USP9X Downregulation Renders Breast Cancer Cells Resistant to Tamoxifen

Cancer Therapeutics, Targets, and Chemical Biology Research

Hendrika M. Oosterkamp1, E. Marielle Hijmans1, Thijn R. Brummelkamp1, Sander Canisius1, Lodewyk F.A. Wessels1, Wilbert Zwart2, and Rene Bernards1

Abstract Tamoxifen is one of the most widely used endocrine agents for the treatment of estrogen receptor a (ERa)– positive breast cancer. Although effective in most patients, resistance to tamoxifen is a clinically significant problem and the mechanisms responsible remain elusive. To address this problem, we performed a large scale loss-of-function genetic screen in ZR-75-1 luminal breast cancer cells to identify candidate resistance genes. In this manner, we found that loss of function in the deubiquitinase USP9X prevented proliferation arrest by tamoxifen, but not by the ER downregulator fulvestrant. RNAi-mediated attenuation of USP9X was sufficient to stabilize ERa on chromatin in the presence of tamoxifen, causing a global tamoxifen-driven activation of ERa-responsive genes. Using a gene signature defined by their differential expression after USP9Xattenuationinthepresenceoftamoxifen,wewereabletodefine patients with ERa-positive breast cancer experiencing a poor outcome after adjuvant treatment with tamoxifen. The signature was specificin its lack of correlation with survival in patients with breast cancer who did not receive endocrine therapy. Overall, our findings identify a gene signature as a candidate biomarker of response to tamoxifen in breast cancer. Cancer Res; 74(14); 3810–20. 2014 AACR.

Introduction levels or activity of ERa coactivators (AIB1), growth factor About 70% of human breast cancers are estrogen receptor a receptors (EGFR, HER2, and IGF1R), kinases (AKT and (ERa)–positive and depend on this hormone receptor for their ERK1/2) or adaptor proteins (BCAR1, c-SRC, and PAK1; proliferation (1), rendering ERa an ideal target for endocrine refs. 3, 6). Loss of CDK10 expression (7) and loss of insu- – treatment. Tamoxifen is one of the most commonly used drugs lin-like growth factor bindingprotein5(IGFBP5)expres- in the management of ERa-positive breast cancer. In early sion (8) can also lead to tamoxifen resistance. Furthermore, breast cancer, 5 years of adjuvant treatment with tamoxifen high levels of lemur tyrosine kinase-3 (LMTK3) or CUEDC2 almost halves the rate of disease recurrence and reduces the protein are associated with tamoxifen resistance (9, 10). a annual breast cancer-related death rate by one-third (2). ER -independent mechanisms can play a role in endocrine Despite this adjuvant treatment with tamoxifen, one-third of therapy resistance, including the NOTCH pathway (11). women still develop recurrent disease in the next 15 years (2), Acquired endocrine resistance develops in a certain propor- a illustrating that endocrine resistance is a major problem in the tion of metastatic ER -positive breast cancer that was management of breast cancer. initially sensitive to palliative tamoxifen treatment, but can Several mechanisms may contribute to tamoxifen resis- alsooccurintheadjuvantsettingwhenapatientrelapses tance. At presentation, not all ERa-positive tumors are while on hormonal therapy. Possible mechanisms of this – – sensitive to tamoxifen. This intrinsic endocrine resistance resistance are upregulation of the PI3K mTOR pathway (12 can be the result of ERa phosphorylation (3–5). In addition, 14) and acquiring activating mutations in ESR1 (15). It is intrinsic resistance is found to correlate with increased nevertheless likely that additional mechanisms contribute to unresponsiveness to endocrine treatment, which remains to be identiﬁed. Authors' Afﬁliations: 1Division of Molecular Carcinogenesis and Cancer Ubiquitination serves a role in both protein degradation and Genomics Center Netherlands; and 2Division of Molecular Pathology, The regulation of protein function (16). The level of protein ubi- Netherlands Cancer Institute, Amsterdam, the Netherlands quitination is highly regulated by two families of enzymes with Note: Supplementary data for this article are available at Cancer Research opposing activities: the ubiquitin ligases, which add ubiquitin Online (http://cancerres.aacrjournals.org/). moieties to proteins and deubiquitinating enzymes (DUB) that H.M. Oosterkamp and E.M. Hijmans contributed equally to this work. remove them (17). The X-linked deubiquitinase USP9X is a Corresponding Authors: Rene Bernards, The Netherlands Cancer Insti- member of the family of DUB enzymes and regulates multiple tute, Plesmanlaan 121, 1066 CX Amsterdam, the Netherlands. Phone: 31- cellular functions by deubiquitinating and stabilizing its sub- 20-5121952; E-mail: [email protected]; and Wilbert Zwart, [email protected] strates. USP9X is involved in a number of key cellular processes, doi: 10.1158/0008-5472.CAN-13-1960 as the knockout of this gene in the mouse is embryonic lethal 2014 American Association for Cancer Research. (18). USP9X has been shown to regulate, among others, cell

3810 Cancer Res; 74(14) July 15, 2014

adhesion molecules like b-catenin and E-cadherin, cell polar- Constructs ity, chromosome segregation, NOTCH, mTOR, and TGFb For retroviral transductions, ZR-75-1 and T47D cells were signaling as well as apoptosis (18–27). stably infected with supernatant of the Phoenix ampho- To elucidate novel mechanisms of tamoxifen resistance in trophic virus packaging cell line transfected with pBabeHy- breast cancer, we performed an shRNA screen in the hor- gro-Ecotropic Receptor or pLZRS-Ecotropic Receptor-IRES- mone-dependent human luminal breast cancer cell line ZR- Neo. Short hairpin sequences targeting USP9X are in Sup- 75-1 to identify genes in which knockdown could induce plementary Table S4. tamoxifen resistance. We report here an unexpected role for USP9X in ERa signaling: loss of USP9X enhances ERa/chro- Immunoprecipitation and immunoblotting matin interactions in the presence of tamoxifen, leading to For coimmunoprecipitation, cells were lysed in ELB (250 tamoxifen-stimulated gene expression of ERa target genes mmol/L NaCl, 0.1% NP-40, 50 mmol/L Hepes pH 7.3) contain- and cell proliferation. ing protease (Roche) and phosphatase inhibitors (Sigma- Aldrich). Supernatants were incubated with antibodies for a Materials and Methods USP9X (ab99343; Abcam), or ER (D-12; Santa Cruz Biotech- nology) coupled to protein G Dynabeads (Life Technologies). Cell lines and culture conditions Mixed normal mouse and rabbit serum (Santa Cruz Biotech- Phoenix cells were cultured in DMEM supplemented with nology) was used as control. For Western blotting the following m 10% FCS, 2 mmol/L glutamine and 100 g/mL penicillin/ antibodies were used: USP9X (ab99343; Abcam), ERa (clone streptomycin. ZR-75-1 and T47D cells were cultured under 1D5; Dako), progesterone receptor (PR; clone 1A6; Novocastra), the same conditions in the presence of 1 nmol/L estradiol. Cell and b-actin (clone AC-74; Sigma-Aldrich). lines were obtained from the ATCC (www.ATCC.org) and used at low passage after receipt from the vendor. Luciferase assay Monoclonal ZR-75-1 cells expressing pRS-USP9X or pRS– Transfection and retroviral infection GFP, were plated in triplicate in 6-well plates and were tran- Phoenix cells were transfected using calcium phosphate siently transfected with 1.75 mg ERE-TATA-luciferase reporter precipitation. Viral supernatant was cleared through a and 0.5 mg pRL-CMV Renilla luciferase (Promega) per well. m ﬁ 0.45- m lter. Target cells were infected twice with the viral 24 hours after transfection, ligand was added. Luciferase m supernatant using polybrene (8 g/mL). For transient trans- activity was determined 48 hours after transfection using the fection of ZR-75-1 cells Lipofectamine 2000 (Life Technologies) Dual Luciferase Reporter Assay System (Promega), with Renilla was used. luciferase as control, with vehicle set at 1.

The shRNA screen and recovery of shRNA inserts Quantitative real-time PCR Ecotropic receptor–containing ZR-75-1 cells were infected Total RNA was isolated using TRIzol (Life Technologies) or with the retroviral NKi pRetroSuper-shRNA library (12,540 using the Quick RNA MiniPrep Kit (Zymo Research). cDNA was shRNA vectors targeting 4,180 genes; ref. 28) or pRS as control. generated using Superscript II with random hexamer primers After puromycin selection, cells were cultured in DMEM with (Life Technologies). The qRT-PCR reaction was performed 1 mmol/L 4OH-tamoxifen for 6 weeks. Genomic DNA of indi- using FastStart Universal SYBR Green Master Mix (Roche) on vidual colonies was isolated using DNAzol (Life Technologies). an AB7500 Fast Real Time PCR system (Applied Biosystems). PCR amplification of the shRNA cassettes was performed using All reactions were run in parallel for GAPDH to control for the the Expand Long Template PCR System (Roche). Products amount of cDNA input. were digested with EcoRI/XhoI and recloned into pRS and PCR primer sequences are shown in Supplementary Table sequenced with Big Dye Terminator (PerkinElmer). Primers S4. are in Supplementary Table S4. RNA expression analysis Cell proliferation analyses RNA-seq reads (14–30 million 50-bp single-end) were For colony formation assays, infected cells were cultured in mapped to the human reference genome (hg19) using TopHat DMEM containing 1 mmol/L 4OH-tamoxifen, 1 nmol/L estra- (29), supplied with a known set of gene models (Ensembl diol, or 10 7 mol/L fulvestrant. After 2 to 6 weeks, cells were version 64). HTSeq count was used to obtain gene expressions. photographed, fixed with 4% formaldehyde, and stained with To identify differentially expressed genes, DEGseq (30) was 0.1% Crystal violet. Plate confluence was assessed by measur- used, P < 0.05. Levels of expressed genes were increased by 1 to ing the percentage of cell-covered surface area. Alternatively, avoid negative values after log2 transformation. Crystal violet was extracted with 10% acetic acid and absor- bance was measured at 600 nm. Chromatin immunoprecipitations For dynamic cell proliferation analyses, cells were seeded in Chromatin immunoprecipitations (ChIP) were performed a clear-bottom 96-well plate and proliferation was determined as described before (31). For each ChIP, 10 mgERa antibody by plate confluency using an IncuCyte life cell imaging device (HC-20; Santa Cruz Biotechnology) and 100 mL of Protein A (Essen BioScience). Three wells were measured per condition. Dynabeads (Life Technologies) were used. For qPCR, primers Bars indicate SD. are in Supplementary Table S4.

www.aacrjournals.org Cancer Res; 74(14) July 15, 2014 3811

Next-generation sequencing and enrichment analysis the comparison between treated and untreated, the Loi and ChIP DNA was amplified as described previously (31). colleagues (35) samples were additionally clustered on the Sequences were generated by the Illumina Hiseq 2000 genome basis of this subset only. This clustering was still able to stratify analyzer (using 50 bp reads), and aligned to the Human patients according to prognosis (log-rank P ¼ 1.3 10 5). The Reference Genome (assembly hg19; February 2009). Enriched directionality of USP9X knockdown tamoxifen classification regions of the genome were identified by comparing the ChIP genes in the good and poor outcome patient groups in the Loi samples with input using MACS (32) version 1.3.7.1. Details on and colleagues (35) cohort is shown in Supplementary Table sequence reads are Supplementary Table S1. S2. Association of the USP9X knockdown tamoxifen signature with outcome after chemotherapy was tested on the subset of Motif analysis, heatmaps, and genomic distributions of the 295 patients in the van de Vijver and colleagues cohort (39) binding events with ER-positive breast tumors, treated with adjuvant chemo- ChIP-seq data snapshots were generated using the Integra- therapy, but not with tamoxifen (n ¼ 56). Patients were tive Genome Viewer IGV 2.1 (www.broadinstitute.org/igv/). stratified in two groups and the difference in survival was Motif analyses were performed through the Cistrome (cis- tested for as described above. The gene signature mapped to trome.org), applying the SeqPos motif tool (33). The genomic 758 probes representing 461 unique genes. PAM50 (40) sub- distributions of binding sites were analyzed using the cis- typing by genefu package was applied to a pooled set of breast regulatory element annotation system (34). If the binding cancer samples (n ¼ 1,570; GEO accession: GSE47561; ref. 41), region is within a gene, 50 untranslated region (UTR), 30UTR, including the Loi and colleagues study, ensuring correct coding exon, or intron are determined. Promoter is defined as assignment of molecular subtypes. 3 kb around RefSeq 50 start. If a binding site is >3 kb away from the RefSeq transcription start site (TSS), it is considered distal intergenic. Results An shRNA screen identifies USP9X as a tamoxifen Survival analyses resistance gene Normalized mRNA expression for four patient series was To identify new genes involved in tamoxifen resistance, a downloaded from GEO: GSE6532 (35), GSE22219 (36), GSE2034 loss-of-function genetic screen was performed in the human (37), and GSE11121 (38). From these, two sets of ER-positive, luminal breast cancer cell line ZR-75-1 that expressed the tamoxifen-treated patients (n ¼ 250, ref. 35; n ¼ 134, ref. 36), murine ecotropic receptor. Cells were infected with retroviral and two sets of ER-positive–untreated patients (n ¼ 209, ref. 37; supernatants containing a selection of the NKi pRS–shRNA n ¼ 158, ref. 38) were extracted, for which follow-up was library (12,540 shRNA vectors targeting 4,180 genes) or pRS as available. ER status for the Schmidt and colleagues data (38) control (Fig. 1A; ref. 28). Library-infected cells and control is absent in the publicly available data and was determined by cells were plated at low density and cultured in DMEM with fitting a two-component Gaussian mixture to the expression of 1 mmol/L 4OH-tamoxifen for 6 weeks. Individual colonies that the ESR1 gene, which was verified to follow a bimodal distri- grew out in the presence of tamoxifen were collected, genomic bution. Probes in the Buffa and colleagues (36), Wang and DNA was isolated and shRNA cassettes were recovered by PCR. colleagues (37), and Schmidt and colleagues (38) data were These shRNA cassettes were subsequently recloned and median-centered before further processing; the Loi and col- sequenced. This led to the identification of USP9X. To confirm leagues data had already been median-centered. The 526 genes that knockdown of USP9X was responsible for the rescue of the of the USP9X knockdown tamoxifen signature were mapped to tamoxifen-induced proliferation arrest, ZR-75-1 cells were the corresponding microarray platforms by selecting all probes infected with high titer shUSP9X and control virus. Prolifera- for matching genes, and ignoring genes not present on the tion in the presence of estradiol (E2) or tamoxifen (4-OHT) was array. For the Loi and colleagues (35) data, these selected 949 determined in a colony formation assay (Fig. 1B). To exclude probe sets represent 488 different genes. For the Buffa and that the escape from tamoxifen-induced proliferation arrest colleagues (36) data, 363 probes were selected representing 295 was the result of an "off-target effect," five additional shRNAs genes and for both the Wang and colleagues (37) and the targeting different regions of the USP9X gene were desi- Schmidt and colleagues (38) datasets, 653 probe sets repre- gned. Figure 1C shows that three shRNAs had the identical senting 391 genes were available. Of note, 254 of the signature phenotype: Infected cells grew out despite tamoxifen treat- genes were present on all three array platforms. Patients were ment. Importantly, only the vectors that suppressed USP9X stratified into two groups by applying a hierarchical complete- mRNA (Fig. 1D) and protein levels (Fig. 1E) induced tamoxifen linkage clustering using Pearson correlation distance, and resistance. To ask whether the rescue from tamoxifen-induced dividing by the first split of the clustering. Significant differ- proliferation arrest is independent of cellular context, we ences in distant metastasis-free survival (DMFS) time between tested two USP9X shRNA vectors for their ability to confer these two groups were tested for using the log-rank test. tamoxifen resistance in a second luminal breast cancer cell Survival times longer than 10 years were right-censored. The line: T47D. Knocking down USP9X in T47D cells enabled cell array platform used for the untreated Wang and colleagues (37) growth in the presence of tamoxifen as well, suggesting that and Schmidt and colleagues (38) datasets provides a subset of USP9X suppression leads to tamoxifen resistance independent the probes available for the treated Loi and colleagues data of the cellular context (Supplementary Fig. S1). Importantly, (653 of 949; ref. 35). To verify that this difference does not affect knockdown of USP9X did not rescue cells from a proliferation

3812 Cancer Res; 74(14) July 15, 2014 Cancer Research

Figure 1. The shRNA screen identifies USP9X involvement in tamoxifen resistance. A, schematic outline of the screen. ZR-75-1 cells stably expressing murine ecotropic receptor were infected with retroviral supernatants containing a selection of the NKi pRS–shRNA library or pRS as control. After selection cells were cultured in DMEM with 1 mmol/L 4OH- tamoxifen for 6 weeks. Tamoxifen- resistant individual colonies were isolated, identifying an shRNA- targeting USP9X. B, knockdown of USP9X rescues tamoxifen- induced growth arrest. ZR-75-1 cells were infected with the USP9X shRNA recovered from the initial screen or pRS–GFP as control. Cells were cultured for 6 weeks in the presence of 1 mmol/L 4OH- tamoxifen or 1 nmol/L estradiol, photographed, fixed, and stained. Plate confluency was determined by calculating the number of cell-covered pixels related to the total surface area. C, USP9X hit validation. ZR-75-1 cells were infected with five independent shRNAs, targeting different regions of the USP9X gene and grown in the presence of 4OH-tamoxifen. Rescue from tamoxifen-induced growth arrest by USP9X knockdown was validated by three independent shRNAs. D, knockdown of USP9X decreases USP9X mRNA levels. E, knockdown of USP9X decreases USP9X protein levels. F, knockdown of USP9X does not rescue fulvestrant-induced growth arrest. ZR-75-1 cells were infected with shUSP9X or shGFP as control. Cells were cultured for 3 weeks in the presence of 1 mmol/L 4OH-tamoxifen or 10 7 mol/L fulvestrant. When colonies appeared,cells were photographed, fixed, and stained. Cell number was measured by crystal violet extraction.

arrest induced by the ER downregulator fulvestrant, illustrat- estrogen-responsive luciferase reporter (ERE-luciferase). ing that shUSP9X effects on cell proliferation are ERa-depen- First, we tested whether knockdown of USP9X increased dent (Fig. 1F). ERa activity under the conditions used in the shRNA screen. Figure 2A shows that USP9X knockdown (USP9XKD) Knockdown of USP9X increases ERa activity through cells have increased luciferase activity, both when cultured direct binding in normal culture media (DMEM/FCS) and when cultured Next, we examined whether the rescue from tamoxifen- in the presence of 4OH-tamoxifen. Next, we performed induced proliferation arrest was the result of increased ERa luciferase assays on cells grown in phenol red-free DMEM, signaling. ZR-75-1 cells stably expressing pRS–USP9X supplemented with 10% charcoal-stripped (and hence or control pRS–GFP were transiently transfected with an steroid-free) serum in the presence of vehicle, estradiol,

www.aacrjournals.org Cancer Res; 74(14) July 15, 2014 3813

Figure 2. Knockdown of USP9X increases ERa activity. A, USP9XKD cells show increased ERE luciferase reporter activity in serum-supplemented DMEM in the absence and presence of 4OH-tamoxifen. Data are representative of three independent experiments. B, knockdown of USP9X increases ERE luciferase activity in hormone-deprived, estradiol- and 4OH-tamoxifen–treated cells. Data are representative of three independent experiments. C, USP9X knockdown in the presence of estradiol increases mRNA levels of the ERa target genes PGR, TFF1, and ERa. Data are representative of three independent experiments. D, knockdown of USP9X increases ERa and PR protein levels in hormone-deprived, estradiol- or 4OH-tamoxifen– treated cells.

4OH-tamoxifen, or a combination of estradiol þ 4OH- physiologic conditions, which was recently also shown using tamoxifen. Figure 2B shows that under all these conditions mass spectrometry by Stanisic and colleagues (43). ERa signaling is about 2.5 times higher in the USP9XKD cell line as compared with the control cell line. To conﬁrm the stim- USP9X loss selectively enhances ERa/chromatin ulating effect of USP9X knockdown on ERa signaling, ZR-75-1 interactions upon 4OH-tamoxifen treatment cells, stably expressing pRS–USP9X or pRS–GFP, cultured Knockdown of USP9X gave rise to both tamoxifen resis- in the presence of indicated ligand, were analyzed for ERa tance and ERa-responsive gene activation. Next, the effects of target gene expression. As shown in Fig. 2C and D, knockdown USP9X knockdown on ERa/chromatin interactions were test- of USP9X resulted in increased mRNA (Fig. 2C) and protein ed for hormone-depleted (vehicle), estradiol and tamoxifen levels (Fig. 2D) of PR, Trefoil factor 1 (TFF1/PS2), and of ERa conditions, using ChIP, followed by high-throughput sequenc- itself (42). ing (ChIP-seq). ZR-75-1 cells stably expressing pRS–USP9X or Given the functional interaction between USP9X and ERa,we pRS–GFP (control) were plated in hormone-depleted medium next tested whether ERa and USP9X physically interact. Figure for 72 hours. Typically, ERa ChIP-seq experiments are per- 3 shows that ERa coimmunoprecipitates with USP9X in estra- formed after a treatment for 45 minutes with ligand (44, 45), diol- and tamoxifen-treated ZR-75-1 cells, demonstrating the even though 3 hours of treatment is also used (46, 47). Because existence of a physical complex of these proteins under USP9X suppression causes long-term resistance to tamoxifen,

3814 Cancer Res; 74(14) July 15, 2014 Cancer Research

USP9XKD cells as well as for the sites selectively induced by USP9X knockdown (Fig. 4G). Collectively, these data show that USP9X knockdown induces ERa-binding events, selectively in the presence of 4OH-tamoxifen, that represent a subset of estradiol-induced sites and do not deviate from normal ERa behavior with respect to genomic distributions and DNA motif enrichment.

USP9X and global gene-expression analyses Our ChIP-seq analyses indicate that USP9X knockdown selectively enhances ERa/chromatin interactions in the pres- Figure 3. Physical interaction between USP9X and ERa. Hormone- ence of tamoxifen that are normally found enriched for estra- deprived ZR-75-1 cells were treated with either E2 or 4-OHT. USP9X Immunoprecipitations were performed with nonimmune serum (ni, lanes diol conditions. We, therefore, tested whether knock- 3 and 6), anti-ERa (lanes 4 and 7), or anti-USP9X (lanes 5 and 8) down in tamoxifen-treated cells would also give rise to a typical antibodies. Western blots were incubated with ERa and USP9X estradiol-responsive gene set. To address this, we performed antibodies. Lanes 1 and 2 show 5% input of the whole-cell lysate. RNA-seq on ZR-75-1 cells stably expressing pRS–USP9X or pRS–GFP (control) that—after hormone depletion for we studied ERa biology after continued ligand treatment 72 hours—were treated for 48 hours with vehicle, estradiol, and the effects of USP9X knockdown thereon. Therefore, the or 4OH-tamoxifen. Estradiol treatment resulted in altered cells were treated with vehicle, estradiol or 4OH-tamoxifen for expression of 8,794 genes as compared with vehicle, whereas 48 hours before the ChIP assay. In control cells, estradiol after 4OH-tamoxifen treatment 1,906 genes were differentially treatment greatly enhanced ERa/chromatin interactions, expressed. All altered transcripts under 4OH-tamoxifen con- although this was far less pronounced when treating the ditions represented a subset of the estradiol-responsive genes cells with 4OH-tamoxifen. USP9X knockdown had no effect (Fig. 5A, left). 4OH-tamoxifen treatment in USP9XKD cells as on ERa/chromatin interactions in vehicle- and estradiol-trea- compared with 4OH-tamoxifen–treated control cells resulted ted cells, but significantly increased chromatin binding upon in an altered expression of 6,210 transcripts, 4,336 of which 4OH-tamoxifen treatment, as exemplified in Fig. 4A. Enhanced were shared with estradiol induction in control cells (Fig. 5A, ERa/chromatin interactions under tamoxifen conditions right). Furthermore, integrating these differentially expressed were also observed when cells were treated for 45 minutes genes in 4OH-tamoxifen–treated USP9XKD cells with the ChIP- (Supplementary Fig. S2). The stabilization of ERa/chromatin seq data showed that a subgroup of these genes, 526 of 4,336 interactions in the presence of 4OH-tamoxifen could be gen- genes (Fig. 5B), is enriched for proximal ERa-binding events, eralized throughout the genome, as illustrated by visualizing with a chromatin-binding event within 20 kb from the TSS. This the union of all peaks found for shGFP (5,855 sites) and window of 20 kb represents the optimal window to identify shUSP9X (6,828 sites) in a heatmap (Fig. 4B) and expressed ERa-responsive genes (48). This is a selective enrichment over in a quantified format in a 2D graph (Fig. 4C). This increased the total genomic background, in which 3,001 of all 45,054 intensity of ERa/chromatin interactions in 4OH-tamoxifen– RefSeq genes were found to have a proximal ERa-binding event treated cells also translated into a significant increase in the (Fisher exact test; P ¼ 8.099E30). Analyzing the raw read- number of chromatin-binding events, representing a subset count of the ChIP-seq experiments under 4OH-tamoxifen of the estradiol-induced binding patterns under the same showed that also the ChIP-seq signal intensity of proximal conditions (Fig. 4D). The subgroups of binding sites for control ERa-binding events was selectively increased after USP9X cells and USP9XKD cells, either or not shared between ligand knockdown (Fig. 5C). conditions (Fig. 4D), were separately analyzed in a heatmap visualization and quantified (Supplementary Fig. S3). A USP9X knockdown tamoxifen gene-expression Comparing control with USP9XKD under the same ligand signature identifies patients with breast cancer with a conditions showed gained sites both for estradiol and 4OH- poor outcome after adjuvant tamoxifen treatment tamoxifen conditions, whereas this was not the case for The RNA-seq analyses revealed that the majority of genes vehicle-treated cells (Fig. 4E). ERa rarely binds promoters that were differentially expressed upon tamoxifen treatment in (5%), and the vast majority of ERa-binding events is found at the USP9XKD cells represented a subgroup of estradiol-induced distal enhancers (44). We could confirm these data for estradiol genes (4,336 of 8,794 genes; see Fig. 5A, right), of which 526 of and 4OH-tamoxifen conditions, both in control and USP9XKD 4,336 genes with a proximal ERa-binding site (Fig. 5B) are cells (Fig. 4F). Vehicle-treated cells showed enrichment of ERa expected to be under the direct control of ERa under 4OH- binding to promoters as was found before (46), which was not tamoxifen conditions. This particular subgroup of genes most influenced by knockdown of USP9X. The gained ERa-binding likely represents a direct ERa target gene signature in contrast events for USP9XKD cells under tamoxifen conditions showed with the (potentially indirectly regulated) genes that were not identical distributions as found for estradiol- and tamoxifen- enriched for ERa binding. Because these directly ERa-regu- treated control cells. De novo DNA motif enrichment analyses lated genes would also be the genes that are directly affected for ERa-binding sites in 4OH-tamoxifen–treated cells provided under tamoxifen-resistant conditions, differential expression ESR motifs, both for sites shared between control cells and of these particular genes in breast tumors could hallmark

www.aacrjournals.org Cancer Res; 74(14) July 15, 2014 3815

Figure 4. USP9X loss selectively enhances ERa/chromatin interactions upon 4OH-tamoxifen treatment. Hormone-deprived monoclonal ZR-75-1 cells, stably expressing pRS–USP9X or pRS–GFP as control, were treated with vehicle (veh), estradiol (E2), or 4OH-tamoxifen (4-OHT), after which ChIP-seq analysis was performed on ERa.A,ERa ChIP-seq signal in shGFP control cells (blue) and shUSP9X cells (red) in the presence of indicated ligand. Tag counts (y-axis) and genomic locations (x-axis) are indicated. B, heatmap visualization, depicting a vertical alignment of all identified peaks of control (shGFP, 5,855 sites, blue) and USP9XKD (shUSP9X, 6,828 sites, red) raw read counts of veh, E2, or 4-OHT–treated cells. Arrowhead, top of the peak and scale bar is indicated. C, read-count quantification of data presented in B showing enrichment of ERa/chromatin interactions in the presence of 4-OHT in the shUSP9X cells compared with the control (shGFP) cells. y-axis, average tag count (arbitrary units). x-axis, distance from center of the peak (2.5 kb, þ2.5 kb). D, Venn diagrams showing a significant increase in the number of ERa/chromatin–binding events in the shUSP9X compared with control shGFP cells in the presence of 4-OHT, representing a subset of the E2-induced binding patterns. Numbers, binding events in each subgroup (veh, blue; E2, red; 4-OHT, green). E, Venn diagrams showing shared and unique peaks for control cells (blue) and shUSP9X cells (red) under vehicle, E2, and 4-OHT conditions. Numbers, binding events in each subgroup. F, genomic distributions of peaks under all tested conditions. Locations are indicated relative to the most proximal genes. 4-OHT shUSP9X unique, unique binding sites in tamoxifen-treated shUSP9X cells compared with shGFP control cells. G, de novo motif enrichment analysis identified ESR motifs enriched for 4-OHT shUSP9X unique peaks and peaks shared between 4-OHT–treated shGFP and shUSP9X cells. A P value for ESR1 motif enrichment was 690.77 (10 log) for both situations.

3816 Cancer Res; 74(14) July 15, 2014 Cancer Research

Figure 5. USP9X and global gene expression analyses. Hormone-deprived monoclonal ZR-75-1 cells stably expressing pRS–USP9X or pRS–GFP as control were treated with vehicle (veh), estradiol (E2), or 4OH-tamoxifen (4-OHT), after which RNA-seq analysis was performed. A, left, Venn diagram showing differentially expressed genes in control cells after treatment with E2 (blue) or 4-OHT (green) compared with vehicle control (P < 0.05). Right, Venn diagram showing differentially expressed genes of E2- versus vehicle-treated control cells (blue) and differentially expressed genes of 4-OHT–treated shUSP9X cells compared with 4-OHT–treated shGFP control cells (red). B, proximal ERa-binding events for the 4,436 differentially expressed, estradiol-regulated genes in 4-OHT–treated shUSP9X cells. ERa-binding events found in 4-OHT–treated control cells (left), 4-OHT– treated shUSP9X cells (middle), or shared between both conditions (right) were analyzed for proximal binding (<20 kb) to TSSs of differentially expressed genes in 4-OHT–treated shUSP9X cells. The y-axis shows absolute number of differentially expressed genes. C, average ERa read-count intensity of ERa chromatin–binding sites in 4-OHT–treated shUSP9X cells compared with shGFP control cells, proximal to (<20 kb) TSS regions of 526 genes, differential expressed between 4-OHT–treated shUSP9X cells and 4-OHT–treated control cells. y-axis, average read-count (a.u.). x-axis, distance from center of the peak (2.5 kb, þ2.5 kb). D, heatmap showing differentially expressed genes between 250 patients with primary ERa-positive breast cancer who received adjuvant tamoxifen. E, A USP9X knockdown tamoxifen gene signature identiﬁes primary ERa-positive breast cancer patients with poor outcome after adjuvant tamoxifen treatment. Kaplan–Meier survival curves for DMFS of patients with primary ERa- positive breast cancer treated with adjuvant tamoxifen in the Loi and colleagues cohort (n = 250; ﬁrst; ref. 35) and in the Buffa and colleagues cohort (n = 134; second; ref. 36). Kaplan–Meier survival curves for DMFS of patients with primary ERa-positive breast cancer that did not receive adjuvant endocrine treatment in the Wang and colleagues cohort (n = 209; third; ref. 37) and in the Schmidt and colleagues cohort (n = 158; fourth; ref. 38). F, distribution of the PAM50 molecular subtypes over the USP9X knockdown tamoxifen gene-expression signature-sensitive and -resistant groups.

www.aacrjournals.org Cancer Res; 74(14) July 15, 2014 3817

tamoxifen unresponsiveness. To test this hypothesis, we inves- signature to identify tamoxifen-treated patients with breast tigated whether these genes were differentially expressed in a cancer with a poor response to therapy. publically available dataset of 250 predominantly postmenopausal patients with primary ERa-positive breast cancer with known outcome (35). All these patients received adjuvant Discussion tamoxifen. For all clinicopathologic parameters, see Supple- We identify here a role for USP9X in regulation of the mentary Table S3. As visualized in a heatmap (Fig. 5D), response to tamoxifen in ERa-positive breast cancer. USP9X unsupervised clustering on the basis of our gene signature affects the stability and activity of numerous regulatory pro- resulted in the identification of two distinct subgroups of teins that influence cell survival. Earlier publications had patients. These subgroups of patients were subsequently ana- attributed a prosurvival role to USP9X in B- and mantle cell lyzed for differential distant metastasis-free survival (DMFS) lymphomas, chronic myeloid leukemia, and multiple myeloma, after adjuvant tamoxifen treatment. Figure 5E, first, shows that through stabilization of MCL1 (49). USP9X can also act as a this gene set identifies a subgroup of patients with breast tumor suppressor and plays a role in oxidative stress–induced cancer with a poor response to tamoxifen treatment (P ¼ 9.4 cell death through ASK1 (apoptosis signal–regulating kinase 1; 10 5). These data could be validated using a second cohort of ref. 18). In addition, USP9X was identified as a tumor-suppres- 134 mostly postmenopausal patients with ERa-positive breast sor gene in pancreatic cancer (50). cancer treated with adjuvant tamoxifen (Fig. 5E, second; P ¼ The findings presented here establish a novel role for 7.3 10 3; ref. 36). We then tested our signature on two cohorts USP9X acting as a mediator of the response to tamoxifen in of mainly postmenopausal patients with ERa-positive breast ERa-positive breast cancer. USP9X knockdown stabilizes cancer (37, 38) who did not receive any adjuvant endocrine ERa–chromatin interactions, enables tamoxifen-induced treatment. Importantly, in these patients the USP9X knock- ERa activation, and stimulates ERa-responsive cell prolif- down tamoxifen gene-expression signature did not correlate eration in the presence of tamoxifen. A gene-expression with outcome, indicating that the gene signature is not prog- signature composed of tamoxifen-responsive genes in nostic (Fig. 5E, right). Patients identified as responsive, based shRNA USP9X cells can identify patients with breast cancer on our USP9X-based gene signature were mostly of the luminal with a poor response to tamoxifen treatment. The tumors A subtype, whereas resistant-classified patients were more identified as tamoxifen-responsive through the USP9X sig- often of the luminal B subtype, as identified by PAM50 (Fig. nature are more often of the luminal A type, whereas the 5F; refs. 40, 41). However, still approximately 30% of luminal B group with the resistant signature shows enrichment for tumors were classified as tamoxifen sensitive and approxi- luminal B-type tumors. Still our gene signature is not merely mately 30% of luminal A tumors were classified as tamoxifen providing a distinction on the basis of the molecular sub- resistant, suggesting that our gene signature can be used to type, because approximately 30% of the tumors classified as identify subgroups of luminal A and luminal B tumors that tamoxifen-resistant were luminal A and a comparable per- benefit most from adjuvant tamoxifen treatment. The gene centage of the tamoxifen-sensitive tumors were luminal B signature was selective for endocrine treatment, because our type. This finding is potentially of clinical relevance, as the signature was not able to classify patients with ERa-positive data indicate that our gene signature can identify luminal A- tumors treated with adjuvant chemotherapy only, albeit this type tumors that respond poorly to tamoxifen treatment group of patients was small (Supplementary Fig. S6; ref. 39). and might benefit from alternative treatment regimes. The majority of genes that are differentially expressed upon Importantly, USP9X knockdown did not induce fulvestrant tamoxifen treatment in the USP9XKD cells were shared with resistance, thus providing an alternative treatment option estradiol induction. However, 1,874 of the differentially for patients classified as tamoxifen-resistant through our expressed genes were not estrogen affected (Fig. 5A, right). signature. These 1,874 genes were not linked with estrogen function or Although USP9X physically interacts with ERa, as was endocrine resistance, but ingenuity pathway analysis enrich- shown before (43), we were not able to show any deubiquiti- ment and gene ontology analysis showed enrichment for nating activity of USP9X toward ERa. We hypothesize that the mitochondrial function (Supplementary Fig. S4A and S4B). role of USP9X in ERa signaling lies in regulating its interactions Expression-based gene signatures for these 1,874 genes with chromatin, potentially facilitating cofactor recruitment did not classify patients in the Buffa and colleagues cohort and/or deubiquitination of such ERa cofactors. [P ¼ 0.25; HR, 1.4; 95% confidence interval (CI), 0.8–2.4; The USP9X knockdown tamoxifen gene-expression signa- ref. 36], whereas our shUSP9X-based gene signature did ture we identify here enables the identification of patients (P ¼ 6.5 10 4; HR, 4.0; 95% CI, 1.7–9.3; Supplementary Fig. with breast cancer with a poor response to adjuvant tamox- S5A and S5B, right). Significance was reached with the Loi and ifen treatment. Because USP9X knockdown directly affects colleagues series (P ¼ 0.035; HR, 2.0; 95% CI, 1.2–3.5; Supple- the biology of ERa, other modes of tamoxifen resistance mentary Fig. S5A, left; ref. 35), which was outperformed by our might be recapitulated by the same gene signature, broad- shUSP9X-based gene signature using the same cohort, even ening the potential applicability of our findings. Important- though the confidence intervals did overlap (P ¼ 9.4 10 5; ly, evaluation of USP9X levels alone did not enable such a HR, 3.1; 95% CI, 1.9–5.2; Supplementary Fig. S5B, left). patient stratification (data not shown). This is not an Cumulatively, we found USP9X knockdown to induce tamox- unexpected finding given that USP9X has many cellular ifen resistance in cell lines, and could apply a downstream gene functions, and the identification of our distinct gene-

3818 Cancer Res; 74(14) July 15, 2014 Cancer Research

expression signature enabled us to exclusively study the Authors' Contributions consequences of USP9X loss on ERa function and tamoxifen Conception and design: H.M. Oosterkamp, E.M. Hijmans, W. Zwart, R. Bernards response. Development of methodology: H.M. Oosterkamp, E.M. Hijmans, R. Bernards Our shUSP9X-based gene signature was generated on Acquisition of data (provided animals, acquired and managed patients, in vitro provided facilities, etc.): H.M. Oosterkamp, E.M. Hijmans, T.R. Brummelkamp, tamoxifen-treated cells , and was subsequently applied W. Zwart on gene-expression data from tumor samples obtained before Analysis and interpretation of data (e.g., statistical analysis, biostatistics, endocrine therapy. This demonstrated that the signature is computational analysis): H.M. Oosterkamp, E.M. Hijmans, T.R. Brummel- in vivo kamp, S. Canisius, W. Zwart, R. Bernards capable of identifying intrinsic tamoxifen resistance . Writing, review, and/or revision of the manuscript: H.M. Oosterkamp, Future clinical studies on tumor specimens collected pre- and E.M. Hijmans, L.F.A. Wessels, W. Zwart, R. Bernards post-neoadjuvant tamoxifen treatment are required to further Administrative, technical, or material support (i.e., reporting or orga- nizing data, constructing databases): H.M. Oosterkamp, E.M. Hijmans investigate directionality of the differentially expressed genes Study supervision: E.M. Hijmans, L.F.A. Wessels before and after tamoxifen treatment in relation to treatment response. Acknowledgments In summary, we report here an unexpected role for USP9X in The authors thank Ron Kerkhoven, Iris de Rink, and Mandy Madiredjo for their assistance. ERa-positive breast cancer, as USP9X loss induces tamoxifen- stimulatory effects on ERa action, leading to tamoxifen resis- USP9X Grant Support tance. Furthermore, we show that a knockdown tamox- This work was supported by grants from the Dutch Cancer Society (KWF). ifen gene expression signature can be used as a potential The costs of publication of this article were defrayed in part by the biomarker to identify patients with ERa-positive breast cancer payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this with a poor outcome after tamoxifen treatment. fact.

Disclosure of Potential Conﬂicts of Interest Received July 11, 2013; revised April 14, 2014; accepted April 14, 2014; No potential conﬂicts of interest were disclosed. published online July 15, 2014.

References 1. Harvey JM, Clark GM, Osborne CK, Allred DC. Estrogen receptor 12. Yamnik RL, Digilova A, Davis DC, Brodt ZN, Murphy CJ, Holz MK. S6 status by immunohistochemistry is superior to the ligand-binding kinase 1 regulates estrogen receptor alpha in control of breast cancer assay for predicting response to adjuvant endocrine therapy in breast cell proliferation. J Biol Chem 2009;284:6361–9. cancer. J Clin Oncol 1999;17:1474–81. 13. Yue W, Fan P, Wang J, Li Y, Santen RJ. Mechanisms of acquired 2. EBCTCG. Effects of chemotherapy and hormonal therapy for early resistance to endocrine therapy in hormone-dependent breast breast cancer on recurrence and 15-year survival: an overview of the cancer cells. J Steroid Biochem Mol Biol 2007;106:102–10. randomised trials. Lancet 2005;365:1687–717. 14. Miller TW, Hennessy BT, Gonzalez-Angulo AM, Fox EM, Mills GB, 3. Musgrove EA, Sutherland RL. Biological determinants of endocrine Chen H, et al. Hyperactivation of phosphatidylinositol-3 resistance in breast cancer. Nat Rev Cancer 2009;9:631–43. kinase promotes escape from hormone dependence in estrogen 4. Michalides R, Griekspoor A, Balkenende A, Verwoerd D, Janssen L, receptor-positive human breast cancer. J Clin Invest 2010;120: Jalink K, et al. Tamoxifen resistance by a conformational arrest of the 2406–13. estrogen receptor alpha after PKA activation in breast cancer. Cancer 15. Merenbakh-Lamin K, Ben-Baruch N, Yeheskel A, Dvir A, Soussan- Cell 2004;5:597–605. Gutman L, Jeselsohn R, et al. D538G mutation in estrogen receptor- 5. Campbell RA, Bhat-Nakshatri P, Patel NM, Constantinidou D, Ali S, alpha: a novel mechanism for acquired endocrine resistance in breast Nakshatri H. Phosphatidylinositol 3-kinase/AKT-mediated activation cancer. Cancer Res 2013;73:6856–64 of estrogen receptor alpha: a new model for anti-estrogen resistance. 16. Salmena L, Pandolfi PP. Changing venues for tumour suppression: J Biol Chem 2001;276:9817–24. balancing destruction and localization by monoubiquitylation. Nat Rev 6. Beelen K, Zwart W, Linn SC. Can predictive biomarkers in breast Cancer 2007;7:409–13. cancer guide adjuvant endocrine therapy? Nat Rev Clin Oncol 2012; 17. Nijman SM, Luna-Vargas MP, Velds A, Brummelkamp TR, Dirac AM, 9:529–41. Sixma TK, et al. A genomic and functional inventory of deubiquitinating 7. Iorns E, Turner NC, Elliott R, Syed N, Garrone O, Gasco M, et al. enzymes. Cell 2005;123:773–86. Identification of CDK10 as an important determinant of resistance 18. Nagai H, Noguchi T, Homma K, Katagiri K, Takeda K, Matsuzawa A, to endocrine therapy for breast cancer. Cancer Cell 2008;13:91–104. et al. Ubiquitin-like sequence in ASK1 plays critical roles in the 8. Ahn BY, Elwi AN, Lee B, Trinh DL, Klimowicz AC, Yau A, et al. Genetic recognition and stabilization by USP9X and oxidative stress-induced screen identifies insulin-like growth factor binding protein 5 as a cell death. Mol Cell 2009;36:805–18. modulator of tamoxifen resistance in breast cancer. Cancer Res 2010; 19. Murray RZ, Jolly LA, Wood SA. The FAM deubiquitylating enzyme 70:3013–9. localizes to multiple points of protein trafficking in epithelia, where it 9. Giamas G, Filipovic A, Jacob J, Messier W, Zhang H, Yang D, et al. associates with E-cadherin and beta-catenin. Mol Biol Cell 2004;15: Kinome screening for regulators of the estrogen receptor identifies 1591–9. LMTK3 as a new therapeutic target in breast cancer. Nat Med 2011; 20. Taya S, Yamamoto T, Kanai-Azuma M, Wood SA, Kaibuchi K. The 17:715–9. deubiquitinating enzyme Fam interacts with and stabilizes beta-cate- 10. Pan X, Zhou T, Tai YH, Wang C, Zhao J, Cao Y, et al. Elevated nin. Genes Cells 1999;4:757–67. expression of CUEDC2 protein confers endocrine resistance in breast 21. Taya S, Yamamoto T, Kano K, Kawano Y, Iwamatsu A, Tsuchiya T, cancer. Nat Med 2011;17:708–14. et al. The Ras target AF-6 is a substrate of the fam deubiquitinating 11. Magnani L, Stoeck A, Zhang X, Lanczky A, Mirabella AC, Wang TL, enzyme. J Cell Biol 1998;142:1053–62. et al.Genome-wide reprogramming of the chromatin landscape under- 22. Theard D, Labarrade F, Partisani M, Milanini J, Sakagami H, Fon EA, lies endocrine therapy resistance in breast cancer. Proc Natl Acad Sci et al. USP9x-mediated deubiquitination of EFA6 regulates de novo U S A 2013;110:E1490–9. tight junction assembly. EMBO J 2010;29:1499–509.

www.aacrjournals.org Cancer Res; 74(14) July 15, 2014 3819

23. Al-Hakim AK, Zagorska A, Chapman L, Deak M, Peggie M, Alessi DR. 37. Wang Y, Klijn JG, Zhang Y, Sieuwerts AM, Look MP, Yang F, et al. Control of AMPK-related kinases by USP9X and atypical Lys(29)/Lys Gene-expression profiles to predict distant metastasis of lymph-node- (33)-linked polyubiquitin chains. Biochem J 2008;411:249–60. negative primary breast cancer. Lancet 2005;365:671–9. 24. Vong QP, Cao K, Li HY, Iglesias PA, Zheng Y. Chromosome alignment 38. Schmidt M, Bohm D, von Torne C, Steiner E, Puhl A, Pilch H, et al. The and segregation regulated by ubiquitination of survivin. Science humoral immune system has a key prognostic impact in node-negative 2005;310:1499–504. breast cancer. Cancer Res 2008;68:5405–13. 25. Choe EA, Liao L, Zhou JY, Cheng D, Duong DM, Jin P, et al. Neuronal 39. van de Vijver MJ, He YD, van't Veer LJ, Dai H, Hart AA, Voskuil DW, et al. morphogenesis is regulated by the interplay between cyclin-depen- A gene-expression signature as a predictor of survival in breast cancer. dent kinase 5 and the ubiquitin ligase mind bomb 1. J Neurosci 2007; N Engl J Med 2002;347:1999–2009. 27:9503–12. 40. Parker JS, Mullins M, Cheang MC, Leung S, Voduc D, Vickery T, et al. 26. Dupont S, Mamidi A, Cordenonsi M, Montagner M, Zacchigna L, Supervised risk predictor of breast cancer based on intrinsic subtypes. Adorno M, et al. FAM/USP9x, a deubiquitinating enzyme essential for J Clin Oncol 2009;27:1160–7. TGFbeta signaling, controls Smad4 monoubiquitination. Cell 2009; 41. Ur-Rehman S, Gao Q, Mitsopoulos C, Zvelebil M. ROCK: a resource for 136:123–35. integrative breast cancer data analysis. Breast Cancer Res Treat 27. Agrawal P, Chen YT, Schilling B, Gibson BW, Hughes RE. Ubiqui- 2013;139:907–21. tin-specific peptidase 9, X-linked (USP9X) modulates activity of 42. Eeckhoute J, Keeton EK, Lupien M, Krum SA, Carroll JS, Brown M. mammalian target of rapamycin (mTOR). J Biol Chem 2012;287: Positive cross-regulatory loop ties GATA-3 to estrogen receptor alpha 21164–75. expression in breast cancer. Cancer Res 2007;67:6477–83. 28. Berns K, Hijmans EM, Mullenders J, Brummelkamp TR, Velds A, 43. Stanisic V, Malovannaya A, Qin J, Lonard DM, O'Malley BW. OTU Heimerikx M, et al. A large-scale RNAi screen in human cells identifies Domain-containing ubiquitin aldehyde-binding protein 1 (OTUB1) new components of the p53 pathway. Nature 2004;428:431–7. deubiquitinates estrogen receptor (ER) alpha and affects ERalpha 29. Trapnell C, Pachter L, Salzberg SL. TopHat: discovering splice junc- transcriptional activity. J Biol Chem 2009;284:16135–45. tions with RNA-Seq. Bioinformatics 2009;25:1105–11. 44. Carroll JS, Liu XS, Brodsky AS, Li W, Meyer CA, Szary AJ, et al. 30. Wang L, Feng Z, Wang X, Wang X, Zhang X. DEGseq: an R package for Chromosome-wide mapping of estrogen receptor binding reveals identifying differentially expressed genes from RNA-seq data. Bioin- long-range regulation requiring the forkhead protein FoxA1. Cell formatics 2010;26:136–8. 2005;122:33–43. 31. Schmidt D, Wilson MD, Spyrou C, Brown GD, Hadfield J, Odom DT. 45. Hurtado A, Holmes KA, Ross-Innes CS, Schmidt D, Carroll JS. FOXA1 ChIP-seq: using high-throughput sequencing to discover protein-DNA is a key determinant of estrogen receptor function and endocrine interactions. Methods 2009;48:240–8. response. Nat Genet 2011;43:27–33. 32. Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, 46. Zwart W, Theodorou V, Kok M, Canisius S, Linn S, Carroll JS. Oestro- et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol 2008; gen receptor-co-factor-chromatin specificity in the transcriptional 9:R137. regulation of breast cancer. EMBO J 2011;30:4764–76. 33. He HH, Meyer CA, Shin H, Bailey ST, Wei G, Wang Q, et al. Nucleosome 47. Theodorou V, Stark R, Menon S, Carroll JS. GATA3 acts upstream of dynamics define transcriptional enhancers. Nat Genet 2010;42:343–7. FOXA1 in mediating ESR1 binding by shaping enhancer accessibility. 34. Ji X, Li W, Song J, Wei L, Liu XS. CEAS: cis-regulatory element Genome Res 2013;23:12–22. annotation system. Nucleic Acids Res 2006;34:W551–4. 48. Fullwood MJ, Liu MH, Pan YF, Liu J, Xu H, Mohamed YB, et al. An 35. Loi S, Haibe-Kains B, Desmedt C, Lallemand F, Tutt AM, Gillet C, et al. oestrogen-receptor-alpha-bound human chromatin interactome. Definition of clinically distinct molecular subtypes in estrogen recep- Nature 2009;462:58–64. tor–positive breast carcinomas through genomic grade. J Clin Oncol 49. Schwickart M, Huang X, Lill JR, Liu J, Ferrando R, French DM, et al. 2007;25:1239–46. Deubiquitinase USP9X stabilizes MCL1 and promotes tumour cell 36. Buffa FM, Camps C, Winchester L, Snell CE, Gee HE, Sheldon H, et al. survival. Nature 2010;463:103–7. microRNA-associated progression pathways and potential therapeu- 50. Perez-Mancera PA, Rust AG, van der Weyden L, Kristiansen G, Li A, tic targets identified by integrated mRNA and microRNA expression Sarver AL, et al. The deubiquitinase USP9X suppresses pancreatic profiling in breast cancer. Cancer Res 2011;71:5635–45. ductal adenocarcinoma. Nature 2012;486:266–70.

3820 Cancer Res; 74(14) July 15, 2014 Cancer Research

Hendrika M. Oosterkamp, E. Marielle Hijmans, Thijn R. Brummelkamp, et al.

Cancer Res 2014;74:3810-3820.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/74/14/3810

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2015/01/06/74.14.3810.DC1

Cited articles This article cites 50 articles, 21 of which you can access for free at: http://cancerres.aacrjournals.org/content/74/14/3810.full#ref-list-1

Citing articles This article has been cited by 6 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/74/14/3810.full#related-urls

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/74/14/3810. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.