they compete for binding to the same (or very close) sites, so explainingtheir opposite effects inmost different cell lines. However, in amore re-cent paper, Ross-Innes et al. have described that RARα cooperates withERα on E2-responsive loci in order to drive the expression of estrogentarget genes [20].

In the present study, we address the mechanism underlying theswitch from cooperative interaction to antagonism shown by theretinoic acid receptor-α with respect to ERα.

2. Material and methods

2.1. Cells

Human breast cancer MCF-7 cells have been routinely grown as al-ready described [6]. To evaluate the effect of E2 and/or RA challenge,cells have been first incubated with phenol red-free DMEM with 0.5%dextran–charcoal-stripped FCS for 6–8 h, and thenwith the sameme-dium containing 5% dextran–charcoal stripped FCS for further 4 days,prior to be challenged for 30 min with 10 nM E2 and/or 100 nM RAaccording to the experimental needs. To inhibit the cAMP-PKA orMEK cascades, H89 (10 μM) or PD98059 (50 μM) has, respectively,been added to cells just prior to estrogen challenge. Rescue of PKA ac-tivity has been done by incubation of H89-treated cells with forskolin(40 μM) for 1 h. According to the experimental design, 10 nM E2 hasbeen added for the last 30 min. To obtain LSD1 knock down withsiRNAs, cells have been transfected with 37.5 ng of the 3′-untranslatedregion (customer service, Qiagen, USA), in medium without serum to afinal concentration of 5 nM and incubation was continued for 72 h. Todetermine the rescue of LSD1 activity in knockdown experimentswith siRNAs, LSD1 full-length cDNA has been inserted into the CMV3xFLAG expression vector (Sigma-Aldrich, USA) and co-transfectedinto MCF-7 cells.

2.1. Chromatin immunoprecipitation (ChIP)

ChIP assays have been carried out as previously reported [9]. Chro-matin has been sonicated in order to obtain fragments of 300–400 bp.For each assay, chromatin from a total of 106 cells has been used. Allbands from ethidium bromide stained gels have been analyzed bydensitometry and quantified with Total Lab ID software. Antibodiesused, as well as sequences of primers and PCR conditions are availableupon request.

2.2. DNA-picked chromatin (DPC)

DPC experiments have been performed as previously reported[10]. MCF-7 cells from semi-confluent 10 cm dishes have beencross-linked and collected into 100 mM Tris–HCl (pH 9.4), 10 mMDTT. Prior to sonication, cell pellets have been re-suspended into0.5 volumes of 1× PBS–0.5% Triton X-100, with 2 μl of RNase A(20 mg/ml). The mixture has been then incubated at RT for 1 h withgentle shaking and at 4 °C for 12–16 h before being washed six timesin 1× PBS and centrifuged. Pellet has been finally re-suspended in300 μl of sonication buffer (50 mM Tris–HCl, pH 8.0, 10 mMEDTA, pH 8.0, 1% SDS, protease inhibitors). After sonication, sampleshave been cleared by centrifugation at 16,000 g at RT for 15 min andsurnatants have been collected and diluted in bufferswith the followingfinal concentration, 0.4% SDS, 0.1% Sarkosyl, 100 mM NaCl, 2 mMEDTA, pH 8.0, 1 mM EGTA, pH 8.0. 100 μl aliquots (correspondingto 1.0–1.5 × 106 cells) have been added with high molar excess(1 μM final concentration) of biotinylated oligonucleotide probe, andhybridization has been conducted as follows: 25 °C for 3 min–70 °Cfor 6 min–38 °C for 60 min–60 °C for 2 min–38 °C for 60 min–60 °Cfor 2 min–38 °C for 120 min–25 °C final temperature. Eluates fromsamples incubated at RT for 1 h and at 65 °C for 10 min havebeen treated with proteinase K and DNA has been extracted with

phenol–chloroform and precipitated with ethanol, prior to be amplifiedwith PCRs. Quantities of PCR products from hybridized DNA have beencalculated as a percent of amplified Input. To estimate the association ofspecific DNA sites with the hybridizing region captured in DPC assays,the proportion of PCR products from any considered region quantifiedas % of the related amplified Input has been compared to the % ofInput represented by the hybridizing site amplified by PCR under thesame experimental conditions. The following 5′-biotinylated oligonu-cleotide has been used as probe in shown experiments:

bcl-2 ERE: GCCAGGCCGGCGACGACTTCTCC.

2.3. Nuclear run-on

Run-on assays have been realized as already described [9]. Briefly,cell nuclei have been extracted by addition of NP-40 lysis buffer(10 mM Tris–HCl, pH 7.44; 10 mM NaCl; 3 mM MgCl2; and 0.5%NP-40) and stored in a glycerol buffer (50 mM Tris–HCl, pH 8.3;40% glycerol; 5 mM MgCl2; 0.1 mM EDTA). Transcription has beencarried-out in the presence of 5 μl of 10 mCi/ml [α-32P]UTP. RNAsamples (10 μg) were hybridized with plasmids (pcDNA3) containingcDNAs of bcl-2 or β-actin, immobilized on nylon filters.

2.4. Western blotting

To determine LSD1 levels, E2-deprived cells were treated with10 nM E2 in the absence or presence of 100 nM RA for 30 min.Thereafter, cells were washed in ice-cold PBS, and dissolved in lysisbuffer (50 mM Tris–HCl, pH 7.4, 1% Nonidet NP-40, 300 mM NaCl,1 mM MgCl2, 2 mM EDTA, 50 mM NaF, 0.1 mM NaVO3, 1 mMglycerophosphate, 2.5 mM sodium pyrophosphate, in the presenceof protease inhibitors and 40 U DNase). Thirty μg of protein waselectrophoresed on 9% SDS–polyacrylamide gel and transferred ontonitrocellulose membranes (Schleicher and Schuell GmbH, Germany).Western blot analysis was carried out with anti-LSD1 or GAPDH(employed to normalize observed results) antibodies, used accordingto manufacturer's protocols.

3. Results and discussion

3.1. RA binding to RARα reverts its cooperation with ERα by preventingLSD1-driven demethylation of H3K9me2

To explore how the retinoic acid receptor-α influences the controlof ERα on survival of estrogen-responsive mammary cells, wehave first assessed by chromatin immunoprecipitation (ChIP) assayswhether RARα joins the other nuclear receptor on the enhancerspresent in the coding region of bcl-2 gene (Fig. 1A) that prevents ap-optosis triggered by different stimuli [21]. This analysis has revealedthat RARα occupies, together with the estrogen receptor, bcl-2 chro-matin in hormone-starved breast cancer MCF-7 cells treated with E2for 30 min (Fig. 1B). We have chosen this cell line, as MCF-7 areestrogen-dependent human cells that harbor well-characterized E2target genes. Thereafter, we have investigated the behavior of thesame receptors in cells added with RA and we have found that,under these conditions, they do not join bcl-2 locus (square box inFig. 1B). However, when cells are concomitantly incubated with RAand E2, we observe the following picture at bcl-2 EREs: both nuclearreceptors still park on chromatin but demethylation of H3K9me2detected in E2-treated cells disappears (Fig. 1B). On the other side, ly-sine 4 in the same histone remains methylated (Fig. 1B), coherentlywith the cyclic nucleosomal changes induced by E2 on target geneswhere H3K4me2 is lost at a later time of hormonal addition, whenthe transcriptional output decreases [22,23]. Therefore, we havesearched for the enzyme that we had already found drives the pro-cess, the LSD1 demethylase [9]. In fact, even though LSD1 has beeninitially associated to demethylation of lysine 4 in histone H3 [24], a

Fig. 1. RA prevents estrogen-induced demethylation of H3K9me2 driven by LSD1. (A) Simplified structure of bcl-2 gene. Colored boxes represent the three exons where the EREsand the polyadenylation (polyA) sites are evidenced. Pr. on the horizontal arrow indicates the promoter with transcription start sites. (B) ChIP analysis to show changes visible atbcl-2 enhancer chromatin, induced by E2 addition (10 nM for 30 min) to hormone-starved MCF-7 cells, in the absence or presence of 100 nM RA. The antibodies used are indicatedon the left of an illustrative gel that, as in the next panels, represents one of the three different experiments. On top is evidenced the DNA region amplified by preliminarily fixed PCRreactions, realized within the range of cycles with exponential duplication of template sequences from sonicated chromatin before immunoprecipitation (Input) as graphically in-dicated on middle top. Antibody to α-tubulin has been used as negative control. All bands from ethidium bromide-stained 1.8% agarose gels have been quantified with Total Lab IDsoftware and plotted in the graph illustrated on the bottom right. Error bars represent SEM. As in section (C) and in the following figures, all graphical values show a statisticalsignificance where 0.05 ≥ p ≥ 0.01. The square box on the top right shows the result of a ChIP experiment searching for recruitment of nuclear receptors to bcl-2 EREs in cells treat-ed with 100 nM RA. (C) Sequential ChIP (Re-IP) analysis from cells primarily incubated with antibodies to ΕR⟨. Chromatin from primary ChIPs (1-IP) has been subsequently pre-cipitated (2-IP) with antibodies to RARα and quantified as % of amplified Input, as represented in graphs on the right.

482 M.N. Ombra et al. / Biochimica et Biophysica Acta 1829 (2013) 480–486

growing body of evidence indicates that, depending upon the cellularcontext, it is also able to relieve transcriptional repression bydemethylating H3K9me and K9me2 [9,25]. As shown in Fig. 1B, weobserve that LSD1 recruitment is enhanced after concomitant addi-tion of estrogens and RA, indicating that its joining to chromatin isnot sufficient for activation. Indeed, we see that even the alreadydescribed E2-dependent oxidation of neighboring DNA, where8-oxo-Gs emerge as a consequence of local creation of ROS generated

Fig. 2. RA prevents E2-elicited LSD1 activation by interfering with chromatin targeting of theH3K9me2, as revealed by use of the competitive inhibitor H89 (10 μM) in ChIPs. Treatmentbcl-2 enhancers that, however, reappears exclusively in chromatin from E2-treated cells. Thegraphics, see Fig. 1 legend. (B) ChIP experiments showing that PKA joins bcl-2 chromatin awith E2-induced assembly of PKA on estrogen-target chromatin, as assessed by ChIP showitreated with both hormones. (D) Run-on assay revealing that inhibition of PKA in cells challproduction. Hormonal effect cannot be bypassed by global PKA activation, as addition of forstranscribing RNA Pol II phosphorylated at serine 5 (Pol II-P) to bcl-2 enhancer site under d

by activation of the demethylase, is inhibited under these conditions(Fig. 1B).

Before deeper investigation, and to be sure that the describedchanges are connected to the presence of liganded-RARα, we havealso confirmed by double ChIP experiments (Re-IPs, where ERα-ChIP has been followed by “release” of receptor-enriched chromatinto be used in a second ChIP with antibodies to RARα) that both recep-tors meet into the same complexes on bcl-2 enhancers, as shown in

PKA catalytic subunit. (A) Activation of PKA is specifically involved in demethylation ofof H89-added MCF-7 cells with forskolin (40 μM) for 1 h rescues K9 demethylation atMEK inhibitor PD98059 (50 μM) has been used as a control of specificity. For reading offter E2 challenge where it can activate LSD1 independently recruited. (C) RA interferesng an increase of kinase targeting to promoters of RA-responsive genes (cyp26) in cellsenged with estrogens results into prevention of the E2-induced increase of bcl-2 mRNAkolin is unsuccessful in the absence of E2. (E) ChIP showing recruitment of the activelyifferent hormonal treatments.

483M.N. Ombra et al. / Biochimica et Biophysica Acta 1829 (2013) 480–486

Fig. 1C. On this regard, it has been hypothesized that, even though adirect interaction is not still thoroughly demonstrated, the presenceof full-length RARα is necessary for its cooperation with ERα where

it may operate as a scaffold protein in the transcriptional complex as-sembled by the estrogen receptor, in order to facilitate specific cofac-tor interactions [20]. In the same manuscript Ross-Innes et al., in

484 M.N. Ombra et al. / Biochimica et Biophysica Acta 1829 (2013) 480–486

agreement with the concept that histone acetylation marks tran-scribed loci [26], have assumed that co-treatment with E2 and RA in-hibits estrogen-induced recruitment of the histone acetyl transferase(HAT) CBP/p300 on target genes [20]. We can confirm this statement:in our experimental conditions, RA prevents E2-induced HAT assem-bly on bcl-2 chromatin with consequent loss of H3K9 acetylation andrescue of methylation at the same residue (Fig. 1B).

However, Hu et al., even though highlighting the role played byCBP/p300 in the establishment of ERα-dependent inter-chromosomalinteractions, report that, in order to achieve complete E2-dependenttranscriptional activation, hormone responsive cells follow a “two-step”program where LSD1 links initially induced genes to the inter-chromatin granules involved in the building of the transcription facto-ries [27]. Notably, this molecular mechanism appears in accord withour previous findings that show LSD1 as a leading player in the controlof chromatin plasticity [9]. Therefore, besides RA-dependent impair-ment of HAT recruitment, on the basis of data showing a direct effectof RA on demethylation of H3K9me2, we assume that LSD1 plays an im-portant role also in the crosstalk between estrogens and RA.

3.2. RA blocks H3K9me2 demethylation by interfering with E2-inducedactivation of protein kinase A (PKA)

To explain the observation that, notwithstanding LSD1 recruitmentto bcl-2 chromatin, H3K9me2 is not demethylated in cells under doublehormonal treatment, we have searched for different candidates able toactivate the demethylase, independently from the estrogen receptoritself. It is well known, in fact, that in addition to the classical genomicpathway, estrogens trigger adenylyl cyclase [28,29] and that the cAMPrise induces migration of the catalytic subunit of PKA to the nucleusinto specific transcriptional compartments visible as fluorescentnuclear speckles [30]. Similar speckles, containing ERα and RNA pol lIcofactors, are seen after E2 challenge [31]. Correspondingly, it hasbeen reported that RA stimulates a rapid increase of cAMP levels andPKA activity in cells that express functional RARα [32]. Moreover,PKA-mediated phosphorylation has been already involved in changingthe ability of transcription factors to interact with LSD1 [33]. Therefore,also considering that analysis of the demethylase sequence reveals acanonical phosphorylation site, we have first analyzed whether PKAcould be involved in estrogen-dependent LSD1 activation.

To this aim, we have measured H3K9me2 demethylation inE2-stimulated cells under conditions of kinase-repressed activity. Asshown in Fig. 2A, the addition of the specific inhibitor H89 preventsdisappearance of dimethyl lysine 9. This effect is specific, as the mito-gen extracellular kinase (MEK) inhibitor PD98059 does not exhibitthe same behavior. Moreover, to be confident that disappearance ofK9me2 depends indeed by inhibition of PKA, we have first confirmedthese findings with a different inhibitor (KT5720, data not shown),and then we have treated H89-added cells with forskolin. Forskolinthat stimulates phosphorylations realized by PKA, rescues in fact de-methylation of H3K9me2 at bcl-2 EREs (Fig. 2A). However, the roleof ERα still remains crucial, as the recovery of the PKA activity is suc-cessful exclusively in the presence of estrogens (Fig. 2A). We postu-late that ERα, when accompanied by unliganded RARα, positionsthe kinase to the right target allowing it to reach the local concentra-tion needed to exert its phosphorylating activity. In addition, we havecorroborated data reported in Fig. 1B where LSD1 recruitment doesnot automatically correspond to its activation: in fact, inhibition ofPKA with H89 results into the prevention of its recruitment to bcl-2chromatin, where LSD1, in E2-challenged cells, remains, nevertheless,still resident (Fig. 2B). Of note, H89 does not preclude targeting ofERα and RARα to the same locus, reinforcing the concept that the as-sembly of PKA onto the transcription complex is mandatory (Fig. 2B).

The effect of treatment with H89 on the disappearance of PKAfrom bcl-2 EREs, is mimicked by RA (Fig. 2C). Therefore, even thougheach hormone is able to induce kinase activity, their concomitant

addition results into reciprocal limitation, at least for estrogen-responsive sites. Since, as also shown in Fig. 2C, RA raises recruitmentof PKA to the promoter of the RA-target cyp26 gene [34], we assumethat liganded RARα interferes with the transcriptional control of ERαby titrating the kinase from E2-responsive sites to RA-dependentgenes. We are aware that whether LSD1 represents a direct phosphory-lation target by PKA or whether the kinase modifies an intermediatesubstrate (i.e., one of the several transcription factors present in thecomplex) needs a more detailed analysis, that will be the focus of ourfurther investigation. We must also emphasize that estrogen-inducedactivation of responsive genes demands the removal of mono-, di-, andtrimethyl lysine 9 in histone H3 while LSD1 is able to remove only thefirst two post-translational modifications. Therefore, we cannot excludethe role eventually played by the H3K9-specific demethylase membersof the family of the JMJC domain-containing (JMJ) hydroxylases [35];or even the differential recruitment of histone methyl-transferasesafter addition of RA and E2 to the same cells [36].

As a further step, we have assessed that the effect of PKA stimulationon demethylation of lysine 9 is strictly linked to bcl-2 transcription: infact, measurement of bcl-2 mRNA production reveals a peak after E2challenge that is prevented by H89 and rescued by further addition offorskolin, even in this case exclusively in the presence of E2 (Fig. 2D).In accordance, we see that RA addition hampers the E2-dependent as-sembly of the active form of RNA polymerase II (RNA Pol II) phosphory-lated at serine 5 to the bcl-2 regulatory locus (Fig. 2E), confirming that itcounteracts the stimulatory effect of estrogens on the expression ofE2-target genes.

3.3. Activation of LSD1 by PKA drives intragenic looping essential for genetranscription

Wehave previously reported that E2-induced transcription associatesto looping between specific regions of hormone-target genes [10]. In par-ticular, we have shown that E2-stimulated bcl-2 expression is accompa-nied by the bridging of the enhancer locus, present within the codingregion, with the promoter and the 3′ polyadenylation sites [10]. There-fore, we have investigated the formation of such loops in cells addedwith both hormones. To this aim, we have utilized the experimentalapproach named DNA-picked chromatin (DPC) where, upon use ofbiotinylated probes, DNA fragments kept near to the hybridizing locusare preferentially captured [10] (Fig. 3A). Upon this analysis, we first seethat the polyadenylation sites are not specifically recovered in the neigh-borhood of enhancers when RA is present togetherwith E2 (Fig. 3B), and,second, that PKA signaling plays a role in the process (Fig. 3C).

As a last step, to verify whether the previously shown effect of RAon gene looping may be due to its prevention of E2-dependent activa-tion of LSD1 via PKA, we have assessed the formation of bridges be-tween the EREs and the 3′-end of bcl-2 gene in cells where LSD1had been knocked-down with specific small interference (si)RNAs(as checked by the assessment of LSD1 protein levels). Lack of thedemethylase inhibits, in fact, formation of ERE/3′-end loops, strictlyresembling what was observed when cells are seeded with both hor-mones, and, fascinatingly, exogenous LSD1 rescues the endogenousprotein only in the absence of RA (Fig. 3D).

4. Conclusions

In conclusion, the findings illustrated above strongly sustain the useof RA in the treatment of estrogen-responsive human breast tumors[37]. They reveal a putative mechanism by which RA counteracts theproliferative effect of estrogens on human breast cancer cells: RA, infact, interferes with the stimulation of PKA activity triggered by E2,being a kinase inducer on its own. It is able, in this way, to overtakethe effect of estrogens, redirecting the activated catalytic subunit ofPKA from E2- to RA targets. Upon the reported molecular interaction,and especially if it will be confirmed for other pivotal E2-responsive

Fig. 3. RA inhibits E2-dependent bcl-2 intragenic looping, essential for gene expression. (A) Simplified graphical illustration of changes induced by E2 on chromatin structure at bcl-2locus, as assessed by DPC [10]. Pr. indicates the promoter region with the transcription start sites. Blank ovals represent transcription factors involved in bridging, dotted lines stayfor sonicated DNA. (B) Evaluation of the presence of loops between bcl-2 enhancer and polyadenylation sites in chromatin from cells seeded with E2 alone or in the presence of RA,as measured by DPC. The probe used to capture specific DNA sequences (indicated on top of each gel) is reported on the left. An illustrative gel is presented. Plots on the right showthe results that have been normalized according either to the quantity of DNA extracted from cross-linked, sonicated chromatin before hybridization (Input), as well as to the re-covery of the hybridizing region, when the rescue of associated fragments has been analyzed, as previously detailed [10]. (C) DPC where involvement of PKA signaling in bcl-2looping has been assessed. (D) DPC highlighting the role played by LSD1 in bcl-2 looping. Demethylase knockdown has been obtained with siRNAs and checked by Westernblot, as shown at the bottom. The rescue of LSD1 has been gained through co-transfection of exogenous protein (lanes 4 and 5).

485M.N. Ombra et al. / Biochimica et Biophysica Acta 1829 (2013) 480–486

genes, as well as in different human breast cancer cell lines actuallyunder investigation, RA may well represent a promising therapeutictool for human mammary tumors.

Acknowledgements

This work has been supported by the Associazione Italiana per laRicerca sul Cancro (AIRC Grant IG 11520: PI Antimo Migliaccio).

References

[1] D.R. Ciocca, M.A. Fanelli, Estrogen receptors and cell proliferation in breast cancer,Trends Endocrinol. Metab. 8 (1997) 313–321.

[2] C. Teixeira, J.C. Reed, M.A.C. Pratt, Estrogen promotes chemoterapeutic drug resis-tance by a mechanism involving bcl-2 proto-oncogene expression in humanbreast cancer cells, Cancer Res. 55 (1995) 3902–3907.

[3] D.J. Mangelsdorf, C. Thummel, M. Beato, P. Herrlich, G. Schutz, K. Umesono, B.Blumberg, P. Kastner, M. Mark, P. Chambon, R.M. Evans, The nuclear receptor su-perfamily: the second decade, Cell 83 (1995) 835–839.

486 M.N. Ombra et al. / Biochimica et Biophysica Acta 1829 (2013) 480–486

[4] M.P. Cosma, Ordered recruitment: gene-specific mechanism of transcription acti-vation, Mol. Cell 10 (2002) 227–236.

[5] D.M. Hockenbery, Z.N. Oltvai, X.M. Yin, C.L. Milliman, S.J. Korsmeyer, Bcl-2 func-tions in an antioxidant pathway to prevent apoptosis, Cell 75 (1993) 241–251.

[6] B. Perillo, A. Sasso, C. Abbondanza, G. Palumbo, 17Beta-estradiol inhibits apopto-sis in MCF-7 cells, inducing bcl-2 expression via two estrogen-responsive ele-ments present in the coding sequence, Mol. Cell. Biol. 20 (2000) 2890–2901.

[7] B.D. Strahl, C.D. Allis, The language of covalent histone modifications, Nature 403(2000) 41–45.

[8] J.C. Rice, S.D. Briggs, B. Ueberheide, C.M. Barber, J. Shabanowitz, D.F. Hunt, Y. Shinkai,C.D. Allis, Histone methyltransferases direct different degrees of methylation to de-fine distinct chromatin domains, Mol. Cell 12 (2003) 1591–1598.

[9] B. Perillo, M.N. Ombra, A. Bertoni, C. Cuozzo, S. Sacchetti, A. Sasso, L. Chiariotti, A.Malorni, C. Abbondanza, E.V. Avvedimento, DNA oxidation as triggered byH3K9me2 demethylation drives estrogen-induced gene expression, Science 319(2008) 202–206.

[10] C. Abbondanza, C. De Rosa, M.N. Ombra, F. Aceto, N. Medici, L. Altucci, B.Moncharmont, G.A. Puca, A. Porcellini, E.V. Avvedimento, B. Perillo, Highlightingchromosome loops in DNA-picked chromatin (DPC), Epigenetics 6 (2011)979–986.

[11] Q. Zhou, M. Stetler-Stevenson, P.S. Steeg, Inhibition of cyclin D1 expression inhuman breast carcinoma cells by retinoids in vitro, Oncogene 15 (1997)107–115.

[12] R. Liu, S. Takayama, Y. Zheng, B. Froesch, G.Q. Chen, X. Zhang, J.C. Reed, X.K. Zhang,Interaction of BAG-1 with retinoic acid receptor and its inhibition of retinoicacid-induced apoptosis in cancer cells, J. Biol. Chem. 273 (1998) 16985–16992.

[13] K. Niederreither, P. Dolle, Retinoic acid in development: towards an integratedview, Nat. Rev. Genet. 9 (2008) 541–553.

[14] M. Petkovich, N.J. Brand, A. Krust, P. Chambon, A human retinoic acid receptorwhich belongs to the family of nuclear receptors, Nature 330 (1987) 444–450.

[15] R.M. Evans, The steroid and thyroid hormone receptor superfamily, Science 240(1988) 889–895.

[16] D.J. Mangelsdorf, R.M. Evans, The RXR heterodimers and orphan receptors, Cell 83(1995) 841–850.

[17] C.K. Glass, M.G. Rosenfeld, The coregulator exchange in transcriptional functionsof nuclear receptors, Genes Dev. 14 (2000) 121–141.

[18] J.S. Carroll, C.A. Meyer, J. Song, W. Li, T.R. Geistlinger, J. Eeckhoute, A.S. Brodsky,E.K. Keeton, K.C. Fertuck, G.F. Hall, Q. Wang, S. Bekiranov, V. Sementchenko, E.A.Fox, P.A. Silver, T.R. Gingeras, X.S. Liu, M. Brown, Genome-wide analysis of estro-gen receptor binding sites, Nat. Genet. 38 (2006) 1289–1297.

[19] S. Hua, R. Kittler, K.P. White, Genomic antagonism between retinoic acid and es-trogen signaling in breast cancer, Cell 137 (2009) 1259–1271.

[20] C.S. Ross-Innes, R. Stark, K.A. Holmes, D. Schmidt, C. Spyrou, R. Russell, C.E.Massie, S.L. Vowler, M. Eldridge, J.S. Carroll, Cooperative interaction betweenretinoic acid receptor-α and estrogen receptor in breast cancer, Genes Dev. 24(2010) 171–182.

[21] A. Gross, J.M. McDonnell, S.J. Korsmeyer, Bcl-2 family members and the mito-chondria in apoptosis, Genes Dev. 13 (1999) 1899–1911.

[22] R. Metivier, G. Penot, M.R. Hubner, G. Reid, H. Brand, M. Kos, F. Gannon, Estrogenreceptor-alpha directs ordered, cyclical, and combinatorial recruitment of cofac-tors on a natural target promoter, Cell 115 (2003) 751–763.

[23] C. Martin, Y. Zhang, The diverse functions of histone lysine methylation, Nat. Rev.Mol. Cell Biol. 6 (2005) 838–849.

[24] Y. Shi, F. Lan, C. Matson, P. Mulligan, J.R. Whetstine, P.A. Cole, R.A. Casero, Y. Shi,Histone demethylation mediated by the nuclear amine oxidase homolog LSD1,Cell 119 (2004) 941–953.

[25] E. Metzger, M. Wissmann, N. Yin, J.M. Muller, R. Schneider, A.H.F.M. Peters, T.Gunther, R. Buettner, R. Schule, LSD1 demethylates repressive histone marksto promote androgen-receptor-dependent transcription, Nature 437 (2005)436–439.

[26] V. Perissi, M.G. Rosenfeld, Controlling nuclear receptors: the circular logic of co-factor cycles, Nature Rev. Mol. Cell. Biol. 6 (2005) 542–554.

[27] Q. Hu, Y.S. Kwon, E. Nunez, M.D. Cardamone, K.R. Hutt, K.A. Ohgi, I.Garcia-Bassets, D.W. Rose, C.K. Glass, M.G. Rosenfeld, X.D. Fu, Enhancingnuclear receptor-induced transcription requires nuclear motor and LD1-dependentgene networking in interchromatin granules, Proc. Natl. Acad. Sci. 105 (2008)19199–19204.

[28] S.M. Aronica, W.L. Kraus, B.S. Katzenellenbogen, Estrogen action via the cAMP sig-naling pathway: stimulation of adenylate cyclase and cAMP-regulated gene tran-scription, Proc. Natl. Acad. Sci. 91 (1994) 8517–8521.

[29] A. Pedram, M. Razandi, E.R. Levin, Nature of functional estrogen receptors at theplasma membrane, Mol. Endocrinol. 20 (2006) 1996–2009.

[30] A. Gallo, E. Benusiglio, I.M. Bonapace, A. Feliciello, S. Cassano, C. Garbi, A.M. Musti,M.E. Gottesman, E.V. Avvedimento, V-ras and protein kinase C dedifferentiatethyroid cells by down-regulating nuclear camp-dependent protein kinase A,Genes Dev. 6 (1992) 1621–1630.

[31] D. Rai, A. Frolova, J. Frasor, A.E. Carpenter, B.S. Katzenellenbogen, Distinctive ac-tions of membrane-targeted versus nuclear localized estrogen receptors in breastcancer cells, Mol. Endocrinol. 19 (2005) 1606–1617.

[32] Q. Zhao, J. Tao, Q. Zhu, P.M. Jia, A.X. Dou, X. Li, F. Cheng, S. Waxman, G.Q. Chen, S.J.Chen, M. Lanotte, Z. Chen, J.H. Tong, Rapid induction of cAMP/PKA pathwayduring retinoic acid-induced acute promyelocitic leukemia cell differentiation,Leukemia 18 (2004) 285–292.

[33] Y. Li, C. Deng, X. Hu, B. Patel, X. Fu, Y. Qiu, M. Brand, K. Zhao, S. Huang, Dynamicinteraction between TAL1 oncoprotein and LSD1 regulates TAL1 function in he-matopoiesis and leukemogenesis, Oncogene 31 (2011) 5007–5018.

[34] E. Sonneveld, C.E. van den Brink, B.M. van der Leede, R.K. Schulkes, M. Petkovich,B. van der Burg, P.T. van der Saag, Human retinoic acid (RA) 4-hydroxylase(CYP26) is highly specific for all-trans-RA and can be induced through RA recep-tors in human breast and colon carcinoma cells, Cell Growth Differ. 9 (1998)629–637.

[35] Y. Tsukada, J. Fang, H. Erdjument-Bromage, M.E. Warren, C.H. Borchers, P. Tempst,Y. Zhang, Histone demethylation by a family of JmjC domain-containing proteins,Nature 439 (2006) 811–816.

[36] M.G. Rosenfeld, V.V. Lunyak, C.K. Glass, Sensors and signals: a coactivator/corepressor/epigenetic code for integrating signal-dependent programs of tran-scriptional response, Genes Dev. 20 (2006) 1405–1428.

[37] F. Darro, P. Cahen, A. Vianna, C. Decaestecker, J.M. Nogaret, B. Leblond, C.Chaboteaux, C. Ramos, M. Pétein, V. Budel, A. Schoofs, B. Pourrias, R. Kiss, Growthinhibition of human in vitro and mouse in vitro and in vivo mammary tumormodels by retinoids in comparison with tamoxifen and the RU-486 anti-progestagen, Breast Cancer Res. Treat. 51 (1998) 39–55.