epigenome-wide src-1 mediated gene silencing represses ...molecular determinants that drive...

Translational Cancer Mechanisms and Therapy

Epigenome-wide SRC-1–Mediated Gene SilencingRepresses Cellular Differentiation in AdvancedBreast CancerElspeth Ward1, Damir Vares�lija1, Sara Charmsaz1, Ailis Fagan1, Alacoque L. Browne1,Nicola Cosgrove1, Sin�ead Cocchiglia1, Siobhan P. Purcell1, Lance Hudson1, Sudipto Das2,Darran O'Connor2, Philip J. O'Halloran3, Andrew H. Sims4, Arnold D. Hill1, andLeonie S. Young1

Abstract

Purpose:Despite the clinical utility of endocrine therapies forestrogen receptor–positive (ER) breast cancer, up to 40% ofpatients eventually develop resistance, leading to disease pro-gression. Themolecular determinants that drive this adaptationto treatment remain poorly understood. Methylome aberra-tionsdrive cancer growthyet the functional roleandmechanismof these epimutations in drug resistance are poorly elucidated.

Experimental Design: Genome-wide multi-omics sequenc-ing approach identified a differentially methylated hub of pro-differentiation genes in endocrine resistant breast cancerpatients and cell models. Clinical relevance of the functionallyvalidated methyl-targets was assessed in a cohort of endocrine-treated human breast cancers and patient-derived ex vivo met-astatic tumors.

Results: Enhanced global hypermethylation was observedin endocrine treatment resistant cells and patient metastasisrelative to sensitive parent cells and matched primary breast

tumor, respectively. Using paired methylation and transcrip-tional profiles, we found that SRC-1–dependent alterations inendocrine resistance lead to aberrant hypermethylation thatresulted in reduced expression of a set of differentiation genes.Analysis of ER-positive endocrine-treated human breasttumors (n ¼ 669) demonstrated that low expression of thisprodifferentiation gene set significantly associated with poorclinical outcome (P ¼ 0.00009). We demonstrate that thereactivation of these genes in vitro and ex vivo reverses theaggressive phenotype.

Conclusions: Our work demonstrates that SRC-1-depen-dent epigenetic remodeling is a 'high level' regulator of thepoorly differentiated state in ER-positive breast cancer.Collectively these data revealed an epigenetic reprogramingpathway, whereby concerted differential DNA methylationis potentiated by SRC-1 in the endocrine resistant setting.Clin Cancer Res; 24(15); 3692–703. �2018 AACR.

IntroductionBreast cancer develops through the accumulation of genetic and

epigenetic abnormalities to chief regulators of cell proliferation,differentiation, andapoptosis. Estrogen receptor (ER) is a key driverofhormone-dependentbreast cancerand its expression is indicativeof good prognosis. Despite the efficacy of endocrine treatment,including tamoxifen and aromatase inhibitors (AI) in ER-positivebreast cancer, acquired therapy resistance is commonand it remains

a major clinical challenge (1). Mechanisms underlying this resis-tanceare complex,highly adaptive, andheterogeneousandcanvaryfrom patient to patient, from primary tometastatic tissue and evenamongst different endocrine treatments. Recent studies of meta-static tissues frompatients that have failedAIs revealed a number ofmutations, including those activating ESR1 (2), as a feature ofresistance.On theotherhand, lossofER function/expression canbefound in 20% of metastatic tumors highlighting the dynamicnature of therapeutic resistance (3, 4).

Endocrine treatment–resistant cancer cells activate pathwayscooperating and interacting with ER, its coregulators and tran-scription factors providing survival advantage and therapeuticescape. One such ER regulator, SRC-1, has been shown to becentral to the ability of ER tumors to adapt and facilitatemetastaticdisease progression (5, 6). Typically, SRC-1 binds to and coacti-vates nuclear receptors such as ER to regulate a network ofproliferation- anddifferentiation-associated genes critical to breastcancer progression (7). Notably, aberrant upregulation of SRC-1has been implicated in the development of endocrine treatmentresistance in breast cancer, where high protein levels correlate withendocrine resistance and poor clinical outcome (8–10). Modula-tions of these endocrine resistant pathways can be driven bygenomic, epigenetic or tumor microenvironment influences.

Although current emphasis for tumor profiling is onmutation-level alterations, these approaches have failed to uncover the

1Endocrine Oncology Research Group, Department of Surgery, Royal College ofSurgeons in Ireland, Dublin, Ireland. 2Molecular and Cellular Therapeutics, RoyalCollege of Surgeons in Ireland, Dublin, Ireland. 3Department of Neurosurgery,National Neurosurgical Center, Beaumont Hospital, Dublin, Ireland. 4AppliedBioinformatics of Cancer Group, University of Edinburgh Cancer Research UKCentre, MRC Institute of Genetics & Molecular Medicine, Western GeneralHospital, Edinburgh, United Kingdom.

Note: Supplementary data for this article are available at Clinical CancerResearch Online (http://clincancerres.aacrjournals.org/).

E. Ward and D. Vares�lija contributed equally to this article.

Corresponding Author: Leonie S. Young, Endocrine Oncology Research Group,Department of Surgery, Royal College of Surgeons in Ireland, Dublin 2, Ireland.Phone: 353-1402-8576; Fax: 353-1402-8551; E-mail: [email protected]

doi: 10.1158/1078-0432.CCR-17-2615

�2018 American Association for Cancer Research.

ClinicalCancerResearch

Clin Cancer Res; 24(15) August 1, 20183692

on March 28, 2020. © 2018 American Association for Cancer Research. clincancerres.aacrjournals.org Downloaded from

Published OnlineFirst March 22, 2018; DOI: 10.1158/1078-0432.CCR-17-2615

http://crossmark.crossref.org/dialog/?doi=10.1158/1078-0432.CCR-17-2615&domain=pdf&date_stamp=2018-7-13

http://clincancerres.aacrjournals.org/

molecular determinants that drive adaptation to treatment. Con-versely, transcriptional and epigenetic reprograming developswith higher frequency and has been observed to functionallyaffect oncogenes and related signaling pathways (11–13). Increas-ing evidence suggests that aberrant DNA methylation of tumorsuppressors and differentiation/developmental genes may repre-sent a major mechanism underlying tumor progression (11). Thediscovery of hypo and hypermethylation (14–16) in cancer hasled to major advancements toward uncovering novel epigeneticdrivers in tumor initiation and progression. Aberrant DNA hyper-methylation is the most prominent epialteration reported incancer, originating in regions marked with repressive histonemarks (e.g., H3K27me3; ref. 17), guided by DNA-methyl trans-ferases (DNMT) (18) and carefully regulated by transcriptio-nal influencers such as polycomb-repressive complexes (PRC;refs. 17, 19) and methyl-CpG-binding domain proteins (MBD;refs. 20–23). These methylome changes are potentially revers-ible making them prime candidates as novel targets for diagno-sis and treatment strategies. Indeed, altered DNA methylationevents in breast cancer have been used to identify potentialbiomarkers (24–26), while DNA methylation signatures can beused to classify tumor subtypes (27–29) or inform endocrineresponse (30–32). Although, several studies have reportedshifts in the epigenetic profile of endocrine-resistant cell linemodels (30, 33, 34), the role of epigenetic dysregulation inendocrine treatment resistance is still poorly understood, as arethe key potentiators of its function.

In this study, we have investigated the methylome profile ofendocrine-resistant tumors and we report on extensive globalepigenomic remodeling events unique to treatment resistantdisease. Our investigation places SRC-1 in a critical position incontrolling the methylation reprograming in endocrine treat-ment–resistant models and identifies it as a necessary componentin the core regulatory circuitry. SRC-10s role as a transcriptionalrepressor was further validated as SRC-1–dependent events medi-ate aberrant methylation leading to reduced expression of a set ofdifferentiation genes. We demonstrate that SRC-1 is pivotal inrecruiting the corepressor complex to a hub of prodifferentiation

genes, thus remodeling breast cancer cells to promote a moreaggressive endocrine treatment–resistant phenotype. Our datasupport a model where epigenetic reprograming toward a poorlydifferentiated cell profile, driven by an oncogenic coregulator, is acrucial step in endocrine treatment resistance.

Materials and MethodsCell culture

The endocrine sensitiveMCF7 cells (ATCC, USA) were culturedin minimum essential Eagle medium (M2279, Sigma) supple-mented with 10 % FCS (F7524, Sigma). The endocrine-resistantLY2 breast cancer cells were a gift from Robert Clarke and werecultured as described previously (35). Each cell line was tested formycoplasma (LT07-118, Lonza), genotyped (SourceBioScience)and authenticated according to ATCC guidelines. The T347x brainmetastatic primary cell line was derived from an ERþ PR� HER2þ

patient tumor, whichwas expanded inNOD-SCIDmice (36). Thetumor was resected, dissociated, and cultured in human breastepithelial cell (HBEC) media as described previously (36) forin vitro experiments.

CRISPR/Cas9, Lentiviral transduction, siRNA transfectionThe LY2 SRC-1 knockout (KO) cell line was created using

CRISPR/CCas9 technology (Santa Cruz Biotechnology) details ofthe transfection procedure are provided in Supplementary Infor-mation. The LY2 luciferase (LY2-Luc), LY2 shSRC-1 KD (knock-down) and LY2 shNT (nontargeting) cells were created by trans-ducing LY2 cells with viral particles as previously described (37).Gene silencing was carried out using predesigned siRNAs directedagainst NTRK2 (144201, Ambion), non-targeting siRNA control(NT siRNA) (AM4611, Ambion), NR2F2 (J-003422-06-0002),CTDP1 (J-009326-080002), SETBP1 (J-013930-18-0002), POU3F2(J-020029-06-0002) and NT siRNA (J.Human-xx-002) (Dharma-con) and transfection was carried out using Lipofectamine 2000(11668-019, Invitrogen) as per manufacturer's instructions.

Gene expressionRNA extractions were performed using the RNeasy Mini Kit

(74106, Qiagen) as per manufacturer's instructions and Super-Script III (18080400, Invitrogen) was utilized for cDNA conver-sion. Gene expression was confirmed by semi qPCR usingpredesigned TaqMan assays (ThermoFischer Scientific) forb-actin(401846), NTRK2 (Hs00178811), NR2F2 (Hs00819630),CTDP1 (Hs00364467), SETBP1 (Hs01098447), POU3F2(Hs00271595), DNMT1 (Hs0094587) and DMNT3A(Hs01027162) on the StepOnePlus Real Time System (AppliedBiosystems). The comparative Ct (DDCt) method was applied toanalyze relative mRNA expression.

Flow cytometryFACSwas performedon the FACSARIA II (BDBiosciences). The

LY2 CRISPR/Cas9 cells (clone 7c and 9c) were sorted for bothCRISPR HDR plasmid (red fluorescent protein) and luciferase(green fluorescent protein). Stem cell analysis was carried out onthe BD FACSCanto II (BD Biosciences). Cell lines which under-went gene silencing were analyzed by flow cytometry 48 hourspost siRNA transfection with NTRK2, NR2F2, CTDP1, SETBP1,and POU3F2. Cells were stained with CD24 (555428; BD Bio-sciences), CD44 (555478; BD Biosciences), EpCam (12-9326-42,Thermo Fisher Scientific) andCD49f (17-0495-82, Thermo Fisher

Translational Relevance

Aberrant DNA methylation-mediated gene silencing fre-quently occurs in cancer. While substantial effort has beendevoted to the elucidation of methylome changes associatedwith the development of breast cancer, comparatively little isknown about the methylome alterations that accompanytreatment resistance and their contribution to the metastaticphenotype. In this study, we addressed this gap by generating acomprehensive epigenomic map of endocrine treatment resis-tance and identified a key potentiator, its effectors and theirmechanistic and functional output. From this study, we estab-lished a methylation molecular marker set of 5 genes whosesilencing mediated tumor aggressiveness. Subsequently, thesemarkers were confirmed to predict metastatic survival from acohort of endocrine treated breast cancer patients. These novelinsights provide vital clues to the epigenetic basis of on-treatment progression in endocrine-resistant breast cancer andcould advance the management of resistant disease.

SRC-1 Mediates Repressive DNA Methylation in Breast Cancer

www.aacrjournals.org Clin Cancer Res; 24(15) August 1, 2018 3693




Scientific) antibodies. Data were analyzed using FlowJo Software(FlowJo).

Mammosphere forming, anchorage independence, 3D acini,and motility assays

Functional assays were performed in the LY2 luc control cellline, LY2SRC-1CRISPR/Cas9KOcells (clones 7c and9c), and LY2shSRC-1 cells 24 hours post gene silencing with NTRK2, NR2F2,CTDP1, SETBP1, POU3F2. All functional assays were carried outwith cells treated with 4-OHT [10�7 mol/L].

Mammosphere culture and analysis was performed as pre-viously described (38). Anchorage independence was analyzedusing the agarose colony formation assay as previouslydescribed (39).

3D Acini assays were performed to assess cellular polarization/organization (40). Cells were cultured for 21 days, then fixed andstained as previously described (8).

Cell migration was carried out using the Cellomics CellMotility Kit (K0800011, Thermo Fisher Scientific) as previouslydescribed (8).

Chromatin immunoprecipitationChromatin immunoprecipitation (ChIP) was performed on

the LY2 cells, LY2-luc and LY2 CRISPR/Cas9 SRC-1 KO cell line(clone 7c) as previously described (4, 41). Full details can befound in Supplementary Methods.

IHCImmunohistochemistry was performed on 5 mmol/L formalin-

fixed paraffin-embedded (FFPE) tumor sections as previouslydescribed using DAKO envisionþ HRP kit (K400611-2, AgilentTechnologies; ref. 4). Full details on antibodies and protocols canbe found in Supplementary Methods.

Explant studiesAn LY2 luciferase cell line xenograft and patient breast cancer

brain metastatic tumors (T347x and T638x) were expanded inNOD-SCID mice. The primary tumors were resected, grown ongelatin sponges (Spongostan, Johnson and Johnson) as previous-ly described (42) and treated with estrogen combined withvehicle, 4-hydroxytamoxifen (4-OHT),RG108anda combinationof 4-OHT and RG108 for 72 hours. Following treatment, tumorpieces were formalin fixed and paraffin embedded for IHC anal-ysis. LY2 cell line–derived xenograft results shown are a represen-tative of three individual experiments, T347x and T638x PDXresults are individual experiments. The viability of the tumors wasevaluated by screening for necrosis of the tissue and using pro-liferation markers to confirm viable, proliferating cells.

Sequencing acquisitionSeqCap Epi–targeted bisulfitemethylation sequencing (Roche)

was performed on breast cancer cell lines MCF7 (n ¼ 2) and LY2(n ¼ 3) cells and in FFPE breast cancer primary and matchedmetastatic patient tumor samples (RCS_4). DNA was extractedusing the DNA/RNA FFPE extraction kit (80234, Qiagen) as permanufacturer's instructions. Further details are available in Sup-plementary Methods.

MeDIP sequencing was carried out in the LY2 shNT (n¼ 2) andLY2 shSRC-1 knockdown (n¼2) cells following 4-OHT treatmentfor 3 hours. DNA extraction, MeDIP library construction andsequencing (50PE) were all performed by Beijing Genomics

Institute (BGI, Hong Kong) following standard protocols on theIllumina platform.

ChIP sequencing was performed in LY2 cells treated withvehicle or 4-OHT for 45 minutes and immunoprecipitated withSRC-1 antibody, as previously described (43).

RNA sequencing was performed on 4 technical replicates ofLY2 shNT shRNA and LY2 shSRC-1 cells treated with 4-OHT for8 hours. RNA extraction and sequencing was carried out byBeijing Genomics Institute (BGI, Hong Kong) as per standardprotocols (3).

Bioinformatic analysisFull details of the bioinformatic analysis undertaken for each of

the sequencingmethods are available in SupplementaryMethods.

Affymetrix microarray analysisData from four published datasets (GSE66532, GSE9195,

GSE17705, and GSE12093; ref. 44) were utilized to generate theranked gene expression heatmap for the SRC-1 target genes in ERþ

tamoxifen-treated patient tumors (n¼ 669). Data is summarizedwith Ensemble alternative CDF (45) and normalized with RobustMulti-array Average (RMA), before integrationusingComBat (46)to remove dataset-specific bias.

Statistical analysisGene expression, in vitro assays and ki67 scores are shown as

mean � SEM. The Student paired t test was used for two groupcomparisons and results for each assay are representative of threeindividual experiments unless otherwise stated and expressed asmean � SEM, �, P < 0.05; ��, P < 0.01; ��, P < 0.001. Treatmentgroups were compared with vehicle or parental cell line unlessotherwise stated. With respect to randomization, for ex vivoexperiments, similar sized tumors were equally divided into thecontrol and experimental groups for subsequent drug treatmentwhich was not blinded. The investigators were not blinded toallocation for ex vivo and IHC analyses. No statistical method wasused to predetermine sample size. Gene Expression-Based Out-come for Breast Cancer Online (GOBO) was applied to analyzeexpression of the SRC-1 target genes (NTRK2, NR2F2, CTDP1,SETBP1, and POU3F2) in the Pam50 breast cancer tumorsubtype (Basal, ERBB2, Luminal A, Luminal B, and Normal like).Kaplan–Meier plots were used as an estimate of distant meta-static free survival (DMFS) in SRC-1 target genes in ERþ patients(n ¼ 669) and recurrence-free survival (RFS) in untreatedpatients (n ¼ 343; ref. 44).

Data availabilityRNA-seq, MeDIP-seq, and SeqCap Epi targeted bisulfite

sequencing data files were deposited and are available on GeneExpression Omnibus GSE99649. Data from tamoxifen-treatedSRC-1 ChIP-seq in LY2 cells is available on Gene ExpressionOmnibus GSE28987.

EthicsWritten and informed consent was acquired prior to collection

of patient tumor tissue under The Royal College of SurgeonsInstitutional Review Board approved protocol (#13/09:ICORG09/07). Mouse experiments were performed in accordance withthe European Communities Council Directive 2010/63/EU andwere reviewed and approved byResearch Ethics Committee underlicense from theHPRA (Health Products and Research Authority).

Ward et al.

Clin Cancer Res; 24(15) August 1, 2018 Clinical Cancer Research3694




ResultsSRC-1 global genemethylation signature in endocrine resistantbreast cancer

Perturbations in DNA methylation profiles may influencetumor initiation, metastatic progression, and resistance to treat-ment. To investigate global aberrant DNA methylation as afunction of treatment-resistant breast cancer, we undertook tar-geted bisulfite sequencing in endocrine-sensitive and -resistantsamples. Density distribution of events examining differentialmethylation revealed increased hypermethylation in endocrine-resistant cells and an ERþ metastatic patient tumor, relative toparent sensitive cells and matched primary tumor tissue, respec-tively (Fig. 1A). Moreover genome wide CpG methylation wasobserved in themetastatic tumor in comparisonwith thematched

primary tissue (Fig. 1A).Having established the role of differentialDNA methylation in acquired resistance the resulting changes toERbindingwas examined as differential ER binding is observed intumors from patients with poor outcome (47). ER occupancy ofCpG islands in endocrine-resistant LY2 cells is greater than in theendocrine-sensitive MCF7 cells. Furthermore, tamoxifen-drivenER/SRC-1 cooccupancy at these sites suggests a role for the steroidreceptor/coactivator complex in mediating these enhanced DNAhypermethylation events (Fig. 1B). This further supports the roleof SRC-1 in transcriptional silencing (37). To understand thecontribution of SRC-1 to global methylation, comprehensivegenome-wide MeDIP-seq was undertaken in the presence andabsence of SRC-1 (shNT and shSRC-1; Supplementary Fig. S1A) inendocrine-resistant cells. Consistent with its role as a coactivatorprotein, hypermethylation was enriched in the absence of SRC-1

A

r2 = 0.903 r2 = 0.7727

MCF7 Primary breast cancerB

rain

met

asta

sis

LY2

B

< 2KbCpGIsland

MethylationSRC1 Peak

E

n = 251 SRC1ChIP-seq binding event

Filterby

n = 736Genesdown

regulatedwith SRC1

n = 9 Differentially methylated

Differentiation

Biological pathway analysis

NTRK2NR2F2 CTDP1 SETBP1

POU3F2

Development

shNTshSRC-1

D

F

0

20

40

60

80

100 Veh4-OHT

0

5

10

15

20

25

0

200

400

600

800

% E

R o

ccup

ancy

at C

pG s

ites

MCF7 LY2

% E

R/S

RC

-1 o

ccup

ancy

at C

pG s

ites

Global CpG sites

Diff

eren

tially

met

hyla

ted

regi

ons

LY2 shNT LY2 shSRC-10

5

10

15

20

% M

ethy

late

d re

gion

s w

ithin

2 kb

of S

RC

-1 a

t CpG

Isla

nds

LY2 shNT LY2 shSRC-1

Hypomethylation

Hypermethylation

C

Figure 1.

SRC-1 global gene methylation signature. A, Colorimetric density plot of the correlation between the global methylation profile of endocrine sensitive MCF7 (n¼ 2)and endocrine-resistant LY2 (n ¼ 3) cell lines and patient primary breast tumor with a matched brain metastatic tumor (RCS_4) using Roche SeqCap Epitargeted bisulfite methylation sequencing. The correlation plot demonstrates an asymmetric density distribution: LY2 cells and the brain metastatic tumorhave increased hypermethylation relative to the MCF7 cells and matched primary tumor sample, respectively. Circos plot of differentially methylated CpGsdetected in brain metastasis compared with primary tumor. B, Analysis of ER and SRC-1 ChIP-seq in MCF7 (n ¼ 2) and LY2 (n ¼ 2) cells. Bar plot demonstrates agreater percentage of ER binding at CpG islands in the resistant LY2 cells in comparison with the sensitive MCF7 cells. ER/SRC-1 DNA binding shows that 74% of ERpeaks have SRC-1 cobound at CpG islands in LY2 treated cells in comparison to 45% in untreated cells. C, The circos plot demonstrates the distributionof the differentiallymethylated regions of shNT (n¼2) and shSRC-1 (n¼ 2)MeDIP-seq in LY2 cells treatedwith 4-OHT across all chromosomesusingLog2 fold changedifference. A bar plot illustrates the differences in differentially methylated regions in shNT and shSRC-1. D, A higher proportion of MeDIP-seq hypermethylatedregions located adjacent to SRC-1 ChIP-seq peaks (<2 Kb) are observed in LY2 shNT cells in comparison to LY2 shSRC-1 cells. E, Volcano plot illustrating thedifferentially expressed genes between shNT and shSRC-1 RNA-seq from 4-OHT treated LY2 cells (fold change > 1, adjusted p-value <0.05, n ¼ 4). F, The736 genes downregulated in shNT (from RNA-seq analysis, Supplementary Table S1) were filtered and 251 genes were identified within 5 kb upstream of thetranscription start site (ChIP-seq analysis, Supplementary Table S2), from 251 genes, nine genes were found with adjacent differentially methylated regions(MeDIP-seq analysis). The chart illustrates the work flow from which nine genes were identified with known role in development and differentiation.






(Fig. 1C). However, SRC-1–dependent hypermethylationevents were also observed throughout the genome (Fig.1C). Of interest, from SRC-1-ChIP-seq analysis in endo-crine-resistant cells (43), we observed an over-representationof methyl marks at CpG islands within 2 Kb of an SRC-1 peak(Fig. 1D).

The initial methylome data suggests that SRC-1–dependentevents result in an altered methlyome profile and may in factsuppress specific gene sets in the resistant setting. Beforeattempting to dissect the underlying mechanism it was impor-tant to determine the identity of these suppressed genes and ifsuch genes were likely to contribute to the resistant state.RNA-seq identified 736 genes downregulated in the presenceof SRC-1 (Fig. 1E). Correspondence analysis and heatmap dis-played separation between shNT and shSRC-1 gene expression(Supplementary Fig. S1B and S1C). Pathway analysis of theSRC-1 suppressed genes revealed a preponderance of genespertinent to cell development and prodifferentiation (Supple-mentary Fig. S2). This is particularly pertinent as a significantfunction of DNA methylation is in modulating differentiationand developmental pathways (48–51). We integrated datafrom the SRC-1- RNA-seq, ChIP-seq, and MeDIP-seq assays toidentify putative genes that were directly suppressed by SRC-1–dependent DNA methylation (Fig. 1F; Table 1). From these 9genes NTRK2, NR2F2, CTDP1, SETBP1, and POU3F2 havedescribed roles in cellular development and differentiation.ChIP and qPCR analysis confirmed these genes as directSRC-1 targets (Supplementary Fig. S3A and S3B).

Functional role of SRC-1 in disease progression in endocrineresistance is mirrored by the roles of NTRK2, NR2F2, CTDP1,SETBP1, and POU3F2 in tumor suppression

In endocrine treatment–resistant cells, SRC-1 is a knownmediator of drug-resistant phenotype (39, 43). Loss of SRC-1expression can lead to resensitization of endocrine-resistantcell lines to tamoxifen treatment (Supplementary Fig. S4A andS4B). To assess the impact of SRC-1 and its suppressed pro-differentiation target genes on tumor progression, we investi-gated the role of the coregulatory protein and each of theindividual prodifferentiation genes in classic mechanisms oftumor aggression including stemness and migration. Expres-sion levels of each of the 5 target genes are elevated in theabsence of SRC-1 (LY2 shSRC-1) in comparison with theparental resistant cells (LY2 shNT) (Supplementary Fig. S3B).

In endocrine-resistant breast cancer cells, CRISPR/Cas9 geneediting was used to specifically knockout SRC-1 (clone 9 (9c) andclone 7 (7c)) andwas confirmedwithmRNA (Supplementary Fig.S4A) and protein expression of SRC-1 (Supplementary Fig. S4B).No effect on protein expression levels of SRC-2 or SRC-3 wasobserved (Supplementary Fig. S5A). Given relative importance ofSRC-3 in breast cancer drug resistance and metastasis, we furtherconfirmed no upregulation of phosphorylated SRC-3 proteinwhen SRC-1 is suppressed (Supplementary Fig. S5C). SRC-1CRISPR/Cas9 KO cells demonstrated enhanced differentiatedCD24þ/44� cell populations in comparison with their control.siRNA was used in LY2 shSRC-1 cells to transiently silence theSRC-1 prodifferentiation target genes NTRK2, NR2F2, CTDP1,SETBP1, and POU3F2 (Supplementary Fig. S5B). Silencing ofNTRK2 and POU3F2 significantly reduced the number ofCD24þ/44 differentiated cells (Fig. 2A). In these endocrine-resis-tant cells, knockout of SRC-1 resulted in reduced self-renewalcapacity (second-generation mammosphere), colony formation,cell organization, andmigration (Fig. 2B–E). In contrast, silencingof each of the SRC-1 target genes displayed loss of cell differen-tiation through increased self-renewal capacity and colony for-mation along with loss of cellular organization (Fig. 2B-D).Furthermore, silencing of the pro-differentiation genes elevatedmigratory capacity of the endocrine-resistant breast cancer cells(Fig. 2E).

NTRK2, NR2F2, CTDP1, SETBP1, and POU3F2 expression inbreast cancer patients

The clinical relevance of the SRC-1 prodifferentiation geneswasexamined in a cohort of breast cancer patients. We used GOBO toanalyze transcript expressionof the gene set inpublished clinicallyannotated primary tumors (44). The genes stratified according toPAM50 subtype (P < 0.0001) with the highest expression levelsobserved in luminal A and normal-like tumors (Fig. 3A). Gene setexpression also associated with ERþ primary breast cancers (P <0.00001; Fig. 3A). In a cohort of ERþ tamoxifen-treated patients(n ¼ 669), ranked sum of SRC-1 suppression genes transcriptsignificantly associated with reduced distant metastatic disease-free survival (P ¼ 0.00009; Fig. 3B and C). The association ofdysregulated gene set with good outcome strongly aligns with itsprodifferentiation role and suggests that its suppression can bedetected in a relatively large subset of human breast cancer andcould contribute to risk assessment for endocrine treatmentresistance. This relationship would appear to be treatment

Table 1. Pro-differentiation genes repressed by SRC-1–dependent DNA methylation

Gene Ensemble ID Gene function Methyl marks

NTRK2 ENSG00000148053 Regulation of neuron survival, proliferation, migrationand differentiation

Intronic

NR2F2 ENSG00000185551 Nuclear receptor involved in neuronal differentiation Downstream (4 kb) of TSS andupstream (380 kb)

CTDP1 ENSG00000060069 Regulates RNA polymerase II, cellular organizationand differentiation

Intronic near TSS/intergenic

SETBP1 ENSG00000152217 DNA replication, differentiation IntronicPOU3F2 ENSG00000184486 Differentiation 2.5 kb upstream of promoterMMP16 ENSG00000156103 Involved in the breakdown of extracellular matrix in normal

physiological processesIntergenic, intronic and exon

NELL2 ENSG00000184613 Cell growth regulation IntronicRASD1 ENSG00000108551 Negatively regulates the transcription regulation activity of the

APBB1/FE65-APP complex via its interaction with APBB1/FE65Intergenic

SDK1 ENSG00000146555 Adhesion molecule that promotes lamina-specific synapticconnections in the retina

Intronic and exon

Ward et al.





dependent as the reverse relationship was observed in ERþ

untreated patients (n ¼ 343) where high transcript expression ofthe gene set associated with reduced recurrence-free survival (P <0.05; Fig. 3C; Supplementary Fig. S6).

To enhance the translational value of our findings and furtherreinforce the roleof aberrantDNAmethylation in this dysregulatedpathway, we employed an ex vivo model of endocrine-resistantmetastatic tumors to evaluate the effect of DNA methylationdisruption (Fig. 3D). These models recapitulate tissue heterogene-ity,morphology, and architecture and create a unique opportunityfor drug efficacy studies and provide a platform for mechanisticstudies. Given our findings that SRC-1 target genes were regulatedthrough inappropriate methylation-dependent silencing, we firstused a DNA methyltransferase inhibitor, RG108, to confirm itscapacity to reexpress the SRC-1 target genes in endocrine-resistantLY2 cells and those derived from endocrine-resistant T347 meta-static tumor (Fig. 3D; Supplementary Fig. S7A). Silencing of theSRC-1 prodifferentiation target genes NTRK2, NR2F2, CTDP1,SETBP1, and POU3F2 resulted in an increase in proliferativecapacity of LY2 shSRC-1 model (Supplementary Fig. S7B). Inaddition, in endocrine-resistant metastatic- cell line–derived froma xenograft tumor (LY2) and endocrine-resistantmetastatic tumors(T347x and T638x, Supplementary Fig. S7C), cultured ex vivo,RG108 treatment over 72 hours had a substantial antitumor effectas demonstrated by a significant decrease in proliferating cells(ki67þ) compared with vehicle-treated tumors (Fig. 3E and F).This proof-of-concept study further supports methylation as anecessary and reversible mechanism promoting tumorigenesis inmultiple models of endocrine treatment resistance.

SRC-1 in combination with a complex of methylators repressesexpression of NTRK2, NR2F2, CTDP1, SETBP1 and POU3F2

We wanted to further delineate the mechanistic pathwayinvolved. We interrogated the DNA methylation of the SRC-1prodifferentiation genes by analyzing methyl sites from Seq CapEpi data from endocrine-resistant LY2 cells and matched primaryand metastatic tumor from an endocrine-resistant patient. Thepercentage methylation indicating the proportion of cytosine'smethylated at each CpG probe is reported for the LY2 cells andmetastatic brain tissue (Fig. 4A). Differential methylated regionswere analyzed from the resistant metastatic tumor and thematched primary. Hypermethylated regions were identifiedupstream of NTRK2, CTDP1, SETBP1, and POU3F2 and a hypo-methylated region upstream of NR2F2 (Fig. 4A). These regionscorresponded to SRC-1peaks as determined fromglobalChIP-seqanalyses in LY2 cells (Fig. 4A).

The expression of the maintenance and de novo methyltrans-ferases (52),DNMT1 andDMMT3a, respectively were found to bereduced in the absence of SRC-1 (LY2 shSRC-1) compared withthe control (LY2 shNT; Fig. 4B). As DNA and histone lysinemethylation systems are highly interrelated and rely mechanisti-cally on each other, we investigated histone methylation at theSRC-1 target genes. Consistent with this, elevated recruitment ofthe histone repression marker H3K27me3, a known mediator ofde novo DNA hypermethylation, to the prodifferentiation geneswas observed (Fig. 4C).

The network of nuclear receptors and coregulator proteins ishighly complex, interconnected, and regulates many transcrip-tionally active regions in the cistrome that are coordinatelyoccupied by multiple nuclear receptors including ER, PR, and AR(53, 54). To further dissect the nuclear receptors' contribution to

the SRC-1 regulated processes, we investigated ER, AR, and PRbinding to the target genes. Enhanced recruitment of ER and itscoactivator protein SRC-1 were observed at each of the prodiffer-entiation genes (Fig. 4D). Interestingly, we found an enhanced,but nonsignificant, recruitment of PR to most of the target genes,but not AR (Fig. 4D; Supplementary Fig. S8A). Treatment withantagonists against ER and PR produced no reduction in therecruitment of SRC-1 to each of the prodifferentiation genes(Supplementary Fig. S8B). Consistent with the specificity ofSRC-10s contribution in this process, no binding of SRC-3 wasdetected at any of the target genes (Supplementary Fig. S8C).

To unravel the complex that may regulate methylation at theseSRC-1 target genes we examined the recruitment ofMBD proteinsto the prodifferentiation target genes. MBD proteins bind meth-ylated DNA and are believed to participate in DNAmethylation–mediated transcriptional repression (55). DNA binding ofMECP2 and MBD2, two MBD family members, to each of thetarget genes was confirmed by ChIP and additionally with ChIP-re-ChIP qPCR indicating cooccupancy (Fig. 4E and F). To deter-mine whether SRC-1 is essential for the recruitment of the meth-ylation-regulatorymodule,we examined recruitment of themeth-yl proteins to the prodifferentiation genes in the absence of SRC-1using the LY2 SRC-1 CRISPR/Cas9 7c KO. Loss of MBD protein-DNA binding at each of the targets was observed in these cells incomparison with the luciferase control endocrine-resistant cells(Fig. 4G).Moreover, SRC-1–dependent recruitment of the histonedeacetylase protein, HDAC2, a known complex partner of theMBD methyl proteins, was also observed in each of the targetgenes, inwhich loss of recruitmentwas again observed after SRC-1KO (Fig. 4I). Interestingly, we detected no such consistent signif-icant binding of the methyl complex to target genes in endocrine-sensitiveMCF7 cells (Supplementary Fig. S8D–S8G). This findingis in line with our previous reports of a reduction of SRC-1-DNAbinding in steroid depleted endocrine-sensitive breast cancer cellsin comparison with the endocrine-resistant phenotype (37).Variable occupancy of the methyl complex and subsequent lossfollowing SRC-1 KO suggests a central role for the transcriptionalregulatory protein in themanagement of themethylome. Togeth-er, these data suggest that SRC-1 plays a regulatory role inorchestrating the operationalmethyl complex at theDNA to bringabout successful functional repression of prodifferentiation targetgenes to enable tumor progression in the context of endocrineresistance (Fig. 4I).

DiscussionA growing body of evidence suggests that breast cancer cells can

develop resistance to endocrine therapy, not only through clonalselection of preexisting progenitor/stem cell like populations andgenetic mutations, but also via aberrant epigenetic regulation ofgene expression. Altered DNA methylation during early carcino-genesis has been associated with dysregulation of key transcrip-tional regulators including p53 (56). Further epigenetic remodel-ing occurs with disease progression and treatment resistance (57).Aberrant methylation has been associated with activation ofcholesterol biosynthesis (34) and decreased gene expression ofclassic ER targets in endocrine-resistant cell line models (32).Consistent with these studies, we observed differential methyla-tion patterns between sensitive and treatment-resistant breastcancer cell lines and patient tumors and uncovered extensivehypermethylation and hypomethylation events unique to






A

B

D

E

C

siNT siNTRK2 siNR2F2

siCTDP1 siSETBP1 siPOU3F2

LY2 shSRC-1 cells

LY2 shSRC-1 cells

LY2 shSRC-1 cells

LY2 shSRC-1 cells

LY2 shSRC-1 cells

LY2 luc CRISPR 9C CRISPR 7C0.0 0

1

2

3

4

0

0.0

0.5

1.0

1.5

1

2

3

4

0.5

1.0

1.5

****

Mam

mos

pher

e fo

rmin

g ef

ficie

ncy

2nd

Gen

(>50

µm

ol/L

)

Mam

mos

pher

e fo

rmin

g ef

ficie

ncy

2nd

Gen

(>50

mm

ol/L

)

LY2 luc CRISPR 9C CRISPR 7C0

200

400

600

800

* *

Ave

rage

col

ony

form

atio

n (>

50 µ

mol

/L)

Ave

rage

col

ony

form

atio

n (>

50 m

mol

/L)

LY2 luc CRISPR 9C CRISPR 7C0.0

0.5

1.0

1.5

******

Mig

ratio

n ar

ea (µ

mol

/L)

0.0

0.5

1.0

1.5

2.0

2.5

Mig

ratio

n ar

ea (m

mol

/L)

LY2 luc CRISPR 9C CRISPR 7C0

2

4

6

8

10

*****

Rel

ativ

e or

gani

zed

acin

ifo

rmin

g ab

ility

Rel

ativ

e or

gani

zed

acin

ifo

rmin

g ab

ility

LY2 luc CRISPR 9C CRISPR 7C0.0

0.5

1.0

1.5

0.0

0.5

1.0

1.52.0

2.5

*R

elat

ive

CD

24+ C

D44

- pop

ulat

ion

Rel

ativ

e C

D24

+ CD

44-

pop

ulat

ion

LY2 P89C

LY2 P87C

LY2

siNTsiNT siNTRK2 siNR2F2 siCTDP1 siSETBP1 siPOU3F2siNTRK2 siNR2F2 siCTDP1 siSETBP1 siPOU3F2

siNT siNTRK2 siNR2F2 siCTDP1 siSETBP1 siPOU3F2




0.0

0.5

1.0

1.5

2.0

**

***

***

Rel

ativ

e C

D49

f-Epc

am+

popu

latio

n

LY2 shSRC-1 cells

Ward et al.





A

C

B

D

ER- ER+

-4-2

02

46

8su

m5

Basal ERBB2 Lum A Lum B Normal-like

sum

5-4

-20

24

68

Low (15)

0 5 10 15

0.0

0.2

0.4

0.6

0.8

1.0 Untreated ER+ population

Time (years)

RFS

High (328)

P = 9E-5

High (609)

(Low (60)

0 5 10 150.0

0.2

0.4

0.6

0.8

1.0

Tamoxifen trx ER+ population

Time (years)

DM

FS

0 Hrs 72 Hrs

Resected tumorfragment is cultured

on 3D matrix

Drug Tx Tx. tumor is fixed and embedded

for IHC

E2+DMSO

E2+RG-108

E2+4-OHT

E2+RG-108+4-OHT

SRC-1

NR2F2SETBP1

NTRK2CTDP1

POU3F2

Epigenetically silencedgenes

Poor outcome Good outcome

Re-activation viamethylation inhibitor

RG-108

(60) (609)

NR2F2

POU3F2CTDP1

SETBP1NTRK2

P cut-offP < 0.05

E

DMSO

Tumor proliferation response (Ki67)

4-OHT RG-108 4-OHT+RG-108

DMSO

4-OHT

RG-108

4-OHT+RG-108

Rel

ativ

e ki

67 p

ositi

vity

FLY2

Endocrine-resistantHigh metastatic potential

DMSO 4-OHT RG-108 4-OHT+RG-108

DMSO

4-OHT

RG-108

4-OHT+RG-108

Rel

ativ

e ki

67 p

ositi

vity

DMSO


4-OHT RG-108

DMSO

4-OHT

RG-108

4-OHT+RG-108

Rel

ativ

e ki

67 p

ositi

vity

x836Tx743T

Endocrine-resistantMetastatic tumor

Endocrine-resistantMetastatic tumor

4-OHT+RG-108


Figure 3.

NTRK2, NR2F2, CTDP1, SETBP1, and POU3F2 expression in breast cancer patients. A, GOBO analysis of combined expression of NTRK2, NR2F2, CTDP1, SETBP1, andPOU3F2, in PAM50 subtypes of breast cancer tumors, showed significantly higher expression in Luminal A subtype, (P < 0.00001). Analysis of combinedexpression of NTRK2, NR2F2, CTDP1, SETBP1, and POU3F2 stratified by estrogen receptor status in all tumors, showed significantly higher expression of theseprodifferentiation genes in the ER positive tumors compared to ER negative tumors, (P < 0.00001). B, Ranked SRC-1 target gene set expression in 669 primarybreast tumors from ER positive 4-OHT-treated patients (44). Colors are log2 mean-centered values (Red ¼ high, Green ¼ low). Data is from four publishedAffymetrix microarray datasets (GSE6532, GSE9195, GSE17705, GSE12093) summarized with Ensembl alternative CDF (45) and normalized with Robust Multi-arrayAverage (RMA), before integration using ComBat (46) to remove dataset-specific bias. White–gray–black bars indicate significance of all possible cut-offpoints fromP¼ 1 to 0.001.C,Kaplan–Meier analysis of distantmetastatic free survival (DMFS) according to expression of the SRC-1 pro-differentiation target genes inERþ 4-OHT–treated patients (n ¼ 669) and Kaplan–Meier analysis of recurrence-free survival (RFS) in untreated patients (n ¼ 343). D, Schematic ofthe ex vivo experimental set up. E, Tumors extracted from LY2 xenografts were assessed for proliferation by IHC analysis of Ki67 quantified using the Aperio digitalpathology imaging, shows significantly less Ki67 in RG108 and RG108/4-OHT–treated groups compared with DMSO control, (n ¼ 10 images/group). F,Patient breast cancer brain metastatic tumor explants (T347x, T638x) were assessed for Ki67 expression and quantified using the Aperio digital pathology imaging.The results show significantly less Ki67 positivity in the RG-108 treated group and in RG108/4-OHT–treated groups in both T347x and T638x tumor explantcompared with DMSO control (n ¼ 10 images/group). Ki67 positive cells indicated with red triangles and negative cells indicated with a green triangle.Results are expressed as mean � SEM (� , P < 0.05; �� , P < 0.01; �� , P < 0.001).

Figure 2.SRC-1 tumorigenic potential, mirrored by NTRK2, NR2F2, CTDP1, SETBP1, and POU3F2 functional role in tumorigenic suppression. A, In SRC-1 CRISPR/Cas9 KO cells(clone 9c and 7c) the CD24þCD44� differentiated cell population is significantly increased compared with control endocrine resistant LY2 luc cell lines(n ¼ 3). In contrast CD24þCD44� and CD49f�Epcamþdifferentiated cell population is significantly decreased after siRNA knock down of NTRK2 and POU3F2 andNTRK2, NR2F2, CTDP1 and POU3F2, respectively, in LY2 shSRC-1 cells (n ¼ 3). B, LY2 SRC-1 CRISPR/Cas9 knock out clones 9c and 7c have significantly lessmammosphere forming efficiency compared to control LY2 luc cell line (n ¼ 3), while the siRNA knock down of NTRK2, NR2F2, CTDP1, SETBP1 and POU3F2 in LY2shSRC-1 cells significantly increases the cell lines mammosphere forming efficiency compared to LY2 shSRC-1 control (n ¼ 3). C, Anchorage independentgrowth in LY2 cell lines was significantly reduced in the absence of SRC-1 (LY2 SRC-1 CRISPR/Cas9 KO 9c and 7c) (n ¼ 3). Anchorage-independent growth wassignificantly increased following siRNA knockdown of all prodifferentiation genes, NTRK2, NR2F2, CTDP1, SETBP1, and POU3F2, in LY2 shSRC-1 cells compared withscramble control (n¼ 3). D, Bar plot and representative images of acini formation from LY2 SRC-1 CRISPR/Cas9 KO (clone 9c and 7c) showed more organized aciniwith superior apico-basolateral structure compared to wild-type LY2 cells (n ¼ 3). In contrast, siRNA knock down of the prodifferentiation genes, NTRK2, NR2F2,CTDP1, SETBP1, and POU3F2, showed decreased level of organization relative to the SRC-1 control cell line (LY2 shSRC-1) (n ¼ 3). Phalloidin 594 (pink color) stainsF-actin and DAPI (blue color) stains the nucleus. Organized acini structures were defined based upon presence of hollow lumen and structured apico-basolaterallayer. E, Scratch assay showed that LY2 SRC-1 CRISPR/Cas9 KO (clone 9c and 7c) were significantly less motile in comparison with wild-type LY2 luc cells (n ¼ 3).KnockdownofNTRK2,NR2F2,CTDP1,SETBP1 andPOU3F2 in the absence of SRC-1 (LY2 shSRC-1)migratory capacity of the cells comparedwith siNT cells (fluorescentbead assay, n ¼ 3). Results are expressed as mean � SEM, � , P < 0.05; �� , P < 0.01; �� , P < 0.001.







Ward et al.




treatment-resistant disease. To date the mechanism of commu-nication between the key transcriptional mechanics of the resis-tant cell with the methylation process to drive the phenotype andpromote disease progression has not been elucidated. We exam-ined the global differential methylation observed between endo-crine-sensitive and -resistantmodels and patient tumors, defininga role for the ER coregulatory protein SRC-1 in mediating generepression, which is both functional and clinically relevant.

Nuclear receptor gene repression is regulated, at least in part,through interactions with coregulatory proteins. The glucocorti-coid receptor can utilize SRC-2 to activate and repress target geneexpression depending on the transcriptional target (58, 59). Morerecently, an amphipathic role for the ER coregulator protein SRC-1has also been described, where SRC-1 the classic steroid receptorcoactivator protein, has been shown to transcriptionally repressthe differentiationmarker CD24 and the apoptotic protein PAWR(37). In this study, employing integrated multi-omics approach,we found specific global SRC-1–dependent hypermethylation,corresponding to regions of transcriptional repression, whichwere confirmed as direct SRC-1 targets. Although the complexinterplay between nuclear receptors and coregulatory proteins isknown to play a significant role in the development of breastcancer, data reported here suggest that SRC-1 repression of thesetarget genes is not dependent on multiple nuclear receptor inter-actions. Importantly, pathway analysis revealed an over represen-tation of these genes in development and differentiation process-es, suppression of which are essential for tumor development.

In this study, we defined five genes with described roles incellular development and differentiation that are direct suppres-sion targets of SRC-1. Aberrant methylation of these genes inendocrine resistant models and in patient tumors was observed.Increased expression of the maintenance and de novo methyl-transferases, DNMT1 and DMMT3a, were seen in the presence ofSRC-1.Moreover, thepresence of the lysinemethylator,H3k27meis consistent with the established mechanistic link between DNAand histone methylation (17) and is indicative of the epigeneticactivity at these regions.

To elucidate the mechanism of repression and the regulatorylink betweenmethylation and the steroid receptor transcriptionalsystem we investigated the methylome pertinent to the targetgenes. MBD2 and MECP2, members of the methyl bindingdomain (MBD) protein family which deciphers the DNA meth-

ylation code (60), were both found to be recruited to the targetgenes. Full suppression capacity of the MBD protein complex isdependent, however, on histone deacetylation (20). Recruitmentto the target genes of the histone deacetylator, HDAC2, a knownMECP2 binding partner (20), was also observed. The dependenceof this regulatory methylome on SRC-1 provides evidence of thecentral role of this transcriptional protein in coordinating theseepigenetic events.

Derepression of the SRC-1 episilenced genes influences a num-ber of key functional pathways whose deregulation is a facet ofendocrine treatment–resistant phenotype. Integration with exist-ing patient datasets and patient survival data (44) revealed thatreduced expression of this gene set associated with poor outcomein tamoxifen-treated population. This was not true for treatment-na€�ve populations suggesting that this is a feature of long-termendocrine treatment. Therefore, SRC-1–dependent methylatedgenes identified here underline key molecular features that dis-tinguish between good outcome and poor outcome in endocrinetreated ERþ patients. Still, these interpretations warrant furtherclinical investigation in larger independent cohorts as methyla-tion of specific genes have the power to be a valuable tool in themanagement of breast cancer (61, 62).

In contrast to mutational modifications, epigenetic alterationsare potentially reversible (63–65). Demethylating agents havedemonstrated therapeutic benefit at low dose long-term treat-ments in solid tumors (66).However, these agents can have broadimpacts on gene expression and also the number of tumor-associated methylated genes could impact its efficacy. To dem-onstrate methylation as a crucial mechanism of the aggressivephenotype observed in our models, we show that DNA demethy-lator treatment reexpressed SRC-1 suppression genes and signif-icantly inhibited proliferation of ex vivo endocrine-resistant met-astatic tumors. Promising observations reported here, warrantfurther studies using a larger cohort of patient tumor samples,providing full clinical relevance of these mechanisms in breastcancer patients.

Taken together data presented here link, for the first time, thekey transcriptional machinery of the endocrine resistant cell withglobal methyl-dependent gene suppression. We demonstrate thatSRC-1 is one of the key orchestrators of the endocrine resistantmethylome, which has consequences that are both functionallyand clinically relevant.

Figure 4.SRC-1 in combination with a complex of methylators drives repressive state of NTRK2, NR2F2, CTDP1, SETBP1 and POU3F2. A, Differentially methylated regions(DMR) of SRC-1 prodifferentiation genes were identified with SeqCap Epi sequencing by comparing primary breast tumor with matched brain metastatictumor (case RCS_4). Plot shows regions of hypermethylation (red) and hypomethylation (blue; meth.diff 30%; q-value < 0.01) found in NTRK2, NR2F2, CTDP1,SETBP1, and POU3F2 genes. Tracks show: CpGs probed (purple), % methylation in LY2 cell line (green), differential methylation in brain metastaticpatient over primary (case RCS_4) (gray), % methylation in case RCS_4 prior to differential analysis (light gray), SRC-1 ChIP-seq peaks in 4-OHT treated LY2 cells(yellow), SRC-1 ChIP-seq peaks in untreated LY2 cells (orange), and RefSeq HG19 gene model. B, mRNA expression levels of DNA methyl transferasesin the presence of SRC-1 (LY2 shNT) compared with its absence (shSRC-1). Expression of de novo transferases DNMT1 and DNMT3A are significantly increased in thepresence of SRC-1 (shNT) in comparison with silenced (shSRC-1) cells (n ¼ 3). C, ChIP assay showed significantly higher recruitment of histone repressionmarker H3K27me3 to prodifferentiation genes in 4-OHT–treated LY2 cells over IgG. D, ChIP assays showed significantly higher recruitment of transcriptionregulators, SRC-1 and ER over IgG. ChIP assay of PR recruitment to the target genes is included. E, Recruitment of methylators, MBD2 and MECP2, to NTRK2,NR2F2,CTDP1, SETBP1, and POU3F2, in 4-OHT–treated LY2 cells over IgG (n ¼ 3). F, ChIP-re-ChIP assay of SRC-1-MBD2 and SRC-1 MECP2 occupancy over SRC-1-IgGcontrol at each of the target genes. G, MBD2 and MECP2 recruitment to prodifferentiation genes in LY2 SRC-1 CRISPR/Cas9 KO (clone 7c) relative to LY2-lucparental cell line (n ¼ 3). H, ChIP assays show significantly higher recruitment of HDAC2 to NTRK2, NR2F2, CTDP1, and SETBP1 in LY2 cells over IgG,which is significantly reduced in LY2 SRC-1 CRISPR/Cas9 KO (clone 7c) cells (n ¼ 3) compared with LY2 luc in NR2F2, CTDP1, SETBP1, and POU3F2. I, Heatmapdemonstrating relative DNA recruitment of ER, SRC-1, and methyl proteins to the SRC-1 prodifferentiation genes in LY2 cells. Cartoon illustrating complexrecruitment of regulatory proteins driving methylation and subsequent repression to SRC-1 prodifferentiation genes. Results are expressed as mean � SEM(� , P < 0.05; �� , P < 0.01; �� , P < 0.001).






Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: E. Ward, D. Vareslija, S. Charmsaz, L.S. YoungDevelopment ofmethodology:E.Ward,D. Vareslija, S. Charmsaz, A.L. Browne,L.S. YoungAcquisition of data (provided animals, acquired and managed patients, pro-vided facilities, etc.): E. Ward, D. Vareslija, A.L. Browne, S. Cocchiglia,S.P. Purcell, L.Hudson, S.Das,D.O'Connor, P.J.O'Halloran, A.D.Hill, L.S. YoungAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): E. Ward, D. Vareslija, S. Charmsaz, A. Fagan,N.S. Cosgrove, A.H. Sims, L.S. YoungWriting, review, and/or revision of the manuscript: E. Ward, D. Vareslija,S. Charmsaz, S. Das, D. O'Connor, P.J. O'Halloran, L.S. Young

Administrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): E. Ward, D. Vareslija, S. Cocchiglia, S.P. Purcell,L. Hudson, A.D. HillStudy supervision: E. Ward, D. Vareslija, D. O'Connor, A.D. Hill, L.S. Young

AcknowledgmentsWe kindly acknowledge the funding support from Irish Cancer Society

Collaborative Cancer Research Centre grant, CCRC13GAL, Breast CancerResearch Foundation, Science Foundation Ireland and Breast Cancer Ireland.

The costs of publication of this articlewere defrayed inpart by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received September 8, 2017; revised February 12, 2018; accepted March 16,2018; published first March 22, 2018.

References1. Clarke R, Tyson JJ, Dixon JM. Endocrine resistance in breast cancer–An

overview and update. Mol Cell Endocrinol 2015;418 Pt 3:220–34.2. Jeselsohn R, Buchwalter G, De Angelis C, Brown M, Schiff R. ESR1 muta-

tions as a mechanism for acquired endocrine resistance in breast cancer.Nat Rev Clin Oncol 2015;12:573–83.

3. McBryan J, Fagan A, McCartan D, Bane FT, Vareslija D, Cocchiglia S, et al.Transcriptomic profiling of sequential tumors from breast cancer patientsprovides a global view of metastatic expression changes following endo-crine therapy. Clin Cancer Res 2015;21:5371–9.

4. Vareslija D, McBryan J, Fagan A, Redmond AM, Hao Y, Sims AH, et al.Adaptation to AI therapy in breast cancer can induce dynamic alterations inER activity resulting in estrogen-independent metastatic tumors. ClinCancer Res 2016;22:2765–77.

5. Qin L, Liu Z, ChenH, Xu J. The steroid receptor coactivator-1 regulates twistexpression and promotes breast cancer metastasis. Cancer Res 2009;69:3819–27.

6. Xu J, Qiu Y, DeMayo FJ, Tsai SY, Tsai MJ, O'Malley BW. Partial hormoneresistance in mice with disruption of the steroid receptor coactivator-1(SRC-1) gene. Science 1998;279:1922–5.

7. Xu J, Wu RC, O'Malley BW. Normal and cancer-related functions of thep160 steroid receptor co-activator (SRC) family. Nat Rev Cancer 2009;9:615–30.

8. McBryan J, Theissen SM, Byrne C, Hughes E, Cocchiglia S, Sande S, et al.Metastatic progression with resistance to aromatase inhibitors is driven bythe steroid receptor coactivator SRC-1. Cancer Res 2012;72:548–59.

9. Redmond AM, Bane FT, Stafford AT,McIlroyM,DillonMF, Crotty TB, et al.Coassociation of estrogen receptor and p160 proteins predicts resistance toendocrine treatment; SRC-1 is an independent predictor of breast cancerrecurrence. Clin Cancer Res 2009;15:2098–106.

10. Al-azawi D, Ilroy MM, Kelly G, Redmond AM, Bane FT, Cocchiglia S, et al.Ets-2 andp160proteins collaborate to regulate c-Myc in endocrine resistantbreast cancer. Oncogene 2008;27:3021–31.

11. Zhao SS, Geybels MS, Leonardson A, Rubicz R, Kolb S, Yan QX, et al.Epigenome-wide tumor DNA methylation profiling identifies novel prog-nostic biomarkers of metastatic-lethal progression in men diagnosed withclinically localized prostate cancer. Clin Cancer Res 2017;23:311–9.

12. Visvanathan K, FacklerMS, Zhang Z, Lopez-Bujanda ZA, Jeter SC, Sokoll LJ,et al. Monitoring of serum DNA methylation as an early independentmarker of response and survival in metastatic breast cancer: TBCRC 005Prospective Biomarker Study. J Clin Oncol 2017;35:751-þ.

13. Priedigkeit N,Hartmaier RJ, ChenY, VareslijaD, BasudanA,Watters RJ, et al.Intrinsic subtype switching and acquired ERBB2/HER2 amplifications andmutations in breast cancer brain metastases. JAMA Oncol 2016;3:666–71.

14. Eden A, Gaudet F, Waghmare A, Jaenisch R. Chromosomal instability andtumors promoted by DNA hypomethylation. Science 2003;300:455.

15. Gaudet F, Hodgson JG, Eden A, Jackson-Grusby L, Dausman J, Gray JW,et al. Induction of tumors in mice by genomic hypomethylation. Science2003;300:489–92.

16. Ting AH, McGarvey KM, Baylin SB. The cancer epigenome–componentsand functional correlates. Genes Dev 2006;20:3215–31.

17. Cedar H, Bergman Y. Linking DNAmethylation and histonemodification:patterns and paradigms. Nat Rev Genet 2009;10:295–304.

18. Rountree MR, Bachman KE, Baylin SB. DNMT1 binds HDAC2 and a newco-repressor, DMAP1, to form a complex at replication foci. Nat Genet2000;25:269–77.

19. Schlesinger Y, Straussman R, Keshet I, Farkash S, Hecht M, Zimmerman J,et al. Polycomb-mediated methylation on Lys27 of histone H3 pre-marksgenes for de novo methylation in cancer. Nat Genet 2007;39:232–6.

20. Nan X, Ng HH, Johnson CA, Laherty CD, Turner BM, Eisenman RN, et al.Transcriptional repression by the methyl-CpG-binding protein MeCP2involves a histone deacetylase complex. Nature 1998;393:386–9.

21. Liu MY, DeNizio JE, Schutsky EK, Kohli RM. The expanding scope andimpact of epigenetic cytosine modifications. Curr Opin Chem Biol 2016;33:67–73.

22. Mellen M, Ayata P, Dewell S, Kriaucionis S, Heintz N. MeCP2 binds to5hmC enriched within active genes and accessible chromatin in thenervous system. Cell 2012;151:1417–30.

23. Bogdanovic O, Veenstra GJ. DNA methylation and methyl-CpG bindingproteins: developmental requirements and function. Chromosoma2009;118:549–65.

24. ChimonidouM, Tzitzira A, Strati A, Sotiropoulou G, Sfikas C, Malamos N,et al. CST6 promoter methylation in circulating cell-free DNA of breastcancer patients. Clin Biochem 2013;46:235–40.

25. Szyf M. DNA methylation signatures for breast cancer classification andprognosis. Genome Med 2012;4:26.

26. Xu Z, Bolick SC, DeRoo LA, Weinberg CR, Sandler DP, Taylor JA. Epigen-ome-wide association study of breast cancer using prospectively collectedsister study samples. J Natl Cancer Inst 2013;105:694–700.

27. Fackler MJ, Umbricht CB, Williams D, Argani P, Cruz LA, Merino VF, et al.Genome-wide methylation analysis identifies genes specific to breastcancer hormone receptor status and risk of recurrence. Cancer Res 2011;71:6195–207.

28. Fang F, Turcan S, Rimner A, Kaufman A, Giri D, Morris LG, et al. Breastcancermethylomes establish an epigenomic foundation for metastasis. SciTransl Med 2011;3:75ra25.

29. Holm K, Hegardt C, Staaf J, Vallon-Christersson J, Jonsson G, Olsson H,et al. Molecular subtypes of breast cancer are associated with characteristicDNA methylation patterns. Breast Cancer Res 2010;12:R36.

30. FanM, Yan PS,Hartman-Frey C, Chen L, PaikH,Oyer SL, et al. Diverse geneexpression and DNAmethylation profiles correlate with differential adap-tation of breast cancer cells to the antiestrogens tamoxifen and fulvestrant.Cancer Res 2006;66:11954–66.

31. Martinez-Galan J, Torres-Torres B, Nunez MI, Lopez-Penalver J, Del MoralR, Ruiz De Almodovar JM, et al. ESR1 gene promoter regionmethylation infree circulating DNA and its correlation with estrogen receptor proteinexpression in tumor tissue in breast cancer patients. BMC Cancer 2014;14:59.

32. Stone A, Zotenko E, Locke WJ, Korbie D, Millar EK, Pidsley R, et al. DNAmethylation of oestrogen-regulated enhancers defines endocrine sensitiv-ity in breast cancer. Nat Commun 2015;6:7758.


Ward et al.




33. Badia E, Duchesne MJ, Semlali A, Fuentes M, Giamarchi C, Richard-Foy H, et al. Long-term hydroxytamoxifen treatment of an MCF-7-derived breast cancer cell line irreversibly inhibits the expression ofestrogenic genes through chromatin remodeling. Cancer Res 2000;60:4130–8.

34. Nguyen VT, Barozzi I, Faronato M, Lombardo Y, Steel JH, Patel N, et al.Differential epigenetic reprogramming in response to specific endocrinetherapies promotes cholesterol biosynthesis and cellular invasion.Nat Commun 2015;6:10044.

35. Bronzert DA, Greene GL, Lippman ME. Selection and characterizationof a breast cancer cell line resistant to the antiestrogen LY 117018.Endocrinology 1985;117:1409–17.

36. Vareslija D, Cocchiglia S, Byrne C, Young L. Patient-derived xenografts ofbreast cancer. Methods Mol Biol 2017;1501:327–36.

37. Walsh CA, Bolger JC, Byrne C, Cocchiglia S, Hao Y, Fagan A, et al. Globalgene repression by the steroid receptor coactivator SRC-1 promotes onco-genesis. Cancer Res 2014;74:2533–44.

38. Shaw FL, Harrison H, Spence K, Ablett MP, Simoes BM, Farnie G, et al. Adetailedmammosphere assay protocol for the quantificationof breast stemcell activity. J Mammary Gland Biol Neoplasia 2012;17:111–7.

39. BrowneAL, Charmsaz S, Vares�lijaD, FaganA,CosgroveN,Cocchiglia S, et al.Network analysis of SRC-1 reveals a novel transcription factor hub whichregulates endocrine resistant breast cancer. Oncogene 2018;37:2008–21.

40. Debnath J,Muthuswamy SK, Brugge JS.Morphogenesis andoncogenesis ofMCF-10Amammary epithelial acini grown in three-dimensional basementmembrane cultures. Methods 2003;30:256–68.

41. O'Hara J, Vareslija D, McBryan J, Bane F, Tibbitts P, Byrne C, et al. AIB1:ERalpha transcriptional activity is selectively enhanced in aromataseinhibitor-resistant breast cancer cells. Clin Cancer Res 2012;18:3305–15.

42. Charmsaz S, Hughes E, Bane FT, Tibbitts P, McIlroy M, Byrne C, et al.S100beta as a serummarker in endocrine resistant breast cancer. BMCMed2017;15:79.

43. McCartan D, Bolger JC, Fagan A, Byrne C, Hao Y, Qin L, et al. Globalcharacterization of the SRC-1 transcriptome identifies ADAM22 as an ER-independent mediator of endocrine-resistant breast cancer. Cancer Res2012;72:220–9.

44. Simoes BM, O'Brien CS, Eyre R, Silva A, Yu L, Sarmiento-Castro A, et al.Anti-estrogen resistance in human breast tumors is driven by JAG1-NOTCH4-dependent cancer stem cell activity. Cell Rep 2015;12:1968–77.

45. DaiM,Wang P, Boyd AD, KostovG, Athey B, Jones EG, et al. Evolving gene/transcript definitions significantly alter the interpretation of GeneChipdata. Nucleic Acids Res 2005;33:e175.

46. Johnson WE, Li C, Rabinovic A. Adjusting batch effects in microarrayexpression data using empirical Bayes methods. Biostatistics 2007;8:118–27.

47. Ross-Innes CS, Stark R, Teschendorff AE, Holmes KA, Ali HR, Dunning MJ,et al. Differential oestrogen receptor binding is associated with clinicaloutcome in breast cancer. Nature 2012;481:389–93.

48. Baylin SB, Ohm JE. Epigenetic gene silencing in cancer - a mechanism forearly oncogenic pathway addiction? Nat Rev Cancer 2006;6:107–16.

49. EaswaranH, Johnstone SE, VanNeste L,Ohm J,Mosbruger T,WangQ, et al.A DNA hypermethylationmodule for the stem/progenitor cell signature ofcancer. Genome Res 2012;22:837–49.

50. Widschwendter M, Fiegl H, Egle D, Mueller-Holzner E, Spizzo G, MarthC, et al. Epigenetic stem cell signature in cancer. Nat Genet 2007;39:157–8.

51. Sasaki H, Matsui Y. Epigenetic events in mammalian germ-cell develop-ment: reprogramming and beyond. Nat Rev Genet 2008;9:129–40.

52. Jones PA, Liang G. Rethinking how DNA methylation patterns are main-tained. Nat Rev Genet 2009;10:805–11.

53. Kittler R, Zhou J, Hua S, Ma L, Liu Y, Pendleton E, et al. A comprehensivenuclear receptor network for breast cancer cells. Cell Rep 2013;3:538–51.

54. Park S, Koo J, Park HS, Kim JH, Choi SY, Lee JH, et al. Expression ofandrogen receptors in primary breast cancer. Ann Oncol 2010;21:488–92.

55. Shen H, Laird PW. Interplay between the cancer genome and epigenome.Cell 2013;153:38–55.

56. Tishkov I, Mushmov D, Milkov V, Lazarov V, Rapundzhieva A. [Thetreatment of patients with idiopathic membranous glomerulonephritiswith cyclosporin A]. Vutr Boles 1989;28:104–6.

57. Lin X, Li J, Yin G, ZhaoQ, Elias D, Lykkesfeldt AE, et al. Integrative analysesof gene expression and DNAmethylation profiles in breast cancer cell linemodels of tamoxifen-resistance indicate a potential role of cells with stem-like properties. Breast Cancer Res 2013;15:R119.

58. Dobrovolna J, Chinenov Y, KennedyMA, Liu B, Rogatsky I. Glucocorticoid-dependent phosphorylation of the transcriptional coregulator GRIP1.Mol Cell Biol 2012;32:730–9.

59. Uhlenhaut NH, Barish GD, Yu RT,DownesM, KarunasiriM, Liddle C, et al.Insights into negative regulation by the glucocorticoid receptor fromgenome-wide profiling of inflammatory cistromes. Mol Cell 2013;49:158–71.

60. Fatemi M, Wade PA. MBD family proteins: reading the epigenetic code.J Cell Sci 2006;119:3033–7.

61. Jones PA, Baylin SB. The fundamental role of epigenetic events in cancer.Nat Rev Genet 2002;3:415–28.

62. Martens JW, Margossian AL, Schmitt M, Foekens J, Harbeck N. DNAmethylation as a biomarker in breast cancer. Future Oncol 2009;5:1245–56.

63. Jones PA. At the tipping point for epigenetic therapies in cancer. J ClinInvest 2014;124:14–6.

64. Navada SC, Steinmann J, Lubbert M, Silverman LR. Clinical developmentof demethylating agents in hematology. J Clin Invest 2014;124:40–6.

65. West AC, Johnstone RW. New and emerging HDAC inhibitors for cancertreatment. J Clin Invest 2014;124:30–9.

66. Jones PA, Issa JP, Baylin S. Targeting the cancer epigenome for therapy.Nat Rev Genet 2016;17:630–41.






2018;24:3692-3703. Published OnlineFirst March 22, 2018.Clin Cancer Res Elspeth Ward, Damir Vareslija, Sara Charmsaz, et al. Cellular Differentiation in Advanced Breast Cancer

Mediated Gene Silencing Represses−Epigenome-wide SRC-1

Updated version

10.1158/1078-0432.CCR-17-2615doi:

Access the most recent version of this article at:

Material

Supplementary

http://clincancerres.aacrjournals.org/content/suppl/2018/03/22/1078-0432.CCR-17-2615.DC1

Access the most recent supplemental material at:

Cited articles

http://clincancerres.aacrjournals.org/content/24/15/3692.full#ref-list-1

This article cites 66 articles, 20 of which you can access for free at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://clincancerres.aacrjournals.org/content/24/15/3692To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/lookup/doi/10.1158/1078-0432.CCR-17-2615

http://clincancerres.aacrjournals.org/content/suppl/2018/03/22/1078-0432.CCR-17-2615.DC1

http://clincancerres.aacrjournals.org/content/24/15/3692.full#ref-list-1

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://clincancerres.aacrjournals.org/content/24/15/3692


epigenome-wide src-1 mediated gene silencing represses ...molecular determinants that drive...

Documents