Biochimica et Biophysica Acta 1813 (2011) 1767–1776

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta

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

Comparative role of acetylation along c-SRC/ETS1 signaling pathway in bonemetastatic and invasive mammary cell phenotypes

Paola Bendinelli a, Paola Maroni b,1, Emanuela Matteucci a,1, Maria Alfonsina Desiderio a,⁎a Dipartimento di Morfologia Umana e Scienze Biomediche “Città Studi”, Molecular Pathology Laboratory, Università degli Studi di Milano, Milano, Italyb Istituto Ortopedico Galeazzi, Milano, Italy

⁎ Corresponding author at: Università degli Studi di MUmana e Scienze Biomediche “Città Studi”, Molecular PMangiagalli, 31, 20133 Milan, Italy. Tel.: +39 02503153

E-mail address: a.desiderio@unimi.it (M.A. Desiderio1 These authors contributed equally to this work.

0167-4889/$ – see front matter © 2011 Elsevier B.V. Aldoi:10.1016/j.bbamcr.2011.06.004

a b s t r a c t

a r t i c l e i n f o

Article history:Received 17 January 2011Received in revised form 16 May 2011Accepted 7 June 2011Available online 29 June 2011

Keywords:c-SrcHDACsCXCR4Breast cancerBone metastasis

Metastatic cells switch between different modes of migration through supramolecular plasticity mechanism(s)still largely unknown. The aimof thepresent paperwas to clarify somemolecular aspects of the epigenetic controlofmigration of 1833-bonemetastatic cells compared toMDA-MB231-parentalmammary carcinoma cells. Activec-Src overexpression enhanced 1833-cell spontaneous migration and CXCR4-mediated chemoinvasion towardCXCL12 ligand. Only inmetastatic cells, in fact, c-Src seemed to stabilize nuclear CXCR4-protein receptor possiblydue to tyrosine phosphorylation, by impairingprotein-degradative smear and causing instead anelectrophoretic-mobility shift; the cytosolic steady-state level of CXCR4 was enhanced, and the protein appeared alsophosphorylated. These findings suggested the triggering of unique signaling pathways in metastasis for homingof breast-cancer cells to congenial environment of specific organs. Microenvironmental stimuli activating c-Srcmight influence Ets1 binding to CXCR4 promoter and consequent transactivation, as well as CXCR4 post-translational regulatory mechanisms such as phosphorylation. Enhancement of Ets1 activity and CXCR4induction by c-Src overexpression were prevented by histone deacetylase (HDAC) blockade. In contrast, HDACinhibition with trichostatin A increased cytosolic phosphorylated CXCR4 expression in MDA-MB231 cells, butEts1 involvement was practically unneeded. c-Src might be suggested as a bio-marker predicting metastasissensitivity patterns to HDAC inhibitors. Rationally designed and individualized therapy may become possible asmore is learned about the target molecules of HDAC's inhibitory agents and their roles, as undertaken for CXCR4that is likely to be crucial for homing, angiogenesis and survival in a c-Src-dependentmanner in bone-metastaticmammary cells.

ilano, Dipartimento Morfologiaathology Laboratory, via Luigi34; fax: +39 0250315338.).

l rights reserved.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

The tyrosine kinase c-Src is involved in the regulation of a range ofcellular processes, including proliferation/survival, adhesion, motilityand bone metabolism. There is evidence for a role of c-Src intumorigenesis and in osteoclastic activity associated with bonemetastasis [1]. c-Src activity is higher in metastasis from colon cancerthan in both primary tumor tissue and in cells with limited metastaticpotential [2–4]. Greater metastatic potential and worse prognosis forcolorectal cancer are, therefore, associated with active c-Src [5].

For bone metastasis of breast cancer few data are available. In amouse model of breast cancer, c-Src inhibition reduces the incidenceof bone metastases, increasing the survival [6]. c-Src-responsivesignature (SRS), consisting of a group of genes that are expressedwhen c-Src is active, characterizes breast carcinomas that relapse to

bone. SRS seems more effective than estrogen-responsive status inidentifying breast cancer patients who would develop bone metas-tases [7].

Recently, we have shown some key molecular mechanismsimplicated in the c-Src role in breast carcinoma invasiveness, considereda prerequisite for metastatic capacity. c-Src mediates transductionsignals downstream of Met-tyrosine kinase, the specific receptor ofhepatocyte growth factor (HGF) [8,9]. HGF/Met couple is critical for thecross-talk between breast cancer metastasis and the bone microenvi-ronment [9,10]. Only in human 1833 clone—with bone metastaticcapacity—HGF stabilizes plasma membrane and nuclear Met, that co-precipitates with phospho-c-Src (pSrc). Differently, in parental-breastcancer MDA-MB231 cells Met is rapidly down-regulated after HGFexposure as well as after histone deacetylase (HDAC) inhibition withtrichostatin A (TSA) [9,11]. In invasive MDA-MB231 cells, HGF alsocauses HDAC3 phosphorylation via c-Src, possibly reducing NF-κBactivity and decreasing CXCR4 protein level [8].

In breast cancer, CXCR4 is the most abundantly expressedchemokine receptor [12], and is observed in 33–55% of primarymammary carcinomas [13,14]. CXCR4 expression is associated withshorter disease-free survival and increased risk of recurrence. CXCR4

1768 P. Bendinelli et al. / Biochimica et Biophysica Acta 1813 (2011) 1767–1776

positivity in the cytoplasm and its phosphorylation are consideredparameters of tumor aggressiveness, with potential value as diagnos-tic and prognostic marker. Site-specific phosphorylation of CXCR4 isdynamically regulated by multiple kinases resulting in both positiveor negative modulation of CXCR4 signaling [15]. Nuclear CXCR4localization seems important for breast tumor progression [13,16,17],and we have shown CXCR4 in nuclei of 1833-bone metastatic cells[10]. CXCR4 protein is detected in 67% of bone metastasis samples,being 3-times more than in visceral metastases [14]. However, CXCR4expression is lower in lymph node metastasis than in primary breastcancers [18].

CXCL12 ligand binding to CXCR4 promotesmigration, invasion andadhesion of breast cancer cells in vitro[13,19]. Noteworthy, HGFdifferently influences spontaneous and CXCL12-stimulated cellmigration depending on the aggressiveness of breast cancer cells, asreported by us for MCF-7 (low invasive) versus MDA-MB231 (highlyinvasive) cells [11]. Because of CXCR4 regulation by cyclooxygenase-2signaling, that mediates the hypoxic response, a crucial role of thismolecular network in the homing and angiogenesis of 1833-bonemetastasis has been suggested [10]. Also, CXCR4 overexpressionseems to enhance the metastatic phenotype of breast cancer cells invivo [20–22]. These effects might be partly due to CXCL12-triggeredtransactivation of CXCR4, mediated by EGF and HER2 receptors in ac-Src-dependent manner [23].

The aim of the present paper was to extend and deepen theknowledge about the role of c-Src, as key molecular regulator ofhuman-bone metastasis of breast cancer, by controlling CXCR4expression through Ets1 activity. To this scope, we comparativelyexamined 1833-bone metastatic clone and the parental-invasiveMDA-MB231 cells. In this context, the role of acetylation on c-Src/Ets1signaling was considered by HDAC blockade with TSA or specificknocking-down. Acetylation results from the balance between theopposite activities of histone acetyltransferases and HDACs. Manynon-histone proteins have been identified to be the substrates ofHDACs, such as proteins involved in transcription (p53, NF-κB, etc.),hormone response, nuclear transport, DNA repair, Wnt signaling(β-catenin) and heat-shock/chaperone response (HSP90) [24,25].HDACs seem to regulate also c-Src activity, and TSA inhibits c-Srcpromoters [11,26]. Generally, increased levels of histone acetylation(hyperacetylation) are associated with enhanced transcriptional activ-ity, whereas decreased levels of acetylation (hypoacetylation) areassociated with repression of gene expression [27,28], even if oppositeregulation can also occur. No data are available on Ets1-activity controlvia acetylation in the bone metastatic process of breast cancer. Thecorrelation with uPA levels and metalloproteinase expression leads tosuppose that Ets1mayplaya role inhumanbreast cancermetastasis andmetastatic cell invasiveness [29,30].

The reasoning of our work was to evaluate in 1833 versus MDA-MB231 cells the correlation and the reciprocal influence of molecularevents, possibly involved in invadopodia formation and functioning.Carcinoma cells utilize invadopodia to degrade extracellular matrixduring invasion and metastasis [31], and endogenous c-Src has beenshown to promote invadopodia formation in response to growthfactors and chemokines. Abl-kinase and the chemokine receptorCXCR4 are downstream of CXCL12, serum growth factors andactivated c-Src [32]. Epigenetic control mechanisms are likely toinfluence tumor aggressive phenotype [33]. Few data are available,however, about the differences in spontaneous and CXCL12/CXCR4-mediated specific chemoinvasion, in the molecular events involvedand in the regulatory acetylation mechanisms occurring betweenpoorly and highly bone-metastatic cells.

Preliminary findings in MDA-MB231 cells indicate a role of Ets1 inCXCR4 expression due to HGF, a microenvironmental signal that mightinfluence acetylation [11], but is unknown the role played by Ets1downstreamof c-Src in the epigenetic control of phenotype plasticity ofmetastatic cells, possibly important for the bone avidity via CXCR4.

2. Materials and methods

2.1. Reagents

TSA was from Sigma Chemical Co. (St. Louis, MO). Recombinant-human CXCL12 and anti-CXCR4monoclonal antibody (MAB-182) werefrom R&D Systems (Abingdon, UK). Anti-Ets1 (C20), anti-vinculin andanti-B23 antibodies were purchased from Santa-Cruz Biotechnology(Santa-Cruz, CA). Anti-c-Src (clone GD11), anti-phospho-c-Src(Tyr416), anti-phosphotyrosine (anti-pTyr) (clone 4G10) and anti-HDAC1were from Upstate Biotechnology (Lake Placid, NY). Anti-HDAC3antibody was from Cell Signalling (Beverly, MA). Fugene 6 was fromRoche Applied Science (Germany). Lipofectamine 2000 was fromInvitrogen (Milan, Italy). pRL-TK (Renilla luciferase) was from Promega(Madison, WI). siRNA sequences, 5′-CGUACGCGGAAUACUUCGATT-3′(siRNA LUC, control); 5′-CAGCGACUGUUUGAGAACCTT-3′ (HDAC1.1)and 5′-CUAAUGAGCUUCCAUACAATT-3′ (HDAC1.3) for siRNA HDAC1;5′-GAUGCUGAACCAUGCACCUTT-3′ for siRNAHDAC3,were synthesizedbyMWG (Ebersberg, Germany). All other chemicals were of the highestgrade available.

2.2. Cell cultures

The human breast carcinoma cells MDA-MB231 and the bonemetastatic 1833 clone, derived from MDA-MB231 cells, were a kindgift of J. Massagué (Memorial Sloan-Kettering Cancer Center, NewYork). The cells were maintained in DMEM containing 10% fetalbovine serum (FBS). TSA (2.5 μM) treatment was performed for 24 hin 10% FBS containing medium [11].

2.3. Plasmids

c-Src wild-type (Srcwt) expression vector was from S. Parson(University of Virginia, Charlottesville, VA); the dominant negative forEts1 (ΔEBHHB, ΔEts1) was from J. Ghysdael (Institut Curie, Orsay,France). The CXCR4(−600/+19)Luc construct was obtained from theoriginal gene reporter CXCR4(−2632/+86)Luc, kindly provided aspGL2 basic-construct by A.J. Caruz (Universidad de Jaen, Madrid,Spain), as previously described [34]; the mutated CXCR4(−600/+19)Luc was prepared by deleting two bases inside the p65-consensus sitecore, as previously described [11]. The gene reporter plasmid NFkBLuccontaining three NF-κB consensus sequences was from M. Hung(Anderson Cancer Center, Huston, TX); pGL3PGK6TKp containing sixHRE was from P.J. Ratcliffe (Welcome Trust Center for Human Genetics,Oxford, UK).

2.4. Matrigel invasion assay

Invasion assays were carried out usingMatrigel invasion chambersfrom BD Biocoat Cellware (Becton Dickinson Labware, Bedford, MA)[34]. Cultured cells, transiently transfected or not with 2.5 μg of Srcwtper 106 cells, or treated for 24 hwith TSA (2.5 μM)were harvested, re-suspended in medium without serum, added (8×104 per well) intriplicate to the upper chambers, and allowed to invade throughMatrigel in the absence of serum in the medium. In someexperiments, CXCL12 (200 ng/ml) was added to the bottom chamberto evaluate chemoinvasion through Matrigel. The cells that hadinvaded the lower surface of the membrane were fixed, stained andcounted.

2.5. Transient transfection assay

The cells, seeded in 24-multiwell plates, were transfected with aDNA/Fugene 6 mixture containing 200 ng of each of the followingconstructs: wild type or mutated CXCR4Luc, NFkBLuc, HRELuc orSrcwt-expression vector. In some experiments, 1 μg per well of ΔEts1

Fig. 1. Invasiveness in Srcwt- and TSA-treated cells. (A) Control and treated cells wereused for Matrigel invasion assay, adding culture medium without serum to thechambers. To estimate spontaneous invasion, we counted (magnification ×200) theinvading cells on the lower side of the membrane after staining. Representative imagesare shown, and the numbers at the bottom are the mean±S.E. of the counts of tenselected fields from three independent experiments. *Pb0.05 vs 1833-control cells.(B) For specific chemoinvasion of TSA-treated MDA-MB231 cells or Srcwt-transfected1833 cells, CXCL12 was added to the bottom chamber in the absence of serum. Thehistograms report the mean±S.E. of the number of migrated cells in ten selected fields.The experiments have been done in triplicate. *Pb0.05; **Pb0.005 vs respective controlcells. °Pb0.05 vs CXCL12-treated MDA-MB231 cells; ΔPb0.05 vs CXCL12-treated 1833cells.

1769P. Bendinelli et al. / Biochimica et Biophysica Acta 1813 (2011) 1767–1776

was co-transfected. For normalization the cells were transfected withpRL-TK (Renilla luciferase) vector, and Firefly/Renilla luciferaseactivity ratios were calculated by the software [34].

2.6. Transfection of siRNAs

Some cells seeded in 6-multiwell plates, were transfectedwith 150nM siRNAs (control, HDAC1 and HDAC3) using Lipofectamine 2000.Two cycles of siRNA transfection were performed at 24 h interval;some cells were co-transfected with Srcwt expression vector after thefirst cycle. The cells, harvested at 72 h, were lysed in urea buffer (8 Murea, 0.1 M NaH2PO4, 0.01 M Tris, pH 8, and proteinase inhibitors) toobtain proteins for Western blot analysis, or processed to obtainnuclear extracts for Electrophoretic mobility shift assay (EMSA) [35].

2.7. EMSA analysis

EMSA was performed as previously described [35]. The oligonu-cleotide sequence used for EMSA is shown in Fig. 3A. The Octamer-1binding site sequence, used for loading control, was: 5′-TGCGAATG-CAAATCACTAGAA-3′. For supergelshift assay, nuclear extracts werepre-incubated with 1 μg of anti-Ets1 antibody for 15 min at roomtemperature, followed by incubationwith the labeled oligonucleotide.

2.8. Immunoprecipitation and Western blot assays

The protein extracts were prepared in the presence of phosphataseinhibitors to prevent the loss of protein-phosphorylation state.Protein immunoprecipitation (1 mg of protein) was performedwith 6.5 μg of anti-CXCR4 antibody using 80 μl of protein A Sepharose(1:1, v:v). Total proteins (100 μg) or nuclear proteins (50 μg) wereanalyzed by Western blots. Immunoblots were performed with anti-CXCR4 (5 μg/ml), anti-pTyr (3 μg/ml), anti-Ets1 (1:1000), anti-cSrc(1 μg/ml) or anti-phospho-Src (2 μg/ml) antibody [35]. To confirmequal loading, themembraneswere sequentially immunoblottedwithanti-vinculin (total extracts) or anti-B23 (nuclear extracts) antibody.After incubation with the appropriate secondary antibody, the signalswere detected using enhanced chemiluminescence kit (ECL or ECL-plus; Amersham Biosciences).

2.9. Statistical analysis

The number of migrated cells, densitometric values, and luciferaseactivity were analyzed by analysis of variance, with Pb0.05 consid-ered significant. Differences from controls were evaluated on originalexperimental data.

3. Results

3.1. Different regulation of invasiveness in MDA-MB231 and 1833 cells

Upon mammary tumor progression, multiple classes of extracel-lular matrix-degrading enzymes are upregulated and activated,responsible for spontaneous mesenchymal movement [36–38].Experiments dealing with the molecular events that might beimplicated in the spontaneous migration were carried out withMDA-MB231 and 1833 cells, to clarify the possible role of c-Src andprotein acetylation, and to consider the relationship with aggressive-ness (Fig. 1). To this purpose, we evaluated the effect of theoverexpression of exogeneous c-Src or of the blockade of class I andII HDACs, to cause acetylation modifications [39]. Experiments wereperformed with Matrigel invasion chamber, that is considered an invitro model system for metastasis [40]. In Fig. 1A, representativeimages are reported showing that Srcwt transfection and TSAtreatment oppositely affected 1833 spontaneous migration, whilebeing ineffective in MDA-MB231 cells.

Then, we studied specific chemoinvasion toward CXCL12, added tothe lower chamber of the Matrigel assay system, as shown in Fig. 1B.In MDA-MB231 cells, CXCL12 in the absence and in the presence ofTSA increased 2.5- and 1.5-fold chemoinvasion, respectively. Srcwttransfection was ineffective in respect to control (data not shown).Differently, in 1833 cells specific chemoinvasion was enhanced 3-foldby Srcwt transfection. Thus, diverse effects of TSA and Srcwt onspontaneous and specific migration were observed in highly invasiveand bone-metastatic cells.

3.2. Evaluation of molecular control mechanisms of CXCR4 expression byTSA, and role of Ets1

To deepen the knowledge on CXCL12-induced cell invasion, someCXCR4-transcriptional and post-translational regulatory mechanisms

1770 P. Bendinelli et al. / Biochimica et Biophysica Acta 1813 (2011) 1767–1776

were evaluated. In this regard, the gene reporter containing humanCXCR4 promoter region spanning from−600 to+19, with 6 Ets1 and1 NF-κB (p65) binding sites, and the p65-mutated construct wereused (Fig. 2A) [11,34]. In a first series of experiments performed underTSA exposure (Fig. 2B–D), CXCR4 transactivation and the role playedby Ets1, as well as the protein expression were comparativelyevaluated in MDA-MB231 and 1833 cells.

As shown in Fig. 2B, the mutation of p65 did not modify basalCXCR4 transactivating activity in MDA-MB231 cells. TSA similarlystimulated the luciferase activity of both the constructs; ΔEts1transfection prevented these TSA-stimulatory effects, also reducingbasal luciferase activity. In 1833 cells, Ets1 blockade was inhibitoryonly on themutated-construct basal activity, being higher than that ofthe wild type. Using total cell extracts (Fig. 2C), in MDA-MB231 cellsTSA enhanced (2.5-fold) CXCR4 protein level (50 kDa band), alsoshifting CXCR4-protein migration possibly due to phosphorylation.Differently, in 1833 cells TSA was ineffective and faint slow migratingbands were observed under basal and treatment conditions. The Ets1protein level of control MDA-MB231 cells, that was higher than that of1833 cells, appeared also phosphorylated, decreasing (−50%) afterTSA treatment.

Fig. 2. Transcriptional and post-transcriptional control of CXCR4 by TSA. (A) Constructs cotransiently transfected with the above reported constructs, were treated with TSA or co-luciferase activity ratios. Columns, mean of three independent experiments done in triplicaTSA-treated cells. (C) Total extracts were used for Western blot analysis. Vinculin was usedMDA-MB231 control value, considered as 1. The experiments were repeated three times windicated the ratio pTyr/CXCR4, obtained by using densitometric values after respective immMB231 control value, considered as 1. The experiments were repeated three times with sim

To better evaluate protein phosphorylation, immunoprecipitateswere performed with anti-CXCR4 antibody blotting with anti-pTyrantibody (Fig. 2D). Noteworthy, under our experimental conditionsTSA doubled pTyr/CXCR4 ratio in MDA-MB231 cells. We cannotexclude that also other types of phosphorylation occurred in TSA-treatedMDA-MB231 cells. In the immunoprecipitates carried out with1833-protein extracts, CXCR4 protein level seemed 6-fold higher thanin MDA-MB231 cells.

3.3. Evaluation of molecular control mechanisms of CXCR4 expression bySrcwt, and role of Ets1

In a second series of experiments, the role of Ets1 in c-Srcoverexpressing cells on CXCR4 luciferase activity and proteinexpression were evaluated (Fig. 3). As shown in Fig. 3A, by co-transfecting 1833 cells with CXCR4Luc wild-type and Srcwt theluciferase activity was strongly enhanced (about 6-fold), and ΔEts1was inhibitory. However, in MDA-MB231 cells co-transfection ofΔEts1 inhibited basal luciferase activity of the wild type- and themutated-promoter constructs.

ntaining wild type and p65-mutated (mut) CXCR4 promoter fragments. (B) The cells,transfected with ΔEts1. The histograms report the absolute values for Firefly/Renillate; bars, S.E. *Pb0.05, **Pb0.005 vs corresponding control value. °Pb0.05, °°Pb0.005 vsfor normalization. The numbers at the bottom indicate the fold-variations relative to

ith similar results. (D) Total extracts were used for immunoprecipitation. Relative pTyrunoblotting. The numbers at the bottom indicate the fold-variations relative to MDA-ilar results.

Fig. 3. Transcriptional and post-transcriptional control of CXCR4 by Srcwt. (A) The cells, transiently transfected with the CXCR4wild type andmutated promoter constructs, were co-transfected with Srcwt in the presence or not of ΔEts1. The histograms report the relative fold-variations for Firefly/Renilla luciferase activity ratios. Columns, mean of threeindependent experiments done in triplicate; bars, S.E. *Pb0.05, **Pb0.005 vs corresponding control value. °°Pb0.005 vs Srcwt-transfected cells. (B) Total extracts were used forWestern blot analysis. Vinculin was used for normalization. The numbers at the bottom indicate the fold-variations relative to MDA-MB231 control value, considered as 1. Theexperiments were repeated three times with similar results. (C) Nuclear protein extracts were used for Western blot analysis. B23 was used for normalization. The asterisk indicatesthe supposed degradation bands. The experiments were repeated three times with similar results.

1771P. Bendinelli et al. / Biochimica et Biophysica Acta 1813 (2011) 1767–1776

Fig. 3B reports that Srcwt transfection in 1833 cells tripled theprotein level of CXCR4 (50 kDa band), and phosphorylation bandsappeared upstream. In MDA-MB231 cells, however, a remarkableelectrophoretic mobility shift of CXCR4 was shown after Srcwttransfection, leading to suppose ubiquitination of CXCR4 in agreementwith Malik and Marchese [41]. Ets1 protein level tripled after Srcwttransfection in the two cell lines, appearing phosphorylated in 1833-treated cells (Fig. 3B).

Further experiments were performed to evaluate post-translationalprotein changes of CXCR4 by Srcwt. Using nuclear extracts (Fig. 3C), weobserved a retardation of electrophoretic mobility of CXCR4 protein incontrol 1833 cells, compared to MDA-MB231 cells. After Srcwttransfection, also a CXCR4-tyrosine phosphorylation band was likelyto appear in 1833-cell nuclei. In contrast, principally in the nuclei ofMDA-MB231 cells, transfected with Srcwt, molecular weight forms ofless than 50 kDa were shown, in agreement with receptor degradation[42].

Based on these results, in 1833 clone Srcwt seemed to phosphor-ylate CXCR4while preventing receptor degradation, at differencewithMDA-MB231 cells. Moreover, Ets1 was important for c-Src-stimulatedCXCR4 transactivation in 1833 cells, even if Ets1 protein level wassimilarly affected by Srcwt in both cell lines. Then, specificexperiments were done, and reported in the following sections(Sections 3.4 and 3.5) to clarify this point.

3.4. Ets1 binding to a consensus sequence of CXCR4 promoter andregulation by HDACs

We determined Ets1 binding activity to an oligonucleotidesynthesized using the consensus sequence of the CXCR4 promoter,shown in Fig. 4A. In the CXCR4 promoter, 5 hypoxia inducible factorresponsive elements (HRE) are present, upstream the Ets1 consensussequences, consistent with a possible interacting regulatory functionof hypoxia inducible factor-1 (HIF-1) and Ets1 [43].

As shown in Fig. 4B, Ets1–DNA binding was markedly elevated inMDA-MB231 cells under basal conditions. Differently, in 1833 cells ahuge increase of Ets1–DNA binding was observed after Srcwttransfection. The specificity of the Ets1–DNA complex was examinedby competition experiments using 50-fold molar excess of unlabelledoligonucleotide (50×), which almost completely suppressed thecomplex formation. Moreover, supergelshift analyses were done withcontrol MDA-MB231 and Srcwt-transfected 1833 cells. The preincuba-tionwith the anti-Ets1 antibody (Ab) disrupted the DNA–Ets1 complex,causing also a shift (ss) beyond giving immunodepletion. Weinvestigated the involvement of class I-HDACs 1 and 3 in Ets1–DNAbinding under basal conditions and Srcwt transfection in the two celllines. By knocking-down the HDACs with the specific siRNAs, Ets1-binding activity in MDA-MB231 cells was partly reduced, while theSrcwt-stimulated Ets1 binding was abolished in 1833 cells. Octamer-1

Fig. 4. HDACs involvement in Ets1–DNA binding after Srcwt transfection. (A) The oligonucleotide consensus sequence for Ets1, present in CXCR4 promoter, was used for EMSA.(B) EMSA analyses of Ets1 and Octamer-1 (Oct-1), using nuclear extracts from control and transfected cells. 50×, specific competition. Supergelshift was performed with anti-Ets1antibody (Ab), and nuclear extracts from control MDA-MB231 and Srcwt-transfected 1833 cells. The experiments were repeated three times with similar results. (C) Western blotsof total cell extracts after transfection of siRNAs for HDACs, and antiluciferase siRNA as control (siRNACTRL). Vinculin was used for normalization. The blots shown are representativeof three independent experiments. (D) Total cell extracts from Srcwt and siRNA-HDACs co-transfected cells were used forWestern blot analysis. Vinculin was used for normalization.The numbers at the bottom indicate the fold-variations relative to MDA-MB231 control value, considered as 1. The experiments were repeated three times with similar results.

1772 P. Bendinelli et al. / Biochimica et Biophysica Acta 1813 (2011) 1767–1776

DNA binding was unaffected by the treatments (data not shown).HDACs knocking down caused the almost complete disappearance ofHDAC proteins (Fig. 4C). To check the reproducibility of the knockingdown, we used two oligonucleotides for HDAC1, i.e. 1.1 and 1.3.

Srcwt transfection enhanced the levels of c-Src (about 5-fold) andof pSrc (about 7-fold) in both the cell lines used (Fig. 4D). pSrc is theactive form of the tyrosine kinase [44]. The siRNAs for HDACs wereeffective in reducing the c-Src protein levels in Srcwt-transfected cells.

Our data suggest that Ets1–DNA binding is activated by c-Src onlyin 1833 cells, involving protein hypoacetylation, consistent withSrcwt-dependent CXCR4 transactivation (Fig. 2B). In MDA-MB231cells, however, HDACs blockade partly reduced Ets1 protein level(Fig. 2C) and the Ets1 binding to CXCR4 promoter (Fig. 4B), stillpermitting a certain CXCR4 expression, even if the functional role, i.e.specific chemoinvasion towards CXCL12, possibly depended onphosphorylation after TSA treatment (Figs. 2C and 1B).

3.5. Transactivating activities of NF-κB and HIF-1 were differentlyregulated by c-Src/HDACs depending on cell aggressiveness

The binding sites of NF-κB (p65) and HIF-1 are adjacent to those forEts1 on CXCR4 promoter (−2060/+19) (Fig. 5), leading to suppose apossible cooperative function. Gene expression depends, in fact, on thespecificity of the stimulus and the interaction of transcription factorsshowing consensus sites on the gene promoter. Transcription factors donot work alone but in cooperation. The transcriptional properties of aparticular factor are influenced either by its position, relative to otherfactors bound to a given promoter, or by the abundance of transcrip-tional cofactors in a given cell type in a certain context [45]. Thisregulatory mechanism might explain differences in Ets1 activity inresponse to Srcwt in the two cell lines used.

As shown in Fig. 5A, the activity of the construct driven by themultimer of NF-κB-consensus sequences was 2.5-fold higher in MDA-

Fig. 5. Transcription-factor transactivating activities in Srcwt-transfected cells, and role of HDACs. (A and B) The cells were transiently transfected with the indicated gene reporters,expression vectors and siRNAs (HDAC1 plus 3). The histograms report the absolute values for Firefly/Renilla luciferase activity ratios. Columns, mean of three independentexperiments done in triplicate; bars, S.E. *Pb0.05; **Pb0.005; ***Pb0.001 vs corresponding control value. ○○Pb0.005 vsMDA-MB231 control value. ΔΔPb0.005; ΔΔΔPb0.001 vs 1833-Srcwt transfected cells. (C) Schematic representation of the different regulatorymechanisms and events controlling CXCR4 expression inMDA-MB231 and 1833 cells, exposed to TSAor Srcwt. HIF-1 is inactive in MDA-MB231 cells [10,35]. TSA inhibits NF-κB activity and activates Ets1 [11].

1773P. Bendinelli et al. / Biochimica et Biophysica Acta 1813 (2011) 1767–1776

1774 P. Bendinelli et al. / Biochimica et Biophysica Acta 1813 (2011) 1767–1776

MB231 than in 1833 cells. Vice versa, HIF-1 was active in 1833 cellsunder basal conditions, in agreementwith our previousworks [10,35].Both Srcwt transfection and HDACs blockade reduced NF-κB luciferaseactivity in MDA-MB231 and 1833 cells. Differently, in 1833 cells Srcwttransfection activated the luciferase activity of HRE-multimer con-struct, while being prevented by concomitant HDACs blockade.

Thus, the possible functional interaction of the transcriptionfactors was examined (Fig. 5B), transfecting the gene reporter drivenby HRE multimer in the presence of ΔEts1. In 1833 cells, the HRE-luciferase activity doubled after c-Src overexpression, and ΔEts1largely prevented this stimulatory effect.

Fig. 5C shows a schematic representation of the data. In MDA-MB231 cells, TSA is known to activate Ets1 and to inhibit NF-κBactivity [11]. CXCR4 might be a gene-target of TSA-stimulated Ets1,even if post-translational regulation involving tyrosine phosphoryla-tion was probably critical for the functional role in chemoinvasion.Srcwt seemed to cause CXCR4 degradation. In 1833 cells, CXCR4 wasinduced by Srcwt, possibly followed by nuclear translocation andprotein phosphorylation/stabilization without degradation.

4. Discussion

We give evidence that c-Src influenced spontaneous invasivenessand CXCR4-mediated chemoinvasion of 1833-bone metastatic cellsfrommammary carcinoma. c-Src seemed to be specifically involved innuclear CXCR4 phosphorylation and stability, impairing degradationof the chemokine receptor at difference with parental MDA-MB231cells. These molecular mechanisms might regulate the signalingdownstream, in addition to the known role played by multiple-serinekinases in the CXCR4 function. Numerous serine-phosphorylationsites have been suggested to control CXCL12- and PKC-mediatedCXCR4 signaling, internalization and degradation [15]. Phosphorylat-ed CXCR4 is the active form of the receptor associated with themetastatic risk [17]. The c-Src-dependent transactivation network,leading to CXCR4 expression and phosphorylation, might influencemorphodynamic and molecular changes allowing the maintenance ofmigratory dissemination, possibly also after abrogation of pericellularproteolysis. In fact, metastasis shows migratory compensationstrategies that make difficult to devise antimetastatic therapies [38].The ability to distinguish phosphorylated CXCR4will be invaluable forthe analysis of the role of CXCR4 in cancer, and the development ofCXCR4 antagonist therapy [46]. CXCR4 is representative of trans-membrane proteins expressed on the surface of normal and/or cancerstem cells. Small molecule compounds or human antibody targeted toCXCR4 have promising preclinical applications for cancer therapeu-tics, focusing on cancer stem cells of the primary lesion as well asmetastasis or recurrence niches, such as bone marrow and peritonealcavity [47].

We suggest that c-Src overexpression was a critical molecular eventcontributing to the remarkable plasticity of 1833-gene expressionprofile, in comparisonwithparentalMDA-MB231 cells.Our presentdatademonstrated that only in 1833metastatic cells c-Src induced CXCR4. Inaddition, in 1833 cells c-Src is a signal transducer of Met, another keyreceptor for bonemetastatization of breast cancer and for the cross-talkwith bone microenvironment [9]. Therefore, through c-Src expression1833 cells might become responsive to CXCL12 and HGF, bonemicroenvironmental stimuli acting via CXCR4 and Met receptors. BothCXCR4 (G protein-coupled receptor, GPCR) and Met (tyrosine kinasereceptor) might influence the metastatic phenotype, and both possiblytrigger critical signaling at nuclear level, even if the data at regard arevery scarce [9,10,48]. The role of other nuclearly-located GPCRs forbioactive lipoids, such as PGE2 and PAF, has never been investigated inbone metastases [49].

Metastatic cells show adaptability to biological and physicalmicroenvironmental stimuli that are likely to control gene transcriptionat epigenetic level, but the molecular mechanisms are still largely

unknown andwould include c-Src/Ets1 signaling.When overexpressed,c-Src enhanced Ets1–DNAbinding to CXCR4 gene promoter, the CXCR4-transactivating activity, the cytosolic protein level and phosphorylationof CXCR4 in 1833 cells. Noteworthy, multiple levels of regulation ofCXCR4 by c-Src occurred in 1833 cells, possibly including tyrosinephosphorylation and the stabilization at nuclear level. Not only anelectrophoretic mobility shift of CXCR4 protein was observed in nucleiof control 1833, when comparedwithMDA-MB231 cells, but also Srcwtstressed the phenomenon giving an upper phosphorylation bandprobably important for the CXCR4 functionality.

This is the first time that nuclear CXCR4-tyrosine phosphorylationin response to c-Src overexpression has been observed with a role inreceptor stabilization by impairing CXCR4 degradation, and in themaintenance of CXCR4-dependent signal responsive to bone-metas-tasis microenvironment. PKC-mediated phosphorylation in Ser-324/5is known, instead, to drive receptor degradation [15]. This regulatorymechanism might be similar to that of β-catenin, i.e. diverse aminoacid phosphorylation in serine or tyrosine might have opposite effectsenhancing or reducing ubiquitin-dependent degradation [9,50].However, phosphorylation seems also responsible for regulatingCXCR4 trafficking into non-lysosomal compartments including recy-cling endosomes [51]. Tyrosine phosphorylation of CXCR4 mightinfluence other two regulatory mechanisms for GPCRs, i.e. thedeubiquitinating enzyme UPS8 and β-arrestin, that promote traffick-ing and degradation of CXCR4 receptor [41,42]. UPS8 loss of functiondiminishes CXCR4 turnover, inducing receptor accumulation onenlarged early endosomes—positive for hepatocyte growth factor-regulated substrate—and affecting ERK signaling and intracellulartrafficking [42]. β-arrestin controls the amount of ubiquitinated-CXCR4 that is degraded in the endosomes [41].

CXCR4 is overexpressed in at least 23 types of cancer [52], withdysregulation of CXCR4 transcription, signaling and trafficking [15].Altogether, multiple serines, threonines and tyrosines can bephosphorylated in response to both ligand binding or activity inparallel signaling pathways, making the definition of an activated,phosphorylated form of the receptor complex [46].

Direct or indirect c-Src-dependent phosphorylation of CXCR4might occur in early endosomes, followed by nuclear transfer, even ifwe cannot exclude intranuclear tyrosine phosphorylation. Endosomesare key signaling “stations” (signaling endosomes), endowed ofmultiple functions, that generate unique signals prohibited at plasmamembrane, permitting signal diversification and specificity [53]. Idealfeatures of endosomes include two or more simultaneous interac-tions, because of limited surface area of endosomes, and microtubule-mediated transport of molecules from plasma membrane to thenucleus. Altogether, these molecular mechanisms might contribute tomodulate the metastatic potential of breast carcinoma cells, byinherent host-derived metastatic risk factor(s) activating CXCR4.

The exact significance of CXCR4 in 1833-cell nuclei is not clear, anda role in specific signaling for breast-carcinoma metastasis migrationcan be suggested. CXCR4 nuclear localization is related to invasive andmetastatic trait of renal carcinoma, exerting transcriptional regulationof genes that control invasion, migration and chemoattraction [54]. Adiffuse immunofluorescence signal for CXCR4 throughout prostatecancer cell, including nuclei, is consistently reported, and CXCR4-mediated migration was through the ERK1/2 pathway [55]. Hassan etal. observe by immunohistochemical analysis cytoplasmic and nuclearexpression of CXCR4 and the phosphorylated form, the latter beingrelated to breast carcinoma progression [17].

We show that hypoacetylationmediated gene-expression enhance-ment of CXCR4 via c-Src/Ets1 activity, possibly by influencing c-Srcpromoter [11,26] and Ets1-binding activities. In 1833 metastatic cells,the knocking down of HDACs 1 and 3 prevented indeed Srcwt-dependent Ets1 binding to the consensus sequence on CXCR4promoter.Consistently, in response to siRNAs for HDACs, 1833-cell migration wasreduced [35]. Inmammarymetastatic cells HDACsmight be coactivators

1775P. Bendinelli et al. / Biochimica et Biophysica Acta 1813 (2011) 1767–1776

of Ets1, as previously shown for HIF-1 and NF-κB [35]. Acetylation stateof histones H3 and H4 is associated with active chromatin and geneexpression, so we would expect the treatment with HDAC inhibitors toinduce a general increase in gene expression profile. However, theavailable data indicate that HDAC inhibitors cause transcriptionalchanges of a limited set of genes, and that the number of upregulatedand downregulated genes is very similar [56]. β-arrestins are endocyticadaptor proteins downstream of activated CXCR4, regulating thesignaling pathway that includes c-Src [15]. In addition, β-arrestinsfunction as nuclear messenger, regulating transcription in several waysalso interactingwith nucleic-acid bindingproteins [53]. Thus,we cannotexclude that β-arrestins play a role also in metastasis nuclear-transactivation pathway.

In conclusion, only in 1833-bone metastatic cells Ets1 wastranscriptionally functional and participated in CXCR4 inductionunder c-Src-overexpression conditions involving HDACs. Based on ourresults, Ets1 seemed to cooperate with other transcription factors onCXCR4 promoter, and was phosphorylated possibly recruiting CREB-binding protein/p300 coactivator [57]. Therefore, these regulatorymechanism(s) dependedon the cell typeand aggressiveness. Consistentwith HIF-1 activation by phosphorylated c-Src, blocked by Ets1-dominant negative, Ets1 might cooperate with HIF-1 to c-Src response.We have shown that 1833 but not MDA-MB231 cells are responsive tohypoxia, because of HIF-1 functionality due to stabilization of α and βsubunits [10]. This network of molecular events might explain the keydifferences between poorly and highly metastatic cells, being possiblyrelated to metastatic mode of migration that reflects supramolecularplasticity mechanisms. In contrast, in MDA-MB231 cells under basalconditions NF-κB activitywas elevated possibly contributingwith Ets1–DNA binding to CXCR4 expression. Srcwt enhanced nuclear CXCR4degradation, while TSA overexpressed cytosolic CXCR4 in a phosphor-ylated form, suggesting the latter as the molecular mechanismpositively affecting chemoinvasion.

Here, we extend and deepen the knowledge on the function of c-Src in relation to the homing of metastatic cells via CXCR4, extensibleto angiogenesis and growth/survival of bone metastasis [58]. It hasbeen suggested that the behavior of cell homing to the bone mayinclude a direct vascular pathway, highly permeable vascularsinusoids, chemotactic factors produced by marrow stromal cellssuch as CXCL12, and synthesis of growth factors by resident cellswithin the bone and the bone marrow, that support the survival andproliferation of metastatic cells [59]. Studies are in progress to clarifythe role of c-Src expression in human specimens of bone metastasis ofbreast cancer as predictive biomarker for the further advancement ofepigenetic targets, with significant impact on defining the clinicalutility of HDAC inhibitors and subset of responsivemetastasis-bearingpatients [60].

Acknowledgments

This study was supported by funds from Ministero Salute, RicercaCorrente—4029 and Fondazione Cariplo—2010-0737.

References

[1] J. Araujo, C. Logothetis, Targeting Src signaling in metastatic bone disease, Int. J.Cancer 124 (2009) 1–6.

[2] M.S. Talamonti, M.S. Roh, S.A. Curley, G.E. Gallick, Increase in activity and level ofpp 60c-src in progressive stages of human colon cancer, J. Clin. Invest. 91 (1993)53–60.

[3] C.A. Cartwright, C.A. Coad, B.M. Egbert, Elevated c-Src tyrosine kinase activity inpremalignant epithelia of ulcerative colitis, J. Clin. Invest. 93 (1994) 509–515.

[4] W. Mao, R. Irby, D. Coppola, L. Fu, M. Wloch, J. Turner, H. Yu, R. Garcia, R. Jove, T.J.Yetman, Activation of c-Src by receptor tyrosine kinases in human colon cancercells with high metastatic potential, Oncogene 15 (1997) 3083–3090.

[5] H. Aligayer, D.D. Boyd, M.M. Heiss, E.K. Abdalla, S.A. Curley, G.E. Gallick, Activationof Src kinase in primary colorectal carcinoma: an indicator of poor clinicalprognosis, Cancer 94 (2002) 344–351.

[6] N. Rucci, I. Recchia, A. Angelucci, M. Alamanou, A. Del Fattore, D. Fortunati, M.Susa, D. Fabbro, M. Bologna, A. Teti, Inhibition of protein kinase c-Src reduces theincidence of breast cancer metastases and increases survival in mice: implicationsfor therapy, J. Pharmacol. Exp. Ther. 318 (2006) 161–172.

[7] X.H. Zhang, Q. Wang, W. Gerald, C.A. Huidis, L. Norton, M. Smid, J.A. Fœkens, J.Massagué, Latent bone metastasis in breast cancer tied to Src-dependent survivalsignals, Cancer Cell 16 (2009) 67–78.

[8] E. Matteucci, E. Ridolfi, P. Maroni, P. Bendinelli, M.A. Desiderio, c-Src/histonedeacetylase 3 interaction is crucial for hepatocyte growth factor-dependentdecrease of CXCR4 expression in highly invasive breast tumor cells, Mol. CancerRes. 5 (2007) 833–845.

[9] S. Previdi, P. Maroni, E. Matteucci, M. Broggini, P. Bendinelli, M.A. Desiderio,Interaction between human-breast cancer metastasis and bone microenviron-ment through activated hepatocyte growth factor/Met and β-catenin/Wntpathways, Eur. J. Cancer 46 (2010) 1679–1691.

[10] P. Maroni, E. Matteucci, A. Luzzati, G. Perrucchini, P. Bendinelli, M.A. Desiderio,Nuclear co-localization and functional interaction of COX-2 and HIF-1αcharacterize bone metastasis of human breast carcinoma, Breast Cancer Res.Treat. (2011), doi:10.1007/s10549-010-1240.

[11] E. Ridolfi, E. Matteucci, P. Maroni, M.A. Desiderio, Inhibitory effect of HGF oninvasiveness of aggressiveMDA-MB231 breast carcinoma cells, and role of HDACs,Br. J. Cancer 99 (2008) 1323–1634.

[12] A.Müller, B.Homey,H. Soto,N.Ge,D. Catron,M.E. Buchanan, T.Mcclanahan,E.Murphy,W. Yuan, S.N. Wagner, J.L. Barrera, A. Mohar, E. Verástegu, A. Zlotnik, Involvement ofchemokine receptors in breast cancer metastasis, Nature 410 (2001) 50–56.

[13] Y.M. Li, Y. Pan, Y. Wei, X. Cheng, B.P. Zhou, M. Tan, X. Zhou, W. Xia, G.N.Hortobagyi, D. Yu, M.C. Hung, Upregulation of CXCR4 is essential for HER2-mediated tumor metastasis, Cancer Cell 6 (2004) 459–469.

[14] N. Cabioglu, A.A. Sahin, P.Morandi, F.Meric-Bernstam, R. Islam, H.Y. Lin, C.D. Bucana,A.M. Gonzalez-Angulo, G.N. Hortobagyi, M. Cristofanilli, Chemokine receptors inadvanced breast cancer: differential expression in metastatic disease sites withdiagnostic and therapeutic implications, Ann. Oncol. 20 (2009) 1013–1029.

[15] J.M. Busillo, S. Armando, R. Sengupta, O. Meucci, M. Buovier, J.L. Benovic, Site-specific phosphorylation of CXCR4 is dynamically regulated by multiple kinasesand results in differential modulation of CXCR4 signaling, J. Biol. Chem. 285(2010) 7805–7817.

[16] O. Salvucci, A. Bouchard, A. Baccarelli, J. Deschênes, G. Sauter, R. Simon, R. Bianchi,M. Basik, The role of CXCR4 receptor expression in breast cancer: a large tissuemicroarray study, Breast Cancer Res. Treat. 97 (2006) 275–283.

[17] S. Hassan, C. Ferrario, U. Saragovi, L. Quenneville, L. Gaboury, A. Baccarelli, O.Salvucci, M. Basik, The influence of tumor–host interactions in the stromal cell-derived factor-1/CXCR4 ligand/receptor axis in determining metastatic risk inbreast cancer, Am. J. Pathol. 175 (2009) 66–73.

[18] H. Shim, S.K. Lau, S. Devi, Y. Yoon, H.T. Cho, Z. Liang, Lower expression of CXCR4 inlymph node metastasis than in primary breast cancers: potential regulation byligand-dependent degradation of HIF-1α, Biochem. Biophys. Res. Commun. 346(2006) 252–258.

[19] E. Matteucci, M. Locati, M.A. Desiderio, Hepatocyte growth factor enhances CXCR4expression favoringbreast cancer cell invasiveness, Exp. Cell Res. 310 (2005)176–185.

[20] Y. Kang, P.M. Siegel, W. Shu, M. Drobnjak, S.M. Kakonen, C. Cordón-Cardo, T.A.Guise, J. Massagué, A multigenic program mediating breast cancer metastasis tobone, Cancer Cell 3 (2003) 537–549.

[21] M.C.P. Smith, K.E. Luker, J.R. Garbow, J.L. Prior, E. Jackson, D. Piwnica-Worms, G.D.Luker, CXCR4 regulates growth of both primary and metastatic breast cancer,Cancer Res. 64 (2004) 8604–8612.

[22] Z. Liang, Y. Yoon, J. Votaw,M.M. Goodman, L.Williams, H. Shim, Silencing of CXCR4blocks breast cancer metastasis, Cancer Res. 65 (2005) 967–971.

[23] N. Cabioglu, J. Summy, C. Miller, N.U. Parikh, A.A. Sahin, S. Tuzlali, K. Pumiglia, G.E.Gallick, J.E. Price, CXCL-12/stromal cell-derived factor-1alpha transactivatesHER2-neu in breast cancer cells by a novel pathway involving Src kinaseactivation, Cancer Res. 65 (2005) 6493–6497.

[24] J.E. Bolden, M.J. Peart, R.W. Johnstone, Anticancer activities of histone deacetylaseinhibitors, Nat. Rev. Drug Discov. 5 (2006) 769–784.

[25] T.Y. Kim, Y.J. Bang, K.D. Robertson, Histone deacetylase inhibitors for cancertherapy, Epigenetics 1 (2006) 14–23.

[26] C.L. Kostyniuk, S.M. Dehm, D. Batten, K. Bonham, The ubiquitous and tissuespecific promoter of the human SRC gene are repressed by inhibitors of histonedeacetylases, Oncogene 21 (2002) 6340–6347.

[27] P.A. Wade, Transcriptional control at regulatory checkpoints by histonedeacetylases: molecular connections between cancer and chromatin, Hum. Mol.Genet. 10 (2001) 693–698.

[28] E.C. Forsberg, E.H. Bresnick, Histone acetylation beyond promoters: long-rangeacetylation patterns in the chromatin world, Bioessays 23 (2001) 820–830.

[29] Y. Buggy, T.M. Maguire, G. McGreal, E. McDermott, A.D.K. Hill, N. O'Higgins, M.J.Duffy, Overexpression of the Ets-1 transcription factor in human breast cancer, Br.J. Cancer 91 (2004) 1308–1315.

[30] A. Furlan, C. Vercamer, X. Desbiens, A. Pourtier, Ets-1 triggers and orchestrates themalignant phenotype of mammary cancer cells within their matrix environment,J. Cell. Physiol. 215 (2008) 782–793.

[31] M. Gimona, R. Buccione, S.A. Courtneidge, S. Linder, Assembly and biological roleof podosomes and invadopodia, Curr. Opin. Cell Biol. 20 (2008) 235–241.

[32] P.S. Smith-Pearson, E.L. Greuber, G. Yogalingam, A.M. Pendergast, Abl kinase arerequired for invadopodia formation and chemokine-induced invasion, J. Biol.Chem. 285 (2010) 40201–40211.

[33] J. Jovanovic, J.A. Rønnemberg, J. Tost, V. Kristensen, The epigenetic of breastcancer, Mol. Oncol. 4 (2010) 242–254.

1776 P. Bendinelli et al. / Biochimica et Biophysica Acta 1813 (2011) 1767–1776

[34] P. Maroni, P. Bendinelli, E. Matteucci, M.A. Desiderio, HGF induces CXCR4 andCXCL12-mediated tumor invasion through Ets1 and NF-κB, Carcinogenesis 28(2007) 267–279.

[35] P. Bendinelli, E. Matteucci, P. Maroni, M.A. Desiderio, NF-κB activation, dependenton acetylation/deacetylation, contributes to HIF-1 activity and migration of bonemetastatic breast carcinoma cells, Mol. Cancer Res. 7 (2009) 1328–1341.

[36] H. Birkedal-Hansen, Proteolytic remodeling of extracellular matrix, Curr. Opin.Cell Biol. 7 (1995) 728–735.

[37] Y. Guo, P. Pakneshan, J. Gladu, A. Slack, M. Szyf, S.A. Rabbani, Regulation of DNAmethylation in human breast cancer, J. Biol. Chem. 277 (2002) 41571–41579.

[38] K. Wolf, I. Mazo, H. Leung, K. Engelke, U.H. von Adrian, E.I. Deryugina, A.Y.Strongin, E.-B. Bröcker, P. Friedl, Compensation mechanism in tumor cellmigration: mesenchymal–ameboid transition after blocking of pericellularproteolysis, J. Cell Biol. 160 (2003) 267–277.

[39] E.E. Cameron, K.E. Bachman, S. Myöhänen, J.G. Herman, S.B. Baylin, Synergy ofdemethylation and histone deacetylase inhibition in the re-expression of genessilenced in cancer, Nat. Genet. 21 (1999) 103–107.

[40] Z. Liang, T. Wu, H. Lou, X. Yu, R.S. Taichman, S.K. Lau, S. Nie, J. Umbreit, H. Shim,Inhibition of breast cancer metastasis by selective synthetic peptide againstCXCR4, Cancer Res. 64 (2004) 4302–4308.

[41] R. Malik, A. Marchese, Arrestin-2 interacts with the endosomal sorting complexrequired for transport machinery to modulate endosomal sorting of CXCR4, Mol.Biol. Cell 21 (2010) 2529–2541.

[42] I. Berlin, K.M. Higginbotham, R.S. Dise, M.I. Sierra, P.D. Nash, The deubiquitinatingenzyme USP8 promotes trafficking and degradation of the chemokine receptor 4at the sorting endosome, J. Biol. Chem. 285 (2010) 37895–37908.

[43] J. Dittmer, The biology of the Ets1 Proto-Oncogene, Mol. Cancer 2 (2003) 1–21.[44] T.J. Yetman, A renaissance for Src, Nat. Rev. Cancer 4 (2004) 470–480.[45] C.J. Fry, P.J. Farnham, Context-dependent transcriptional regulation, J. Biol. Chem.

274 (1999) 29583–29586.[46] B.M.Woerner, N.M.Warrington, A.L. Kung, A. Perry, J.B. Rubin, Widespread CXCR4

activation in astrocytomas revealed by phospho-CXCR4-specific antibodies,Cancer Res. 65 (2005) 11392–11399.

[47] M. Katoh, M. Katoh, Integrative genomic analyses of CXCR4: transcriptionalregulation of CXCR4 based on TGFβ, Nodal, Activin signaling and POU5F1, FOXA2,

FOXC2, FOXH1, SOX17, and GFI1 transcription factors, Int. J. Oncol. 36 (2010)415–420.

[48] E. Matteucci, P. Bendinelli, M.A. Desiderio, Nuclear localization of active HGFreceptor Met in aggressive MDA-MB231 breast carcinoma cells, Carcinogenesis 30(2009) 937–945.

[49] F. Gobeil, A. Fortier, T. Zhu, M. Bossolasco, M. Leduc, M. Grandbois, N. Heveker, G.Bkaily, S. Chemtob, D. Barbaz, G-protein-coupled receptors signaling at the cellnucleus: an emerging paradigm, Can. J. Physiol. Pharmacol. 84 (2006) 287–297.

[50] E. Matteucci, E. Ridolfi, M.A. Desiderio, Hepatocyte growth factor differentlyinfluences Met-E-cadherin phosphorylation and downstream signaling pathwayin two models of breast cells, Cell. Mol. Life Sci. 63 (2006) 2016–2026.

[51] A. Kumar, K.N. Kremer, D. Dominguez, M. Tadi, K.E. Hedin, Gα13 and Rho mediateendosomal trafficking of CXCR4 into Rab11+ vesicles upon stromal cell-derivedfactor-1 stimulation, J. Immunol. 186 (2011) 951–958.

[52] F. Balkwill, The significance of cancer cell expression of the chemokine receptorCXCR4, Semin. Cancer Biol. 14 (2004) 171–179.

[53] G. Scita, P.P. Di Fiore, The endocytic matrix, Nature 463 (2010) 464–473.[54] L. Wang, Z. Wang, B. Yang, Q. Yang, L. Wang, Y. Sun, CXCR4 nuclear localization

follows binding of its ligand SDF-1 and occurs in metastatic but not primary renalcell carcinoma, Oncol. Rep. 22 (2009) 1333–1339.

[55] M.A. Chetram, V. Odero-Marah, C.V. Hinton, Loss of PTEN permits CXCR4-mediated tumorigenesis through ERK1/2 in prostate cancer cells, Mol. Cancer Res.9 (2011) 90–102.

[56] S. Ropero, M. Esteller, The role of histone deacetylases (HDACs) in human cancer,Mol. Oncol. 1 (2007) 19–25.

[57] M.L. Nelson, H.-S. Kang, G.M. Lee, A.G. Blaszczak, D.K.W. Lau, L.P. Mcintosh, B.J.Graves, Ras signaling requires dynamic properties of Ets1 for phosphorylation-enhanced binding to coactivator CBP, Proc. Natl Acad. Sci. 107 (2010)10026–10031.

[58] B.A. Teicher, S.P. Fricker, CXCL12 (SDF1)/CXCR4 pathway in cancer, Clin. CancerRes. 16 (2010) 2927–2931.

[59] B. Furusato, A. Mohamed, M. Uhlèn, J.S. Rhim, CXCR4 and cancer, Pathol. Int. 60(2010) 497–505.

[60] C. Stresemann, F. Lyko, Modes of action of the DNA methyltransferase inhibitorsazacytidine and decitabine, Int. J. Cancer 123 (2008) 8–13.