HES6 drives a critical AR transcriptional programme to induce castration-resistant prostate cancer through activation of an E2F1- mediated cell cycle network Ramos-Montoya, A., Lamb, A. D., Russell, R., Carroll, T., Jurmeister, S., Galeano-Dalmau, N., Massie, C. E., Boren, J., Bon, H., Theodorou, V., Vias, M., Shaw, G. L., Sharma, N. L., Ross-Adams, H., Scott, H. E., Vowler, S. L., Howat, W. J., Warren, A. Y., Wooster, R. F., ... Neal, D. E. (2014). HES6 drives a critical AR transcriptional programme to induce castration-resistant prostate cancer through activation of an E2F1-mediated cell cycle network. EMBO Molecular Medicine, 6(5), 651-61. https://doi.org/10.1002/emmm.201303581 Published in: EMBO Molecular Medicine

Document Version: Publisher's PDF, also known as Version of record

Queen's University Belfast - Research Portal: Link to publication record in Queen's University Belfast Research Portal

Publisher rights © The Authors, 2014 This is an open access article published under a Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution and reproduction in any medium, provided the author and source are cited.

General rights Copyright for the publications made accessible via the Queen's University Belfast Research Portal is retained by the author(s) and / or other copyright owners and it is a condition of accessing these publications that users recognise and abide by the legal requirements associated with these rights.

Take down policy The Research Portal is Queen's institutional repository that provides access to Queen's research output. Every effort has been made to ensure that content in the Research Portal does not infringe any person's rights, or applicable UK laws. If you discover content in the Research Portal that you believe breaches copyright or violates any law, please contact [email protected].

Download date:28. Sep. 2021 Published online: April 14, 2014

Research Article

HES6 drives a critical AR transcriptional programme to induce castration-resistant prostate cancer through activation of an E2F1-mediated cell cycle network

Antonio Ramos-Montoya1,§, Alastair D Lamb1,2,*,§, Roslin Russell3, Thomas Carroll4, Sarah Jurmeister1, Nuria Galeano-Dalmau1, Charlie E Massie1, Joan Boren3, Helene Bon1, Vasiliki Theodorou3,†, Maria Vias3, Greg L Shaw1,2, Naomi L Sharma1,2,#, Helen Ross-Adams1, Helen E Scott1, Sarah L Vowler4, William J Howat5, Anne Y Warren1,6, Richard F Wooster7,‡, Ian G Mills1,8,9,¶ & David E Neal1,2,10,**,¶

Abstract Subject Categories Cancer; Urogenital System DOI 10.1002/emmm.201303581 | Received 16 October 2013 | Revised 6 March Castrate-resistant prostate cancer (CRPC) is poorly characterized 2014 | Accepted 10 March 2014 | Published online 14 April 2014 and heterogeneous and while the androgen receptor (AR) is of EMBO Mol Med (2014) 6: 651–661 singular importance, other factors such as c-Myc and the E2F family also play a role in later stage disease. HES6 is a transcription co-factor associated with stem cell characteristics in neural tissue. Introduction Here we show that HES6 is up-regulated in aggressive human prostate cancer and drives castration-resistant tumour growth in Prostate cancer is the most common non-cutaneous malignancy in the absence of ligand binding by enhancing the transcriptional men in western countries (CRUK, 2010). Death results from the activity of the AR, which is preferentially directed to a regulatory development of castrate-resistant prostate cancer (CRPC). Several network enriched for transcription factors such as E2F1. In the mechanisms have been proposed to explain how this aggressive clinical setting, we have uncovered a HES6-associated signature phase of the disease develops. Most depend on continued androgen that predicts poor outcome in prostate cancer, which can be phar- receptor (AR) signalling in the context of enhanced secondary macologically targeted by inhibition of PLK1 with restoration of pathways, while others are independent of conventional AR activity sensitivity to castration. We have therefore shown for the first (Pienta & Smith, 2005; Lamb et al, 2014). This heterogeneity time the critical role of HES6 in the development of CRPC and explains why, although virtually all primary locally advanced pros- identified its potential in patient-specific therapeutic strategies. tate cancer can be treated effectively with androgen deprivation therapy (ADT), CRPC remains difficult to treat, and those therapies Keywords androgen receptor; castrate-resistant prostate cancer; gene that have been developed are only effective for a limited duration in expression signature; HES6; PLK1 a limited set of patients (Yap et al, 2011). It is therefore critical that

1 Uro-Oncology Research Group, Cancer Research UK Cambridge Institute, University of Cambridge, Li Ka Shing Centre, Cambridge, UK 2 Department of Urology, Addenbrooke’s Hospital, Cambridge, UK 3 Cancer Research UK Cambridge Institute, University of Cambridge, Li Ka Shing Centre, Cambridge, UK 4 Bioinformatics Core Facility, Cancer Research UK Cambridge Institute, University of Cambridge, Li Ka Shing Centre, Cambridge, UK 5 Histopathology/ISH Core Facility, Cancer Research UK Cambridge Institute, University of Cambridge, Li Ka Shing Centre, Cambridge, UK 6 Department of Pathology, Addenbrooke’s Hospital, Cambridge, UK 7 Cancer Metabolism Drug Discovery, Oncology R&D, Collegeville, PA, USA 8 Prostate Cancer Research Group, Nordic EMBL Partnership, Centre for Molecular Medicine Norway (NCMM), University of Oslo, Oslo, Norway 9 Departments of Cancer Prevention and Urology, Institute of Cancer Research and Oslo University Hospitals, Oslo, Norway 10 Department of Oncology, University of Cambridge, Cambridge, UK *Corresponding author. Tel: +44 1223 331940; Fax: +44 1223 769007; E-mail: [email protected] **Corresponding author. Tel: +44 1223 331940; Fax: +44 1223 769007; E-mail: [email protected] §These authors contributed equally to this work. ¶These senior authors contributed equally to this work. †Current address: Institute of Molecular Biology and Biotechnology (IMBB), Foundation for Research and Technology-Hellas (IMBB-FORTH), Heraklion, Crete, Greece ‡Current address: Blend Therapeutics, Watertown, MA, USA #Current address: Nuffield Department of Surgical Sciences, University of Oxford, Roosevelt Drive, Oxford, UK

ª 2014 The Authors. Published under the terms of the CC BY 4.0 license EMBO Molecular Medicine Vol 6 |No5 | 2014 651 Published online: April 14, 2014 EMBO Molecular Medicine HES6 drives prostate cancer progression Antonio Ramos-Montoya et al

we understand more about the driving processes in CRPC, so that 2006), C4-2 and C4-2b cells (Thalmann et al, 1994) as models of patient-specific mechanisms can be targeted. CRPC, we found evidence of significant increases in HES6 transcript Various oncogenic factors have been implicated in the natural levels in the C4-2 derivatives compared to LNCaP controls (Supple- history of prostate cancer. The AR has generally been accepted as mentary Fig S2A). To test the functional relevance of HES6 in the the pre-eminent driving protein in prostate cancer and, even in transition to CRPC, we stably overexpressed HA-tagged HES6 in hormone resistant prostate cancer, it is still implicated in several androgen-sensitive prostate cancer LNCaP and DuCaP cells, which escape mechanisms based on a non-dihydrotestosterone-bound-AR induced reduced sensitivity to bicalutamide (Supplementary Fig (Pienta & Smith, 2005; Lamb et al, 2014). c-Myc has been found S2B–D). We then proceeded to surgical castration using luciferase- to be amplified in more than half of refractory prostate cancers labelled LNCaP cells (LNCaP-LM) subcutaneously injected into the (Bernard et al, 2003; Clegg et al, 2011) and has been identified as flanks of NOD-SCID gamma (NSG) mice. Graft size was monitored one of the key players in cell potency (Takahashi et al, 2007), but by bioluminescent imaging over 3 months. Overexpression of HES6 the factors which mediate its downstream effects in late-stage was sufficient to maintain normal tumour growth (Fig 1C) after prostate cancer are still not well understood. Deregulation of E2F castration. Control grafts in castrated conditions were small and activity and early cell cycle progression has also been implicated in pale, while the comparable HES6 grafts were large and vascular aggressive prostate cancer (Sharma et al, 2010), but the precise role (Fig 1D). Knock-down of HES6 by lentiviral shRNA in C4-2b cells of E2Fs in cancer is dependent on a balanced interplay with other (Supplementary Fig S2E) led to significant attenuation of tumour regulatory factors and these interactions in prostate cancer need to growth in both full and castrated conditions (Fig 1E and Supplemen- be elucidated. tary Fig S2F). These data show that HES6 is sufficient to produce a The study of transcription factors in cell pluripotency has shown castration-resistant phenotype. that a small number of master regulator transcription factors can radically change cell phenotype (Takahashi et al, 2007). HES6 has Nuclear AR and cell proliferation are maintained in HES6 grafts been identified as a regulator of stem cell fate, specifically in neuro- in castrate conditions genesis (Kageyama et al, 2005; Murai et al, 2011), myogenesis (Malone et al, 2011) and gastrulation (Murai et al, 2007, 2011). In We next assessed intensity of nuclear AR and cell proliferation by addition, it has been described as a marker of the neuroendocrine immunohistochemical analysis of the LNCaP grafts. Normally, the phenotype in prostate cancer (Vias et al, 2008) and in metastatic AR shuttles in and out of the nucleus according to the extent of breast cancer (Hartman et al, 2009). Most recently, work demon- ligand activation, and following removal of testosterone, the AR strating the cooperation of HIF-1a and FoxA2 in metastatic neuro- localizes to the cytoplasm (Azzi et al, 2006; Cutress et al, 2008). We endocrine prostate cancer has identified HES6 as an important factor demonstrated cytoplasmic relocation of the AR in the native mouse in the development of these tumours (Qi et al, 2010). prostate (Supplementary Fig S3A) and in control grafts following Here, we evaluate the role of HES6 in CRPC and propose a mech- castration. By contrast, the AR was predominantly nuclear in anism of action in which HES6 induces persistent AR signalling and castrated HES6 xenografts (Supplementary Fig S3A, quantified enhanced E2F1-mediated cell cycle activity. In the clinical setting, in Supplementary Fig S3B), suggesting a role for nuclear AR in we describe a HES6-associated gene signature that correlates with HES6-driven castration-resistant grafts. Staining using the prolifera- poor outcome and outline a possible therapeutic strategy to target tion marker Ki67 showed uniform increases in the castrated this programme of resistance. HES6-overexpressing tissue (Supplementary Fig S3A, quantified in Supplementary Fig S3C).

Results HES6 rescues AR activity at a subset of Androgen Receptor Binding Sites (ARBS) which is also enhanced in CRPC HES6 is regulated by AR and c-Myc In order to explain our discovery of nuclear AR with increased levels Using the androgen-dependent LNCaP cell line and chromatin of HES6 in castration, we measured AR binding to chromatin by immunoprecipitation sequencing (ChIPseq), we showed that two ChIPseq. We found during bicalutamide treatment that HES6 was potent oncogenes in prostate cancer, the AR (Massie et al, 2011) able to maintain many AR binding sites that were lost in the EV and c-Myc, bind upstream of the HES6 coding sequence (Supple- control cells. In such inhibitory conditions, we were able to divide mentary Fig S1A) to positively regulate levels of HES6 mRNA AR binding sites (ARBS) into three distinct classes: first, those peaks (Fig 1A, Supplementary Fig S1B and C) and confirmed that overex- that were lost during AR inhibition in both EV and HES6 cells were pression of c-Myc overcomes growth inhibition by bicalutamide defined as “lost”; second, those that disappeared in EV during bica- (Bernard et al, 2003) (Supplementary Fig S1D), an effect that can be lutamide treatment but were clearly present with HES6 overexpres- reversed by knock-down of HES6 (Supplementary Fig S1E). sion were defined as “rescued”; third, those that were never totally lost in the EV control condition with bicalutamide but were clearly HES6 initiates androgen-independent growth enhanced by the presence of high HES6 were defined as “enhanced” (Fig 2A, Supplementary Table S1). Of these 3 groups, we observed HES6 levels are found to be raised in several independent clinical the strongest difference in average peak height between HES6 and datasets comparing aggressive disease (metastases or CRPC) with EV in those defined as “rescued” (Fig 2B). This included well- primary tumour and benign disease (Fig 1B). Using the bicaluta- known AR responsive genes such as KLK2 and KLK3 (Fig 2C, mide/castration-resistant derivatives LNCaP-Bic (Hobisch et al, Supplementary Fig S4A). We made comparison with ChIPseq data

652 EMBO Molecular Medicine Vol 6 |No5 | 2014 ª 2014 The Authors Published online: April 14, 2014 Antonio Ramos-Montoya et al HES6 drives prostate cancer progression EMBO Molecular Medicine

A B Transcript Expression Dataset 4 HES6 mRNA 4 HES6 mRNA Grasso: 2495 Taylor: 15132 Tomlins: 14044 Varambally: 226446_at ) ) HES6 mRNA ● 3 CT CT * ΔΔ ΔΔ *# 2 2 ● ● groups 2 1 ● Benign 0 Primary Zscore Metastatic −1 /CRPC 1 1 ● CS CS EV Myc −2 ++ ● # Veh R1881 −3 n= 28 59 32 29 131* 19 10 22 9 676 Relative expression (2 0.5 Relative expression (2 0.5 Benign Primary CRPC Benign Primary Mets Benign Primary CRPC Benign Primary Mets Prostate Cancer Type

EV HES6HA C 35 LNCaP-LM Xenograft αHA EV 30 EV Full HES6 Full αTubulin

Full 25 EV Castr HES6 Castr HES6 20

15 Castrated EV 10

Relative size (arbitrary units) 5 Castrated HES6 * 0

173114 Weeks 024681012 Weeks post engraftment

D Full Castrated E LNCaP−LM final weights 60 C4-2b-LM Xenograft 50 shNT Full

EV 40 shHES6 Full

30 OWJ116 OWJ52

Weight (g) Weight * 20 * 10 0.0 0.2 0.4 0.6 0.8 1.0 1.2 HES6

EV HES6 EV HES6 Relative size (arbitrary units) (n=5) (n=5) (n=5) (n=5) 0 OWJ91 OWJ64 1 2 3 4 5 6 7 Full Castrated Weeks post engraftment

Figure 1. HES6 induces castration-resistant growth. A Changes in HES6 mRNA levels on alteration of AR and c-Myc activity in LNCaP cells, shown by quantitative PCR. Cells were starved in Charcoal Stripped Foetal Bovine Serum (CS) for 72 h and treated with 10 nM R1881 (AR agonist) for 16 h to check the AR regulation of HES6. Cells transfected with a pcDNA4-Myc vector were grown in Full Serum (FS) conditions; n = 3, error bars represent mean Æ s.e.m.; *P = 0.016, #P = 6.9E-6 by t-test. B Box and whisker dot plot showing HES6 expression is increased in metastatic disease (Mets) and castrate-resistant prostate cancer (CRPC) in several public datasets (Tomlins et al, 2005; Varambally et al, 2005; Taylor et al, 2010; Grasso et al, 2012). *P = 8.8E-6, #P = 7.9E-5 by limma with Benjamini–Hochberg adjustment for multiple testing for comparison of Mets to Benign. C Xenografts of HES6-overexpressing LNCaP cells are resistant to castration. LNCaP cells expressing luciferase/YFP (LNCaP-LM) and transduced to overexpress HES6 were grafted subcutaneously (2 × 106 cells) in NSG mice, and growth was monitored by bioluminescent imaging for 11 weeks. Host mice were castrated at 3 weeks (grafts approximately 100 mm3); n = 5; error bars represent mean Æ s.e.m; P = 0.0002 by t-test at week 11 for comparison of EV Castr to HES6 Castr. EV = Empty vector. Full = testes present. Castr = castration. D Visual appearance of harvested grafts and mean weights. One xenograft of each group is shown as an illustration. Evidence of significant difference in mean graft weight in castrated conditions (0.017 Æ 0.009 g versus 0.503 Æ 0.158 g). Scale bars = 5 mm; n = 5; error bars represent mean Æ s.e.m.; *P = 0.0001 by t-test. E Androgen-insensitive C4-2b-LM xenografts showed growth attenuation with constitutive lentiviral knock-down of HES6 (shHES6) compared to non-targeting controls (shNT). n = 5; error bars represent mean Æ s.e.m.; *P = 0.004 by t-test at week 7. See also Supplementary Figs S1–S3.

Source data are available online for this figure.

ª 2014 The Authors EMBO Molecular Medicine Vol 6 |No5 | 2014 653 Published online: April 14, 2014 EMBO Molecular Medicine HES6 drives prostate cancer progression Antonio Ramos-Montoya et al

ABAR ChIPseq C D EV HES6 EV HES6 Average AR peaks Example AR binding sites Comparative peak sets Bic --++ 7 Enhanced chr19 Z-score Z-score HES6_Veh -0.92 -1.57 Enh q13.31 q13.33 q13.42 -0.61 -1.05 EV_Veh -0.31 -0.52 51,340 kb 51,360 kb 51,380 kb 0.00 0.00 0.31 0.52 [0 - 16] 0.61 1.05 0.92 1.57 Tx Sensitive CRPC Lost Rescued Enhanced HES6_Bic HNPC E2F1 (global) et al

EV_Bic EV Ram-M cFos_K562 Rescued 0123456 Gata1_Igm_K562

Rescued [0 - 16] HNF4_HEPG

HES6_Veh Vehicle cMyc_std_HeLa

EV_Veh E2F1_std_HeLa

HES6 Fos_Hela

Max_SYDH HES6_Bic [0 - 16] cJun_Hela (Encode) EV_Bic SP1_A549_etoh public ChIP data public ChIP 0.0 0.5 1.0 1.5 2.0 Reads per Million(RPM) FoxA1_HepG

1.0 Lost AP2A_std_HeLa HES6_Veh Lost Stat3_IgR_HeLa [0 - 16] EV_Veh TNF_NFKB_gm91

TCF7L2_UCD_HeLa Bicalutamide BRG1_Hela HES6 EV -ve control HES6_Bic HES6 Tissue EV_Bic groups AR ChIP 0.0 0.2 0.4 0.6 0.8 (Sharma 0 500 1000 1500 2000 (PSA) KLK3 KLK2 et al) Base Pairs n = 4 4 3 4 40 E All ARBS Lost Rescued G LNCaP-HES6HA Promoter Region 0.35%0.35% 0.7% 35 shNT Veh (2000-500bp) shNT Bic 1 μM 10000upstream_To_5000upstream 30 5000upstream_To_2000upstream shAR Veh 2000upstream_To_500upstream 500upstream_To_500downstream 25 shAR Bic 1 μM 500downstream_To_2000downstream 2000downstream_To_Gene_End F Intergenic 20 AR ChIP 15

1.0 Confluence (%)

10 * log 5 FKBP5 CONTROL 1 CONTROL 2 HIF1A TLE4 HES6 TP63 CAMKK2 SPRY1 RFC1 KLK3 (PSA) GDF15 NKX3-1 TMPRSS2 KLK2 KLK15 0.0 0 C4-2b-shNT 0 25 50 75 100 125 150 175 C4-2b-shHES6 Time post plating (h)

Figure 2. HES6 modulates AR binding in bicalutamide and castration resistance with a shift in AR regulome. A Heatmap of normalized signal at peak summits in a window Æ 1 kb. Binding site classes were established by grouping highly reproducible peak calls into those that were still present with empty vector (EV) in 1 lM bicalutamide (Bic) but “enhanced” by HES6, “rescued” by HES6 or “lost” in both conditions. Average binding site location was established from mean location of peak summits across biological replicates and intensity of signal calculated as mean of the RPM normalized read counts. All conditions were in biological quadruplicate as shown with the exception of EV in bicalutamide which was in triplicate. Veh = vehicle (ETOH). B The mean normalized AR signal as profiles across binding sites seen in (A). This illustrates the global differences between conditions within the different AR binding site classes. C Example of binding events at KLK2 and KLK3 (PSA) where binding is markedly reduced with bicalutamide but rescued with overexpression of HES6. D Heatmap showing binding site co-occurrence for “lost,”“rescued” and “enhanced” groups compared to global E2F1 binding sites as well as Encode Consortium ChIPseq sets for well-known regulatory transcription factors. Comparisons were also made with our human tissue AR ChIPseq (Sharma et al, 2013). Co-occurrence was calculated as the total base-pair overlap of occupancy between the test and query binding site sets normalized to the total base-pair occupancy of the query set. To account for the differing lengths of the test sets, co-occurrence scores were normalized to Z-scores within each test set . Tx = treatment, HNPC = hormone naïve prostate cancer. E Genomic distribution analysis of “rescued” versus “lost” ARBS compared to whole genome shows increased AR binding in proximal promoters. F Selected AR binding sites were validated by ChIP in the castrate-resistant C4-2b derivative of LNCaP cells. Knock-down of HES6 reduced AR binding enrichment at all selected sites compared to control sites with no ChIPseq evidence of AR binding. n = 3; CAMKK2 P = 0.021, SPRY1 P = 0.026 , RFC1 P = 0.056, KLK3 P = 0.041, GDF15 P = 0.008, NKX3-1 P = 0.006, TMPRSS2 P = 0.017, KLK2 P = 0.023, HIF1A P = 0.031, KLK15 P = 0.009, TLE4 P = 0.026, HES6 P = 0.018,TP63 P = 0.01, FKBP5 P = 0.006 by t-test. G HES6-overexpressing LNCaP cells were transduced with a pSicoR lentivirus to stably knock-down AR (40% knock-down). Constitutive lentiviral knock-down of the AR limits HES6-overexpressing LNCaP growth and abrogates the HES6-driven bicalutamide-resistant phenotype. Vehicle (Veh) is ETOH and bicalutamide (Bic) 1 lM; n = 3; error bars represent mean Æ s.e.m.; *P = 0.023 by t-test for comparison of shAR Bic 1 lM to shNT Bic 1 lM. See also Supplementary Fig S4 and Tables S1 and S2.

654 EMBO Molecular Medicine Vol 6 |No5 | 2014 ª 2014 The Authors Published online: April 14, 2014 Antonio Ramos-Montoya et al HES6 drives prostate cancer progression EMBO Molecular Medicine

for other transcription factors including E2F1, c-Myc, FOXA1 and activity and investigated potential interactions between HES6 and STAT3 (Fig 2D, accessed through Encode Consortium), and found a E2F1. Given the persistence of nuclear AR with HES6, we also evalu- strong enrichment in the “rescued” group, raising the possibility of ated interactions with the AR. We demonstrated physical interaction concerted activity between the AR and these factors. Furthermore, between HES6 and E2F1, as well as between the AR and E2F1 analysis of previous data assessing ARBS in human tissue (Sharma (Fig 3G). These findings indicate that HES6 may enhance concerted et al, 2013) revealed a similar pattern of transcription factor overlap E2F1 and AR activity in androgen-deprived conditions. in CRPC tissue, as well as untreated and potentially aggressive E2F1 is a final common regulator for G1/S cell cycle transition primary prostate cancer (Fig 2D). Interrogation of the ARBS in an (Cam & Dynlacht, 2003). We therefore assessed by FACS analysis unbiased manner with MEME (Machanick & Bailey, 2011) identified the effect of overexpressing HES6 on cell cycle distribution in LNCaP similar motif enrichment (Supplementary Fig S4B and Table S2). cells and observed an increase in the fraction of cells in both S and Analysis of ARBS genomic distribution uncovered a twofold G2/M phases (Supplementary Fig S8A). To test whether this was increase in ARBS located 500–2,000 bp upstream of transcriptional due to more rapid initiation of cell cycle activity, we observed cell start sites (TSS) in the “rescued” set, consistent with a small shift of cycle fractions after synchronization (Supplementary Fig S8B-D) AR binding towards proximal promoters (Fig 2E). and found more rapid cell cycle re-entry in HES6-overexpressing We confirmed that this persistent AR signalling was HES6- cells in both normal conditions and bicalutamide. dependent by demonstrating evidence of reduced AR binding at several selected ARBS with stable knock-down of HES6 in C4-2b HES6 has a corresponding signature associated with aggressive cells (Fig 2F). Finally, we verified that HES6-driven androgen insen- clinical disease sitivity is still AR-dependent by knocking down the AR in HES6- overexpressing cells, with reversal of the phenotype (Fig 2G, To assess the clinical relevance of these findings, we used publicly Supplementary Fig S4C). Collectively this suggests modulation of available data (Varambally et al, 2005) to make comparisons the AR regulome by HES6. between benign, primary and metastatic prostate cancer. We found that the cell cycle genes associated with HES6 expression (from HES6, AR and E2F1 co-operate in driving a cell cycle-related Fig 3A) were strongly represented in the metastatic group (Supple- tumour-enhancing network mentary Fig S9). We interrogated these data to identify transcripts most strongly correlated with HES6 and selected the top 1.5% We used expression microarrays to identify differentially expressed according to a cut-off correlation coefficient of 0.9 (Supplementary genes (DEGs) associated with HES6 overexpression in the castration- Fig S10A). From these 788 genes, we selected those that were also resistant xenografts (Fig 3A, Supplementary Fig S5) and Ingenuity present in the castrated HES6-overexpressing LNCaP xenograft DEG ® Pathway Analysis (IPA, Ingenuity Systems, www.ingenuity.com) list. The combination of these two sets of data (one from human to infer the functional consequences (Supplementary Fig S6A). samples, one from xenografts overexpressing HES6) generated a Surprisingly, we did not find any evidence for a neuroendocrine “HES6-associated signature” of 222 transcripts (Supplementary phenotype (Supplementary Fig S5) but rather a predominance for Table S4). In order to validate this signature, we studied three addi- cell cycle pathways. A combined cell cycle network from the input tional publicly available datasets (Tomlins et al, 2005; Taylor et al, gene list revealed several dominant clusters including E2Fs, CDKs, 2010; Grasso et al, 2012) and a published cell cycle progression Cyclins and Aurora kinases (AURK) (Supplementary Fig S6B). signature (CCP) in CRPC (Cuzick et al, 2011). Allowing for differ- In order to identify potential transcriptional intermediaries, we ences in array coverage, we found that genes on this HES6-associ- assessed enriched binding motifs by in silico analysis (Supplemen- ated signature list were consistently enriched in DEG lists comparing tary Fig S6C and Table S3) of the DEGs from castrated xenografts as metastatic or castrate-resistant tumours with benign or normal tissue well as HES6-overexpressing LNCaP cells. This identified strong (Supplementary Fig S10B and C). These HES6-correlated genes effec- enrichment for E2F family binding sites and specifically for tively clustered men into two groups following radical prostatectomy E2F1 (Fig 3B). Furthermore, we found a strong overlap (44%) (Taylor et al, 2010) (Fig 4A). Kaplan–Meier analysis at a mean between DEGs in castrated xenografts and E2F1 transcriptional follow-up of 4.5 years (range 2–149 months) showed clear partition- targets identified by ChIP-seq in LNCaP cells (Fig 3C). To identify ing for men with low versus high HES6 (Fig 4B). Use of the n = 222 whether the presence of HES6 has a specific effect on E2F function, HES6-associated signature further enhanced clustering and was even we performed ChIPseq for E2F1 (n = 4) in LNCaP cells and found more effective in differentiating those patients destined for early that HES6 increased average E2F1 binding activity in the presence of relapse (Fig 4C). This signature is more powerful in this capacity bicalutamide (Fig 3D). Given our previous findings suggesting a shift than PSA and Gleason score (Fig 4D and Supplementary Fig S11). In in AR binding towards promoter regions when “rescued” by HES6 in the absence of good HES6 antibodies that are reliable for immunohis- bicalutamide, we investigated whether AR and E2F1 could be inter- tochemistry (Hartman et al, 2009), we checked protein expression of acting and whether HES6 was able to modulate such interaction. We the HES6-associated cell cycle regulator polo-like kinase 1 (PLK1) in found that the ARBS belonging to the “enhanced” and “rescued” representative samples of benign, primary prostate cancer and CRPC groups showed a clear increase in AR occupancy at E2F1 binding tissue (Fig 4E), identifying strong expression in CRPC tissue only. sites (Fig 3E). Enrichment of E2F1 binding in cells overexpressing We quantitatively assessed PLK1 expression in a large novel human HES6 was validated at promoters of selected E2F1 transcriptional tissue microarray incorporating 61 men, some of whom had andro- targets (Fig 3F and Supplementary Fig S7) and was found not to be a gen naı¨ve prostate cancer while others had hormone-relapsed consequence of increased E2F1 protein levels (Fig 3F and Supple- disease. We found clear evidence for an increase in PLK1 expression mentary Fig S5). We reasoned that HES6 may be enhancing E2F in the prostates of men with CRPC (Fig 4F).

ª 2014 The Authors EMBO Molecular Medicine Vol 6 |No5 | 2014 655 Published online: April 14, 2014 EMBO Molecular Medicine HES6 drives prostate cancer progression Antonio Ramos-Montoya et al

A Color Key B 2.0 E2F1 C

EV HES6 −1 1.0 E2F1 ChIPseq DLL3 STEAP2 in LNCaP NDUFB4 −2012 LDHD Z-score 0.0 (10,083 binding sites) NETO1 DDC Enrichment plot: V$E2F1_Q6_01 Xenograft DEGS NCAM2 STAT3 0.45 MYO1B 0.40 HES6 Castr C5orf53 BRDT 0.35 v’s EV Castr FAM171A1 0.30 DFNA5 0.25 NOTCH3 (2,727) CDKN2C 0.20 ABP1 0.15 8,886 NRIP1 0.10 1,530 ZNF680 1,197 TSPYL1 0.05 TRPM4 0.00 NOTCH1 -0.05 CDC2L2 SOX4 Enrichment score (ES) -0.10 PDXP GAPDH GDF15 C7orf68 PGK1 LDHA VEGFA 0 5,000 10,000 15,000 20,000 25,000 30,000 VLDLR JMJD1A Rank in Ordered Dataset NDUFA4L2 BIRC3 ADAM9 NOS3 ANGPT2 KLF6 ASNS D E2F1 ChIPseq E AR occupancy at E2F1 sites CLDN8 HES4 MAGEA8 Groups HDAC7 Enhanced LOC653288 HES6_Veh, n=4 LOC653178 LOC728403 EV_Veh, n=4 Rescued TSPY2 1.5 0.010 TSPY2 Lost MAGEA1 FFAR2 PKIB GRPR HES6_Bic, n=4 SNORD13 1.0 MGP ASAP2 EV_Bic, n=4 0.005 HES1 EZH2 PLK1 E2F2 RCC2 BIRC5 CDCA5 malised Peak Counts

NUSAP1 r 0.000 CCNA2

CDC20 0.0 0.5

CDK4 Reads per Million(RPKM) No ZWILCH 0 200 400 600 800 1000 −5000 −2500 0 2500 5000 PTTG1 TRIP13 ACOT7 Base Pairs Distance_from_Peak RAD51C AMACR TSPO FABP6 F CCDC84 CLK2 CCNL1 PRR7 LNCaP ABCA2 E2F1 ChIP v’s Input KLF4 CENPT EV HES6HA CDKN2A 2-fold CELSR3 CCDC130 CDCA2 αE2F1

SHMT2 log CENPN SLC38A1 CDK10 Í E4F1 No change αHA(HES6) CONTROL UBE2C BUB1 AURKB CDC2 AURKA CDK4 CDC20 SDCCAG3 Myc CENPM E2F4 ACSM3 CDCA7 LNCaPEV UBE2I αTubulin NOVA1 CDCA8 LNCaPHES6HA CDKN3 KIF20A CENPA MELK UBE2C PTTG3P RACGAP1 CDC45L G SN IP TOP2A MCM10 ASPM MCM5 MCM7 RNASEH2A BUB1 E2F1 HES6 E2F1 AR IgG AR IgG Input AURKB HES6 KIAA0101 CDCA3 TK1 CENPM TACC3 αHA PRC1 MCM2 PBK (HES6) 2’ 1’ 1’1’ CDC25C CENPE RRM2 MLF1IP FOXM1 TIMELESS C9orf58 AIF1L αE2F1 NCAPG DLGAP5 2’ 30’5’’ 30’ CEP55 AURKA KIF11 CDC2 MYC RFC4 MCM4 αAR ZMYND19 RFC3 1’ 1’ 1’ CDK2 5’’

Figure 3. HES6 enhances a cell cycle network through concerted activity with E2F1 and AR. A Selection of differentially expressed genes (DEGs) in castrated xenografts of HES6-overexpressing LNCaP cells (orange) versus castrated EV controls (green); n = 5. B GSEA motif analysis revealed enrichment of E2F1 in DEGs associated with HES6-overexpressing cells. Xenograft GSEA-NES-value = 2.25 and cells in vitro GSEA-NES- value = 2.09. C Overlap of the DEGs from castrated HES6-overexpressing LNCaP xenografts with E2F1 targets. Nearest gene (Æ 25 kb) overlap with DEGs from castrated (Castr) HES6- overexpressing LNCaP xenografts is shown; hypergeometric P-value = 2.53E-46. D Average enrichment of E2F1 at target sites is increased with HES6 compared to EV in bicalutamide (Bic) but not vehicle (Veh, ETOH). All ChIPseqs n = 4. E AR ChIPseq occupancy at E2F1 binding sites is increased in HES6 “enhanced” and “rescued” peaks compared to “lost” (see Fig 2). F Heatmap showing E2F1 ChIP. HES6 increased binding of E2F1 to differentially expressed genes in the castrated xenografts of HES6-overexpressing LNCaP cells. Each ChIP n = 3; Myc P = 0.052, AURKB P = 0.057, CDC2 P = 0.12, AURKA P = 0.087, UBE2C P = 0.165, CENPM P = 0.049, CDK4 P = 0.025, BUB1 P = 0.068, CDC20 P = 0.087 by t-test. G HES6 and AR co-immunoprecipitate with E2F1. IPs for AR, E2F1, HES6 and control IgG were performed using HES6HA-overexpressing LNCaP cell extracts and blotted for HA, E2F1 and AR. n = 5. One representative experiment is shown for illustration. SN = supernatant. IP = immunoprecipitate. Different time exposures of the same membrane are shown for each blot section to better visualize the whole experiment and are indicated in the lower-left corner (‘= minutes and “= seconds). Arrows indicate co-IP of HES6 and E2F1. See also Supplementary Figs S5 –S8 and Table S3.

Source data are available online for this figure.

656 EMBO Molecular Medicine Vol 6 |No5 | 2014 ª 2014 The Authors Published online: April 14, 2014 Antonio Ramos-Montoya et al HES6 drives prostate cancer progression EMBO Molecular Medicine

Selective pharmacological inhibition of this mechanism of how, during castration or AR inhibition, HES6 overexpression can androgen independence modulate the AR regulome, maintaining chromatin binding at a proportion of ARBS in the absence of hormone stimulation. We also Finally, we investigated whether inhibition of HES6-driven cell cycle show c-Myc to be another regulator of HES6 transcription, and so regulating genes could alter growth in these castration-resistant have identified two potent oncogenes involved upstream of HES6 in HES6-overexpressing cells. We found that these cells, though resis- this mechanism of androgen independence. tant to bicalutamide, were responsive to inhibition with the PLK1 Our study identifies E2F1 as an important intermediary in this inhibitor GSK461364A (Gilmartin et al, 2009) with a greater than process, with E2F-regulated cell cycle factors accounting for a large 50% reduction in cell proliferation (Supplementary Fig S12A). proportion of differentially expressed genes in these castrate- Indeed, concomitant application of this inhibitor with bicalutamide resistant tumours. This cell cycle enhancing framework fits with further reduced the rate of cell growth suggesting a synergistic effect other recent studies providing evidence for the centrality of cell to rescue the androgen-dependent phenotype. These findings were cycle genes in explaining mechanisms of resistance (Sharma et al, recapitulated in vivo with intraperitoneal injection of drug dramati- 2010) and predicting poor survival in the clinical setting (Cuzick cally slowing castrate-resistant growth of LNCaP-LM-HES6 et al, 2011). We identify a novel interaction between HES6 and xenografts (Fig 4G). PLK1 inhibition also reduced growth of E2F1 and E2F1 and the AR and show enhancement of E2F1 activity castrate-resistant AR-positive cells (C4-2b) and AR-negative cells that seems to result from protein complex formation rather than (PC3) (Supplementary Fig S12B and C) with greater effects on PC3 increased E2F1 expression as previously found in breast cancer cells on isogenic introduction of the AR (Nelius et al, 2007). This (Hartman et al, 2009). We propose that this mechanism may suggests that PLK1 inhibition reverses HES6-driven castration resis- account for maintained prostate cancer cell growth through tance and raises the possibility of synergy with AR activity. concerted activity with the AR in HES6-driven androgen insensitiv- ity, where the AR persistently binds to chromatin at a subset of binding sites and seems to be essential for the survival of the resis- Discussion tant cells, which cannot withstand AR knock-down. This finding is consistent with previous work assessing the impact of AR knock- Clinical progression to castrate-resistant prostate cancer (CRPC) down on castration-resistant derivatives of this model (Snoek et al, remains a major clinical problem and is the cause of death for most 2009). men dying of prostate cancer. Half of men on LHRH monotherapy Although we did not find evidence for a classical neuroendocrine will progress within 2 years of treatment (Hellerstedt & Pienta, phenotype, recent studies have suggested a broader relevance for 2002), and survival after the onset of metastatic CRPC does not neuroendocrine differentiation with powerful transcription factors usually extend beyond 6–12 months (Petrylak et al, 2004). Recent such as Myc and cell cycle regulators such as AURKA being impli- developments in second-/third-line prostate cancer therapeutics cated in development of therapy resistant tumours bearing neuro- with wide-ranging modes of action have reinforced the concept that endocrine characteristics (Beltran et al, 2011). castrate-resistant prostate cancer is a heterogeneous disease with We used a robust independent model to derive a HES6-associated multiple mechanisms of resistance (de Bono et al, 2010, 2011; Kantoff gene signature which contained several factors, including UBE2C et al, 2010; Nilsson et al, 2011; Scher et al, 2012; Ryan et al, 2013). (Wang et al, 2009), EZH2 (Xu et al, 2012), and cell cycle progres- It is therefore important that cancer researchers delineate fully sion factors such as CDC20, TK1, PLK1 and PTTG1 already shown these different mechanisms. Our study outlines a mechanism of to be important in CRPC (Cuzick et al, 2011). We have found that resistance centred on a single transcription co-factor and shows how HES6-driven genes are strongly associated with aggressive human hypothesis-driven investigation of cell function, combined with prostate cancers at their presentation, demonstrating that this large-scale genomic correlation, can deliver a candidate oncogenic phenomenon is not confined to the laboratory setting or to cell lines factor in a meaningful biological context with the potential to alone. Furthermore, in patients undergoing radical treatment, this uncover new therapeutic avenues that could improve the treatment signature is associated with early biochemical relapse after surgery. of men who are no longer responsive to current drugs. We have identified PLK1 as a possible target within this gene We have identified HES6 as a transcription co-factor that is able signature with an inhibitor giving promising results in rescuing an to alter prostate cancer cell phenotype so fundamentally that these androgen-sensitive phenotype and halting CRPC cell growth. When cells are able to grow in the absence of testosterone. HES6 is better combined with our data suggesting poor survival in patients who known for its functions in the nervous system, where it is thought have raised levels of HES6 and its associated genes, this raises the to promote neuronal differentiation (Kageyama et al, 2005; Murai possibility that such molecular information could be used to stratify et al, 2011), but its role in the prostate cell is not yet well under- patients for trials of a PLK1 inhibitor. In future, a more personalized stood. Here we have seen that HES6, a driver of androgen indepen- treatment strategy could be pursued with men who display a high- dence, is itself regulated by the AR. This suggests that the AR risk molecular signature being monitored closely and offered early regulome includes factors that are not necessarily required for adjuvant treatment, while those at lower risk could be managed growth in a normal environment, but which can be recruited in an more expectantly in order to avoid unnecessary intervention. altered environment where the cell has to rely on other pathways Taken together, our results identify the critical role of the HES6 for survival (Mills, 2014). Recently, it has been shown that ETS transcription co-factor in maintaining AR activity in CRPC with factors such as ERG markedly increase AR binding in mouse pros- concerted enhancement of E2F1 activity to maintain cell growth in tate tissue and mediate robust transcriptional changes in PTEN null the absence of testosterone and thereby offer a fresh treatment prostate cancer cells (Chen et al, 2013). In our study, we describe paradigm in combating this disease.

ª 2014 The Authors EMBO Molecular Medicine Vol 6 |No5 | 2014 657 Published online: April 14, 2014 EMBO Molecular Medicine HES6 drives prostate cancer progression Antonio Ramos-Montoya et al

A HES6 signature Post-prostatectomy Relapse No Relapse BC BCR event HES6 CDT1 ACOT7 WDR34 KIF26A

PPFIA4 1.0 C9orf142 Cluster 1 n= 107 DNMT3B Low HES6, n=105 SCRIB C9orf100 SCRIB CELSR3 TK1 High HES6, n=35 Cluster 2 n= 33 ACOT7 DNMT3B DNMT3B CENPA CDCA5 CENPA AURKB p=7.03 E-06 RAD54L p=0.018 FSCN1 OIP5 ACOT7 BIRC5 GTSE1 CDC20 ACOT7 BIRC5 BIRC5 CDCA8 CABLES2 PLK1 ESPL1 TACC3 LIG1 CDKN2A CA9 ZNF367 CDCA3 NTHL1 CDC25C SPAG11B SPAG11B CDC25C DNMT3B NDUFA4L2 FANCA KIF2C SPAG11B SPAG11B GINS2 SLC25A19 SPAG11B LMNB2 STC2 CDKN2A SLC25A19 CDKN2A FLAD1 RNASEH2A FAM54A SLC25A19 FEN1 FAM54A MND1 KRT80 KRT80 RCCD1 UHRF1 RCCD1 ASF1B MCM2 PFKFB4 UHRF1 FANCA E2F8 SGOL1 SPAG11B UBE2C FOXM1 POLQ SGOL1 SPAG11B DEPDC1B

FOXM1 Probability of Freedom FLAD1 KIF20A CCNF CDCA4 SGOL1 EXO1 PTTG1 E2F2 from Biochemical Recurrence FOXM1 EXO1 EXO1 CEP55 KIF14 0.0 0.2 0.4 0.6 0.8 1.0 PABPC1L 0.0 0.2 0.4 0.6 0.8 SPRR1A ONECUT2 UBE2C CDCA4 KIF4A SGOL1 SGOL1 SGOL1 UBE2T 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 C7orf68 CARHSP1 RAB40C SOX4 UBE2C C7orf68 DLGAP5 Time (months) ASPM Time (months) BUB1 CARHSP1 CDC25B CDC25B CDKN3 SPAG5 TRIP13 PBK UBE2C DDX11 CCNA2 DDX11 PROX1 PLK4 CDC25B DDX11 Hazard Ratio (of Biochemical Relapse) UBE2C CASC5 D CASC5 HMMR HMMR CENPE WDR54 CENPF NCAPG TIMELESS 0.751.00 2.00 3.00 4.00 5.00 6.00 7.00 RRM2 ANLN NUF2 PRC1 TPX2 MELK UBE2C MAD2L2 NUF2 PTTG3P GPR19 PRC1 MLF1IP CRIP2 TYMS NUSAP1 NUSAP1 PRC1 CCNB2 PLCXD1 TTK NUP62CL Diagnostic Age − 62.51:53.8575 C3orf26 CDCA7 CCNE2 KIAA0101 LMNB1 CDCA7 KIF11 FANCB APOBEC3B CENPK RAD51AP1 FANCB NUP210 EZH2 EZH2 Surgical Margin Status − Positive:Negative KIAA0101 MYO19 MYO19 TOP2A MCM4 RACGAP1 RACGAP1 GAS2L3 YY1 IL17RB RACGAP1 CHEK1 NAV1 CHEK1 MCM4 STMN1 CSTF2 Pre-Diagnostic Biopsy PSA − 9.36:4.5 CCDC88C SHMT2 NEK2 STMN1 SGOL2 RUVBL1 PPM1G TMPO GTPBP2 STMN1 GGH MCM7 DNMT1 UBN1 IRF2BP2 IRF2BP2 Gleason Score − >6:≤6 AURKA CCNB1 UBN1 SLBP AURKA MCM7 RFC4 POLE2 CHEK1 CEP135 FBXO5 AURKA AURKA ANK2 NDE1 AP1S2 HES6-associated n=222 signature clusters − 2:1 SPC25 AURKA AURKA CNTN4 CNKSR3 SRGAP2 RFC4 RCC2 0.9 0.7 0.8 PPM1G ANK2 0.99 0.95 SRGAP2 ECT2 CNTN4 KPNA2 CASP2 CNTN4 CASP2 WDR67 GPI SKP2 YARS E CCDC99 USP22 EIF2C2 TUBA1C KNTC1 TUBA1B MEST MEST YEATS2 RFC5 HsPLK1 SKP2 MEST AATF POLA2 RFC5 MAD2L1 HMGB2 EXOSC10 RBL1 SMC4 RBL1 ATAD2 HIST1H4C SMC4 FXR1 FXR1 FXR1 PRIM1 ENO1 EXOSC10 TJP2 GART TMEM48 LZTFL1 CEP78 GART TJP2 PLOD2 FAM13A CEP78 TMPO PLOD2 TMPO FAM13A FAF1 SLC38A1 PSMB7 SLC38A1 TCERG1 TCERG1 TRIT1 PARP2 CDK5RAP2 GMPS PARP2 CDK5RAP2 NR2C1 RRM1 NUP155 NUP155 CSE1L PBRM1 PBRM1 PBRM1 NR2C1 SUZ12 DYM CDKAL1 STAG1 GSM527923 GSM527879 GSM527916 GSM527867 GSM527877 GSM527903 GSM527858 GSM527868 GSM527870 GSM527937 GSM528003 GSM527968 GSM527918 GSM527913 GSM527914 GSM527899 GSM527864 GSM527931 GSM527889 GSM527927 GSM527861 GSM527920 GSM527925 GSM527973 GSM527910 GSM527926 GSM527939 GSM527921 GSM527900 GSM527988 GSM527897 GSM527862 GSM527909 GSM527946 GSM527943 GSM527953 GSM527944 GSM527930 GSM527917 GSM527966 GSM527908 GSM527886 GSM527873 GSM527887 GSM527869 GSM527872 GSM527874 GSM527881 GSM527893 GSM527911 GSM527888 GSM527859 GSM527951 GSM527982 GSM527972 GSM527912 GSM527885 GSM527986 GSM527960 GSM527956 GSM528000 GSM527880 GSM527904 GSM528012 GSM527865 GSM527947 GSM527941 GSM527924 GSM527992 GSM527902 GSM527876 GSM527898 GSM527901 GSM527928 GSM527945 GSM527892 GSM527905 GSM527907 GSM527922 GSM527906 GSM527970 GSM527989 GSM527860 GSM527933 GSM527985 GSM527958 GSM527863 GSM527991 GSM527891 GSM527935 GSM527940 GSM527915 GSM527987 GSM527896 GSM527984 GSM527959 GSM527971 GSM527875 GSM527878 GSM527890 GSM527866 GSM527957 GSM527980 GSM527962 GSM527884 GSM527882 GSM527936 GSM527974 GSM527952 GSM527934 GSM528011 GSM528001 GSM528013 GSM527996 GSM527895 GSM528002 GSM528014 GSM528004 GSM527950 GSM528009 GSM527995 GSM527994 GSM528007 GSM528010 GSM527999 GSM527954 GSM527965 GSM527955 GSM528006 GSM527883 GSM527942 GSM527990 GSM527932 GSM527938 GSM527961 GSM527975 GSM527871 GSM527919 GSM527949 GSM527894 GSM527979 GSM527977 GSM527963 GSM527993 GSM527998 GSM527997 GSM527948 GSM528005 GSM528008 GSM527929 cluster 1 cluster 2 Benign (BPH) Primary CaP CRPC F HRTMA - PLK1 G 14 LNCaP-LM-HES6HA Xenograft

n= 10 35 7 25 32 s) p=2.189e-13 it 100% Vehicle CASTRATED

un 12 Score= ay r

Intensity (0-3) x it 10 75% % +ve nuclei (0-4) 4 8 50%

2 inesc. (arbr 6 Castration m

25% 1 lu * 4 Bio % of Total Tumour Cores Tumour % of Total 0% 0

ative 2 t e CaP

R GSK461364A benign benign 0 Primary on CAB Matched on LHRH 02468101214 Progression Progression Independent Weeks post engraftment

Figure 4.

658 EMBO Molecular Medicine Vol 6 |No5 | 2014 ª 2014 The Authors Published online: April 14, 2014 Antonio Ramos-Montoya et al HES6 drives prostate cancer progression EMBO Molecular Medicine

◀ Figure 4. HES6 is associated with aggressive clinical disease targetable by PLK1i. A HES6-associated signature (n = 222, see Supplementary Fig S10 and Table S4 for generation of n = 222 HES6-associated signature) used to cluster 140 prostatectomy specimens (Taylor et al, 2010). See also Supplementary Fig S9. B Kaplan–Meier biochemical relapse-free survival analysis by recursive partitioning to high and low expression of HES6. Logrank P-value. C Kaplan–Meier biochemical relapse-free survival analysis by two cluster partitioning based on expression of HES6-associated signature (n = 222). Logrank P-value. D Graphical representation of proportional hazard ratios for biochemical relapse (BCR) calculated by Cox regression models used to estimate the association with relapse of several key variables: age, surgical margin status, PSA, Gleason score and the clusters generated by the HES6-associated signature. Both the signature and the Gleason score were shown to have an independent prognostic effect (P = 0.0071 and P = 0.0208), but not PSA (P = 0.7384), SMS (P = 0.1467) and age of diagnosis (P = 0.9180). See also Supplementary Fig S11. E Protein expression of HES6-responsive gene PLK1 shows strong staining in CRPC alone, compared to primary prostate cancer (CaP) and benign control (benign prostatic hyperplasia, BPH). Scale bars represent 250 lm. Magnified windows = 100 lm2. F Quantification of PLK1 expression in a hormone-relapsed tissue microarray (HRTMA). Cores were taken from 45 men with varying degrees of androgen insensitivity undergoing channel transurethral resection of the prostate (chTURP), including both tumour areas and neighbouring benign regions where available. A number of independent benign and primary CaP men were also included. Tumour cores were scored by two trained individuals assessing staining intensity and percentage of positive nuclei. There was increased PLK1 staining in men resistant to androgen therapy; P-value = 2.189E-13 by Fisher’s exact test. LHRH = luteinizing hormone releasing hormone. CAB = complete androgen blockade. G LNCaP-LM-HES6HAHA xenografts were sensitized to castration on treatment with selective polo-like kinase 1 (PLK1) inhibitor GSK461364A 50 mg/kg administered i.p. at a dose rate of q2dx12 versus vehicle. Host mice were castrated at 3 weeks (grafts approximately 100 mm3); n = 8; error bars represent mean Æ s.e.m.; *P = 0.007 by Mann–Whitney U-test at week 12. See also Supplementary Figs S9–S12 and Table S4.

Materials and Methods performed using Ingenuity Pathway Analysis (IPA, Ingenuity Systems, Redwood City, CA). For enrichment and motif analysis, we Detailed methodology and details of antibodies, primers, probes and used GSEA (Mootha et al, 2003; Subramanian et al, 2005). oligonucleotides are described in Supplementary Methods. Human prostate tissue samples Plasmids Ethical approval for the use of samples and data collection was c-Myc was amplified from an IMAGE clone (Lawrence Livermore granted by the local Research Ethics Committee under ProMPT National Laboratory, Livermore, CA) and then inserted at the (Prostate Mechanisms for Progression and Treatment) “Diagnosis, BamH1 restriction site of the lentiviral inducible pLVX-Tight-Puro investigation and treatment of prostate disease” (MREC 01/4/061). vector (Clontech, Mountain View, CA) by IN-Fusion Advantage (Clontech) homologous fusion. Lentiviral pLVX-Teton-Genet vector Statistics (Clontech) was used to stably express the tetracycline-controlled transactivator. The human HES6 cDNA sequence was amplified and Data from independent experiments were reported as the inserted into the retroviral vector pBabe-puro (Cell Biolabs, San mean Æ s.e.m. Student’s t-test analysis or Mann–Whitney U-test Diego, CA). For knock-down of HES6, annealed oligonucleotides were performed to determine statistical significance in replicate were inserted into the lentiviral pSicoR vector (Ventura et al, 2004) comparisons and Fisher’s exact test for proportionate analyses. with a puromycin resistance cassette. Knock-down of the AR was achieved using the lentiviral pSicoR vector with a blasticidin resis- Study approval tance cassette. All animal procedures were carried out in accordance with Univer- Mice sity of Cambridge and Cancer Research UK guidelines under UK Home Office project licenses 80/2301 and 80/2435. For human Immunocompromised NSG male mice (Charles River, Wilmington, material, informed written consent was received from participants MA) were used for tumour implantation. GSK461364A was adminis- prior to inclusion in the study under ethics committee number tered intraperitoneally 50 mg/kg at a dose rate of q2dx12. Vehicle MREC 01/4/061. was water with 2% Cremophor-EL and 2% N,N-dimethylacetamide (DMA) (Sigma-Aldrich, St. Louis, MO). Mice were maintained in the Accession numbers Cancer Research UK Cambridge Institute Animal Facility. All experi- ments were performed in accordance with national guidelines and Study data are deposited in NCBI GEO under accession number regulations, and with the approval of the animal care and use GSE36526. committee at the institution. Supplementary information for this article is available online: Gene expression data http://embomolmed.embopress.org

IlluminaBeadChip HumanWG-12 (version 3 and 4; Illumina, San Acknowledgements Diego, CA) were used for the gene expression. Publicly available We thank the core facilities at the Cancer Research UK Cambridge Institute led prostate cancer data sets were downloaded from GEO (Gene Expres- by James Hadfield (Genomics), Matt Eldridge (Bioinformatics) and Allen Hazel- sion Omnibus). Biological function and network generation was hurst (BRU), as well as Jodi Miller (IHC), Ellie Pryor (BRU) and Jason Fox (BRU).

ª 2014 The Authors EMBO Molecular Medicine Vol 6 |No5 | 2014 659 Published online: April 14, 2014 EMBO Molecular Medicine HES6 drives prostate cancer progression Antonio Ramos-Montoya et al

were responsible for overseeing the manuscript. All authors discussed the The paper explained results and reviewed the manuscript.

Problem Conflict of interest Prostate cancer is the most common non-cutaneous cancer in men. The authors declare that they have no conflict of interest. Radical prostatectomy or hormone therapy is commonly used to treat early stage or androgen-sensitive disease. However, relapse remains a serious clinical problem with approximately one-third of men devel- oping recurrent disease within 5 years after prostatectomy and half of References men developing hormone resistance within 2 years of commencement of hormone treatment. We noticed that HES6 levels tend to be high in Azzi L, El-Alfy M, Labrie F (2006) Gender differences and effects of sex aggressive disease and wished to discover whether this transcription steroids and dehydroepiandrosterone on androgen and oestrogen alpha co-factor plays a role in the development of resistance and could be receptors in mouse sebaceous glands. Br J Dermatol 154: 21 – 27 used for stratification of targeted treatment. Beltran H, Rickman DS, Park K, Chae SS, Sboner A, MacDonald TY, Wang Y, Results Sheikh KL, Terry S, Tagawa ST et al (2011) Molecular characterization of In this study, we report that HES6 maintains AR chromatin binding at neuroendocrine prostate cancer and identification of new drug targets. a critical subset of sites, which are enriched for cell cycle regulatory Cancer Discov 1: 487 – 495 genes under the control of E2F1, induces resistance to anti-androgens Bernard D, Pourtier-Manzanedo A, Gil J, Beach DH (2003) Myc confers and castration and predicts poor outcome in the clinical setting. We androgen-independent prostate cancer cell growth. J Clin Invest 112: also identify a protein interaction between the AR and E2F1 and – between HES6 and E2F1 and show that by drugging a kinase within 1724 1731 the HES6 regulome, namely PLK1, we are able to restore response to de Bono JS, Logothetis CJ, Molina A, Fizazi K, North S, Chu L, Chi KN, Jones RJ, anti-androgens. Goodman OB Jr, Saad F et al (2011) Abiraterone and increased survival in metastatic prostate cancer. N Engl J Med 364: 1995 – 2005 Impact de Bono JS, Oudard S, Ozguroglu M, Hansen S, Machiels JP, Kocak I, Gravis G, This study demonstrates the relevance of HES6 in the development of Bodrogi I, Mackenzie MJ, Shen L et al (2010) Prednisone plus cabazitaxel androgen resistance and brings forward our understanding of how other factors co-operate with the master regulator of prostate cancer, or mitoxantrone for metastatic castration-resistant prostate cancer the AR, to maintain essential pathways for cell survival. A novel set of progressing after docetaxel treatment: a randomised open-label trial. end-stage human prostate cancers have been assessed in this study, Lancet 376: 1147 – 1154 raising the possibility that judicious combination of hormone therapy Cam H, Dynlacht BD (2003) Emerging roles for E2F: beyond the G1/S with inhibition of the HES6 regulome offers a new treatment para- transition and DNA replication. Cancer Cell 3: 311 – 316 digm for high-risk prostate cancer patients who can be selectively identified and targeted for early treatment. Chen Y, Chi P, Rockowitz S, Iaquinta PJ, Shamu T, Shukla S, Gao D, Sirota I, Carver BS, Wongvipat J et al (2013) ETS factors reprogram the androgen receptor cistrome and prime prostate tumorigenesis in response to PTEN Michaela Frye, Lee Fryer, Mohammad Asim, Hayley Whitaker, Carrie Yang, Joana loss. Nat Med 19: 1023 – 1029 Borlido, Vinny Zecchini, Roheet Rao, Ajoeb Baridi, Satoshi Hori, Maxine Tran, Lui Clegg NJ, Couto SS, Wongvipat J, Hieronymus H, Carver BS, Taylor BS, Shiong, Karan Wadhwa, Steve Hawkins, Oscar M Rueda and Ivan Valle Rosado Ellwood-Yen K, Gerald WL, Sander C, Sawyers CL (2011) MYC cooperates provided valuable discussion and assistance. We also thank the support staff of with AKT in prostate tumorigenesis and alters sensitivity to mTOR the Uro-Oncology Group in the University of Cambridge, Department of Oncol- inhibitors. PLoS ONE 6:e17449 ogy including Jo Burge, Sara Stearn, Marie Corcoran and Ann George. Olga CRUK (2010) Cancer Research UK cancerstats. http://www. Volpert kindly provided cells. Critical advice on the project and manuscript was cancerresearchuk.org/cancer-info/cancerstats/types/prostate/incidence/ gratefully received from Jason Carroll and Ashok Venkitaraman. We are grateful Cutress ML, Whitaker HC, Mills IG, Stewart M, Neal DE (2008) Structural to study volunteers for their participation and to staff at the Welcome Trust basis for the nuclear import of the human androgen receptor. J Cell Sci Clinical Research Facility, Addenbrooke’s Clinical Research Centre, Cambridge 121: 957 – 968 for their help in conducting the study. This work was supported by a Cancer Cuzick J, Swanson GP, Fisher G, Brothman AR, Berney DM, Reid JE, Mesher D, Research UK project grant (DEN), the Cambridge Biomedical Research Centre Speights VO, Stankiewicz E, Foster CS et al (2011) Prognostic value of an and National Institute of Health Research UK (ADL, DEN), the DOH HTA (ProtecT RNA expression signature derived from cell cycle proliferation genes in grant), the MRC (ProMPT grant), GlaxoSmithKline (ADL), Prostate Action (ARM & patients with prostate cancer: a retrospective study. Lancet Oncol 12: ADL) and a Raymond and Beverley Sackler Studentship (ADL). Hutchison 245 – 255 Whampoa Limited, the University of Cambridge and the National Cancer Gilmartin AG, Bleam MR, Richter MC, Erskine SG, Kruger RG, Madden L, Research Prostate Cancer: Mechanisms of Progression and Treatment Hassler DF, Smith GK, Gontarek RR, Courtney MP et al (2009) Distinct (PROMPT) collaborative (Grant code G0500966/75466) funded tissue collections concentration-dependent effects of the polo-like kinase 1-specific inhibitor in Cambridge. The Addenbrooke’s Human Research Tissue Bank is supported GSK461364A, including differential effect on apoptosis. Cancer Res 69: by the NIHR Cambridge Biomedical Research Centre. 6969 – 6977 Grasso CS, Wu YM, Robinson DR, Cao X, Dhanasekaran SM, Khan AP, Quist Author contributions MJ, Jing X, Lonigro RJ, Brenner JC et al (2012) The mutational landscape of ARM and ADL conducted most of the experiments, analysis and writing of the lethal castration-resistant prostate cancer. Nature 487: 239 – 243 manuscript. RR, TC, SJ, NGD, CEM, JB, HB, VT, HRA, HES, SV and WH generated Hartman J, Lam EW, Gustafsson JA, Strom A (2009) Hes-6, an inhibitor of data, performed analyses or both. ADL, NLS, GS and AW provided prostate Hes-1, is regulated by 17beta-estradiol and promotes breast cancer cell samples. RFW provided compounds. IGM and DEN supervised the project and proliferation. Breast Cancer Res 11:R79

660 EMBO Molecular Medicine Vol 6 |No5 | 2014 ª 2014 The Authors Published online: April 14, 2014 Antonio Ramos-Montoya et al HES6 drives prostate cancer progression EMBO Molecular Medicine

Hellerstedt BA, Pienta KJ (2002) The current state of hormonal therapy for enzalutamide in prostate cancer after chemotherapy. N Engl J Med 367: prostate cancer. CA Cancer J Clin 52: 154 – 179 1187 – 1197 Hobisch A, Fritzer A, Comuzzi B, Fiechtl M, Malinowska K, Steiner H, Bartsch Sharma A, Yeow WS, Ertel A, Coleman I, Clegg N, Thangavel C, Morrissey C, G, Culig Z (2006) The androgen receptor pathway is by-passed in prostate Zhang X, Comstock CE, Witkiewicz AK et al (2010) The retinoblastoma cancer cells generated after prolonged treatment with bicalutamide. tumor suppressor controls androgen signaling and human prostate cancer Prostate 66: 413 – 420 progression. J Clin Invest 120: 4478 – 4492 Kageyama R, Ohtsuka T, Hatakeyama J, Ohsawa R (2005) Roles of bHLH Sharma NL, Massie CE, Ramos-Montoya A, Zecchini V, Scott HE, Lamb AD, genes in neural stem cell differentiation. Exp Cell Res 306: 343 – 348 MacArthur S, Stark R, Warren AY, Mills IG et al (2013) The androgen Kantoff PW, Higano CS, Shore ND, Berger ER, Small EJ, Penson DF, Redfern receptor induces a distinct transcriptional program in castration-resistant CH, Ferrari AC, Dreicer R, Sims RB et al (2010) Sipuleucel-T immuno- prostate cancer in man. Cancer Cell 23: 35 – 47 therapy for castration-resistant prostate cancer. N Engl J Med 363: 411 – 422 Snoek R, Cheng H, Margiotti K, Wafa LA, Wong CA, Wong EC, Fazli L, Nelson Lamb AD, Massie CE, Neal DE (2014) The transcriptional programme CC, Gleave ME, Rennie PS (2009) In vivo knockdown of the androgen of the androgen receptor (AR) in prostate cancer. BJU Int 113: 358 – 366 receptor results in growth inhibition and regression of well-established, Machanick P, Bailey TL (2011) MEME-ChIP: motif analysis of large DNA castration-resistant prostate tumors. Clin Cancer Res 15: 39 – 47 datasets. Bioinformatics 27: 1696 – 1697 Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, Malone CM, Domaschenz R, Amagase Y, Dunham I, Murai K, Jones PH (2011) Paulovich A, Pomeroy SL, Golub TR, Lander ES et al (2005) Gene set Hes6 is required for actin cytoskeletal organization in differentiating enrichment analysis: a knowledge-based approach for interpreting C2C12 myoblasts. Exp Cell Res 317: 1590 – 1602 genome-wide expression profiles. Proc Natl Acad Sci USA 102: Massie CE, Lynch A, Ramos-Montoya A, Boren J, Stark R, Fazli L, Warren A, 15545 – 15550 Scott H, Madhu B, Sharma N et al (2011) The androgen receptor fuels Takahashi K, Tanabe K, Ohnuki M, Narita M, Ichisaka T, Tomoda K, Yamanaka prostate cancer by regulating central metabolism and biosynthesis. EMBO S(2007) Induction of pluripotent stem cells from adult human fibroblasts J 30: 2719 – 2733 by defined factors. Cell 131: 861 – 872 Mills IG (2014) Maintaining and reprogramming genomic androgen receptor Taylor BS, Schultz N, Hieronymus H, Gopalan A, Xiao Y, Carver BS, Arora VK, activity in prostate cancer. Nat Rev Cancer 14: 187 – 198 Kaushik P, Cerami E, Reva B et al (2010) Integrative genomic profiling of Mootha VK, Lindgren CM, Eriksson KF, Subramanian A, Sihag S, Lehar J, human prostate cancer. Cancer Cell 18: 11 – 22 Puigserver P, Carlsson E, Ridderstrale M, Laurila E et al (2003) Thalmann GN, Anezinis PE, Chang SM, Zhau HE, Kim EE, Hopwood VL, PGC-1alpha-responsive genes involved in oxidative phosphorylation are Pathak S, von Eschenbach AC, Chung LW (1994) Androgen-independent coordinately downregulated in human diabetes. Nat Genet 34: 267 – 273 cancer progression and bone metastasis in the LNCaP model of human Murai K, Philpott A, Jones PH (2011) Hes6 is required for the neurogenic prostate cancer. Cancer Res 54: 2577 – 2581 activity of neurogenin and NeuroD. PLoS ONE 6:e27880 Tomlins SA, Rhodes DR, Perner S, Dhanasekaran SM, Mehra R, Sun XW, Murai K, Vernon AE, Philpott A, Jones P (2007) Hes6 is required for MyoD Varambally S, Cao X, Tchinda J, Kuefer R et al (2005) Recurrent fusion of induction during gastrulation. Dev Biol 312: 61 – 76 TMPRSS2 and ETS transcription factor genes in prostate cancer. Science Nelius T, Filleur S, Yemelyanov A, Budunova I, Shroff E, Mirochnik Y, Aurora A, 310: 644 – 648 Veliceasa D, Xiao W, Wang Z et al (2007) Androgen receptor targets Varambally S, Yu J, Laxman B, Rhodes DR, Mehra R, Tomlins SA, Shah RB, NFkappaB and TSP1 to suppress prostate tumor growth in vivo. Int J Chandran U, Monzon FA, Becich MJ et al (2005) Integrative genomic and Cancer 121: 999 – 1008 proteomic analysis of prostate cancer reveals signatures of metastatic Nilsson S, O’BryanTear C, Bolstad B, Lokna A, Parker CC (2011) Alkaline progression. Cancer Cell 8: 393 – 406 phosphatase (ALP) normalization and overall survival in patients with Ventura A, Meissner A, Dillon CP, McManus M, Sharp PA, Van Parijs L, bone metastases from castration-resistant (CRPC) treated with radium-223 Jaenisch R, Jacks T (2004) Cre-lox-regulated conditional RNA interference [abstract]. J Clin Oncol 29:A4620 from transgenes. Proc Natl Acad Sci USA 101: 10380 – 10385 Petrylak DP, Tangen CM, Hussain MH, Lara PN Jr, Jones JA, Taplin ME, Burch Vias M, Massie CE, East P, Scott H, Warren A, Zhou Z, Nikitin AY, Neal DE, PA, Berry D, Moinpour C, Kohli M et al (2004) Docetaxel and estramustine Mills IG (2008) Pro-neural transcription factors as cancer markers. BMC compared with mitoxantrone and prednisone for advanced refractory Med Genomics 1: 17 prostate cancer. N Engl J Med 351: 1513 – 1520 Wang Q, Li W, Zhang Y, Yuan X, Xu K, Yu J, Chen Z, Beroukhim R, Wang H, Pienta KJ, Smith DC (2005) Advances in prostate cancer chemotherapy: a new Lupien M et al (2009) Androgen receptor regulates a distinct transcription era begins. CA Cancer J Clin 55: 300 – 318; quiz 323-305 program in androgen-independent prostate cancer. Cell 138: 245 – 256 Qi J, Nakayama K, Cardiff RD, Borowsky AD, Kaul K, Williams R, Krajewski S, Xu K, Wu ZJ, Groner AC, He HH, Cai C, Lis RT, Wu X, Stack EC, Loda M, Liu T Mercola D, Carpenter PM, Bowtell D et al (2010) Siah2-dependent et al (2012) EZH2 oncogenic activity in castration-resistant prostate concerted activity of HIF and FoxA2 regulates formation of cancer cells is Polycomb-independent. Science 338: 1465 – 1469 neuroendocrine phenotype and neuroendocrine prostate tumors. Cancer Yap TA, Zivi A, Omlin A, de Bono JS (2011) The changing therapeutic landscape Cell 18: 23 – 38 of castration-resistant prostate cancer. Nat Rev Clin Oncol 8: 597 – 610 Ryan CJ, Smith MR, de Bono JS, Molina A, Logothetis CJ, de Souza P, Fizazi K, Mainwaring P, Piulats JM, Ng S et al (2013) Abiraterone in metastatic License: This is an open access article under the prostate cancer without previous chemotherapy. N Engl J Med 368: terms of the Creative Commons Attribution 4.0 138 – 148 License, which permits use, distribution and reproduc- Scher HI, Fizazi K, Saad F, Taplin ME, Sternberg CN, Miller K, de Wit R, tion in any medium, provided the original work is Mulders P, Chi KN, Shore ND et al (2012) Increased survival with properly cited.

ª 2014 The Authors EMBO Molecular Medicine Vol 6 |No5 | 2014 661