Tiffany A. Katz Shauna N. Vasilatos Emily Harrington Steffi Oesterreich Nancy E. Davidson Yi Huang

Transcription

1 Brest Cner Res Tret (214) 146:99 18 DOI 1.17/s PRECLINICAL STUDY Inhiition of histone demethylse, LSD2 (KDM1B), ttenutes DNA methyltion nd inreses sensitivity to DNMT inhiitorindued poptosis in rest ner ells Tiffny A. Ktz Shun N. Vsiltos Emily Hrrington Steffi Oesterreih Nny E. Dvidson Yi Hung Reeived: 23 My 214 / Aepted: 26 My 214 / Pulished online: 13 June 214 Ó Springer Siene+Business Medi New York 214 Astrt Inresing evidene suggests tht dysfuntion of histone lysine demethylse is ssoited with norml hromtin remodeling nd gene silening, ontriuting to rest tumorigenesis. In silio nlysis shows tht the newly identified histone demethylse lysine-speifi demethylse 2 is highly expressed in rest ner, espeilly in invsive tumors. However, it is urrently unknown how LSD2 regultes hromtin remodeling nd gene expression regultion in rest ner. Using short hirpin RNA, we stly knoked down LSD2 (LSD2-KD) in MDA-MB-231 rest ner ells. LSD2-KD led to umultion of H3K4me1/2 without hnging methyltion levels of other key histone lysine residues, suggesting tht Eletroni supplementry mteril The online version of this rtile (doi:1.17/s ) ontins supplementry mteril, whih is ville to uthorized users. T. A. Ktz S. N. Vsiltos E. Hrrington S. Oesterreih N. E. Dvidson (&) Y. Hung UPMC Cner Reserh Pvilion, University of Pittsurgh Cner Institute, Suite 5, 515 Centre Ave, Pittsurgh, PA 15232, USA e-mil: dvidsonne@upm.edu T. A. Ktz E. Hrrington S. Oesterreih N. E. Dvidson Y. Hung Deprtment of Phrmology & Chemil Biology, University of Pittsurgh, Pittsurgh, PA, USA T. A. Ktz S. N. Vsiltos E. Hrrington S. Oesterreih N. E. Dvidson Y. Hung Women s Cner Reserh Center, University of Pittsurgh, Pittsurgh, PA, USA Y. Hung (&) Mgee Womens Reserh Institute, Room 46, 24 Crft Ave, Pittsurgh, PA 15213, USA e-mil: yih26@pitt.edu LSD2 ts s on fide H3K4 demethylse in rest ner ells. LSD2-KD resulted in deresed olony formtion nd ttenuted glol DNA methyltion in MDA- MB-231 ells. Additionlly, tretment with the DNMT inhiitor, 5-z-deoxyytidine (DAC), synergistilly inresed mrna expression of errntly silened genes importnt in rest ner development, inluding PR, RAR, ER, SFRP1, SFRP2, nd E-dherin in LSD2-KD ells. Furthermore, LSD2-KD ells re more suseptile to ell deth thn srmle ontrols, nd omined tretment with trnylypromine, n LSD2 inhiitor, nd DAC resulted in synergisti growth inhiition of rest ner ells. DNMT inhiition y DAC in LSD2-KD ells led to internuleosoml DNA frgmenttion, enhned PARP levge nd inresed su-g1 poptoti ell popultion. These results demonstrte n importnt role for LSD2 in regultion of DNA methyltion nd gene silening in rest ner, nd suggest tht inhiition of LSD2 in omintion with DNA methyltrnsferse inhiition represents novel pproh for epigeneti therpy of rest ner. Keywords Brest ner LSD2 DNA methyltion DNMT inhiitor Comintion therpy Apoptosis Arevitions LSD Lysine-speifi demethylse KD Knokdown HDAC Histone deetylse DAC Deitine TCP Trnylypromine DNMT DNA methyltrnsferse PGR (PR) Progesterone reeptor ESR1 (ER) Estrogen reeptor SFRP Sereted frizzled relted protein

2 1 Brest Cner Res Tret (214) 146:99 18 RAR CDH1 Retinoi id reeptor E-dherin gene trnsription, nd therpeuti response. We demonstrte for the first time tht LSD2-KD leds to inresed suseptiility to DAC nd enhned reexpression of silened genes. Introdution Mterils nd methods The groundreking disovery tht histones n e demethylted highlights the pervsive nd dynmi nture of post-trnsltionl modifitions of hromtin. The reversiility of histone methyltion y histone demethylses provides novel nd promising trget for therpeuti intervention. The mine oxidse fmily of histone demethylses onsists of LSD1 (KDM1A or AOF2) nd LSD2 (KDM1B or AOF1). Both re FAD-dependent nd ple of demethylting mono- nd di-methylted lysine 4 of histone H3 (H3K4me1/2) [1 3]. While LSD1 nd LSD2 shre 33 % mino id sequene homology in the mine oxidse domin [3] nd trget the sme enzymti sustrte (H3K4me1/2), reent evidene indites distint funtions for eh. LSD2 lks tower domin whih LSD1 uses for protein protein intertions. LSD2 lso ontins n N-terminl CW-type zin finger domin whih is sent in LSD1 [3]. LSD1 is lrgely ssoited with promoter regions of genes [4], while LSD2 ssoites more with oding regions [5]. LSD2 does not ind histone deetylses (HDACs) or the REST orepressor 1, while LSD1 is known to ind oth [3, 4, 6]. Therefore, LSD2 nd LSD1 hve distint funtions. However, the preise role of LSD2 in rest ner tumorigenesis hs not een fully investigted. A reent study hs reported tht LSD2 is needed for de novo DNA methyltion during emryoni development [7]. Anorml DNA methyltion is frequently ssoited with dysregulted histone tivity tht ontriutes to errnt gene silening in rest tumors nd is ompnied y poor prognosis [8]. Silening of importnt genes suh s the nuler reeptors estrogen reeptor, progesterone reeptor, nd retinoid id reeptor (RAR), nd tumor suppressor genes suh s E-dherin (CDH1) nd sereted frizzled relted proteins promote rest ner progression nd resistne to trgeted therpy [4, 9 12]. LSD1 inhiition lso leds to reexpression of silened genes, nd signifintly ttenutes rest ner ell growth [12 14]. Crosstlk etween HDACs nd other key hromtin modifiers is supported y our reent work showing tht LSD1 interts with HDACs to regulte gene expression nd rest ner ell growth [15]. On the sis of these reent findings, we ddressed the role of LSD2 in rest ner nd its rosstlk with other hromtin modifiers. We exmined the funtionl link etween LSD2 nd different key epigeneti enzymes in hromtin remodeling, Regents nd ell ulture onditions Cells were mintined in DMEM (Corning ellgro, Tewksury, MA) with 5 % fetl ovine serum (omplete medi), nd 4 lg/ml G418 t 37 C in humidified tmosphere with 5 % CO 2. Experiments were performed in omplete medi without G418. Live ell imges were tken on Zeiss Axiovert 4C using Motim 2 digitl mer nd Moti Imges Plus 2. quisition softwre. LSD2 shrna tretment nd stle ell line genertion MDA-MB-231 or MCF7 ells were trnsfeted with one of 4 short hirpin RNA (shrna) or srmle shrna (SA Biosienes, Vleni, CA) using Attrtene (SA Biosienes). Cells were then plted in 1 m dishes t low density in medi ontining G418 (8 lg/ml for MDA- MB-231, nd 6 lg/ml for MCF7) for seletion. Singleell olonies were tested for LSD2 knokdown y quntittive PCR. Two olonies expressing the lowest LSD2 mrna levels were developed from shrna 2 nd shrna 3 nd used throughout this mnusript. Western lotting Nuler extrts were prepred using the NE-PER Kit (Thermo Sientifi, Rokford, IL), seprted y SDS- PAGE, trnsferred to nitroellulose memrne, nd immunolotted using LSD1, H3K4me1, H3K4me2, H3K4me3, H3K9me1, H3K9me2, H3K9me3, AH3K9, H3K27me1, H3K36me2, H3K27me3, AH4K16, H3, PCNA (Millipore, Billeri, MA), LSD2 (Novus Biologils, Littleton, CO), -tin (Agent, Sn Diego, CA), DNMT1 (provided y Dr. Willim Nelson t Johns Hopkins University), PR (Snt Cruz Biotehnologies, Snt Cruz, CA), nd PARP (Ative Motif, Crlsd, CA) speifi ntiodies. Li-Cor (Linoln, NE) Odyssey Bloking Buffer nd seondry ntiodies were used. Bnds were snned on the Li-Cor CLx Imger nd quntified using Imge Studio 2.1 Softwre (Li-Cor). Colony formtion Srmle nd LSD2-KD MDA-MB-231 ells were plted in 6-well pltes (5 ells/well) in omplete medi. After

3 Brest Cner Res Tret (214) 146: dys, ells were stined with rystl violet (Sigm), dried overnight, nd olonies were ounted. Demethyltion ssys Srmle LSD2 KD LSD2 LSD1 Atin Nuler extrts were used in two ELISA-like ssys MethylFlsh 5-methylytosine (5-mC) Quntifition Kit (olorimetri, sensitivity = 1 nm of 5-mC), nd Epigense LSD1 Demethylse tivity Kit (fluorometri, sensitivity = 2 ng purified protein) (Epigentek, Frmingdle, NY) ording to mnufturer s reommendtions DNA frgmenttion After tretment with DAC (5-z-2 -deoxyytidine, Deitine, Cymn Chemil, Ann Aror, MI) for 96 h, DNA ldder frgments were prepred s desried previously [16], nd nlyzed y grose (2 %) gel eletrophoresis. Propidium iodide stining Cells were trypsinized, fixed in 7 % ethnol, entrifuged, nd wshed. The ell pellet ws then resuspended in 5 lg/ ml propidium iodide (Sigm) ontining 1 lg/ml RNseI (Rohe, Indinpolis, IN). Smples were nlyzed on the Auri C6 (BD Biosienes, Sn Jose, CA) in the University of Pittsurgh Cner Institute Cytometry Fility. BD CSmpler Softwre ws used to ssess ell yle. Srmle 2 RNA nlysis Totl RNA ws isolted from ells using the RNesy Kit (Qigen, Vleni, CA) ording to mnufturer s speifitions nd treted with DNseI (Rohe). 3 lg of RNA ws reverse trnsried nd quntittive rel-time PCR ws performed using 3 ll of DNA s previously desried using Tqmn proes (ABI) (Suppl Tle 1). Crystl violet nd drug omintion index nlysis To otin IC 5 vlues for hemotherpeuti drugs, srmle or LSD2-KD ells were plted in 96-well pltes nd treted with inresing doses of DAC, SAHA (Vorinostt, Cymn Chemil), ropltin, 4-OH-tmoxifen, lptini, doxoruiin, ABT-888, plitxel, or TCP (Sigm, St. Louis, MO). To otin the omintion index (CI), prentl MDA-MB-231 ells were treted with inresing doses of DAC or TCP for 12 h nd IC 5 vlues for eh were otined fter 4 experiments (DAC = 5.3 lm, TCP = lm, rtio = 1:19.87). Comintoril doses of 49, 29, 19,.759,.59, nd.259 were lulted using the IC 5 vlues. Cells were then simultneously Fig. 1 Stle LSD2-KD leds to deresed olony formtion. A representtive immunolot of MDA-MB-231 srmle ontrol nd LSD2-KD nuler extrts proed for LSD2, LSD1, nd -tin protein is shown. LSD2-KD nd srmle ontrol ells were llowed to form olonies for 2 weeks nd olonies were ounted. Brs represent the men of 3 experiments ± SEM (t test, = p \.1). A representtive imge from 3 olony formtion ssys for eh lone treted with TCP nd DAC using the omintoril doses for 12 h. Cells were stined with rystl violet, dried overnight, rystls were dissolved in.1 M sodium itrte, nd red t 45 nm. Clusyn softwre (Biosoft, Cmridge, UK) ws used to lulte IC 5 nd CI vlues. The Chou Tlly medin effet/ci model ws used to determine synergy, dditivity, or ntgonism of omintion therpy [17]. Sttistis GrphPd Prism 5. or Exel softwre ws used to determine the sttistil differenes etween vrious experimentl nd ontrol groups through one-wy or two-wy nlysis of vrine or Student s t test.

4 12 Brest Cner Res Tret (214) 146:99 18 Results Inhiition of LSD2 redues olony formtion of rest ner ells The Onomine dtse shows tht LSD2 is highly expressed in invsive rest ner ompred to norml tissue in two dt sets ( nd is ltered in 11 % of humn rest ners ( (Supp Fig 1) rising the question of whether enhned LSD2 expression in rest ner plys role in tumor ell growth. To ddress this question, we developed stle LSD2-KD ells using shrna in the rest ner ell line, MDA-MB-231. LSD2 trgeting shrna effetively redued endogenous LSD2 protein expression over 9 % without effeting LSD1 protein levels in two LSD2-KD olonies whih rose from single ells (Fig. 1). The two LSD2-KD lones hosen for use throughout these studies hd the lowest levels of LSD2 protein nd were reted using two different shrnas with unique sequenes. In 2D olony formtion ssy loss of LSD2 led to 25 5 % redution in olonies formed ompred to srmle ontrol MDA-MB-231 ells in oth LSD2-KD lones (Fig. 1, ). These results demonstrte survivl promoting role for LSD2 in rest ner ells. LSD2 speifilly demethyltes H3K4me1/2 in rest ner ells Our reent study hs shown tht trnsient suppression of LSD2 mrna expression y sirna in MDA-MB-231 ells inreses nuler levels of H3K4me2, ut fils to lter the level of AH3K9 [12]. In nother study using dendriti ells, LSD2 hs een shown to demethylte H3K9me2 t speifi gene promoters [18]. To investigte the impt of stle LSD2 defiieny on nuler levels of H3K4me1/2 nd other histone mrks two pprohes were utilized; n ELISA-like ssy to determine the demethylse tivity of LSD2-KD nd srmle nuler extrts, nd western lot nlysis to ssess glol histone mrks. The in vitro ELISA-like ssy quntitted H3K4 demethyltion y LSD2 using short 21 mino id H3K4me peptides s sustrtes. This ssy demonstrted tht signifintly less H3K4 is demethylted y nuler protein lystes from LSD2-KD ells ompred to srmle ontrol ounterprts (Fig. 2). Western lot nlysis onfirmed this finding showing tht LSD-KD led to signifint inrese in oth H3K4me1 nd H3K4me2, without ffeting glol H3K9me2 levels (Fig. 2). The inresed H3K4 methyltion in LSD2-KD ells ws ompnied y 3 % redution in Aetyl H3K9, hromtin mrk ssoited with tive trnsription. H3K27me2, H3K36me1 & 3 proteins were undetetle, while levels of H3K4me3, H3K9me1, H3K9me2, H3K9me3, H3K27me1, Methylted H3K4 (Perent of Srmle) Protein Expression (Correted y PCNA) d Reltive mrna Level Perent of Control (%) e H3K4me1 H3K4me2 H3K4me3 H3K9me1 H3K9me2 H3K9me3 H3K27me1 H3K27me3 H3K36me2 AH3K9 AH4K16 H3 H3K27me3, AH4K16, nd H3K36me2 remin unhnged, suggesting tht LSD2 ts s on fide demethylse speifi to H3K4me1/2 in rest ner ells (Fig. 2, ) Srmle SAHA (µm) Fig. 2 Stle LSD2-KD is ssoited with H3K4 methyltion nd hs little effet on the Jmj-C histone demethylses or HDAC tivity. MDA-MB-231 srmle or LSD2-KD nuler extrts were used s input for n ELISA-like H3K4 demethyltion ssy (Epigentek). Brs represent the men of 3 experiments ± SEM (t test, = p \.1). Nuler extrts from srmle or LSD2-KD MDA-MB-231 ells were proed using the indited ntiodies y western lot. Imges depited re representtive of 3 independent experiments. Imge Studio softwre ws used to quntitte western lot nd intensity. Brs represent the men of 3 experiments ±SEM (t test, = p \.5). d Jmj mrna expression ws ssessed y quntittive PCR using speifi Tqmn proes in srmle nd LSD2-KD MDA- MB-231 ells. Brs represent the men of 3 experiments ±SEM (t test). e Srmle nd LSD2-KD MDA-MB-231 ells were treted with inresing doses of SAHA for 72 h nd rystl violet ssy ws performed. Eh point represents the men of 3 experiments ±SEM

5 Brest Cner Res Tret (214) 146: m levels (fold) Reltive DNMT1 PCNA DNMT3L PCNA Fig. 3 LSD2-KD redues glol DNA methyltion in rest ner ells without hnging DNMT1 protein levels. 5-Methylytosine (5- mc) levels in genomi DNA from srmle ontrol, or LSD2-KD MDA-MB-231 ells were mesured y n ELISA-like ssy (Epigentek). Brs represent the men of 3 experiments ±SEM (t test, = p \.5). DNMT or TET1 mrna levels were nlysed y quntittive PCR using Tqmn gene expression ssys. Brs represent the men of 3 experiments ± SEM (t test, = p \.5). A representtive western lot imge of 3 experiments of DNMT1 nd DNMT3L protein expression in nuler lyste is shown. In ll pnels t tests were used (n = 3, = p \.5) We lso exmined the effet of LSD2 defiieny on mrna expression of JmjC-domin-ontining histone demethylses whih tlyze Fe nd -ketoglutrtedependent histone demethyltion. JARID1B (KDM5B) is n H3K4 demethylse speifilly trgeting H3K4me3/2 [19], while JMJD2B (KDM4B) ntgonizes H3K9me3/2 nd H3K36me3/2 [2, 21]. Loss of LSD2 hs no effet on mrna expression of these enzymes (Fig. 2d). Reently, we demonstrted tht LSD1 interts with HDACs in rest ner ells nd stle knokdown of LSD1 repressed mrna expression of most HDAC isozymes. Additionlly, HDAC inhiition led to signifint growth inhiition nd poptoti deth in LSD1-KD MDA- MB-231 ells [15]. To understnd whether LSD2, in onert with LSD1, interts with HDACs in humn rest ner ells, we exmined the effet of LSD2-KD on ellulr sensitivity in response to the HDAC inhiitor SAHA, finding similr sensitivity in srmle nd LSD2-KD ells (Fig. 2e). These dt indite tht LSD2 my not funtion in ssoition with HDAC tivity in the wy LSD1 does in rest ner ells. Inhiition of LSD2 redues glol DNA methyltion In ner ells, DNA hypermethyltion frequently ts in ollortion with norml histone modifitions ulminting in deresed hromtin tivting mrks. However, the impt of histone demethylses on DNA methyltion in ner ells hs not een fully investigted. We ssessed glol DNA methyltion in LSD2-KD ells using n ELISA-like ssy to detet 5-mC levels in genomi DNA. Glol DNA methyltion ws signifintly deresed in oth LSD2-KD lones to similr extent s DAC tretment (Fig. 3). Quntittive RT-PCR results showed tht LSD2- KD inresed the mrna expression of DNMT1 nd DNMT3L without ltering mrna levels of DNMT3 nd DNMT3 or TET1 (Fig. 3). DNMT3L is DNMT-like protein whih inds diretly to DNMT3 nd DNMT3. However, neither DNMT1 nor DNMT3L protein expression ws mrkedly hnged y LSD2-KD s evidened y western lot (Fig. 3). These dt suggest tht loss of LSD2 redues DNA methyltion likely through lokde of DNMT tivity rther thn downregultion of the protein expression of DNMTs. Loss of LSD2 enhnes DNMT inhiitor-indued reexpression of errntly silened genes in rest ner ells In ner ells, the oupny of H3K4me2 is typilly found to e t low level in the promoters of epigenetilly silened genes tht re frequently ssoited with DNA hypermethyltion [22, 23]. The omintion of LSD2-KD with DNMT inhiition leds to enhned reexpression of severl epigenetilly silened ndidte genes in rest ner ells inluding PR, RAR, SFRP1, SFRP2, ER, nd CDH1 (Fig. 4). Reexpression of PR, SFRP1, nd SFRP2 rehed sttistil signifine in oth LSD2-KD lones, while reexpression of RAR ws signifint only in lone 2 nd CDH1 reexpression ws gretly enhned only in LSD2-KD lone 2. ER gene reexpression ws gretly enhned in oth lones ut did not reh sttistil signifine. PR ws the most signifintly reexpressed gene y omined inhiition of LSD2 nd DNMT. PR plys n importnt role in rest ner iology nd hs een shown to e silened in rest ner due to DNA methyltion

6 14 Brest Cner Res Tret (214) 146:99 18 PR/tin Reltive PR/tin Reltive RARβ/tin Reltive RARβ/tin Reltive SFRP1/tin Reltive SFRP1/tin Reltive SFRP2/tin Reltive SFRP2/tin Reltive ERα/tin Reltive ERα/tin Reltive E-dherin/tin Reltive E-dherin/tin Reltive Untreted Srmle LSD2 KD DAC Srmle LSD2 KD T47D (+) PRB PRA Atin Reltive Density (PRA/Atin) Srmle 2 Untreted DAC [24], nd this silening is known to e ssoited with worse prognosis [25]. Next, we investigted further if the reexpression of PR mrna trnslted to enhned protein expression. Tretment with 1 lm DAC for 72 h led to roust inrese of PRA protein reexpression in oth LSD2- KD lones, while PRB levels were unffeted (Fig. 4). Quntittion in Fig. 4 shows tht DAC tretment leds to n inrese in PRA of out 2-fold in srmle 1 ompred

7 Brest Cner Res Tret (214) 146: Fig. 4 Comined LSD2-KD nd DNMT inhiition gretly enhnes reexpression of epigenetilly silened ndidte genes. Srmle or LSD2-KD MDA-MB-231 ells were treted with DAC (1 lm) or vehile for 48 h nd RNA ws olleted. Reltive fold hnge expression is presented in oth srmle nd LSD2-KD lones 1 nd 2 for the genes indited. Two-wy ANOVA ws implemented to determine sttistilly signifint differenes. Brs represent the men of 3 experiments ± SEM (t test, = p \.5, = p \.1, = p \.1). A representtive immunolot of MDA-MB- 231 srmle, or LSD2-KD lones 1 & 2 ± DAC (1 lm) or vehile for 72 h nd proed for PR nd tin protein is shown. PRA nds were quntitted nd the men ± SE of 3 experiments is presented to untreted with no hnge in srmle 2, nd n pproximte 7-fold inrese in protein in eh LSD2-KD lone. These dt indite tht omined inhiition of LSD2 nd DNMT leds to reexpression of the epigenetilly silened PRA gene whih trnsltes to inresed protein expression. LSD2-KD renders rest ner ells more suseptile to poptosis in response to DNMT inhiition The pprent synergy etween the LSD2-KD nd DNMT inhiitor for gene re-expression rised the importnt question of whether suh n effet might lso trnslte into therpeuti effiy in rest ner. To ddress this issue, MDA-MB-231 LSD2-KD ells were treted with DAC for 12 h, nd ell numer ws ssessed y rystl violet stining. MDA-MB-231 ells stly expressing LSD2 shrna were more sensitive to DAC-indued growth inhiition s evidened y signifintly deresed IC 5 vlues (Fig. 5). A similr result ws oserved in MCF-7 LSD2-KD ells, suggesting tht loss of LSD2 signifintly sensitizes rest ner ells to DAC-indued growth inhiition nd exerts similr effet in different sutypes of rest ner ells (Fig. 5). Next, we investigted the omintoril effet of trnylypromine, n identified LSD2 inhiitor, nd DAC on ell growth using the medin effet/ci model s desried in Mterils nd methods setion. Using onomitnt 12 h tretment shedule of gents, signifint synergisti growth inhiition (CI \ 1) ws oserved t low, medin, or higher dose omintion (ED 5, 75 nd 9) (Fig. 5). This result lerly suggests tht omintion therpy trgeting LSD2 nd DNMT exhiits gret synergy in inhiiting growth of rest ner ells. To determine if DAC-indued growth inhiition ws due to ell deth y poptosis, PARP levge nd ell yle prmeters were nlyzed. PARP levge ws indued y DAC tretment in oth srmle nd LSD2-KD ells nd quntittive nlysis showed DAC indued pproximtely 2-fold more PARP levge in LSD2-KD ells ompred to srmle ounterprts (Fig. 6). LSD2- KD in omintion with DAC tretment led to signifint Growth Inhiition (% of Control) Growth Inhiition (% of Control) Comintion Index DAC + TCP Srmle MDA-MB-231 MCF-7 Srmle 2 LSD2 KD MDA-MB-231 LSD2 Atin ED5 ED75 ED9 Frtionl Growth Inhiition MCF-7 Srmle LSD2 KD Fig. 5 LSD2-KD ells disply inresed sensitivity to DAC. MDA- MB-231 nd MCF7 srmle nd stle LSD2-KD ells were treted with inresing doses of DAC, nd rystl violet ssy ws performed fter 12 h. Eh point represents the men of 3 experiments ±SEM. Clusyn softwre ws used to lulte IC 5 vlues. Prentl MDA-MB-231 ells were treted simultneously with inresing doses of DAC nd TCP t 49, 29, 19,.759,.59, nd.259 of eh IC 5 (DAC = 21.2, 1.6, 5.3, 4., 2.7, nd 1.3 lm; TCP = , , 582.3, 436.7, 291.2, nd lm, DAC = 5.3 lm, TCP = lm, rtio = 1:19.87) for 12 h nd stined with rystl violet. Men ± SEM of omintion index vlues from 3 experiments is presented. Vlues \1 re defined s synergism nd vlues [1 s ntgonism indution of DNA frgmenttion, typil feture of poptoti ell deth, in oth LSD2-KD lones (Fig. 6). Furthermore, flow ytometry nlysis indited tht the su

8 16 Brest Cner Res Tret (214) 146:99 18 Untreted DAC Sr1 KD1 Sr1 KD C-PARP Atin Reltive Density Untreted DAC L Sr1 KD1 Sr1 KD1 Untreted DAC L Sr2 KD2 Sr2 KD2 % of Cells G2/M S G/G1 Su-G1 Sr1 KD1 Sr1 KD1 Untreted DAC d Untreted Srmle1 LSD2-KD1 Srmle1 DAC LSD2-KD1 e Untreted Srmle1 LSD2-KD1 Srmle1 DAC LSD2-KD1 Fig. 6 Inresed DAC sensitivity in LSD2-KD ells is due to indution of poptosis. MDA-MB-231 LSD2-KD nd srmle ells were treted with 1 lm DAC for 96 h. A representtive western lot of leved PARP with the reltive density elow eh nd is shown. A representtive imge of 3 experiments of frgmented DNA inditive of poptosis ws nlysed y eletrophoresis. Propidium iodide stined ells nlyzed y flow ytometry. Perentge of ell yle distriution ws quntitted (n = 4, twowy ANOVA, = p \.1, = p \.1, = p \.5). d Representtive imges of MDA-MB-231 srmle nd LSD2-KD ells in the presene or sene of 1 lm DAC fter 96 h G/G1 popultion ontining poptoti ells ws signifintly enhned in the LSD2-KD ells treted with DAC ompred to DAC-treted srmle ells (p \.5) (Fig. 6). As shown in Fig. 6d, LSD2-KD ells hd 2 % less ells in G/G1 popultion thn srmle ells whih is replited in DAC-treted ells (p \.1). LSD2-KD ells treted with DAC umulted in S phse (p \.1), with only slight inrese of ell numers in G2/M phse. Representtive imges of srmle nd LSD2-KD ells ±1 lm DAC for 96 h lerly indite tht inhiition of LSD2 sensitizes MDA-MB-231 ells to DAC-indued poptosis (Fig. 6e). To determine if deresed growth in LSD2-KD ells in response to DAC reflets generl defiieny in survivl fter n insult, or if it is speifi to DAC, we ssessed sensitivity of srmle nd LSD2-KD ells to pnel of linilly used rest ner therpeuti regents. LSD2-KD ells were not more sensitive to ny other rest ner therpeutis tested inluding 4-OH-tmoxifen, plitxel, doxoruiin, ropltin, or ABT-888 (PARP inhiitor). Thus, the inresed sensitivity to growth inhiition in LSD2-KD ells is speifi to the DNMT inhiitor, DAC (Tle 1 nd Supp Fig. 2). Disussion DAC is n effiious therpy for leukemi nd myelodysplsti syndrome nd numer of studies suggest tht Tle 1 LSD2-KD MDA-MB-231 ells hve deresed IC 5 vlue in response to DAC, ut no other nti-neoplsti drug tested Chemo drug Dose rnge Srmle IC5 LSD2 KD IC5 Deitine.5 1 lm OH-tmoxifen.1 4 lm Plitxel.1 2 nm ABT mm Doxoruiin.1 2 lm.1.1 SAHA.1 1 lm 1.4(S1) & 3(S2) 2.9(KD1) & 2.2(KD2) IC 5 vlues were lulted using Clusyn softwre nd t test ws used to determine signifint differenes epigeneti gents, like DAC, exert their effet in prt through the reexpression of epigenetilly silened genes. However, the suess of DAC nd other epigeneti modifiers s tretment for solid tumors inluding rest ner hs een more limited. To improve the potentil of DNMT inhiitors to t s effetive ntitumor gents in rest ner, it is neessry to etter understnd the epigeneti mehnisms y whih DNMT tivity is regulted. The potentil of developing novel nd effetive omintion strtegies to improve the effiy of urrent epigeneti gents in rest ner tretment is lso worthy of study. Our studies showed tht LSD2-KD sensitizes rest ner ells to the DNMT inhiitor, DAC, through indution of S phse ell yle rrest nd poptosis s evidened

9 Brest Cner Res Tret (214) 146: y enhned PARP levge, DNA frgmenttion, nd perent of ells in the poptoti pek y flow ytometry. We suggest novel strtegy to overome the refrtoriness of solid tumors to DAC through omintion therpy with LSD2 inhiition. Inhiiting LSD2 nd DNMT y TCP nd DAC resulted in synergisti growth inhiition, demonstrting similr effet of phrmologil inhiition nd shrna-medited LSD2-KD. The mehnisms underlying LSD2 s effets on the effiy of DNMTi re not ler. A likely explntion is tht H3K4 methyltion y LSD2 inhiition mkes genomi DNA more essile to the DNMT inhiitor. Ooi et l. [26] demonstrted tht H3K4 methyltion strongly inhiits the inding of DNMT3L to H3, suggesting n importnt role for H3K4 methyltion in DNMT tivity. However, inresed H3K4 methyltion is lerly not the only hromtin ltertion ontriuting to redued DNMT tivity. Interestingly, inresed sensitivity ws not oserved in ny linilly used nti-neoplsti gents with different mehnisms of tion ssessed, implying unique role for LSD2 in medition of ntitumor effiy of DNMTi in rest ner ells. Further studies re needed to understnd the preise mehnisms underlying the distint roles of histone demethylse fmily memers in regultion of DNMT tivity in rest ner ells. In this study, we explored norml histone methyltion in rest ner, nd the possiilities for reversing these ltertions s novel therpeuti trgets. We demonstrted for the first time tht inhiition of novel FAD-dependent histone demethylse, LSD2, y shrna signifintly redues olony formtion in rest ner ells. Although the preise role of LSD2 in rest tumorigenesis remins unler, these results provide evidene for growth promoting role for LSD2 in rest ner nd point to potentil utility of LSD2 inhiition s n effetive therpeuti pproh for this disese. We hve demonstrted tht LSD2 is on fide histone demethylse speifi for H3K4me1/2. Reent studies hve suggested role for LSD2 in H3K9 methyltion [5, 18, 27], ut we oserved no hnge in H3K9 methyltion in LSD2- KD rest ner ells. Unlike LSD1, the inresed H3K4 methyltion y LSD2 depletion ws not ompnied y inresed etyl-h3k9, suggesting tht LSD2 my not e funtionlly ssoited with HDACs. This is further supported y our findings tht LSD2-KD filed to hnge expression of the mjority of HDAC isozymes (dt not shown) nd did not lter the ellulr sensitivity to the HDAC inhiitor, SAHA. We lso found tht the level of glol DNA methyltion is signifintly ttenuted in LSD2-KD. The loss of LSD1 demethylse tivity ws reported to diretly result in redued levels of DNMT1 nd glol DNA methyltion during mouse emryogenesis [28]. Loss of LSD2 in ooytes resulted in deresed ility to methylte DNA during oogenesis [7]. This suggests tht LSD2 might e required for the stiliztion or tivity of DNMTs. The preise mehnisms y whih inhiition of LSD2 glolly dereses DNA methyltion remin elusive. H3K4me2 is ritil histone mrk tht is ssoited with open hromtin nd tive gene trnsription. H3K4 methyltion is redued in the promoters of numer of epigenetilly silened genes suh s tumor suppressor genes tht my led to tumorigenesis [29]. Enhned expression of LSD2 in rest ner ells depresses levels of H3K4me2 providing suitle hromtin environment for DNMT reruitment. Ooytes from LSD2-defiient mie disply signifintly inresed H3K4 methyltion nd filed to reruit DNMTs t importnt imprinted genes, suggesting ritil role of LSD2 in estlishing DNA methyltion t imprinted gene loi during oogenesis [7]. Comined inhiition of LSD2 nd DNMT led to roust retivtion of epigenetilly silened genes in rest ner ells, inluding severl nuler reeptors (PR, ER, nd RAR) nd tumor suppressor genes (SFRP1, SFRP2 nd CDH1). Aerrnt silening of these genes hs een implited in rest tumor development nd resistne to trgeted therpies [4, 11]. In triple negtive rest ner (TNBC), the sene of ER nd PR my ontriute to resistne to hormonl therpy. Indution of these nuler ftors y omined inhiition of LSD2 nd DNMT suggests novel epigeneti mehnism ontriuting to errnt loss of these genes s well s potentil therpeuti intervention in the tretment of TNBC through restortion of sensitivity to hormonl therpy. In sum, our studies provide solid evidene tht LSD2 is key regultor of DNA methyltion nd ntitumor effiy of DNMTi in rest ner ells. Inhiition of LSD2 promotes the poptoti response of rest ner ells to DNMTi. Bsed on these findings, we onlude tht omined inhiition of LSD2 nd DNMT represents novel nd effetive therpeuti trget for rest ner therpy. It is ntiipted tht these dt will dvne novel pproh of trgeting multiple epigeneti pthwys in rest ner nd led to the development of novel inhiitors of histone lysine demethylses, whih will e more effetive thn urrent strtegies in rest ner therpy. Aknowledgments Prtilly supported y Brest Cner Reserh Foundtion, Smuel Winters Foundtion, nd Competitive Medil Reserh Fund of UPMC. These studies used the UPCI Genomis Core Fility supported y P3CA4794. Conflits of interest The uthors delre no ompeting interests.

10 18 Brest Cner Res Tret (214) 146:99 18 Ethil stndrd All experiments omply with the urrent lws of the United Sttes of Ameri. Referenes 1. Shi Y, Ln F, Mtson C, Mullign P, Whetstine JR, Cole PA, Csero RA (24) Histone demethyltion medited y the nuler mine oxidse homolog LSD1. Cell 119(7): Lee MG, Wynder C, Cooh N, Shiekhttr R (25) An essentil role for CoREST in nuleosoml histone 3 lysine 4 demethyltion. Nture 437(757): Krytinos A, Forneris F, Profumo A, Ciossni G, Bttglioli E, Bind C, Mttevi A (29) A novel mmmlin flvin-dependent histone demethylse. J Biol Chem 284(26): doi:1. 174/j.M Hung Y, Nyk S, Jnkowitz R, Dvidson NE, Oesterreih S (211) Epigenetis in rest ner: wht s new? Brest Cner Res 13(6):225. doi:1.1186/r Fng R, Brer AJ, Xu Y, Rutenerg M, Leonor T, Bi Q, Ln F, Mei P, Yun GC, Lin C, Peng J, Cheng D, Sui G, Kiser UB, Shi Y, Shi YG (21) Humn LSD2/KDM1/AOF1 regultes gene trnsription y modulting intrgeni H3K4me2 methyltion. Mol Cell 39(2): doi:1.116/j.molel Yng Z, Jing J, Stewrt DM, Qi S, Ymne K, Li J, Zhng Y, Wong J (21) AOF1 is histone H3K4 demethylse possessing demethylse tivity-independent repression funtion. Cell Res 2(3): doi:1.138/r Cione DN, Su H, Hevi S, Gy F, Lei H, Bjko J, Xu G, Li E, Chen T (29) KDM1B is histone H3K4 demethylse required to estlish mternl genomi imprints. Nture 461(7262): doi:1.138/nture Mersin H, Yildirim E, Bereroglu U, Gulen K (28) The prognosti importne of triple negtive rest rinom. Brest 17(4): doi:1.116/j.rest Munster PN, Thurn KT, Thoms S, Rh P, Levi M, Miller A, Melisko M, Ismil-Khn R, Rugo H, Mosser M, Minton SE (211) A phse II study of the histone deetylse inhiitor vorinostt omined with tmoxifen for the tretment of ptients with hormone therpy-resistnt rest ner. Br J Cner 14(12): doi:1.138/j Pthirj TN, Sterns V, Oesterreih S (21) Epigeneti regultion in estrogen reeptor positive rest ner role in tretment response. J Mmmry Glnd Biol Neoplsi 15(1): doi:1.17/s Sterns V, Zhou Q, Dvidson NE (27) Epigeneti regultion s new trget for rest ner therpy. Cner Invest 25(8): doi:1.18/ Hung Y, Vsiltos SN, Bori L, Shw PG, Dvidson NE (212) Inhiitors of histone demethyltion nd histone deetyltion ooperte in regulting gene expression nd inhiiting growth in humn rest ner ells. Brest Cner Res Tret 131(3): doi:1.17/s Lim S, Jnzer A, Beker A, Zimmer A, Shule R, Buettner R, Kirfel J (29) Lysine-speifi demethylse 1 (LSD1) is highly expressed in ER-negtive rest ners nd iomrker prediting ggressive tumor iology. Crinogenesis 31(3): doi:1.193/rin/gp Zhu Q, Hung Y, Mrton LJ, Woster PM, Dvidson NE, Csero RA Jr (212) Polymine nlogs modulte gene expression y inhiiting lysine-speifi demethylse 1 (LSD1) nd ltering hromtin struture in humn rest ner ells. Amino Aids 42(2 3): doi:1.17/s Vsiltos SN, Ktz TA, Oesterreih S, Wn Y, Dvidson NE, Hung Y (213) Crosstlk etween lysine speifi demethylse 1 (LSD1) nd histone demethylses medites ntineoplsti effiy of HDAC inhiitors in rest ner ells. Crinogenesis 34 (6): doi:1.193/rin/gt Hung Y, Hger ER, Phillips DL, Dunn VR, Hker A, Frydmn B, Kink JA, Vlsins AL, Reddy VK, Mrton LJ, Csero RA Jr, Dvidson NE (23) A novel polymine nlog inhiits growth nd indues poptosis in humn rest ner ells. Clin Cner Res 9(7): Chou TC, Tlly P (1984) Quntittive nlysis of dose effet reltionships: the omined effets of multiple drugs or enzyme inhiitors. Adv Enzyme Regul 22: vn Essen D, Zhu Y, Sni S (21) A feed-forwrd iruit ontrolling induile NF-kppB trget gene tivtion y promoter histone demethyltion. Mol Cell 39(5): doi:1. 116/j.molel Sewrd DJ, Cuerley G, Kim S, Shonewld M, Zhng L, Tripet B, Bentley DL (27) Demethyltion of trimethylted histone H3 Lys4 in vivo y JARID1 JmjC proteins. Nt Strut Mol Biol 14(3): doi:1.138/nsm12 2. Ktoh Y, Ktoh M (27) Comprtive integromis on JMJD2A, JMJD2B nd JMJD2C: preferentil expression of JMJD2C in undifferentited ES ells. Int J Mol Med 2(2): Fodor BD, Kuiek S, Yonezw M, O Sullivn RJ, Sengupt R, Perez-Burgos L, Oprvil S, Mehtler K, Shott G, Jenuwein T (26) Jmjd2 ntgonizes H3K9 trimethyltion t perientri heterohromtin in mmmlin ells. Genes Dev 2(12): doi:1.111/gd Bylin SB, Ohm JE (26) Epigeneti gene silening in ner mehnism for erly onogeni pthwy ddition? Nt Rev Cner 6(2): doi:1.138/nr MGrvey KM, Vn Neste L, Cope L, Ohm JE, Hermn JG, Vn Criekinge W, Shueel KE, Bylin SB (28) Defining hromtin pttern tht hrterizes DNA-hypermethylted genes in olon ner ells. Cner Res 68(14): doi:1.1158/ CAN Lpidus RG, Ferguson AT, Ottvino YL, Prl FF, Smith HS, Weitzmn SA, Bylin SB, Iss JP, Dvidson NE (1996) Methyltion of estrogen nd progesterone reeptor gene 5 CpG islnds orreltes with lk of estrogen nd progesterone reeptor gene expression in rest tumors. Clin Cner Res 2(5): Pthirj TN, Shetty PB, Jelinek J, He R, Hrtmier R, Mrgossin AL, Hilsenek SG, Iss JP, Oesterreih S (211) Progesterone reeptor isoform-speifi promoter methyltion: ssoition of PRA promoter methyltion with worse outome in rest ner ptients. Clin Cner Res 17(12): doi:1.1158/ ccr Ooi SK, Qiu C, Bernstein E, Li K, Ji D, Yng Z, Erdjument- Bromge H, Tempst P, Lin SP, Allis CD, Cheng X, Bestor TH (27) DNMT3L onnets unmethylted lysine 4 of histone H3 to de novo methyltion of DNA. Nture 448(7154): doi:1.138/nture Holmes A, Roseulin L, Shurr C, Wxin H, Lmert S, Zrtiegui M, Mrtienssen RA, Arngioli B (212) LSD1 nd LSD2 ontrol progrmmed replition fork puses nd imprinting in fission yest. Cell Rep 2(6): doi:1.116/j.elrep Wng J, Hevi S, Kursh JK, Lei H, Gy F, Bjko J, Su H, Sun W, Chng H, Xu G, Gudet F, Li E, Chen T (29) The lysine demethylse LSD1 (KDM1) is required for mintenne of glol DNA methyltion. Nt Genet 41(1): doi:1. 138/ng MGrvey KM, Fhrner JA, Greene E, Mrtens J, Jenuwein T, Bylin SB (26) Silened tumor suppressor genes retivted y DNA demethyltion do not return to fully euhromti hromtin stte. Cner Res 66(7): doi:1.1158/ can