Sample Thesis Title with a Concise and Accurate

IDENTIFICATION OF THE POLYCOMB PROTEIN CBX2 AS A POTENTIAL DRUG TARGET IN ADVANCED PROSTATE CANCER AND BEYOND

Pier-Luc Clermont

B.Sc. (Honours), McGill University, 2012

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES

(Interdisciplinary Oncology)

THE UNIVERSITY OF BRITISH COLUMBIA

(Vancouver)

August 2015

Abstract Globally, prostate cancer (PCa) represents the most commonly diagnosed cancer in men. While localized PCa can often be cured, all patients with metastatic disease inevitably develop castration-resistant prostate cancer (CRPC) or neuroendocrine prostate cancer (NEPC). Increasing evidence suggests that epigenetic alterations involving the Polycomb Group (PcG) family drive PCa progression. Although the PcG protein CBX2 is required for prostate development, its implication in human cancer remains unexplored. I therefore hypothesized that CBX2 may become deregulated during PCa progression and induce transcriptional programs promoting PCa aggressiveness.

Using patient-derived xenografts and clinical datasets, I have explored the epigenetic landscape of advanced PCa and identified the Polycomb Group protein and epigenetic reader CBX2 as a potential drug target. First, CBX2 was significantly up-regulated in metastatic and castration- resistant PCa tissues. Furthermore, CBX2 overexpression predicted lower overall survival and correlated with numerous adverse prognostic factors. In addition, CBX2 depletion induced proliferation arrest and apoptosis in metastatic PCa cell lines, implying that CBX2 is required for PCa cell survival. Microarray analysis conducted after CBX2 silencing revealed that CBX2 regulates many genes controlling cellular proliferation and differentiation. Given the rising incidence of NEPC in advanced PCa, I analyzed whether CBX2 was also involved in NEPC pathogenesis. Strikingly, CBX2 was consistently the most highly up-regulated epigenetic regulator across multiple clinical and xenograft tumor tissues. Furthermore, I derived a list of 185 genes down-regulated in NEPC that was preferentially enriched in PcG target genes and predicted poor clinical outcome, in line with a critical function for CBX2 in late-stage PCa.

Since CBX2 has never been linked to human cancers, I conducted a comprehensive meta-analysis of CBX2 across many tumor types using previously published clinical data. Strikingly, these studies demonstrated that the CBX2 locus is rarely inactivated or down-regulated. In contrast, CBX2 was frequently amplified and over-expressed in many common tumors, where it correlated with metastatic dissemination and poor clinical outcome. Overall, this work identifies CBX2 as novel epigenetic driver of cancer progression and investigates the therapeutic potential of CBX2 in advanced solid malignancies. ii

Preface

Tumor tissues were obtained from patients through a protocol approved by the Clinical Research Ethics Board of the University of British Columbia (UBC) and the BC Cancer Agency (BCCA). All patients signed a consent form approved by the Ethics Board (UBC Ethics Board #: H09- 01628 and H04-60131; VCHRI #: V09-0320 and V07-0058). Animal care and experimental procedures were carried out in accordance with the guidelines of the Canadian Council of Animal Care (CCAC) under the approval of the Animal Care Committee of University of British Columbia (permit #: A10-0100).

Portions of Chapter 1 have been adapted from [Clermont PL], Crea F, Helgason CD. Chapter 22: Trithorax genes in prostate cancer. In "Advances in Prostate Cancer" (ISBN 978-953-51-0932-7; edited by Gerhard Hamilton), InTech, 2013. I am the original writer of all portions included. Crea F and Helgason CD critically revised the manuscript.

A version of Chapter 2 is being prepared for publication. [Clermont PL], Crea F, Chiang YT, Lin D, Zhang A, Wang JZL, Parolia A, Wu R, Xue H, Wang Y, Ding J, Thu KL, Lam WL, Shah SP, Collins CC, Wang Y, Helgason CD. (2015, in preparation) Identification of the epigenetic reader CBX2 as a potential drug target in advanced prostate cancer. I am first author of this manuscript. I designed experiments, collected and analyzed data, and wrote the manuscript. Lin D, Wu R, Xue H, Wang Y, and Wang Y were involved in the generation and characterization of the patient- derived xenograft models. Crea F, Chiang YT, Zhang A, Wang JZL, Parolia A, Ding J, and Thu KL provided assistance with experimental procedures and analysis. Lam WL, Shah SP, Collins CC, Wang Y, and Helgason CD critically revised the manuscript.

A version of Chapters 1 and 3 has been accepted for publication. [Clermont PL], Lin D, Crea F, Wu R, Xue H, Wang Y, Thu KL, Lam WL, Collins CC, Wang Y, Helgason CD. (2015) Polycomb-mediated silencing in neuroendocrine prostate cancer. Clinical Epigenetics. I am first author of this manuscript. I designed experiments, collected and analyzed data, and wrote the manuscript. Lin D, Wu R, Xue H, Wang Y, and Wang Y were involved in the generation and characterization of the patient-derived xenograft models. Crea F and Thu KL provided assistance

iii

with experimental procedures and analysis. Lam WL, Collins CC, Wang Y, and Helgason CD critically revised the manuscript.

A version of Chapters 1 and 4 has been published. [Clermont PL], Sun L, Crea F, Zhang A, Parolia A, Thu KL, Lam WL, Helgason CD. Genotranscriptomic meta-analysis of the polycomb gene CBX2 in human cancers: Initial evidence of an oncogenic role. British Journal of Cancer. 2014 Oct 14;111(8):1663-72. I am first author of this manuscript. I designed experiments, collected and analyzed data, and wrote the manuscript. Sun L, Crea F, Zhang A, Parolia A and Thu KL provided assistance with experimental procedures and analysis. Lam WL and Helgason CD critically revised the manuscript.

Table of Contents Abstract ...... ii

Preface ...... iii

Table of Contents ...... v

List of Tables ...... ix

List of Figures ...... xi

List of Abbreviations ...... xiii

Acknowledgements ...... xv

Dedication ...... xvii

1. Introduction ...... 1

1.1. Prostate cancer ...... 1

1.1.1. Overview ...... 1

1.1.2. Prostate development and physiology ...... 3

1.1.3. Prostate tumorigenesis and metastasis ...... 4

1.1.4. Clinical management ...... 5

1.1.5. Castration-resistant prostate cancer ...... 6

1.1.6. Neuroendocrine prostate cancer ...... 8

1.1.7. Experimental models ...... 9

1.2. Polycomb group complexes ...... 11

1.2.1. Epigenetic control of transcription ...... 11

1.2.2. Composition and molecular mechanisms ...... 13

1.2.3. Functions in development and cancer ...... 15

1.2.4. Polycomb-mediated silencing in prostate cancer ...... 16

1.2.5. CBX gene family ...... 18

1.3. CBX2 ...... 19

1.3.1. Structure and function ...... 19

1.3.2. Roles in development ...... 21

1.4. Thesis theme and rationale ...... 23

1.5. Hypotheses and specific aims ...... 24

2. Identification of CBX2 as a potential drug target in advanced prostate cancer ...... 26

2.1. Introduction ...... 26

2.2. Materials and methods ...... 27

2.2.1. Patient-derived xenograft models ...... 27

2.2.2. Bioinformatic database analysis ...... 27

2.2.3. Cell culture ...... 28

2.2.4. qRT-PCR ...... 28

2.2.5. Western blot ...... 30

2.2.6. Microscopy ...... 30

2.2.7. CBX2 siRNA knockdown ...... 30

2.2.8. Caspase 3-7 activity ...... 30

2.2.9. MTT analysis ...... 31

2.2.10. Microarray analysis ...... 31

2.2.11. Immunohistochemistry ...... 31

2.2.12. Statistical analysis ...... 31

2.3. Results ...... 32

2.3.1. CBX2 is over-expressed in metastatic PCa ...... 32 vi

2.3.2. Elevated CBX2 levels associate with clinical PCa progression ...... 35

2.3.3. CBX2 depletion induces cell death in advanced PCa cell lines ...... 40

2.3.4. Analysis of CBX2-regulated genes ...... 44

2.4. Discussion ...... 52

3. Polycomb-mediated silencing in neuroendocrine prostate cancer ...... 56

3.1. Introduction ...... 56

3.2. Methods ...... 57

3.2.1. Clinical expression datasets ...... 57

3.2.2. Gene lists ...... 58

3.2.3. Immunohistochemistry ...... 58

3.2.4. Patient-derived xenografts ...... 58

3.2.5. Statistical analysis ...... 59

3.3. Results ...... 59

3.3.1. Expression profiling of epigenetic regulators in NEPC ...... 59

3.3.2. PcG gene expression in LTL patient-derived xenografts ...... 63

3.3.3. Polycomb silencing and neuroendocrine-associated repression signature ...... 71

3.4. Discussion ...... 74

4. CBX2 meta-analysis in human cancers: evidence of a widespread oncogenic role ...... 77

4.1. Introduction ...... 77

4.2. Materials and methods ...... 78

4.2.1. COSMIC database analysis ...... 78

4.2.2. Oncomine database analysis ...... 78

4.2.3. Human protein atlas database analysis ...... 78 vii

4.2.4. Analysis of the minimal common region of CBX2 amplification ...... 79

4.2.5. Statistical analysis ...... 79

4.3. Results ...... 79

4.3.1. Genomic analysis of the CBX2 locus ...... 79

4.3.2. Transcriptomic analysis of CBX2 expression in human cancers ...... 84

4.3.3. Clinical correlations of differential CBX2 expression ...... 88

4.4. Discussion ...... 95

5. Conclusions ...... 98

5.1. Summary of findings ...... 98

5.2. Main conclusions ...... 99

5.3. Strengths and limitations ...... 101

5.4. Overall significance and impact ...... 103

5.5. Future research directions ...... 104

Bibliography ...... 108

Appendix ...... 130

viii

List of Tables Table 1.1: Core members of PRC1 and PRC2 in drosophila, mouse, and human ...... 14

Table 2.1: qRT-PCR primers ...... 29

Table 2.2: Expression of CBX2 in metastatic and primary prostate tumors compiled in Oncomine

...... 34

Table 2.3: Number and frequency of genomic alterations affecting the CBX2 locus in four independent PCa cohorts from cBIO portal ...... 35

Table 2.4: Multivariate analysis of variance correlating CBX2 and clinicopathological features in the MSKCC cohort ...... 38

Table 2.5: Concentration and purity of RNA used for microarray analysis ...... 46

Table 2.6: List of cancer-related CRGs whose expression correlates with that of CBX2 in the

MSKCC dataset ...... 48

Table 2.7: Expression of CBX2-regulated genes known to be implicated in mitosis following siRNA-mediated CBX2 depletion ...... 49

Table 2.8: Biological processes associated with the list of CRGs ...... 51

Table 2.9: Diseases and functions associated with CRGs ...... 52

Table 3.1: Distribution of 147 investigated epigenetic regulators across different epigenetic modifications, activities, and transcriptional effects ...... 60

Table 3.2: Up-regulation of 22 selected EpRs in metastatic compared to non-metastatic prostate cancer across five independent clinical datasets ...... 62

Table 3.3: Literature-reported direct interactions between PcG complexes and transcriptional repressors up-regulated by at least 1.5 fold in both the clinical NEPC cohort and the 331R/331 model...... 63

Table 3.4: Patient-derived xenograft models from Living Tumor Lab used for comparative

NEPC/PCa analysis and their immunohistologic features ...... 64

Table 3.5: List of top 12 Literature-derived concepts most significantly associated with NEARS

...... 73

Table 3.6: Correlations between down-regulation of NEARS and poor prognostic factors in clinical prostate tumors ...... 74

Table 4.1: List of tumor types harboring a frequency of CBX2 amplification higher than 3% . 81

Table 4.2: Studies with differential CBX2 expression between cancerous and normal tissues .. 86

Table 4.3: Spearman correlation between CBX2 and CDKN2A/B in the datasets displaying

CBX2 up-regulation in malignant compared to normal tissues ...... 88

Table 4.4: List of studies with differential CBX2 expression between metastatic and primary tumors ...... 89

Table 4.5: MANOVA of CBX2 in TCGA colon dataset ...... 91

Table 4.6: MANOVA of CBX2 in Bild lung dataset ...... 91

Table 4.7: MANOVA of CBX2 in Curtis breast dataset ...... 91

Table 4.8: Cox proportional hazards regression for CBX2 in Curtis breast dataset ...... 92

List of Figures Figure 1.1: The clinical progression of prostate cancer ...... 2

Figure 1.2: CBX2 gene and protein ...... 21

Figure 2.1: CBX2 is over-expressed in metastatic PCa ...... 33

Figure 2.2: Hormonal regulation of CBX2 ...... 37

Figure 2.3: CBX2 mRNA expression in normal human tissues (The Human Protein Atlas) ..... 39

Figure 2.4: Relative CBX2 expression in metastatic PCa cell lines LNCaP and C4-2 compared to benign control (BPH1) assessed by qRT-PCR ...... 40

Figure 2.5: CBX2 depletion induces proliferation arrest and apoptosis in advanced PCa cell lines

...... 42

Figure 2.6: Morphology of LNCaP and C4-2 cells following CBX2 depletion ...... 43

Figure 2.7: Gene expression profiling of CBX2-regulated genes (CRGs) ...... 45

Figure 2.8: Assessment of RNA quality by Bioanalyzer in C4-2 cells treated with non-targeting siRNA or CBX2-specific siRNA ...... 47

Figure 3.1: Differential expression of epigenetic regulators in NEPC ...... 61

Figure 3.2: Coordinated increase in PcG gene expression ...... 65

Figure 3.3: CBX2 and EZH2 protein expression in NEPC ...... 67

Figure 3.4: Regulation of PcG proteins CBX2 and EZH2 in lung cancer subtypes ...... 69

Figure 3.5: Expression of CBX2, EZH2, and RB1 in tumor tissue derived from prostate adenocarcinoma or NEPC ...... 70

Figure 3.6: Diagram summarizing the establishment and analysis of a 185-gene

“Neuroendocrine-Associated Repression Signature” (NEARS) derived from NEPC models using the Oncomine resource ...... 71

Figure 4.1: Extremely rare occurrence of genetic mutations disrupting CBX2 function in human cancers (COSMIC database) ...... 80

Figure 4.2: Percentage of CBX2 point mutations in different human cancer types (COSMIC database) ...... 81

Figure 4.3: Mutational analysis of CBX2 protein in human cancers ...... 82

Figure 4.4: List of tumor types harboring a frequency of CBX2 amplification higher than 3%

(COSMIC database) ...... 83

Figure 4.5: Marked up-regulation of CBX2 in cancerous compared to normal tissues (Oncomine database)...... 85

Figure 4.6: CBX2 over-expression is not associated with CDKN2A silencing but co-occurs with

CDKN2B down-regulation in colorectal cancer only ...... 87

Figure 4.7: Differential CBX2 expression associates with clinic-pathologic features ...... 90

Figure 4.8: CBX2 as the driver within the minimal common region of its amplicon ...... 94

xii

List of Abbreviations AC Adenocarcinoma ADT Androgen-deprivation therapy AR Androgen receptor ASO Antisense oligonucleotide BCA Bicinchoninic acid CBX Chromobox CHGA Chromogranin A ChIP Chromatin immunoprecipitation CNG Copy number gain CRG CBX2-regulated gene CRPC Castration-resistant prostate cancer CK Cytokeratin DHT Dihydrotestosterone DMSO Dimethyl sulfoxide DNMT DNA methyltransferase ECL Electrochemiluminescence ELISA Enzyme-linked immunosorbent assay EpR Epigenetic regulator ESC Embryonic stem cell H2AK119ub Ubiquitinated histone H2A lysine 119 H3K27me3 Trimethylated histone 3 lysine 27 H&E Hematoxylin and eosin HDAC Histone deacetylase HOX Homeobox IHC Immunohistochemistry Kb Kilobase KO Knock-out LncRNA Long non-coding RNA MBD Methyl-CpG-binding domain

xiii

MCR Minimal common region MTT 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide NEPC Neuroendocrine prostate cancer NOD-SCID Nonobese diabetic-severe combined immunodeficiency NSE Neuron-specific enolase PCa Prostate cancer PDX Patient-derived xenograft PcG Polycomb group PRC Polycomb repressive complex PSA Prostate-specific antigen PSADT PSA doubling time OS Overall survival RIPA Radioimmunoprecipitation assay SCLC Small cell lung cancer SqCC Squamous cell carcinoma SYP Synaptophysin TNM Tumor, node, metastasis TRAMP Transgenic adenocarcinoma of mouse prostate UGE Urogenital sinus epithelium UGM Urogenital sinus mesenchyme

xiv

Acknowledgements

First and foremost, I would like to express my deepest gratitude to my research supervisor Dr. Cheryl D. Helgason for her phenomenal mentorship throughout my graduate studies. She has believed in me since day one and gave me the freedom to pursue all of my many interests, both scientific and personal. I will always remember our long conversations about research and life in her office. I consider them the cornerstone of my development in graduate school. I have received more from her than I can ever give back, and for that I am eternally grateful.

Dr. Helgason was also one of four members of my supervisory committee, all of whom deserve much recognition. It was a pleasant surprise to realize that Dr. Cheryl Helgason, Dr. Yuzhuo Wang, Dr. Martin Hirst and Dr. Wan Lam were not only great scientists but also great people. They provided amazing guidance throughout this adventure while challenging me intellectually, which allowed me to become a better researcher. I also thank the IOP directors serving during my degree, Dr. Victor Ling who initially encouraged me to join the program and Dr. Angela Brooks- Wilson who greatly helped me throughout my studies, notably with science outreach projects.

In addition to great mentors, I was also blessed with amazing colleagues. Particularly, I want to sincerely thank Dr. Francesco Crea, who I consider both a friend and a role model. Despite his success and busy schedule, he has always taken the time to answer my many questions and listen to my incessant ideas. I hope we will continue to try pushing back the limits of epigenetics together; it has been a real pleasure. In addition, I also had the chance to work alongside very talented undergraduate students who have provided much help to this work and who I hope will also remain colleagues in the future.

Throughout my graduate studies, I closely collaborated with Dr. Yuzhuo Wang’s lab, which has been a vital part of my work and development. First, I must acknowledge the constant support provided by Dr. Wang himself. He was the first to direct me towards a PhD back when my mind was set on a master’s degree. I have learned a great deal from him, and he is the badminton player I hope to one day become. I am also thankful to his students, post-doctoral fellows, and

staff members who have all contributed to this work one way or another, and who have all become great friends in the process.

Next, I would like to acknowledge the tremendous work of our many collaborators: Dr. Wan L. Lam, Dr. Kelsie L. Thu, Dr. Steven Jones, Dr. Fraser Hof, Dr. Colin C. Collins, Dr. Lei Sun, Jiarui Ding, and Dr. Sohrab P. Shah. I am also thankful to all the sources of funding that made this work possible, notably Prostate Cancer Canada and the Canadian Cancer Society Research Institute, as well as all the people who donated their own money to enable our research. Lastly, I am grateful to the cancer patients who bravely agreed to have their tumors analyzed for research purposes. I promise that I have given my best throughout this project in the hope that your generous act can one day benefit others.

Finally, my heart goes out to all the people in Québec who have supported me long before this adventure started. I am privileged to have an incredible brother, two amazing parents, as well as loving family members and friends who deserve much credit for this work. I am looking forward to celebrating with you in a near future.

xvi

Dedication

I dedicate this thesis to my friend Simon Lessard, a young man who has inspired me through the hardest times and whose memory I carry in my heart always.

xvii

1. Introduction 1.1. Prostate cancer 1.1.1. Overview

Globally, cancer represents a significant health problem that arises in more than 14 million people and causes more than 8 million deaths per year [1]. It is now well accepted that cancer is not a single disease, but rather a heterogeneous spectrum of neoplasms that share key hallmarks [2, 3]. While these general features can be found in virtually all malignancies, specific tumor types present more important clinical problems. Currently, neoplasms arising from the breast, lung, prostate, and colon represent about 50% of cancers diagnosed in the western world [4]. Since they occupy a large proportion of the socio-economic burden attributed to cancer, research should be directed at improving their management.

Prostate cancer (PCa), also referred to as prostatic adenocarcinoma, is the most commonly diagnosed cancer in men and represents a leading cause of cancer-related deaths worldwide [5]. Localized PCa presents as an androgen-dependent tumor which is typically slow-growing and is often cured by first line therapy, which usually consists of surgery or radiation therapy [6]. Although most PCas are indolent, approximately 30% of the tumors are metastatic at presentation or acquire resistance to first-line treatment [7]. For the past 30 years, androgen- deprivation therapy (ADT) has remained the standard of care to treat recurrent or metastatic disease [8]. Unfortunately, all patients on ADT eventually become refractory to hormonal treatments and develop castration-resistant prostate cancer (CRPC) (Figure 1.1) [9]. Alternatively, resistance to ADT may be achieved through transdifferentation of PCa into neuroendocrine prostate cancer (NEPC), another highly aggressive prostate malignancy with a particularly poor prognosis [10].

At present, two major problems complicate the clinical management of PCa. First, despite a number of existing prognostic tools, it remains difficult to determine whether a particular primary prostate tumor will stay localized or will progress to a metastatic and lethal state [11- 13]. The second problem is that castration-resistant disease can only be palliated despite the introduction of novel AR-targeting drugs and chemotherapeutic agents [14]. The overall survival

for CRPC varies between 9 and 30 months [15] while NEPC patients typically succumb to their disease about a year following diagnosis [16]. Together these neoplasms account for virtually all PCa-related deaths [17], reflecting an urgent need to develop novel therapeutic strategies with higher clinical efficacy for advanced disease.

Figure 1.1: The clinical progression of prostate cancer (PCa). Localized PCa is a hormone-sensitive disease that can often be cured by surgery or radiation. Metastatic or recurrent PCa is treated with castration, which induces cancer regression. After a variable amount of time, the tumor recurs as castration-resistant prostate cancer (CRPC). Treatments for CRPC are only palliative in nature and include suppressors of AR activity (abiraterone and MDV3100) and taxane-based chemotherapy (docetaxel and cabazitaxel).

1.1.2. Prostate development and physiology

Prostate development is initiated when prostatic buds begin to grow from the urogenital sinus epithelium (UGE) into the urogenital sinus mesenchyme (UGM) at about 10-12 weeks of fetal development [18, 19]. At this point, androgen receptors (ARs) present within the UGM are activated by testicular androgens, resulting in the secretion of key growth factors and cytokines that in turn induce UGE budding, proliferation, and differentiation [19, 20]. Importantly, tissue- recombination experiments have demonstrated that epithelial AR activity is not required for epithelium proliferation but is necessary to maintain secretory functions [21, 22]. As development proceeds, the UGE gives rise to prostate cells which are positive for both basal and luminal markers [23]. These progenitor cells terminally differentiate into prostatic basal cells expressing cytokeratin 5 (CK5), CK14, CK19, p63, and GSTpi or into luminal cells displaying AR, CK8, and CK18 immunoreactivity [23]. On the other hand, the UGM differentiates into prostatic smooth muscle cells and fibroblasts, which forms a dense fibromuscular stroma encapsulating the epithelium [24, 25]. Additionally, rare neuroendocrine cells are scattered within the prostate and take part in homeostatic interactions with basal and luminal cells [18, 26].

At birth, the human prostate weighs about 2g and remains about that size until puberty, when a pubertal elevation in serum testosterone levels stimulate AR activity, which in turn induces prostate growth until it reaches the size of a walnut (about 20g) [19]. Anatomically, the adult prostate is located between the bladder and the penis, with the urethra running through the prostate [27]. The prostate is subdivided into three main areas: the central, peripheral, and transition zones [28]. From a histological standpoint, the prostate represents a tubuloalveolar gland made up of ducts lined with a pseudostratified columnar epithelium which is surrounded by a dense fibromuscular stroma composed mostly of smooth muscle cells [29]. The primary function of the prostate is to secrete the main constituents of the seminal fluid, including seminogelin, fibronectin, lactoferrin, and prostate-specific antigen (PSA) [30]. The secretory ability, as well as the survival of adult luminal prostate cells, has been shown to be dependent on androgens, particularly dihydrotestosterone (DHT) [31]. Accordingly, castration causes regression of the prostate via apoptosis, implying a fundamental role for hormonal regulation in normal prostatic cells [21, 32]. 3

1.1.3. Prostate tumorigenesis and metastasis

Over time, the prostate is subjected to a number of inflammatory insults and oxidative damage that predispose prostate cells to molecular abnormalities [33, 34]. As a result, genomic alterations can accumulate in a stepwise fashion and disrupt homeostatic epithelium-stroma interactions, thereby leading to prostatic cell de-differentiation and tumorigenesis [35]. Multiple lines of evidence support a model in which PCa is derived from a precursor lesion called prostate intraepithelial neoplasm, a majority of which arise from cells located in the peripheral zone of the prostate [36-38]. A number of factors influence the risk of developing PCa [39]. Age is by far the biggest risk factor for PCa incidence, with a majority of patients presenting after the age of 65 [40, 41]. Other reported risk factors include family history [42], race [40], diet [43], exposure to high levels of testosterone [44], and other environmental agents [45]. Overall, prostate tumorigenesis represents a highly complex phenomenon that likely results from the combination of multiple genetic and epigenetic alterations accumulated during the aging process.

Molecular pathology of primary PCa is an active area of investigation and a number of genes involved in PCa initiation and progression have been identified [46]. Collectively, these studies have demonstrated that the AR represents a central driver of PCa [47]. Outside of AR activation, the molecular landscape of PCa is characterized by extensive heterogeneity, both within and across prostate tumors [48]. The most frequent genetic events are TMPRSS-ERG translocation [49], 8q24 amplification [50], PTEN loss [51], and NKX3.1 loss [52]. Interestingly, it is the epigenetic silencing of GSTP1 by DNA methylation which represents the most frequent molecular aberration in PCa, found in about 90% of patients [53]. Since GSTP1 neutralizes potential DNA-damaging agents, its inactivation favors the accumulation of genetic mutations that can further promote PCa progression [54, 55].

While most prostate tumors stay confined to the prostate, a sizeable fraction (~20%) will eventually disseminate and cause lethal distant metastases [56]. PCa metastasis is a multi-step process that requires extensive phenotypic plasticity as multiple selective barriers hamper the spread of tumor cells [57, 58]. Bone is the most common metastatic site of PCa, with osseous metastases found in more than 90% of disseminated PCa patients [59]. Bone metastases can

induce excruciating symptoms such as pain, pathological fractures, nerve compression syndromes, and hypercalcemia [60]. Additionally, a smaller fraction of PCa patients will develop metastases in the lung, liver, pleura, and adrenal glands [7]. The sequence of molecular events leading to PCa metastasis remains incompletely understood, implying the need to identify novel drivers of metastatic dissemination [61].

1.1.4. Clinical management

The clinical management of PCa is highly interdisciplinary and can be divided into detection, diagnosis, and treatment of the disease. Typically, PCa screening is carried out through digital- rectal exam and PSA testing [62]. PSA is an AR-regulated gene which is normally secreted in the prostatic lumen to ultimately integrate within the seminal fluid [63]. Since PCa cells lose their normal polarity, PSA can instead be secreted into the blood and its levels can be quantified using enzyme-linked immunosorbent assay (ELISA) [64]. Thus, since PCa cells display robust AR activity, higher PSA levels correlate with increased probability of disease [63]. However, the PSA test remains controversial as it can lead to overtreatment given that a large number of prostate tumors diagnosed via PSA screening remain clinically asymptomatic [65]. Patients with more aggressive PCas may present with symptoms such as urinary issues, weight loss, fatigue, and pain [66]. Currently, it is recommended that men aged 55 to 69 years who are at average risk and asymptomatic should undergo PSA screening [67].

If the presence of PCa is suspected, a prostate biopsy is definitive for diagnosis. The biopsy consists of removing 12-14 small needle-core samples from pre-determined locations within the prostate [68]. The procedure is typically performed with the help of imaging modalities such as ultrasound or magnetic resonance imaging [69, 70]. The harvested tissue is subsequently analyzed by haematoxylin and eosin (H&E) staining as well as immunohistochemistry (IHC) [71]. If the diagnosis of PCa is made, the Gleason score and Tumor, Node, Metastasis (TNM) score is used to help define patient prognosis [72]. The Gleason score consists of microscopic analysis of PCa morphology and differentiation to establish tumor grade, with a higher score correlating with higher grade [73]. TNM staging is another widely used method to establish patient prognosis which relies on the size of the tumor, as well as the number of nodes and the

presence of distant metastatic disease. [74]. Other important factors employed in determining prognosis include the status of surgical margins, the PSA level at diagnosis, and the PSA doubling time (PSADT) [75, 76]. Despite these methods, evaluating the prognosis of PCa patients remains a challenge and new biomarkers are highly desirable to improve patient care [77].

A critical factor in treatment selection is the presence of metastasis, as primary and metastatic tumors are treated very differently. Localized PCa is often effectively cured by current therapies, while disseminated tumors are inevitably fatal and can only be palliated [14, 78]. In early stage tumors, oncologists may recommend that their patients forego any therapy and instead be followed through active surveillance [79]. However, in more than 75% of cases, localized PCa will be treated by surgery or radiotherapy [78]. For patients who progress following first line therapy or present with metastatic disease, the standard of care is ADT, which may be achieved surgically or more frequently chemically [8]. ADT works by reducing serum testosterone levels through interference with the hypothalamic–pituitary–adrenal axis [80]. ADT inhibits AR activity in PCa cells, resulting in widespread apoptosis [8, 81]. While almost all PCas have a significant initial response to ADT, virtually all tumors will eventually become refractory to hormonal treatments and progress to a lethal state referred to as CRPC [9].

1.1.5. Castration-resistant prostate cancer

After a latency period of about two to three years, prostate tumors treated with ADT will re- emerge as CRPC [82, 83]. The median overall survival of CRPC patients is about 16-18 months, and novel therapeutic agents recently approved for CRPC can only extend survival a few months at best (see Figure 1.1) [84]. Currently, the recommended first line of treatment for patients who develop their first recurrence after ADT but remain asymptomatic is abiraterone, a potent inhibitor of androgen biosynthesis [14, 85]. Patients with extensive metastases or that progress on abiraterone will be offered docetaxel, a taxane-based chemotherapeutic drug [86]. Once resistance to docetaxel is achieved, options become limited for patients but include AR- suppressing drugs such as abiraterone [87] or MDV3100 [88] and more recently cabazitaxel [89], another taxoid derivative. In addition, bone-targeted agents may be offered to prevent skeletal-

related events associated with bone metastases, which are found in a majority of CRPC patients [90].

There is ample evidence demonstrating that ligand-dependent AR transactivation can persist during ADT through various mechanisms, resulting in castration resistance [91]. First, AR itself can be amplified and over-expressed [92]. Alternatively, co-factors that enhance AR activity are also found to be over-expressed in CRPC, enabling sustained AR activity despite low androgen levels [93]. In addition, point mutations within AR have been found to stimulate its transactivation even upon binding with an AR antagonist, thereby causing resistance [94]. Another strategy employed to sustain AR signaling is to increase androgen biosynthesis by adrenal glands or the tumor itself, leading to extra- and intra-tumoral production of androgens able to sustain AR transactivation [95, 96]. Thus, CRPC cells rapidly acquire resistance to drugs suppressing ligand-dependent AR activity [14], implying that androgen-independent mechanisms represent critical drivers of CRPC progression and drug resistance.

Castration resistance may also be achieved by additional mechanisms that are independent of ligand-dependent AR activity. For example, CRPC cells often express alternative AR splice variants that lack the ligand-binding domain and therefore remain constitutively active in the absence of androgens [97]. Furthermore, AR may be stimulated by specific post-translational modifications such as phosphorylation, which also induces its transactivation in the absence of ligand binding [98, 99]. These modifications are notably catalyzed by members of oncogenic signal transduction pathways, which has sparked much interest in understanding the link between hormonal regulation and molecular signaling [100, 101]. Along these lines, there is a growing body of evidence demonstrating that CRPC cells can activate pro-survival pathways that bypass the need for AR activity [102-104]. Notably, recent molecular profiling of advanced PCa has revealed recurrent activation of the PI3K/AKT, RAS/RAF, MYC, and many growth factor signaling pathways, all of which are thought to drive disease progression in an AR-independent manner [46, 105].

1.1.6. Neuroendocrine prostate cancer

Further complicating the clinical problem of PCa progression, CRPC cells may acquire resistance to hormonal therapy by evolving into lethal NEPC [106]. This has important clinical implications because NEPC cells are highly aggressive and do not express AR, which renders them insensitive to hormonal therapies offered in late-stage disease [106]. At present, the experimental evidence supports a model in which NEPC is clonally derived from prostate adenocarcinoma via neuroendocrine transdifferentiation [10]. Notably, it has been demonstrated that the frequency of TMPRSS-ERG rearrangement is similar in CRPC and NEPC, supporting the idea that NEPC is clonally derived from prostate adenocarcinoma [107]. Moreover, in vitro studies have demonstrated that androgen-sensitive LNCaP cells can reversibly acquire neuroendocrine features in response to androgen depletion [108]. It is therefore anticipated that the incidence of NEPC will greatly increase as a consequence of the recent approval of AR- suppressing agents, indicating an urgent need to improve NEPC detection and management [106].

At present, NEPC represents the most common extra-pulmonary neuroendocrine tumor and can be found in up to 25% of CRPC patients [106]. In addition, a small fraction of NEPC may arise de novo [109]. The aggressiveness of NEPC represents a significant clinical problem as current treatments with platinum-based chemotherapy only elicit transient responses [110]. Most studies report a median overall survival of 10-13 months in NEPC patients [111]. NEPC is also distinct from prostatic adenocarcinoma as it preferentially metastasizes to visceral organs and less frequently to osseous structures [112]. Furthermore, NEPC cells have absent to low AR expression and therefore disease progression occurs without eliciting an increase in PSA [113]. In addition, biopsies are rarely performed in late stage patients, further contributing to NEPC under-diagnosis [106]. Taken together, these factors make NEPC a very difficult neoplasm to detect and treat, reflecting a need to develop more effective biomarkers and therapies.

From a histological standpoint, NEPC is characterized by the presence of small cells with a prominent nucleus and scant cytoplasm [114]. Typical features of neuroendocrine tumors include the presence of cytoplasmic eosinophilic granules, hyperchromatic nucleus, salt-and-pepper

chromatin, and frequent mitoses [115]. Given their very high proliferative rate, NEPC cells are often found arranged in a monomorphic pattern with very little space between each tumor cell [116]. IHC can be conducted to diagnose NEPC using antibodies specific to neuroendocrine markers, which include chromogranin A (CHGA) and synaptophysin (SYP) [116]. In order to improve detection and diagnosis of NEPC, serum levels of specific neuroendocrine markers have been evaluated as biomarkers but have offered only limited success [117, 118].

While the molecular landscape of NEPC remains widely uncharacterized, alterations in specific genes have been reported [119]. For example, genetic inactivation of the retinoblastoma (RB) and p53 tumor suppressors is highly recurrent in NEPC [120, 121]. Emerging literature also supports a role for amplification and over-expression of AURKA and MYCN [122, 123]. Interestingly, NEPC also shares similarities with CRPC, notably PI3K/AKT activation and RB loss [17] suggesting that molecular alterations present in CRPC are selected for and retained in NEPC. In addition to genetic alterations, epigenetic events such as over-expression of the chromatin regulator EZH2 have been reported in NEPC [123]. Consequently, there is a growing interest in identifying epigenetic alterations that drive NEPC progression as they could represent attractive therapeutic targets.

1.1.7. Experimental models

A number of pre-clinical and clinical strategies have been developed to study PCa. The advent of high throughput and low-cost profiling techniques has enabled the genomic and transcriptomic analysis of numerous clinical tumor samples [124]. These results have been compiled in publically available datasets that researchers can use to generate new hypotheses or validate findings observed in pre-clinical studies. For example, cBIO portal [125], Oncomine [126], and KMplotter [127] can be mined to identify or validate molecular changes associated with human cancers. However, these analyses must be complemented with further studies in pre-clinical testing. As such, animal models developing spontaneous tumors such as dogs and rats have been employed for pre-clinical investigation [128]. Mice do not develop spontaneous PCa but some genetically-engineered mice can give rise to prostate tumors [129]. The two most widely used murine models include the transgenic adenocarcinoma of mouse prostate (TRAMP) induced by

constitutive expression of SV40 T antigen [130] as well as the PTEN deletion mice [131]. These models offer the advantage of having the tumor and the stroma from the same species, however, the biology of spontaneous animal tumors strongly differs from human PCa, making them sub- optimal models [132].

To circumvent the species-specific nature of PCa, a number of human cell lines have been developed. To date, most of the PCa research has been conducted using the PC3, DU145, LNCaP, and C4-2 cell lines [133]. These allow for easy genetic manipulations in an in vitro system. However, PC3 and DU145 represent cell lines that display low expression of classical adenocarcinoma and neuroendocrine genes [133]. These tumor cells referred to as “double negative” comprise a minority of CRPC tumors and therefore do not reflect the biology observed in the clinic, which limits their translational relevance. Thus, the isogenic LNCaP/C4-2 model has remained popular as it represents a validated and clinically-relevant model of PCa progression, and other metastatic sublines have also been generated from this model [134]. LNCaP was originally derived from a lymph node metastasis and subsequently implanted into a castrated mouse, which gave rise to a castrate-resistant cell line C4-2 following ADT [135]. Both LNCaP and C4-2 express AR, but only LNCaP exhibits androgen-responsive growth [135]. Moreover, C4-2 xenografts display higher tumor formation and produce more metastatic foci in vivo, consistent with the idea that androgen-independent cells are inherently more aggressive [136].

The establishment of PCa cell lines is difficult and thus the number of cell lines available is limited [137]. In addition, cells cultured in vitro lack the microenvironment that plays fundamental roles in defining PCa cell phenotype [138]. To address this issue, xenografts consisting of human PCa cell lines were developed in mice, where the microenvironment more accurately reflects clinical conditions. Cell line xenografts offer the advantage of conducting genetic manipulation in vitro and subsequently observing the phenotypic effects in vivo. In mice, three main sites have been used for xenografting: subcutaneous, orthotopic, and subrenal [132]. While subcutaneous xenografts are easy to implant and monitor, they are poorly vascularized and therefore poorly reproduce the PCa tumor microenvironment [132]. On the other hand, orthotopic models provide more realistic tumor-stroma relationship but are technically 10

challenging and do not allow for serial transplantation [138]. Providing many advantages over all other approaches, the subrenal capsule has emerged as the preferred site of xenograft. The subrenal capsule is vascularized, thus offering high rates of engraftment and the possibility of serial transplantation, although these experiments typically take longer to conduct and tumor growth is more difficult to monitor [139].

One major limitation of cell line xenografts is that they are homogeneous and lack a human stromal microenvironment, which plays critical roles in the molecular regulation of PCa progression [138]. In recent years, technological advances have allowed the direct implantation of patient tumor tissue into mice, giving rise to patient-derived xenograft (PDX) models. In contrast to cell lines, PDXs allow for interactions between the tumor and its stroma [139]. Many studies have demonstrated that the genomic profile of PDXs is highly homologous to the tumor patient tissue from which they were derived and remains very stable after serial transplantation, providing an optimal model to investigate the molecular basis of PCa progression [139]. However, one limitation of PDX models is that genetic manipulation cannot be done prior to grafting, thereby eliciting the need to combine experiments in PDX models with other in vivo and in vitro models. Nonetheless, studies using isogenic PDX models of PCa implanted in the subrenal capsule have revealed a high degree of genomic similarity between metastatic prostate tumors and their localized counterpart, suggesting that epigenetic mechanisms may be driving disease progression [139]. In particular, epigenetic alterations involving the Polycomb Group (PcG) family of transcriptional repressors have been associated with the development of PCa metastasis and castration resistance [140, 141].

1.2. Polycomb group complexes 1.2.1. Epigenetic control of transcription

Epigenetics refers to the study of heritable changes in gene expression that are not related to changes in DNA sequence [142, 143]. Epigenetic regulation is thought to provide cells with phenotypic plasticity, allowing them to alter their gene expression patterns in response to environmental cues [144]. As a consequence, such changes play important roles in developmental and pathological processes [145]. The key unit of epigenetic regulation is the

nucleosome, which represents a macromolecular structure composed mostly of DNA and histone proteins [146]. Each nucleosome is made up of an octameric core of histone proteins around which about 147 base pair of DNA can assemble [147, 148]. The interaction between DNA and histones is modulated by covalent modifications that can be catalyzed on histone N-terminal tails [149]. In addition, DNA can also be methylated, which influences the interactions between DNA and histones [150, 151]. Important histone marks include methylation, acetylation, ubiquitination, sumoylation, phosphorylation, ADP-ribosylation, citrullination, and biotinylation [152]. These modifications therefore regulate the three-dimensional organization of nucleosomes into chromatin, which in turn controls the transcriptional competency of surrounding loci [146].

Epigenetic regulation is conferred by a distinct code of covalent DNA and histone modifications mediated by three major types of proteins: ‘‘writers’’, ‘‘erasers’’, and ‘‘readers’’ of the epigenetic code [142, 147]. Writer proteins catalyze the addition of covalent marks that can be recognized by reader proteins, which can trigger changes in chromatin structure [153]. Reflecting the plastic nature of the epigenome, these histone modifications are reversible due to the existence of eraser proteins, which mediate removal of chemical modifications [154]. The histone code consequently provides cells with a highly plastic regulatory system that allows the integration of external cues into transcriptional responses [155].

Recent advances in DNA sequencing technologies have allowed unprecedented insights into the genomic landscape of human tumors [156]. A unifying theme is emerging: the majority of cancers display extensive inter- and intra-tumoral genetic heterogeneity, with very few neoplasms resulting from a single recurrent genomic aberration [157, 158]. In parallel with those observations, many lines of evidence suggest that epigenetic alterations on their own or in combination with genetic lesions drive cancer initiation and progression [159, 160]. In the past decade, a number of chromatin regulators known to play key molecular functions in embryonic development have been identified as novel cancer-promoting genes [140]. Since epigenetic events do not alter the DNA sequence itself, they are reversible and thus the molecules responsible for inducing or reactivating epigenetically silenced genes could provide attractive therapeutic opportunities [161].

In PCa, emerging evidence suggests that aberrant methylation of histone H3 represents a central driver of disease progression [162, 163]. A number of reports have shown that histone methylation patterns are associated with metastatic spreading and poor clinical outcome in PCa [162]. More specifically, the PcG family of epigenetic repressors has been intimately linked to PCa progression [140, 164]. Disruption of aberrant PcG-mediated silencing has been shown to induce PCa cell death in pre-clinical models [164], suggesting that individual PcG proteins may represent attractive drug targets.

1.2.2. Composition and molecular mechanisms

The Polycomb Group (PcG) family of genes encodes key regulators of embryonic development and accordingly PcG genes have been conserved throughout evolution [165]. They were first discovered in mutagenesis screens in drosophila, where mutations in PcG genes led to severe developmental defects [166]. PcG proteins assemble into two main Polycomb Repressive Complexes, PRC2 and PRC1 (Table 1.1) [167, 168]. In drosophila, PRC2 is composed of E(z), Esc, and Su(z), for which there is only one isoform of each. The drosophila PRC1 members include Pc, Ph, Psc, and dRing [169]. Interestingly, all PRC1 genes have undergone duplications during evolution, giving rise to multiple homologs [165]. For example, there are 5 human homologs of Pc, all of which share similarities and differences, increasing the diversity of molecular interactions involving PRC1. It is thought that genetic changes in loci encoding PcG members have contributed to generating key phenotypic adaptations that allowed species evolution over time [170].

Table 1.1: Core members of PRC1 and PRC2 in drosophila, mouse, and human.

Complex Drosophila Mouse Human Esc Eed EED Ezh1 EZH1 PRC2 E(z) Ezh2 EZH2 Su(z)12 Suz12 SUZ12 M33 CBX2 MPc2 CBX4 Pc Cbx6 CBX6 Cbx7 CBX7 MPc3 CBX8 Pcgf1 PCGF1 Pcgf2 PCGF2 Pcgf3 PCGF3 PRC1 Psc Bmi1 BMI1 Pcgf5 PCGF5 Pcgf6 PCGF6 Phc1 PHC1 Ph Phc2 PHC2 Phc3 PHC3 Ring1 RING1 dRing Rnf2 RNF2

In the classical model of PcG-mediated repression, PRC2 trimethylates histone H3 at lysine 27 (H3K27me3) through the SET domain of its catalytic subunit EZH2 [171], which promotes transcriptional repression of target genes [172]. In a coordinated fashion, H3K27me3 can be directly recognized by one of five chromodomain-containing proteins (CBX2, 4, 6, 7, 8), which subsequently recruit PRC1 to chromatin by simultaneously interacting with the E3 ubiquitin

ligases RING1A or RING1B [173]. This interaction brings RING1A/B and PRC1 to H3K27me3 sites, where RING1A/B ubiquitinates lysine 119 on histone H2A (H2AK119ub), a chromatin mark associated with gene silencing [174, 175]. In addition, PRC recruitment at chromatin is also regulated by interactions with long non-coding RNAs such as HOTAIR and ANRIL, some of which have been shown to bind CBX proteins [176, 177]. Thus, CBX proteins serve as the molecular bridge between PRC2 and PRC1 activity, implying a fundamental role in orchestrating PcG-mediated silencing [178].

Additionally, PRCs can also act independently of each other as recent chromatin immunoprecipitation (ChIP) studies have shown that PRC1 can silence genomic regions which are not marked by H3K27me3 [179, 180]. Furthermore, PRC1 complexes containing the protein RYBP and those that include a CBX subunit were found to be mutually exclusive [181]. Notably, the activity of the PRC1-RYBP complex has been shown to be mediated by H2A ubiquitination while CBX-containing PRC1 complexes were devoid of ubiquitin ligase activity despite containing Ring1b and retaining gene silencing properties [180, 182]. This function has been partly attributed to the ability of PRC1 to induce chromatin compaction [183-185]. Alternatively, studies have also demonstrated that PRC1 localization may also affect the genomic distribution of PRC2 and H3K27me3 [186]. Taken together, mechanistic studies have revealed that the composition of PcG complexes varies extensively from one cellular context to the other, and that this combinatorial complexity is reflected in their diverse molecular mechanisms of action [168].

1.2.3. Functions in development and cancer

Assembling into two main PcG complexes (PRC1/2), PcG proteins regulate several hundred genes, many of which encode transcription factors involved in cell fate decisions [187]. PcG genes are thought to be essential in both stem and differentiated cell types, and most likely are responsible for allowing the alterations in gene activity that accompany changes in cell identity [188]. In undifferentiated cells, PcG proteins are highly expressed and maintain lineage-specific genes in a transcriptionally repressed state [189]. The classic example of PcG-mediated repression involves the regulation of the HOX loci [190]. In ESCs, PcG complexes are found at

HOX genes, which are transcriptional drivers of differentiation [191]. As embryogenesis occurs, HOX genes undergo loss of PcG-mediated silencing, which allows the expression of specific HOX transcripts [191]. Based on the timing and location of their expression, HOX proteins will specify segmental identity within the developing animal [192]. Thus, mutations in PcG genes lead to abnormal regulation of HOX gene expression, resulting in segmental defects known as homeotic transformations [166]. These phenotypes are also observed upon germline deletion of some PcG members in mammals [193], implying a fundamental role for PcG-mediated silencing during embryonic development across numerous species.

Importantly, incorrect regulation of PcG complexes has also been demonstrated to play inherent roles in cancer initiation and progression [140, 176, 194]. In human tumors, many embryonic transcriptional programs are regulated by PcG proteins and push tumor cells towards a more aggressive state [187]. Notably, PcG complexes silence key tumor suppressors in cancer and down-regulation of PcG targets correlates with poor prognosis in many cancer types [195]. In hematopoietic malignancies, both gain-of-function and loss-of-function mutations in PcG genes have been identified, suggesting some level of context-dependency [196, 197]. Although PcG proteins may act through different pathways in different cancer types, PcG gain of function generally associates with an undifferentiated cellular state and aggressive clinical behavior in solid tumors [187, 198].

1.2.4. Polycomb-mediated silencing in prostate cancer

In the past decade, accumulating evidence has demonstrated that epigenetic alterations involving the PcG family are important drivers of PCa progression [140, 199]. Notably, altered PcG- mediated silencing has been recognized as a key event mediating tumor metastasis and drug resistance through aberrant epigenetic regulation [200, 201]. Accordingly, down-regulation of a PcG repressive signature is associated with poor prognosis in PCa [195]. Furthermore, pharmacologic inhibition of PRC2 activity induces PCa cell death, further establishing a functional requirement for PcG activity [164]. However, the molecular mechanisms through which PcG complexes become aberrantly targeted to specific tumor suppressive loci remain

poorly understood. To date, EZH2 and BMI1 represent the two PcG genes most strongly implicated in the progression of PCa [141].

A well established oncogene, EZH2 is the catalytic subunit of PRC2 which catalyzes methylation of H3K27. A number of studies have demonstrated that EZH2 becomes over-expressed at both the mRNA and protein levels during PCa progression, which in turn leads to aberrant regulation H3K27me3 distribution [140, 202]. Importantly, elevated EZH2 levels significantly correlate with metastatic dissemination, castration resistance, and poor overall survival in PCa patients [140]. Moreover, Crea et al. have shown that pharmacological targeting of EZH2 could reduce tumorigenicity and metastatic ability in advanced PCa cells, further supporting a role for EZH2 in CRPC [164]. In addition, there is evidence indicating that altered PRC2 activity also affects the regulation of HDACs [203, 204], potentially stabilizing an altered epigenetic state driving disease progression. Interestingly, emerging evidence suggests that the sequence specificity of EZH2 is controlled by interactions with long non-coding RNAs (lncRNAs) [171, 205, 206]. Studies have shown that EZH2 activity lies downstream of oncogenic regulators such as AKT and E2F [207, 208], providing a possible link between EZH2 over-expression and aggressive tumor growth.

In addition to aberrant PRC2 regulation, dysregulated activity of PRC1 has also been linked to PCa progression. To date, most studies have focused on BMI1, which is known to enable replicative and long term self-renewal properties in PCa cells [209]. BMI1, also known as PCGF4, does not possess intrinsic enzymatic activity but it takes part in important protein- protein interactions that regulate PRC1 activity [210]. BMI1 expression is elevated in PCa cells and is also linked to clinical prognosis [211]. Notably, increased BMI1 activity has been observed upon PTEN deletion, which represents a frequent event in disease pathogenesis [212]. It was found that BMI1 silencing potentiated the effect of docetaxel, a key drug in CRPC management, via modulation of the anti-oxidant response [213]. While the tumor-promoting roles of BMI1 have been well documented, very little is known about how BMI1 and PRC1 get recruited to chromatin.

1.2.5. CBX gene family

Representing the molecular bridge between PRC2 and PRC1, CBX proteins have become increasingly recognized as active players in cancer initiation and progression. The human genome encodes 5 distinct CBX proteins (CBX2, 4, 6, 7, 8) which share a conserved N-terminal chromodomain responsible for H3K27me3 binding [165]. Individual CBX family members display non-homologous sequences in their C-terminus, accounting for their non-redundant binding partners and functions [214]. Distinct CBX proteins interact with a wide range of macromolecules including DNA, non-coding RNAs, and numerous other proteins [215, 216]. Furthermore, individual CBX family members can be differentially expressed, undergo alternative splicing, harbor distinct post-translational modifications and lie under the control of different microRNAs (miRNAs) [217]. These regulatory steps act in concert to control the function of each CBX protein and contribute to the complexity of PRC1 activity and sequence specificity [218].

An increasing number of reports link dysregulation of CBX proteins to human tumorigenesis. CBX4 possesses SUMO activity and is thought to play important roles in cellular proliferation and DNA damage repair in some human tumors [219]. Conversely, SNPs in CBX6 have been reported in genome-wide association studies of bladder cancer although their functional implications have not been fully elucidated [220]. CBX7 has been the most extensively investigated CBX family member and exhibits cancer type-specific activity in human tumors. Most studies report a widespread oncosuppressive function [221], while other investigations suggest a tumor-promoting role [222]. Importantly, antagonists of the CBX7 chromodomain have recently been developed and therefore similar chromodomain-targeting strategies could be employed to inhibit other CBX proteins [223, 224]. Finally, CBX8 appears to be essential in hematological malignancies, specifically in MLL/AF9 leukemogenesis [225] in addition to its ability to silence the CDKN2A locus [226]. While these four CBX proteins have been well characterized in cancer, the contribution of CBX2 to human tumorigenesis and metastasis has remained unexplored.

1.3. CBX2 1.3.1. Structure and function

CBX2 represents a member of the human CBX family and is known to play critical functions in embryonic development [227]. CBX2 is the structural and functional counterpart of the drosophila Pc, which is the only CBX homolog in the fruit fly [228, 229]. The mouse homolog of CBX2 is M33 [229], and re-introduction of M33 in Pc mutants partially rescues the mutant phenotype, suggesting an evolutionarily conserved function [230]. In humans, the CBX2 gene is located at the tip of 17q25.3 close to a telomeric region [227]. At this locus, CBX2 is flanked by CBX4 and CBX8, which are thought to have evolved from Pc following a gene duplication event [165]. The CBX2 gene is transcribed as two isoforms, CBX2.1 and CBX2.2 [231]. CBX2.1 represents the full length protein, while CBX2.2 lacks a large portion of the C-terminus [231]. For the purpose of this thesis, CBX2.1 will be referred to as CBX2 unless otherwise stated. The full length CBX2 carries many molecular features that distinguish it from the short version, and the literature supports primarily the former as the functional counterpart [232].

The chromodomain is an important structural feature of CBX2 since it allows chromatin binding [173]. More specifically, the chromodomain binds H3K27me3 which is located within an alanine-arginine-lysine-serine (ARKS) sequence on histone H3 [173]. Deletion of the CBX2 chromodomain impairs CBX2 chromatin binding, implying an important functional role [214]. Moreover, CBX2-containing PRC1 complexes are enriched at H3K27me3 sites compared to PRC1 harboring other CBXs [214]. However, not all H3K27me3 sites are bound by CBX2, indicating that additional factors regulate the genomic localization of CBX2 [232]. CBX2 sequence specificity can partly be explained by the set of proteins interacting with CBX2 in a given cellular context, which would favor CBX2 recruitment at particular genomic loci [181]. As described above, it is important to note that there exist many members of each PRC1 gene family. CBX2 has been shown to be preferentially associated with RING1B, BMI1, and PCGF2, as demonstrated by immunofluorescence co-localization and reciprocal co-immunoprecipitation studies [181, 233]. It was demonstrated that this association is mediated through a C-terminal domain called the Pc box, which serves as an adaptor for protein-protein interactions with other

PRC1 members [165, 214, 234]. Thus, through the combined function of its N-terminal chromodomain and its C-terminal Pc box, CBX2 can mediate PRC1 recruitment at chromatin.

While CBX2 shares common features with other CBX proteins, it also distinguishes itself from other family members by the presence of key structural features that confer CBX2 a particularly critical role in epigenetic regulation (Figure 1.2) [165]. CBX2 harbors an AT-hook region, a DNA-binding domain that preferentially recognizes repetitive sequences rich in AT repeats [165]. This domain has been shown to facilitate binding of the chromodomain to H3K27me3 in regions of repetitive AT sequences, notably at heterochromatic regions [235]. While PRC1 has long been associated with chromatin compaction, emerging evidence suggests that this activity could be mediated through a short positively charged region within the CBX2 protein [183]. These findings complement earlier studies demonstrating that during metaphase, CBX2 is enriched in distinct foci on chromosomes and overlaps regions of Hoechst staining, indicating that CBX2 is localized at regions of condensed chromatin [214, 232, 236]. Further supporting a role in chromosomal compaction, CBX2 is involved in X inactivation and can be found at heterochromatic sites throughout the inactivated X chromosome [214, 237]. Interestingly, phosphorylation of CBX2 induces its nuclear translocation and increases its selectivity for H3K27me3, in line with cell cycle-regulated regulation of CBX2 activity [238, 239]. Overall, the CBX2 protein carries many features that make it a unique factor in PcG-mediated repression.

Figure 1.2: CBX2 Gene and Protein. A) Functional domains within the full length CBX2 protein. B) 3D structure of CBX2 chromodomain (source: Protein Data Bank, #2D9U) C) 3D structure of CBX2 chromodomain interacting with H3K27me3 peptide (source: Protein Data Bank, #3H91).

1.3.2. Roles in development

Germline deletion of CBX2 induces important developmental defects in animal models [228]. Initial functional characterization of CBX2 was conducted in drosophila, where mutations in Pc induce abnormalities in the anterior-posterior axis known as homeotic transformations [240]. However, much of the literature relating to CBX2 is derived from studies of M33-knock out (M33-KO) mice, where germline deletion of CBX2 induces severe developmental defects [193]. More than 90% of M33-KO mice die post-natally within 4 weeks, and those who survive weigh only a quarter of their wild type counterpart [228]. As in drosophila, homeotic transformations are observed in M33-KO mice, although they mostly implicate the skeletal system in murine 21

models [228]. In addition, M33-KO mice are unique in that XY individuals are unable to achieve development of the male urogenital system [193]. While these phenotypes indicate a role in differentiation, there is also extensive evidence demonstrating that proliferation is impaired in M33-KO mice, as suggested by the reduced weight and hypocellular nature of many organs [228]. M33-KO mice give rise to mice with some traits such as homeotic transformations which overlap those of BMI1-KO mice, suggesting that these phenotypes are mediated through PRC1 activity [241]. However, some phenotypes are exclusive to M33-KO mice, implying a non- redundant role for CBX2.

Consistent with the phenotypes observed in M33-KO mice, emerging data indicate that CBX2 plays an important role in the transcriptional control of cellular differentiation [242]. Recent evidence demonstrates that CBX2 is silenced in ESCs but is induced upon differentiation, which promotes the expansion of multipotent progenitor cells [217, 218]. This is consistent with the homeotic transformations affecting the skeletal system of M33-KO mice, which are thought to arise from early progenitor cells with high replicative and migrative ability [241, 242]. While homeotic defects can be seen in mice deficient for other PcG members, the sex-reversal observed in M33-KO mice with an XY karyotype is the most unique and striking [193]. M33-KO mice that are karyotypically XY cannot undergo development of the male sexual and urogenital systems and phenotypically present as females [193]. It has also been reported that a human with female anatomy carried an XY karyotype [243]. Genomic analyses revealed homozygous mutations of the CBX2 gene, in line with the sex-reversal observed in M33-KO mice [193]. Since development of the male urogenital system is tightly linked to AR stimulation, it is likely that CBX2 activity lies under hormonal control [244, 245]. Additionally, M33-KO mice with an XX karyotype also exhibit sexual abnormalities such as small ovaries and lack of vaginal orifice, further supporting a role for CBX2 in cells that are under hormonal regulation [193].

In addition to its role in differentiation, CBX2 has also been shown to promote cellular cell cycle progression and mitosis [232]. First, M33-KO animals have retarded growth [228] and many organs derived from M33-KO mice such as the spleen and adrenal glands are hypocellular [246], consistent with a role for CBX2 in cell proliferation. Second, many cells including fibroblasts and lymphocytes derived from M33-KO mice have severely impaired proliferation [228]. Further 22

supporting an important function for CBX2 in proliferation, studies have shown that CBX2 becomes phosphorylated and translocates into the nucleus of proliferating hepatocytes in regenerating liver [238]. Importantly, it has been shown that this activity is regulated via the E2F/RB1 axis, providing a key link between CBX2 and cell cycle progression [247, 248]. In addition, M33-KO mice have impaired S phase progression and have a senescent phenotype associated with silencing of the CDKN2A and CDKN2B tumor suppressive loci [248]. Since studies have shown that CBX2 phosphorylation also increases its affinity to H3K27me3 in addition to inducing its nuclear translocation, this idea is consistent with cell-cycle-dependent regulation of CBX2 activity [238, 239].

1.4. Thesis theme and rationale

The theme of this thesis is the identification of critical epigenetic alterations that drive PCa progression in the search of novel therapeutic targets for advanced disease. Mounting evidence demonstrates that aggressive prostate malignancies are characterized by extensive epigenetic deregulation, thereby providing novel therapeutic avenues [140, 249]. Notably, aberrant transcriptional repression mediated by complexes of the PcG family (PRC1 and PRC2) were reported to strongly contribute to PCa progression [211]. These epigenetic complexes are thought to drive tumor aggressiveness by maintaining the epigenetic silencing of key tumor suppressor genes through various mechanisms including aberrant histone modifications and nucleosomal compaction [188, 250]. To date, the tumor-promoting roles of PcG members EZH2 (PRC2) and BMI1 (PRC1) have been well described in PCa [141]. However, little is known about how other PcG proteins influence PCa metastasis and castration-resistance.

The 5 human CBX proteins are responsible for targeting and recruiting PRC1 to chromatin, and therefore their relative expression holds particular importance in orchestrating PcG-mediated silencing [218]. CBX proteins share a conserved N-terminal chromodomain but harbor nonhomologous sequences in their C-terminus, which underlie their structural and functional differences [214]. Thus, over-expression of one CBX family member would alter PRC1 composition and therefore preferentially recruit PRC1 at the target site of this CBX member [218]. Accordingly, aberrant regulation of one CBX protein could represent a key epigenetic

alterations driving PCa progression to a lethal state. Given the recent advances in targeting epigenetic readers [224, 251], identifying a recurrently altered CBX member could allow new insights into the pathophysiology of advanced PCa and provide novel therapeutic strategies for advanced disease.

The PcG protein CBX2 represents an epigenetic reader whose role has yet to be described in human cancer. Since CBX2 can mediate the recruitment of PRC1 at sites of H3K27me3, it could be an important downstream mediator of the oncogenic effects of aberrant EZH2 activity [173]. In line with this idea, CBX2 function has been linked to the E2F/RB1 axis regulating proliferation, which is also known to regulate EZH2 [248]. Furthermore, CBX2 is required for the development of the male urogenital system [193]. This is relevant as multiple lines of evidence support the idea that PCa progression results from abnormal tumor-stroma interactions which induce altered epigenetic states leading to aberrant silencing of tumor suppressor genes [195, 252]. Since epigenetic alterations are reversible, genes abnormally silenced by CBX2 in advanced PCa could be re-expressed upon CBX2 inhibition, thereby causing proliferation arrest and cell death. Given these data and the recent success in targeting human chromodomains, CBX2 therefore represents a particularly relevant target to investigate in the context of PCa progression.

1.5. Hypotheses and specific aims The goal of this research project is to identify novel epigenetic drug targets for the treatment of metastatic and castration-resistant PCa, a disease for which the prognosis is dismal [14]. The PcG protein and epigenetic reader CBX2 has been previously implicated in normal prostate development [193]. While many publications have reported a fundamental role for CBX2 in regulating cellular proliferation and differentiation [228], the role of CBX2 in PCa and other human malignancies remains unexplored. Based on this information, I hypothesize that:

1) CBX2 undergoes a clinically-relevant up-regulation in advanced prostate malignancies. 2) Elevated CBX2 expression functionally contributes to PCa progression by inhibiting transcription of critical tumor suppressive genes, thereby activating oncogenic pathways.

To address this hypothesis, I will use clinical, biological, and computational approaches to characterize the role of CBX2 in advanced PCa and other malignancies according to the following aims:

Aim 1: Identify differential expression and clinical correlations of CBX2 in advanced PCa.

Aim 2: Analyze the phenotypic effects of CBX2 silencing in metastatic and castration-resistant PCa cells.

Aim 3: Dissect the molecular pathways underlying the cancer-driving activity of CBX2.

Aim 4: Assess the molecular profile of CBX2 in human cancers.

Overall, the results of this study will provide important insights into the epigenetic mechanisms regulating the growth of aggressive PCa cells and will contribute to the development of improved strategies for treatment of disseminated disease. Importantly, this thesis features the first reported investigation into the molecular function and therapeutic potential of CBX2 in human malignancies, thereby adding a new player to the growing field of cancer epigenetics.

2. Identification of CBX2 as a potential drug target in advanced prostate cancer 2.1. Introduction PCa represents the most commonly diagnosed solid tumor in the world [5]. While localized disease can be effectively treated with surgery or radiotherapy, metastatic PCa can only be palliated [253]. Despite the introduction of novel therapeutic agents for late-stage patients [86, 254], CRPC remains an incurable malignancy and thus a better understanding of its molecular drivers is required to identify new drug targets. Mounting studies have demonstrated that epigenetic alterations significantly contribute to PCa progression [161], suggesting that the PCa epigenome may harbor clinically relevant therapeutic targets that warrant further investigation.

Emerging evidence suggest that epigenetic dysregulation mediated by the Polycomb Group (PcG) family of transcriptional repressors plays a critical role during PCa progression [140]. Conserved throughout evolution, PcG proteins assemble in two main complexes, PRC1 and PRC2 [165]. In PCa, it has been shown that EZH2 is over-expressed and that silencing of PcG target genes can predict poor clinical outcome in PCa patients [140]. However, the role of individual PcG members during PCa progression and their contribution to CRPC has yet to be evaluated.

Since CBX proteins bridge the activity of PRC2 and PRC1, they represent critical regulators of PcG-mediated silencing. While implications of CBX2 remain unexplored in cancer, multiple lines of evidence demonstrate that CBX2 represents a critical regulator of cellular differentiation and proliferation [228]. In mouse models, germline M33 deletion induces high rates of postnatal lethality accompanied by homeotic transformations, general hypocellularity, and sex reversal [228, 243]. Notably, it was shown that XY subjects with homozygous M33 inactivation were unable to undergo development of the male urogenital system [193], implying an important function in early prostate development. From a molecular standpoint, CBX2 is regulated in a cell cycle-dependent manner which involves interplay with the E2F/RB1 axis [248].

With the aim of identifying new epigenetic drug targets, we analyzed the molecular profiles of PcG family members in patient-derived PCa tissues implanted in the subrenal capsule of Nonobese diabetic-severe combined immunodeficiency (NOD-SCID) mice. Using validated in

vitro and in vivo models, we demonstrate that the PRC1 member CBX2 is recurrently over- expressed in metastatic and androgen-independent PCa cells, and that elevated CBX2 expression predicts poor clinical outcome. Furthermore, we show that CBX2 depletion induces PCa cell death and proliferation arrest by regulating the expression of a key subset of genes, suggesting that CBX2 may emerge as a novel therapeutic target for advanced PCa.

2.2. Materials and methods 2.2.1. Patient-derived xenograft models

As previously reported, the Living Tumor Lab (LTL, www.livingtumorlab.com) has developed a collection of high-fidelity PDXs implanted into the subrenale capsule of NOD-SCID mice [139]. We used the LTL313B/LTL313H model to investigate the role of CBX2 in metastasis and the LTL313B/BR model to assess the implications of CBX2 in drug-resistant CRPC [139]. Tumor tissues were obtained from patients through a protocol approved by the Clinical Research Ethics Board of the University of British Columbia (UBC) and the BC Cancer Agency (BCCA). All patients signed a consent form approved by the Ethics Board (UBC Ethics Board #: H09-01628 and H04-60131; VCHRI #: V09-0320 and V07-0058). Animal care and experimental procedures were carried out in accordance with the guidelines of the Canadian Council of Animal Care (CCAC) under the approval of the Animal Care Committee of University of British Columbia (permit #: A10-0100). The microarray gene expression data for these tumor lines have been previously deposited in the NCBI Gene Expression Omnibus (GEO) and are freely available under the accession number GSE41193.

2.2.2. Bioinformatic database analysis

The Oncomine database was used to compare the expression of CBX2 between metastatic and non-metastatic PCa [126]. Data was acquired in an unbiased fashion by compiling all the Oncomine studies with significantly altered CBX2 expression (p≤0.05). The cBIO portal (www.http://www.cbioportal.org) was used to assess the genomic alterations affecting the CBX2 locus in PCa. In addition, the MSKCC dataset [46] was extracted from cBIO portal since it provides detailed clinical information for a large cohort of 150 PCa patients. Using this dataset, correlation between CBX2 and all other genes were calculated using the Pearson and Spearmann 27

correlation tests. For CBX2 mRNA levels in normal tissues, we extracted data from The Human Protein Atlas (http://www.proteinatlas.org/) [255]. All data was queried between March 2015 and May 2015.

2.2.3. Cell culture

All cell lines were maintained in RPMI 1640 growth medium (GIBCO, Grand Island, USA) o supplemented with 10% fetal bovine serum (GIBCO, Grand Island, USA) at 37 C and 5% CO2. For the androgen depletion experiment, LNCaP cells were initially plated in conditions described above for 24 hours, following which media was changed to RPMI 1640 (GIBCO, Grand Island, USA) supplemented with charcoal-stripped FBS (GIBCO, Grand Island, USA), which has the property of being completely free of steroid hormones [108]. This charcoal-stripped media was then itself supplemented with DHT (10 nM) or not, and the cells were harvested at 6h, 24h, and 48h after media change for qPCR analysis.

2.2.4. qRT-PCR

RNA was extracted using the RNeasy Kit (Qiagen) according to the manufacturer’s protocol. NanoDrop technology (ND-1000, NanoDrop) was used to quantify extracted RNA, which was subsequently subjected to reverse transcription using the QuantiTect Kit (Qiagen). Quantification of cDNA was done using primers from Integrated DNA Technologies (see Table 2.1 for sequences) and SYBR Green Universal Master Mix (KAPA Biosystems) on an ABIPrism 7900HT platform (Applied Biosystems) as per the manufacturers’ instructions.

Table 2.1: qRT-PCR primers.

Gene Direction Sequence CBX2 Forward ATCGAGCACGTATTTGTCAC CBX2 Reverse AGTAATGCCTCAGGTTGAAG CENPF Forward GAGGACCAACACCTGCTACC CENPF Reverse GGCTAGTCTTTCCTGTCGGG CEP55 Forward CCGTTGTCTCTTCGATCGCT CEP55 Reverse GGCTTCGATCCCCACTTACT DICER1 Forward TGAAATGCTTGGCGACTCCT DICER1 Reverse GCCAATTCACAGGGGGATCA FOXM1 Forward ATAGCAAGCGAGTCCGCATT FOXM1 Reverse AGCAGCACTGATAAACAAAGAAAGA HPRT1 Forward GGTCAGGCAGTATAATCCAAAG HPRT1 Reverse CGATGTCAATAGGACTCCAGATG INPP5A Forward TGTGACCGCATCCTCATGTC INPP5A Reverse TGATTCGGAAGGCCAGGAAC ITGB8 Forward TTTGTCTGCCTGCAAAACGA ITGB8 Reverse GCACAGGATGCTGCATTTGA MKI67 Forward TGAGCCTGTACGGCTAAAACA MKI67 Reverse GGCCTTGGAATCTTGAGCTTT PIK3R1 Forward GATTCTCAGCAGCCAGCTCTGAT PIK3R1 Reverse GCAGGCTGTCGTTCATTCCAT TERT Forward GAGAACAAGCTGTTTGCGGG TERT Reverse AAGTTCACCACGCAGCCATA TIMP2 Forward GCGGTCAGTGAGAAGGAAGT TIMP2 Reverse GGAGGGGGCCGTGTAGATAA

2.2.5. Western blot

Cell lysis was done using radioimmunoprecipitation assay (RIPA) buffer supplemented with a protease inhibitor cocktail (Roche). Bicinchoninic acid (BCA) protein assay (Thermo Fisher Scientific) was conducted to quantify protein concentrations in the resulting lysates. 15µg of proteins were run on a 10% sodium dodecyl sulfate polyacrylamide gel, transferred to a nitrocellulose membrane (Bio-Rad), and subjected to Western blot analysis. Primary rabbit antibodies specific to CBX2 (Thermo Fisher Scientific, Cat # PA5-30996, 1:1000) and actin (Thermo Fisher Scientific, Cat # PA1-16889, 1:4000) were incubated overnight at 4 oC, and goat anti-rabbit secondary antibody (Thermo Fisher Scientific, Cat # 31460, 1:15 000) was detected using electrochemiluminescence (ECL) kit (Thermo Fisher Scientific) according to the manufacturer’s protocol.

2.2.6. Microscopy

Light microscopy images were obtained using the Axiovert 40 CFL (Zeiss) and the Axioplan 2 (Zeiss).

2.2.7. CBX2 siRNA knockdown

24 hours after seeding, cells at a confluency of 30-50% were treated with 8nm CBX2-specific or non-targeting siRNA (ON-TARGET plus siRNA, Dharmacon). Lipofectamine 2000 (Invitrogen) was used as the transfecting agent according to the manufacturer’s protocol, and cells were subjected to functional assays 1 to 5 days post-transfection.

2.2.8. Caspase 3-7 activity

72 hours after CBX2 or non-targeting siRNA treatment in LNCaP and C4-2 (as described earlier), the relative caspase 3/7 activity was assessed using the Caspase-Glo 3/7 assay (Promega) according to the manufacturer’s protocol and chemiluminescence was measured with a spectrophotometer (Thermo Fisher Scientific).

2.2.9. MTT analysis

At 1, 3, and 5 days post-treatment, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) solution (5mg/ml, Sigma) was added to media and incubated for 3.5 hours, after which the cells were solubilized with dimethyl sulfoxide (DMSO) and absorbance was read at 570nm using a spectrophotometer (Thermo Fisher Scientific).

2.2.10. Microarray analysis

RNA was extracted from C4-2 cells treated with CBX2-specific or non-targeting siRNA 55 hours post-treatment in triplicate, using the RNA isolation protocol described above in the qRT- PCR section. RNA quality was assessed using the Agilent 2100 Bioanalyzer. Samples were subjected to microarray analysis using the Agilent human GE 8x60 v1 array at the Laboratory for Advanced Genomic Analysis (LAGA) in Vancouver, BC. Differential gene expression was quantified using T Test unpaired unequal variance (Welch) and p values were corrected for multiple testing using the Benjamini-Hochberg correction (p≤0.05).

2.2.11. Immunohistochemistry

The preparation of paraffin-embedded tissue sections and IHC were carried out as previously described [139, 256]. A CBX2-specific primary antibody was used (rabbit polyclonal, Pierce) and was recognized by a goat anti-rabbit secondary antibody (Vector Laboratory).

2.2.12. Statistical analysis

Unsupervised hierarchical clustering and multivariate analysis of variance (MANOVA) were conducted using the R statistical package. Computational analyses of CBX2-regulated transcripts were carried out with the Interactive Pathway Analysis software (Qiagen). Unless otherwise mentioned, all analyses were conducted using p≤0.05 as the significance threshold with the GraphPad Prism software (version 6), where error bars represent the standard deviation of three independent replicates.

2.3. Results 2.3.1. CBX2 is over-expressed in metastatic PCa

As the first step to identify putative therapeutic targets for advanced PCa, we analyzed the expression of core PcG genes in the LTL313H/LTL313B patient-derived model of metastatic and non-metastatic PCa [139]. LTL313B and LTL313H represent two xenograft models that were derived from two independent needle biopsies of the same primary PCa tumor (Figure 2.1A). This unique PDX pair captures the intratumoral heterogeneity of clinical PCa as LTL313H consistently gives rise to metastases when implanted in the mouse subrenal capsule while LTL313B always stays local to the grafting site [139]. Thus, this model provides a unique experimental system to identify differential expression of PcG genes between distinct foci of different metastatic ability within a single primary prostate tumor [139].

Microarray profiling of PcG gene expression demonstrated that the chromodomain-containing protein and known regulator of male urogenital system development CBX2 was the most highly up-regulated PcG transcript in LTL313H compared to LTL313B (Figure 2.1B). To validate these results, we assessed CBX2 expression in both tumor lines using qRT-PCR, which confirmed that CBX2 expression was 3.2 fold higher in LTL313H compared to LTL313B (Figure 2.1C, p<0.0001, student’s t test). Consistent with mRNA levels, CBX2 protein expression was undetectable in LTL313B while LTL313H showed strong CBX2 immunostaining, in line with a possible role in PCa dissemination (Figure 2.1D).

In order to ensure that the over-expression of CBX2 in metastatic PCa was not solely a property specific to the LTL313B/LTL313H xenograft model, we assessed the expression of CBX2 in primary and metastatic tumors from PCa patients using the Oncomine database. As observed in the xenografts, CBX2 expression was significantly higher in metastatic compared to non- metastatic tumors in 5 independent clinical cohorts (Table 2.2; Figure 2.1E, all p≤0.05, student’s t test). Within those studies, the p values of differential CBX2 expression between distant and localized disease ranged from 8.32x10-3 to 1.68x10-12, accounting for a total of 312 patients. Importantly, we could not find a single study in which CBX2 was significantly down-regulated in metastatic tissues (Table 2.2). Thus, the CBX2 up-regulation observed in the

LTL313B/LTL313H PDX model was also recapitulated in patient tumors, implying that CBX2 may play a role in metastatic PCa.

Figure 2.1: CBX2 is over-expressed in metastatic PCa. A) Establishment of the LTL313B/LTL313H PDX model of metastatic PCa B) Microarray analysis of core PcG family members in the LTL 313H/LTL313B xenograft model C) Confirmation of CBX2 mRNA up-regulation in the LTL313H tumor line by qRT-PCR (p≤0.05, student’s t test) D) Confirmation of CBX2 protein up-regulation in the LTL313H tumor line by IHC E,F) Elevated CBX2 mRNA levels in metastatic PCa compared to non- metastatic clinical samples from Oncomine studies: Chandran (n=31), Taylor 3 (n=150) and Grasso (n=30) (all p≤0.01 in primary compared to metastatic PCa, student’s t test).

Table 2.2: Expression of CBX2 in metastatic and primary prostate tumors compiled in Oncomine.

CBX2 in Cohort # Patients P Value Metastasis Over-expressed Chandran 31 1.68x10-12 Over-expressed Taylor 3 150 6.86x10-6 Over-expressed Varambally 13 1.26x10-4 Over-expressed Grasso 30 4.71x10-3 Over-expressed Yu 88 8.32x10-3 Under-expressed - - -

Given the elevated expression of CBX2 in PCa, we set out to determine whether any genetic aberrations may be underlying this process. We queried 4 independent patient cohorts that had data for both copy number changes and mutations. The first striking observation was that not a single point mutation could be found within the CBX2 locus in any of the four datasets, which were comprised of a total of 329 patients (Table 2.3). Additionally, only 3 out of 329 patients (0.9%) were found to have a CBX2 copy number loss (CNL). Similarly, only 5 out of 329 patients (1.5%) exhibited CBX2 copy number gain, which is not sufficient to account for the CBX2 up-regulation observed in clinical PCa (Table 2.2). Thus, the rarity of genomic disruption of CBX2 protein is not the cause of its up-regulation in PCa, suggesting that its elevated levels themselves result from epigenetic mechanisms.

Table 2.3: Number and frequency of genomic alterations affecting the CBX2 locus in four independent PCa cohorts from cBIO portal.

PCa Dataset & Journal PMID # Patients % Mut % CNG % CNL MSKCC - Cancer Cell 2010 20579941 103 0 2 1 Michigan - Nature 2012 22722839 61 0 5 0 Broad/Cornell - Nat. Gen. 2012 22610119 109 0 0 0 Broad/Cornell - Cell 2013 23622249 56 0 0 4 Total - 329 0 2 1

2.3.2. Elevated CBX2 levels associate with clinical PCa progression

Since metastatic PCa patients typically progress to CRPC [9], we wanted to investigate whether CBX2 was also involved in the progression to CRPC. To investigate this question, we took advantage of another patient-derived xenograft model in which the primary tumor line, LTL313B, was subjected to ADT (Figure 2.2A) [139]. As observed in the clinic, ADT elicited a significant reduction in tumor volume shortly after castration. However, the tumor developed resistance and eventually re-emerged as the CRPC tumor line LTL313BR [139]. LTL313BR retains important properties of clinical CRPC such as androgen-independent growth, resistance to hormonal therapy and docetaxel, and expression of PSA [139].

As the first step to link CBX2 and CRPC pathogenesis, we quantified the expression of CBX2 in the LTL313B/LTL313BR xenograft model and observed that CBX2 expression was elevated in LTL313BR relative to LTL313B using qRT-PCR (Figure 2.2B, p<0.001, student’s t test). Furthermore, IHC staining revealed that CBX2 protein levels were undetectable in LTL313B while LTL313BR exhibited strong CBX2 nuclear staining (Figure 2.2C). To confirm the results obtained in the 313B/BR model, we assessed the expression of CBX2 in a panel of PCa PDX models that were either androgen-dependent (n=10) or androgen-independent (n=5) available at the Living Tumor Laboratory. In line with the 313B/BR model, CBX2 expression was significantly higher in androgen-independent PDXs (Figure 2.2D, p<0.05, student’s t test), consistent with a potential role in castration-resistant disease.

To complement the observations made in PDX models, we conducted in vitro studies investigating the androgenic regulation of CBX2. CBX2 expression was assessed in vitro using androgen-responsive LNCaP cells subjected to removal and addition of dihydrotestosterone (DHT), a potent AR agonist. In LNCaP cells, CBX2 mRNA levels significantly increased after 48 hours of culture in androgen-depleted media, as assessed by qRT-PCR (Figure 2.2E, p<0.001, student’s t test). Accordingly, this dramatic effect was not observed in cells supplemented with DHT, suggesting that a decrease in ligand-induced AR transactivation reversibly stimulates CBX2 expression.

After observing elevated CBX2 levels in a number of advanced PCa tissues, we sought to determine whether CBX2 over-expression correlated with the clinical progression of PCa. First, we conducted multivariate analysis of variance (MANOVA) to associate the expression of CBX2 with specific clinicopathologic features in clinical PCa patients. In this multivariate model, we analyzed whether CBX2 up-regulation also correlated with clinical progression of PCa using the Taylor dataset obtained through the cBIO Portal since it provides detailed clinical information for a large cohort of 150 PCa patients (Table 2.4). This analysis revealed that elevated CBX2 levels were significantly correlated with lower patient age (p<0.05), higher Gleason grade (p<0.05), and a positive nodal status (p<0.005). All these variables are themselves indicators of poor prognosis in patients, further supporting a role for CBX2 in clinically aggressive disease. Taken together, these findings demonstrate that the expression of CBX2 likely reflects the aggressive potential of prostate tumors, suggesting that CBX2 could be functionally involved in PCa progression and therefore might represent a putative therapeutic target in CRPC.

Figure 2.2: Hormonal regulation of CBX2. A) Establishment of the LTL313B/LTL313BR patient- derived xenograft model of CRPC B) Assessment of CBX2 mRNA levels in the LTL313B/LTL313BR xenograft model by qRT-PCR (p≤0.05, student’s t test) C) IHC staining of CBX2 in the LTL313B and LTL313BR xenografts 20x). Images are representative of multiple fields taken from 2 independent experiments D) Levels of CBX2 mRNA in androgen-dependent (AD, n=10) and androgen-independent (AI, n=5) PDXs from the LTL (p≤0.05, student’s t test) E) CBX2 mRNA levels in LNCaP cells cultured in charcoal-stripped media in the presence or absence of DHT supplementation (10nM, p≤0.05, student’s t test).

Table 2.4: Multivariate analysis of variance correlating CBX2 and clinicopathological features in the MSKCC cohort (MANOVA, ***p≤0.001; *p≤0.05).

Factor F value P value Significance Age 4.8235 0.030674 * Extension 1.9261 0.131084 Gleason 5.5086 0.021142 * Nodal 15.4775 0.000165 *** Race 0.7067 0.55053 SemVesicle 0.0262 0.871665 SurgMargins 0.0839 0.772712 T stage 0.5005 0.607944

Optimally, good drug targets should have limited effects on normal cells under physiological conditions. If not, the corresponding drug would likely interfere with normal processes, which would result in undesirable side effects [257]. With the assumption that lower target expression correlates with decreased functional requirement in normal cells [257], we next sought to assess the relative expression of CBX2 expression in normal tissues using data derived from The Human Protein Atlas, which provides a detailed expression catalogue of almost every gene in a number of human tissues [255]. As a whole, analysis of CBX2 mRNA levels revealed strikingly low to absent CBX2 expression across almost all normal tissues (Figure 2.3). The only exception was the testis, which expressed much higher levels of CBX2 compared to the average of other tissues. Overall, the low CBX2 expression in normal tissues coupled with the significant up- regulation of CBX2 in metastatic and castration-resistant PCa provided the rationale to further investigate CBX2 function in vitro.

Figure 2.3: CBX2 mRNA expression in normal human tissues (The Human Protein Atlas).

2.3.3. CBX2 depletion induces cell death in advanced PCa cell lines

To evaluate the functional requirements of CBX2 in advanced PCa cells, we conducted siRNA- mediated knockdown of CBX2. First, we quantified the expression of CBX2 in LNCaP and C4-2 cells compared with benign prostate cells (BPH1) [258]. Androgen-independent C4-2 cells displayed CBX2 mRNA levels 41 times higher than BPH1 while LNCaP exhibited a 9-fold up- regulation in CBX2 expression (Figure 2.4, p≤0.05 for LNCaP and C4-2 compared to BPH1, student’s t test). We then used an siRNA-mediated knockdown approach to reduce CBX2 levels in LNCaP and C4-2 cell lines, which was validated 48h post-siRNA transfection both at the mRNA and protein levels by qPCR and western blot, respectively. In LNCaP cells, both CBX2 mRNA and protein levels were reduced by more than 90% following siRNA treatment (Figure 2.5A,C). For C4-2 cells, CBX2-specific siRNA induced a 60% reduction in CBX2 mRNA levels while CBX2 protein reached undetectable levels (Figure 2.5B, D). Approximately 55 hours following transfection, both LNCaP and C4-2 cells treated with CBX2-specific siRNA started exhibiting notable morphological changes not observed in cells treated with non-targeting siRNA. In both the LNCaP and C4-2 lines, cells started to round up and lose their spindle-like appearance, leaving very few viable cells 4 days after siRNA treatment (Figure 2.6). These results suggest that CBX2 is involved in regulating cellular proliferation and/or apoptosis.

Figure 2.4: Relative CBX2 expression in metastatic PCa cell lines LNCaP and C4-2 compared to BPH1 assessed by qRT-PCR (LNCaP VS BPH1 and C4-2 VS BPH1, p≤0.05, student’s t test).

To quantify the extent of cell viability loss resulting from CBX2 depletion, we conducted an MTT analysis on LNCaP and C4-2 cells treated with mock, non-targeting control, or CBX2- specific siRNA. MTT assay confirmed a significant reduction in cell viability 5 days following CBX2 knockdown in both cell lines (Figure 2.5E, F, p<0.0001, student’s t test). More specifically, the proliferation arrest induced by CBX2 depletion started to appear after 3 days of siRNA treatment and culminated in a dramatic decrease in cell viability after 5 days in both cell lines, thus confirming the microscopic observations. Taken together, these results indicate that CBX2 is required for optimal cellular proliferation and that inhibiting its activity abrogates PCa cell viability.

To explore the possibility that CBX2 might regulate apoptotic cell death, caspase 3/7 activity (a marker of apoptosis) was analyzed in LNCaP and C4-2 cells treated with either control or CBX2-specific siRNA for 72h. Notably, CBX2 depletion induced a 3.7 (p<0.001) and 2.3 fold increase in caspase 3/7 activity in LNCaP and C4-2, respectively (Figure 2.5G, H, p<0.001, student’s t test), suggesting that the elevated CBX2 expression observed in metastatic PCa promotes cell survival. Taken together, these findings demonstrate that CBX2 is functionally involved in the regulation of PCa cell morphology, proliferation, and apoptosis.

Figure 2.5: CBX2 depletion induces proliferation arrest and apoptosis in advanced PCa cell lines. A, B) Confirmation of CBX2 mRNA knockdown in LNCaP and C4-2 cells by qPCR following siRNA- mediated depletion (8nM, siCBX2 VS siCTRL, p≤0.05, student’s t test C, D) Confirmation of CBX2 protein knockdown in LNCaP and C4-2 cells (8nM) E,F) MTT analysis of cell viability following CBX2 silencing in LNCaP and C4-2 cells (8nM, 5 days, p<0.0001, student’s t test) G,H) Assessment of caspase 3-7 activity in LNCaP and C4-2 cells following CBX2 depletion (8nM, days, p<0.0001, student’s t test).

Figure 2.6: Morphology of LNCaP and C4-2 cells following CBX2 depletion (Images are representative of 3 independent experiments, 96h post-siRNA treatment (8nM), 20x for large image, 40x for small image).

2.3.4. Analysis of CBX2-regulated genes

Given the striking phenotypes observed upon CBX2 depletion, we further investigated the molecular mechanisms controlled by CBX2 in CRPC. To identify CBX2-regulated genes (CRGs), we used microarray profiling in RNA extracted from C4-2 cell lines treated with control or CBX2-specific siRNA, all in triplicates. RNA was extracted 55h hours after siRNA transfection, a time point where CBX2 expression is reduced in siCBX2-treated cells but prior to when these cells display abnormal proliferation and morphology (Figure 2.7A). RNA quality was assessed via nanodrop technology and showed that all replicates had A280/A230 and A260/A230 higher than 2.0, indicating high purity (Table 2.5). Furthermore, Bioanalyzer was used as an added quality control measure. The Bioanalyzer program assesses RNA degradation on a scale of 0 to 10, with 10 having the least degradation. All six samples had an RIN value higher than 9.4 out of 10, indicating high quality and minimal degradation (Figure 2.8).

Figure 2.7: Gene expression profiling of CBX2-regulated genes (CRGs). A) Experimental design of microarray analysis B) Validation of CBX2 silencing in samples subjected to microarray analysis (all p≤0.05, student’s t test) C) Unsupervised hierarchical clustering of genes differentially-expressed following CBX2 knockdown. D) Differential expression of up-regulated CRGs confirmed by qRT-PCR in CBX2-depleted C4-2 cells (all p≤0.05, student’s t test) E) Differential expression of down-regulated CRGs confirmed by qRT-PCR in CBX2-depleted C4-2 cells (all p≤0.05, student’s t test).

The RNA extracted from C4-2 cells was reverse transcribed to generate cDNA used to validate that CBX2 expression was down-regulated upon CBX2 knockdown by qRT-PCR. As expected, CBX2 levels were about 80% lower in C4-2 cells treated with CBX2 siRNA compared to control, indicating successful knockdown (Figure 2.7B). After validating the quality of the RNA and knockdown, we proceeded to microarray analysis using the Agilent platform. Using an unpaired t test with Benjamini-Hochberg correction, we identified 544 transcripts that were differentially expressed upon CBX2 silencing and were termed CBX2-regulated genes (CRGs). Among them, 232 were up-regulated and 312 were down-regulated (Figure 2.7A). Unsupervised hierarchical clustering revealed that the up-regulated and down-regulated genes have distinct expression patterns which are extremely consistent across all replicates (Figure 2.7C). Thus, all quality control measures appear to indicate that the microarray profiling was conducted successfully.

Table 2.5: Concentration and purity of RNA used for microarray analysis (ND-1000, NanoDrop).

Concentration Sample A280/A230 A260/A230 (ng/µl) siCTRL1 538.7 2.06 2.13 siCTRL2 361.1 2.03 2.19 siCTRL3 529.8 2.09 2.12 siCBX2 1 362.8 2.05 2.17 siCBX2 2 450.5 2.03 2.01 siCBX2 3 414.4 2.03 2.07

Figure 2.8: Assessment of RNA quality by Bioanalyzer in C4-2 cells treated with non-targeting siRNA or CBX2-specific siRNA (score ranges from 0 to 10, with 10 being the highest possible RNA quality).

Within the 544 CRGs, we first analyzed individual genes to investigate whether known cancer- related genes were significantly modulated. Interestingly, we found a number of important regulators of cell proliferation and metastasis. Notably, ITGB8, DICER1, INPP5A, PIK3R1 are key tumor suppressors that were among up-regulated CRGs following CBX2 knockdown (Table 2.6). In addition, significant up-regulation of these genes in CBX2-depleted cells was also validated using qRT-PCR (Figure 2.5D, p≤0.05, student’s t test) To determine whether these genes were also correlated with CBX2 in clinical PCa, we analyzed if they showed significant association with CBX2 expression. We report that these transcripts were inversely correlated with CBX2 expression in the MSKCC PCa patient cohort described earlier [46] (R2 range: -0.40 to -0.64, p<0.0001, Pearson and Spearman correlation). Conversely, the tumor-associated proteins MKI67, FOXM1, CENPF, TERT, and CEP55 were down-regulated following CBX2 depletion and were positively correlated with CBX2 expression in clinical datasets (Table 2.6, R2 range: 0.43 to 0.84, p<0.0001, Pearson and Spearman correlation). Moreover, qRT-PCR 47

confirmed that mRNA levels of these five genes were indeed reduced, further validating the reproducibility of transcriptomic changes detected through microarray analysis (Figure 2.5E, p≤0.05, student’s t test).

Table 2.6: List of cancer-related CRGs whose expression correlates with that of CBX2 in the MSKCC dataset (n=150).

Gene Fold Change Pearson R2 Spearman R2 ITGB8 9.5 -0.41 -0.44 DICER1 3.4 -0.48 -0.44 INPP5A 2.3 -0.48 -0.52 TIMP2 2.0 -0.4 -0.4 PIK3R1 1.6 -0.62 -0.64 MKI67 -3.0 0.45 0.56 FOXM1 -3.0 0.69 0.76 CENPF -3.1 0.43 0.48 TERT -2.5 0.82 0.84 CEP55 -3.5 0.6 0.68

Several key components of the mitotic machinery were also significantly down-regulated upon CBX2 silencing. These mitotic genes included members of the following group of genes: centromere proteins (CENPE, CENPF, CENPJ, CENPM, CENPO), kinesin family (KIF22, KIF23), Structural Maintenance of Chromosomes (SMC2, SMC4), as well as Spindle And Kinetochore Associated Complex Subunit (SKA1, SKA2, SKA3) (Table 2.7). The inability to undergo mitosis caused by widespread down-regulation of proteins involved in mitotic integrity could therefore partly explain the strong proliferative defect induced by CBX2 knockdown.

Table 2.7: Expression of CBX2-regulated genes known to be implicated in mitosis following siRNA- mediated CBX2 depletion (conditions described in Figure 2.7A, p≤0.05, unpaired t test with Benjamini-Hochberg correction).

P Value Gene Regulation Fold Change P Value (corrected) CCNB1 up 2.4 5.0E-04 4.7E-02 CENPE up 3.0 1.6E-04 3.5E-02 CENPF up 3.1 2.7E-04 3.8E-02 CENPJ up 2.1 1.9E-04 3.5E-02 CENPM up 3.1 3.8E-04 4.3E-02 CENPO up 2.2 4.0E-04 4.3E-02 CEP55 up 3.5 5.1E-04 4.7E-02 CHEK1 up 3.5 3.7E-04 4.2E-02 KIF22 up 2.5 1.8E-04 3.5E-02 KIF23 up 2.2 1.5E-04 3.5E-02 KNTC1 up 2.2 3.8E-04 4.3E-02 MAD2L1 up 2.7 2.0E-04 3.6E-02 MCM3 up 2.4 3.8E-04 4.3E-02 MKI67 up 3.0 6.3E-04 5.0E-02 MLF1IP up 3.0 3.5E-04 4.2E-02 NCAPG up 3.8 3.9E-04 4.3E-02 SKA1 up 3.0 1.4E-04 3.5E-02 SKA2 up 2.1 6.2E-04 4.9E-02 SKA3 up 2.8 3.5E-05 2.6E-02 SMC2 up 2.7 1.4E-04 3.5E-02 SMC4 up 2.5 2.1E-04 3.6E-02 TROAP up 2.9 5.7E-04 4.8E-02 TTK up 2.7 6.3E-04 5.0E-02

Next, we assessed the biological processes and functions associated with CRGs. Using the IPA software set at the analysis of “biological processes and functions”, a significant link between CBX2 and cell cycle progression was observed. Out of the top 13 processes most significantly correlated with CRGs (inclusion criteria: odds ratio>2, p<0.05), 11 were directly involved in the regulation of cell cycle progression (Table 2.8). These included “DNA replication and chromosome cycle”, “Mitotic chromosome condensation”, “Mitotic sister chromatid segregation”, and “G1/S transition of mitotic cell cycle”. The other two processes were “Nucleotide-excision repair” and “DNA repair”, which were again related to DNA integrity, in line with roles previously described for CBX2. As the next step, we analyzed the disease and functions associated with CRGs and found that the top three hits were related to the cell cycle or to cellular organization (Table 2.9). The top diseases associated with CRGs were “Cancer”, “Developmental disorder”, and “Hereditary disorder”. Overall, the properties of CRGs are consistent with the strong proliferation arrest observed upon CBX2 depletion as well as phenotypes observed in M33-KO mice.

Table 2.8: Biological processes associated with the list of CRGs (cutoff: odds ratio > 2, Oncomine).

Odds Rank Concept Name P-Value Ratio 1 DNA replication and chromosome cycle 2.3E-06 39.0 2 Mitotic chromosome condensation 8.2E-04 23.2 3 Mitotic sister chromatid segregation 8.2E-04 23.2 4 G1/S transition of mitotic cell cycle 1.0E-02 7.7 5 Nucleotide-excision repair 1.0E-02 7.7 6 Mitosis 1.5E-06 6.5 7 DNA repair 4.7E-08 6.3 8 DNA replication 2.7E-06 6.1 9 Cytokinesis 7.2E-07 5.8 10 Chromosome organization and biogenesis 1.0E-03 4.6 11 Cell Cycle 2.3E-05 3.9 12 Regulation of cell cycle 4.0E-03 2.6 13 Intracellular signaling cascade 6.1E-04 2.4

Table 2.9: Diseases and functions associated with CRGs (IPA, disease and functions analysis).

Rank Category p-value 1 Cell Cycle 2.57E-12 - 1.72E-02 2 Cellular Assembly and Organization 5.51E-11 - 1.72E-02 3 DNA Replication, Recombination, and Repair 5.51E-11 - 1.72E-02 4 Cancer 5.86E-10 - 1.71E-02 5 Developmental Disorder 1.60E-08 - 1.70E-02 6 Hematological Disease 1.60E-08 - 1.03E-02 7 Hereditary Disorder 1.60E-08 - 1.70E-02 8 Organismal Injury and Abnormalities 1.60E-08 - 1.72E-02 9 Gastrointestinal Disease 3.55E-08 - 8.34E-03 10 Reproductive System Disease 4.46E-08 - 1.70E-02 11 Cell Death and Survival 2.08E-07 - 1.72E-02 12 Organismal Survival 5.32E-07 - 2.65E-05 13 Cell Morphology 1.55E-06 - 1.72E-02

2.4. Discussion Despite numerous large-scale sequencing efforts, very few genetic mutations are recurrently found in PCa, suggesting that epigenetic alterations likely contribute to PCa progression [259]. Recent studies have highlighted a critical role for the PcG family of epigenetic repressors in PCa cell survival and metastasis [211]. We therefore analyzed the expression of all PcG members in paired primary/metastatic patient-derived xenografts and clinical datasets of PCa. Our results demonstrate that CBX2 is the most highly up-regulated PcG member across multiple models of metastatic and castration-resistant PCa, and that CBX2 abundance correlates with poor clinical outcome. Moreover, CBX2 depletion induced PCa cell death in vitro, which was accompanied by differential expression of key genes regulating PCa progression. Taken together, these results position CBX2 as a putative drug target in advanced PCa.

CBX2 up-regulation was first identified in our paired non-metastatic (LTL313B) and metastatic (LTL313H) PDXs implanted into the subrenal capsule of NOD-SCID mice [139]. A particular feature of the LTL313B/H model is that both tumor lines originate from different foci of a single localized tumor, thus properly recapitulating the intratumoral heterogeneity observed in clinical PCa [139]. Based on this model, our results suggest that a small population of CBX2-expressing PCa cells within the primary tumor is the likely seed of metastatic dissemination. In addition, we have also shown that CBX2 expression is elevated in metastatic tumors compared to those remaining local to the prostate. This is in accordance with our in vitro studies, which demonstrate that CBX2 depletion induced death in two metastatic PCa cell lines.

Based on its association with metastatic dissemination and poor clinical outcome, CBX2 warrants further investigation as a biomarker for aggressive PCa. We found that elevated CBX2 levels independently predicted high grade, high stage, early biochemical recurrence, metastatic dissemination, and overall survival in PCa patients. In the 313B/313H model, we detected differential CBX2 expression in different intratumoral foci of a primary PCa obtained through needle biopsy, consistent with the notion that small subpopulations of epigenetically distinct cells drive tumor progression [260, 261]. Therefore, we propose that positive CBX2 immunostaining in at least one core biopsy could be incorporated as an unfavorable prognostic marker into currently used algorithms such as TNM staging and Gleason score. Currently, a major clinical challenge lies in identifying patients who will develop lethal, disseminated PCa and those who will not progress to metastatic disease, the latter of which would benefit from less aggressive treatment options [262]. In these challenging cases, the added prognostic value inferred from CBX2 status could improve patient stratification and thus quality of life.

There is solid evidence demonstrating that PCa cells with high proliferative ability tend to give rise to more metastatic and aggressive disease [263, 264]. For example, it is well known that a low PSA doubling time and a high Ki67 index, both of which are markers of high proliferation, correlate with poor clinical outcome [264-266]. In line with the idea that CBX2 promotes tumor progression, the biological processes and functions associated with CRGs were intricately related with proliferation. Furthermore, Ki67 was itself down-regulated upon CBX2 silencing, indicating that CBX2 positively regulates Ki67 expression. These properties are consistent with 53

phenotypic features of CBX2-deficient animals which arise as a result of a proliferative block [228]. Further linking CBX2 and cell cycle progression, analysis of CRGs revealed that a large number of proteins involved in mitotic spindle assembly are significantly down-regulated upon CBX2 silencing. In the literature, there is evidence demonstrating that CBX2 directly contributes to cell cycle progression through its association with condensed chromatin [232, 235]. Here, we expand on this function and show that, in addition, CBX2 also ensures integrity of cell division indirectly via the regulation of genes involved in mitotic spindle assembly.

A striking phenotype of M33-KO mice is that XY subjects are unable to undergo development of the male urogenital system, implying a key role for CBX2 in the establishment of male sexual differentiation [193]. While this feature suggests that CBX2 may cooperate with AR activity, our data indicates that CBX2 is antagonistically related to ligand-dependent AR signaling. In this chapter, we report that CBX2 is preferentially up-regulated in androgen-independent cells and that ligand-dependent AR activation is associated with reduced CBX2 levels. This is validated by the marked increase in CBX2 mRNA and protein in the castration-resistant 313BR xenograft line compared to 313B, and the decrease in CBX2 expression following DHT addition in LNCaP cells. Given the pro-survival properties conferred by CBX2 in vitro, we posit that CBX2 up- regulation may serve as an adaptive mechanism to bypass the apoptotic response induced by AR inhibition. This hypothesis also predicts that CBX2 may also be involved in resistance to pharmacologic AR suppression, which is currently the standard of care for CRPC patients who have progressed after docetaxel treatment [14].

In conclusion, the prognostically significant up-regulation of CBX2 coupled to the massive PCa cell death induced by CBX2 depletion suggest that it may represent a potential therapeutic target for CRPC. As previously mentioned, CRPC patients are becoming increasingly susceptible to neuroendocrine transdifferentiation into NEPC as a result of treatment with novel AR suppressors [16]. Since NEPC is highly aggressive, there is unique value in identifying therapeutic targets with efficacy in CRPC that could also impair NEPC pathogenesis. Two important properties of NEPC include high proliferative rate and androgen-independent growth, both of which have been associated with CBX2 up-regulation in this chapter. Thus, investigating

the regulation of CBX2 in NEPC may lead to promising therapeutic approaches that can benefit patients with late-stage disease by simultaneously blocking the progression of CRPC and NEPC.

3. Polycomb-mediated silencing in neuroendocrine prostate cancer 3.1. Introduction Recently, potent AR antagonists have been approved to treat metastatic CRPC. As a result, there is an increased selective pressure for cells that can thrive in the absence of AR activity, which favors the transdifferentiation of PCa into AR-negative NEPC [16]. In the last chapter, we demonstrated that CBX2 was up-regulated in CRPC, an androgen-independent malignancy, and that removal of androgen hormones induced CBX2 expression in androgen-responsive PCa cells. Since these data indicate that CBX2 may harbor therapeutic potential in the clinical setting of CRPC, it would be valuable if CBX2 inhibition also impairs the progression of NEPC, which is in dire need of therapeutic options. With a median survival of less than a year, NEPC represents the most aggressive prostate malignancy and only a small fraction of NEPC patients benefit from current treatments [268]. Since NEPC represents an emerging clinical problem, identifying its molecular drivers consequently represents a critical task.

The LTL has previously established the first in vivo model of ADT-induced NEPC using PDXs implanted in the mouse subrenal capsule [139]. The tumor tissue was initially harvested from a primary PCa patient and grafted into NOD-SCID mice, giving rise to an adenocarcinoma (LTL331) characterized by the expression of AR and PSA [139]. Upon castration of the host, LTL331 regressed to undetectable disease and stayed dormant for many months. However, resistance to castration was achieved and the recurrent tumor (LTL331R) harbored typical NEPC features, including expression of neuroendocrine markers such as CHGA and SYP [139]. Comparative genomic hybridization revealed that the original LTL331 and the relapsed LTL331R tumor lines share a remarkably similar genetic profile [139]. Interestingly, NEPC is histologically and molecularly similar to small cell lung cancer (SCLC), another aggressive neuroendocrine malignancy characterized by epigenetic alterations involving PcG family members [269, 270].

To date, dysregulation of PcG-mediated silencing has been observed in many aggressive tumor types but has not been studied in NEPC. However, several lines of evidence have shown that PcG-mediated silencing plays important roles in the differentiation of cells into neuronal lineages [271-273]. For example, PcG genes are required for neurogenesis and neural stem cell survival 56

[274-276], implying that they may regulate pathways involved in neuroendocrine-like differentiation. Furthermore, altered expression of PcG transcripts have been reported in a number of aggressive brain tumors, suggesting a potential role in regulating neuronal lineages [277, 278]. Taken together, these findings are consistent with the idea that aberrant transcriptional regulation mediated by PcG complexes may be involved in NEPC pathogenesis.

Since a comprehensive analysis of epigenetic regulators (EpRs) has never been done in NEPC, we conducted comparative gene expression analysis between LTL331R and LTL331, as well as in a clinical NEPC cohort, to identify EpRs that were differentially expressed in NEPC. Our data demonstrate that multiple PcG family members are over-expressed in NEPC, notably CBX2 and EZH2. Consistent with these results, we derived a neuroendocrine-associated repression signature (NEARS) that predicted aggressive disease progression and was enriched in PcG targets. Overall, our results support a clinically-relevant function for CBX2 and altered PcG- mediated silencing in NEPC, further supporting the idea that CBX2 may emerge as a novel drug target.

3.2. Methods 3.2.1. Clinical expression datasets

Originating from the work of Beltran et al., the clinical NEPC cohort contained 7 NEPC tumors and 30 PCas which contained less than 10% stroma, as confirmed by a certified pathologist [123]. Prognostic analysis of NEARS, as well as the selected list of EpRs, was conducted using PCa datasets available from the Oncomine resource (www.oncomine.com) [279], which encompassed more than 3800 patients. Clinical parameters assessed for differential gene expression included grade, metastasis, outcome, and stage. Analysis of literature-derived concepts correlated with NEARS was also done through the Oncomine resource, and the final list was unbiasedly determined using the lowest p values of associated concepts. Expression profiles of lung malignancies were generated by The Clinical Lung Cancer Genome Project (CLCGP) and Network Genomic Medicine (NGM) [280] (http://www.uni-koeln.de/med- fak/clcgp/).

3.2.2. Gene lists

We established a list of targetable EpRs based on the following inclusion criteria: 1) Being involved either in DNA methylation, histone acetylation or histone lysine methylation and; 2) Function as a writer, eraser, or reader of the epigenetic code. EpRs regulating DNA methylation were also subdivided into the same functional categories as those established for histone modifications (i.e. writer=DNA methyltransferase (DNMT), eraser=ten eleven translocation (TET), reader=methyl CpG-binding domain (MBD)). Analysis of relevant literature [281-289] was conducted to identify such candidates, which were subsequently assessed in our NEPC datasets. In a similar way, a list of PcG genes was also derived from the literature using recent review papers written by authorities in the field [168, 290, 291]. We also derived a list of repressors that directly interact with PcG proteins based on literature findings. Finally, NEARS was established by combining the 185 genes down-regulated in all three of the following datasets: 1) LTL331R/LTL331, 2) Clinical NEPC/PCa, and 3) LTL331/all LTL PCa [139].

3.2.3. Immunohistochemistry

Establishment of paraffin-embedded tissue sections and immunostaining were conducted as previously described [139, 256]. Detection was done using primary antibodies specific to PSA (rabbit polyclonal, Dako), SYP (mouse monoclonal, Dako), CBX2 (rabbit polyclonal, Pierce) and EZH2 (rabbit monoclonal, Cell Signaling), as well as a goat anti-rabbit secondary antibody (Vector Laboratory).

3.2.4. Patient-derived xenografts

As previously described [139], the Living Tumor Lab (www.livingtumorlab.com) has established a bank of high-fidelity patient-derived xenografts. Tumor tissues were obtained from patients through a protocol approved by the Clinical Research Ethics Board of the University of British Columbia (UBC) and the BC Cancer Agency (BCCA). All patients signed a consent form approved by the Ethics Board (UBC Ethics Board #: H09-01628 and H04-60131; VCHRI #: V09-0320 and V07-0058). Animal care and experimental procedures were carried out in accordance with the guidelines of the Canadian Council of Animal Care (CCAC) under the 58

approval of the Animal Care Committee of University of British Columbia (permit #: A10- 0100). In this study, we used microarray data derived from 10 PCa and 3 NEPC tumor lines, all of which retain the classical histological features of their respective subtype. The microarray gene expression data for these tumor lines have been previously deposited in the NCBI Gene Expression Omnibus (GEO) and are freely available under the accession number GSE41193.

3.2.5. Statistical analysis

Unless otherwise mentioned, all analyses were conducted using p≤0.05 as the significance threshold with the GraphPad Prism software (version 6), where error bars represent the standard deviation of three independent replicates.

3.3. Results 3.3.1. Expression profiling of epigenetic regulators in NEPC

To uncover potential epigenetic targets in NEPC, we set out to identify up-regulated EpRs in the LTL331R/LTL331 xenograft model, as well as in a clinical NEPC dataset containing gene expression profiling of PCa and NEPC patient tumors [123]. We initially established a list of EpRs using criteria that would maximize the translational application of identified targets. For these reasons, we restricted our list to the epigenetic writers, erasers, and readers regulating histone acetylation and lysine methylation, as well as DNA methylation [292]. Furthermore, the selected genes were also functionally classified into those associated with transcriptional activation or repression, and EpRs for which the transcriptional role remains unclear. Using a panel of recent comprehensive reviews, we derived a list of 147 EpRs that we subsequently analyzed in our NEPC expression datasets (Table 3.1).

Table 3.1: Distribution of 147 investigated epigenetic regulators across different epigenetic modifications, activities, and transcriptional effects.

Criteria Epigenetic Regulator Distribution DNA Histone Histone Epigenetic methylation acetylation methylation Modification 13 (9%) 65 (44%) 69 (47%) Epigenetic Writer Eraser Reader

Activity 53 (36%) 40 (27%) 54 (37%) Activation Repression Unclear Transcriptional Effect 72 (49%) 66 (45%) 9 (6%) TOTAL 147 genes

To investigate the epigenetic landscape of NEPC, we assessed the differential expression of our EpR list in the clinical NEPC cohort and the LTL331R/LTL331 microarray dataset [123, 139]. First, we determined if there was preferential up-regulation of readers, writers, or erasers and observed no significant difference (Figure 3.1A, non-significant, Kruskal-Wallis test). The same analysis was conducted investigating factors affecting different chromatin modifications (histone acetylation and methylation, DNA methylation) and demonstrated that no expression differences could be detected between EpRs associated with these chemical marks (Figure 3.1B, non- significant, Kruskal-Wallis test). However, we found that the mRNA levels of transcriptional repressors were significantly higher than that of activators in NEPC, which may result in a more repressed chromatin state that potentially regulates neuroendocrine differentiation (Figure 3.1C, p<0.05, Kruskal-Wallis test). Next, starting from our 147 EpR list described earlier (Table 3.1), we selected 22 genes that were up-regulated in both the LTL331R/LTL331 model and the clinical cohort (cutoff fold change (FC) in both >1.5 since no p values were available given the single measurement). Expression of these 22 EpRs was significantly correlated between the clinical cohort and the LTL331R model (Figure 3.1D, R2=0.48, p<0.001, Spearman test), 60

indicating that gene expression in our xenograft model accurately reproduced the transcriptional profiles observed in the clinical setting.

Figure 3.1: Differential expression of epigenetic regulators in NEPC. Average fold change of individual EpRs grouped into distinct A) epigenetic modifications, B) epigenetic activities, and C) transcriptional effects in clinical NEPC/AC and the 331R/331 model (Error bars represent standard deviation, p≤0.05, Kruskal-Wallis test) D) Expression correlation of 22 EpRs up-regulated by more than 1.5-fold in clinical NEPC/AC and the 331R/331 model E) Expression levels of individual up-regulated EpRs in clinical NEPC/AC and the 331R/331 model.

Interestingly, most of the selected EpRs were preferentially involved in transcriptional repression, accounting for 68% of all selected EpRs (Figure 3.1E). To assess the clinical relevance of these genes, we investigated whether their elevated expression was associated with specific parameters of prostate cancer progression in the Oncomine database [279]. Using stringent inclusion criteria (p<0.005, odds/ratio>5, top 10% over-expressed), we identified five independent Oncomine studies in which this gene list was significantly up-regulated in disseminated prostate tumors (Table 3.2).

Table 3.2: Up-regulation of 22 selected EpRs (From Figure 3.1E) in metastatic compared to non- metastatic prostate cancer across five independent clinical datasets (Oncomine Analysis, primary VS metastatic).

ONCOMINE - METASTATIC VS PRIMARY PCa List of Up-regulated Epigenetic Regulators (22 genes) Dataset P-Value Odds Ratio Rank Grasso 4.69E-07 13.4 Top 5% OE Taylor 2.04E-05 7.5 Top 10% OE LaTulippe 1.80E-04 13.7 Top 5% OE Varambally 8.70E-04 5.2 Top 10% OE Lapointe 0.004 6.5 Top 10% OE

To further characterize these epigenetic alterations, we focused on individual genes that were aberrantly regulated in both the clinical NEPC cohort and in LTL331R. An important finding was that the PcG H3K27me3 reader CBX2 was the most highly over-expressed transcript in both datasets (Figure 3.1E, FC 331R/331=8.2, FC NEPC/PCa=10.2). Interestingly, the H3K27me3 writer EZH2 was the second most highly up-regulated transcript (Figure 3.1E, FC 331R/331=3.4, FC NEPC/PCa=9.2), implying that H3K27me3 and its downstream epigenetic effects may be potentiated in the molecular context of NEPC. Of note, the selected gene list also included two other PRC1-associated CBX proteins, CBX6 and CBX8 (Figure 3.1E), further supporting a role for dysregulated PcG-mediated silencing during neuroendocrine

transdifferentiation. In addition to the increase in PcG genes themselves, we also observed that 73% of non-PcG repressors in our list of up-regulated EpRs have been reported to directly interact with at least one PcG member (Table 3.3). These PcG-interacting proteins were mainly involved in DNA methylation (DNMT1, DNMT3A, DNMT3B, MBD1) and histone methylation (SUV39H1, DOT1L, CHD5, CBX3), suggesting that the up-regulation of other EpRs may contribute to the effect of altered PcG-mediated silencing in NEPC [181, 204].

Table 3.3: Literature-reported direct interactions between PcG complexes and transcriptional repressors up-regulated by at least 1.5 fold in both the clinical NEPC cohort and the 331R/331 model.

PcG Reference Repressor Interaction (PMID) DNMT1 Yes 16357870 DNMT3B Yes 16357870 SUV420H2 No - DOT1L Yes 23891621 CHD5 Yes 23948251 SUV39H1 Yes 12101246 CBX3 Yes 22325352 CBX1 No - HDAC5 No - DNMT3A Yes 16357870 MBD1 Yes 17428788

3.3.2. PcG gene expression in LTL patient-derived xenografts

Since the most aberrantly expressed transcripts were members of the PcG family (Figure 3.2A, p<0.001, student’s t test), we compared the expression of 36 known PcG genes in ten PCa and three NEPC xenografts from the LTL using previously published microarray data (properties

described in Table 3.4) [139]. To ensure that our xenograft models retained typical features of their respective subtype, we assessed expression of molecular markers specific to PCa (AR, PSA) and NEPC (SYP, CHGA) in the investigated LTL tumor lines. As expected, expression of these markers segregated perfectly between the two malignancies. AR and PSA were selectively up-regulated 489- and 124-fold in PCa over NEPC, respectively (Figure 3.2B, p<0.0001, student’s t test). In contrast, SYP and CHGA respectively exhibited a 21- and 854-fold enrichment in NEPC tumor lines compared to adenocarcinoma models (Figure 3.2B, p<0.0001, student’s t test).

Table 3.4: Patient-derived xenograft models from Living Tumor Laboratory used for comparative NEPC/AC analysis and their immunohistologic features (Figure 3.2, see Lin et al. 2014 [139])

Number Subtype Tumor Line ID AR PSA CHGA SYP

1 PCa LTL310 Yes Yes No No 2 PCa LTL311 Yes Yes No No 3 PCa LTL313A Yes Yes No No 4 PCa LTL313B Yes Yes No No 5 PCa LTL313C Yes Yes No No 6 PCa LTL313D Yes Yes No No 7 PCa LTL313H Yes Yes No No 8 PCa LTL331 Yes Yes No No 9 PCa LTL412 Yes Yes No No 10 PCa LTL418 Yes Yes No No 1 NEPC LTL331R No No Yes Yes 2 NEPC LTL352 No No Yes Yes 3 NEPC LTL370 No No Yes Yes

Having confirmed that our patient-derived xenografts were transcriptionally representative of each subtype, we assessed the expression of PcG genes in the same PDX models (see Table 3.4 for description of each tumor line). Of the 36 PcG genes that were queried, ten were significantly

up-regulated in NEPC xenografts compared to adenocarcinoma (Figure 3.2C, p<0.05, student’s t test). CBX2 and EZH2 were again the two most highly over-expressed genes with fold changes of 5.8 and 4.7, respectively (Figure 3.2C, p<0.0001, Mann-Whitney U test). In addition, overabundance of CBX6 and CBX8 transcripts was also observed in neuroendocrine tumor lines (Figure 3.2C, p<0.05, student’s t test), consistent with our previous findings. Notably, all core PRC2 members (EED, EZH1, EZH2, SUZ12) were significantly up-regulated in NEPC (Figure 3.2C, p<0.05, student’s t test). In addition, we found a significant correlation between expression of PcG genes in the LTL331R/LTL331 model and in the clinical cohort (Figure 3.2D, R2=0.68, p<0.0001, Spearman test), validating reproducible PcG up-regulation across all investigated datasets.

Figure 3.2: Coordinated increase in PcG gene expression. A) Average fold change of non-PcG and PcG genes from unselected 147 EpR list and selected 22 EpR list (NEPC=10, AC=3 (see Table 3.4) 65

p≤0.05, student’s t test) B) Expression of typical prostate neuroendocrine (SYP, CHGA) and adenocarcinoma (AR, PSA) markers in LTL xenograft models (adenocarcinoma=10, NEPC=3, p≤0.05, student’s t test) C) Coordinated up-regulation of core PRC1 and PRC2 members led by CBX2 and EZH2 in LTL xenograft models (NEPC=10, AC=3 (see Table 3.4), p≤0.05, student’s t test) D) Significant correlation between PcG gene expression changes in all LTL xenograft models (NEPC=10, AC=3 (see Table 3.4) and 331R/331 model (Pearson score).

Since we found that CBX2 and EZH2 were consistently the most highly up-regulated EpRs in our NEPC models, we further investigated the molecular profiles of these two PcG members. Using immunohistochemistry (IHC), we analyzed CBX2 and EZH2 protein expression in the LTL331R/LTL331 model. We first used antibodies to PSA and SYP to confirm that LTL331 and LTL331R retain the histological features of PCa and NEPC, respectively (Figure 3.3). As expected, PSA expression was very strong in LTL331 while LTL331R displayed undetectable levels. In contrast, SYP immunoreactivity was strictly restricted to LTL331R, in line with the expected histological profiles of each tumor line. Having validated our positive and negative controls for each subtype, we analyzed CBX2 and EZH2 protein levels in both tumor lines (Figure 3.3). In accordance with high mRNA levels, CBX2 exhibited strong immunoreactivity in LTL331R while displaying only weak positivity in LTL331. Similarly, EZH2 protein expression was extremely high in basically all LTL331R cells, although most LTL331 cells also displayed moderate to strong EZH2 immunostaining. Taken together, these data demonstrate that CBX2 and EZH2 are highly over-expressed at both the mRNA and protein levels in NEPC.

Figure 3.3: CBX2 and EZH2 protein expression in NEPC. Immunohistochemical analysis in LTL331 and LTL331R (20x) A) Histological markers respective to prostate adenocarcinoma (PSA, images are representative of duplicate experiments) B) Histological marker respective to NEPC (SYP, images are representative of duplicate experiments) C) Immunostaining of CBX2 (images are representative of duplicate experiments) D) Immunostaining of EZH2 (images are representative of duplicate experiments).

Focusing on EZH2 and CBX2, we investigated whether elevated expression of these two PcG members also occurred in small cell lung cancer (SCLC), since it represents a neuroendocrine

malignancy that closely resembles NEPC histologically and molecularly [293, 294]. The relative mRNA levels of CBX2 and EZH2 were therefore assessed in SCLC and compared to two non- small cell lung cancer (NSCLC) subtypes, lung adenocarcinoma (AC) and squamous cell carcinoma (SqCC) [295]. We report that both CBX2 and EZH2 were significantly up-regulated in SCLC compared to NSCLC [280] (Figure 3.4A,B, p<0.0001 for both genes, Kruskall-Wallis test). In addition, CBX2 and EZH2 expression was strongly correlated in a clinical SCLC cohort (Pearson score=0.59, Figure 3.4C), suggesting that these two PcG proteins likely act in concert in SCLC.

Figure 3.4: Regulation of PcG proteins CBX2 and EZH2 in lung cancer subtypes. A) Differential expression of CBX2 mRNA in lung adenocarcinoma (AC), squamous cell carcinoma (SqCC), and small cell lung cancer (SCLC) (p≤0.05, Kruskal-Wallis test) B) Differential expression of EZH2 mRNA in lung adenocarcinoma (AC), squamous cell carcinoma (SqCC), and small cell lung cancer (SCLC) (p≤0.05, 69

Kruskal-Wallis test) C) Correlation between CBX2 and EZH2 mRNA levels in SCLC using (Pearson correlation=0.59, all data from CLGCP).

Another similarity between NEPC and SCLC is that both malignancies were reported to have frequent RB1 inactivation, which contributes to their high proliferative rate [120, 296]. Thus, we investigated the RB1 status in our patient-derived xenograft models. As expected, RB1 expression was relatively high in almost all PCa xenografts while the expression of CBX2 and EZH2 was considerably lower (Figure 3.5). Conversely, the expression of RB1 was undetectable in the NEPC tumor lines LTL352 and LTL370 while LTL331R exhibited a modest increase. Genomic characterization of these models revealed that both LTL352 and LTL370 had homozygous RB1 deletion while LTL331R had a monoallelic loss and a hemyzygous mutation in RB1 (Wyatt et al., unpublished). Taken together, these data suggest that aberrant PcG-mediated silencing may cooperate with RB1 inactivation to sustain the high proliferative rate of NEPC.

Figure 3.5: Relative mRNA expression of CBX2, EZH2, and RB1 in tumor tissue derived from prostate adenocarcinoma or NEPC (microarray analysis, Living Tumor Lab [139], properties of each tumor line is described in Table 3.4). 70

3.3.3. Polycomb silencing and neuroendocrine-associated repression signature

Since our initial analysis revealed that epigenetic repressors, in particular the PcG genes, were significantly up-regulated in NEPC, we investigated whether we could infer molecular and clinical information from the genes down-regulated in NEPC. We therefore derived a gene signature by combining the list of genes whose expression decreased by at least two-fold (50%) in three independent scenarios: 1) LTL331R vs LTL331; 2) clinical NEPC vs PCa; and 3) LTL331 vs all other LTL PCa. The latter was investigated under the hypothesis that LTL331 may be “predisposed” to transdifferentiate compared to other PCa tumor lines. Thus, using this 50% threshold, we established a list of 185 genes silenced in NEPC, which we termed the Neuroendocrine-Associated Repression Signature (NEARS, Figure 3.6).

Figure 3.6: Diagram summarizing the establishment and analysis of a 185-gene “Neuroendocrine- Associated Repression Signature” (NEARS) derived from datasets originating from NEPC models using the Oncomine resource. Results of the concept analysis are shown in Table 3.5 and results of the prognostic analysis can be found in Table 3.6. 71

We first used the Oncomine database [279] to identify molecular “concepts” (that is, sets of genes derived from previously published experiments) that were significantly associated with NEARS. In line with the striking up-regulation of PcG genes, this analysis revealed that, of the thousands of concepts present on Oncomine, 6 of the top 12 concepts were directly linked to PcG-mediated silencing (Table 3.5, odds ratios=3.7 to 5.0, p=3.4x10-9 to 1.2x10-15). These 6 concepts specifically overlapped with target genes of known PcG members CBX8, SUZ12, and EED, as well as H3K27me3. Of note, these concepts were derived from experiments conducted either in embryonic stem cells or in embryonic fibroblasts, consistent with the role of PcG complexes in undifferentiated cells [189]. As an unexpected indicator of quality control, we also found that two concepts strongly linked to NEARS included “down-regulated genes in prostate cancer after androgen ablation therapy” and “up-regulated genes in prostate cancer cells in response to synthetic androgen R1881”, two concepts sharing 19 genes (odds ratios=20.5, p<3.662x10-18). These concepts describe genes that are likely AR-regulated [297], and therefore their down-regulation is expected in NEPC cells given their lack of AR expression. Taken together, our results demonstrate that NEARS is enriched in PcG targets and preferentially contains genes regulated by AR transactivation.

PcG gain of function usually correlates with poor patient prognosis and our results suggest that the NEARS preferentially contains many PcG targets. Thus, we sought to determine whether NEARS was also associated with clinical parameters indicative of aggressive PCa progression. Using the Oncomine database, we analyzed the differential expression of NEARS in tumors of different metastatic potential, grade, and prognosis. Strikingly, significant down-regulation of NEARS was observed in metastatic compared to primary prostate malignancies in six independent datasets (Table 3.6, odds ratios=2.4-6.0, p=4.5x10-4-2.0x10-26). Likewise, NEARS under-expression was also recorded in 8 clinical PCa datasets (Table 3.6, odds ratios=2.2-7.6, p=3.4x10-4-1.9x10-8). Finally, 6 additional datasets displayed reduced NEARS expression in patients with poor clinical outcome (Table 3.6, odds ratios=2.4-15.2, p=2.0x10-3-1.0x10-11), further supporting the idea that down-regulation of PcG target genes correlates with aggressive disease progression. Overall, our results demonstrating that NEARS contains many conserved PcG genomic targets and that multiple PcG genes are highly up-regulated in NEPC strongly

suggest that altered PcG-mediated silencing plays a key role in establishing and maintaining the NEPC phenotype.

Table 3.5: List of top 12 Literature-derived concepts most significantly associated with NEARS (Oncomine Concept Analysis, see Figure 3.6).

NEARS Literature-Defined Oncomine Concept Analysis

PcG- Odds Rank Concept P Value Q Value Related Ratio

1 CBX8 target genes in human embryonic fibroblasts X 1.2E-15 9.6E-12 5.0

Up-regulated genes in neutrophils compared to other 2 7.4E-13 3.8E-10 6.5 blood cells

Down-regulated in human embryonic stem cells vs 3 1.3E-13 5.8E-10 6.9 differentiated counterparts

Down-regulated genes in prostate cancer after 4 4.9E-12 1.5E-8 20.5 androgen ablation therapy

5 SUZ12 target genes in human embryonic stem cells X 5.2E-12 1.5E-8 6.4

Trimethylated H3K27 target genes in human 6 X 2.5E-11 4.8E-8 5.9 embryonic stem cells

7 Drugbank Targets – FDA approved 5.6E-11 1.0E-8 12.3

Up-regulated genes in weakly invasive colon cancer 8 5.6E-11 1.0E-7 12.3 cells

Polycomb Group target genes in human embryonic 9 X 7.0E-10 8.4E-7 7.1 stem cells

Up-regulated genes in prostate cancer cells in 10 9.1E-10 9.5E-7 11.0 response to synthetic androgen R1881

11 EED target genes in human embryonic fibroblasts X 1.8E-9 1.8E-6 5.4

Trimethylated H3K27 target genes in human 12 X 3.4E-9 2.8E-6 3.7 embryonic fibroblasts

Table 3.6: Correlations between down-regulation of NEARS and poor prognostic factors in clinical prostate tumors (Oncomine Clinical Outcome Analysis, see Figure 3.6).

Clinical Parameters Associated with NEARS

Analysis Concept Dataset Odds Ratio P value Percentile Samples Metastasis Lapointe Prostate 2.4 1.3E-04 Top 10% 71 Metastasis LaTulippe Prostate 2.4 4.5E-04 Top 10% 32 Metastasis Yu Prostate 3.5 2.0E-07 Top 10% 88 1o VS Met Metastasis Grasso Prostate 3.8 5.6E-13 Top 10% 94 Metastasis Vanaja Prostate 4.6 8.4E-18 Top 10% 32 Metastasis Taylor Prostate 3 6.0 2.0E-26 Top 10% 150 Advanced Gleason Score Wallace Prostate 2.2 3.4E-04 Top 10% 65 Advanced Gleason Score Vanaja Prostate 2.4 6.6E-06 Top 10% 27 Advanced Gleason Score Lapointe Prostate 3.4 1.8E-05 Top 5% 61 Advanced Gleason Score Tomlins Prostate 3.6 1.9E-08 Top 10% 30 Grade Advanced Gleason Score Taylor Prostate 3 3.6 4.4E-08 Top 5% 130 Advanced Gleason Score Yu Prostate 4.4 4.1E-07 Top 5% 61 Advanced Gleason Score Setlur Prostate 7.6 2.8E-04 Top 1% 353 High Grade Bittner Prostate 2.5 3.3E-04 Top 5% 46 Dead at 3 years Setlur Prostate 2.4 2.0E-03 Top 10% 358 Recurrence at 5 years Taylor Prostate 3 3.1 1.3E-09 Top 10% 61 Recurrence at 3 years Taylor Prostate 3 3.5 1.0E-11 Top 10% 107 Outcome Recurrence at 5 years Nakagwa Prostate 10.1 1.0E-03 Top 10% 592 Dead at 5 years Setlur Prostate 10.7 2.9E-06 Top 1% 363 Recurrence at 3 years Nakagwa Prostate 15.2 7.9E-04 Top 5% 594

3.4. Discussion NEPC is an incurable malignancy which is expected to become more prevalent given the widespread use of potent AR-targeting drugs, which positively select for AR-negative NEPC cells [10]. However, the lack of suitable pre-clinical NEPC models has limited investigations into the molecular underpinnings of NEPC, therefore hampering therapeutic development of novel agents to treat this deadly disease. To circumvent this issue, we took advantage of a high fidelity, patient-derived xenograft model retaining classical NEPC features observed in the clinic 74

[139] that allows us to investigate the molecular mechanisms involved in NEPC transdifferentiation. Using this model, we identified many PcG genes that are dysregulated in NEPC, a finding that was also observed in a clinical cohort and in additional patient-derived NEPC xenografts from the LTL. Moreover, we derived a NEARS that was predictive of PcG gain of function and aggressive disease progression. Based on these results, we propose that aberrant PcG-mediated silencing contributes to NEPC pathogenesis and that disrupting PcG activity may emerge as a valuable therapeutic strategy in NEPC.

In line with the up-regulation of PcG genes observed in NEPC, we derived a list of genes recurrently silenced in multiple NEPC models (NEARS) that was strongly associated with PcG- mediated repression. Moreover, silencing of NEARS in localized PCa tissue predicts poor clinical outcome, consistent with the notion that PcG-driven tumors are highly aggressive. Interestingly, NEARS contained many PcG target genes in ESCs, implying that PcG activity in NEPC may suppress epithelial differentiation through silencing of genes specifying specialization into prostatic tissues, particularly since NEARS included some AR-regulated genes. In line with this idea, epigenetic reprogramming of a similar nature has been repeatedly observed in many aggressive tumor types featuring aberrant PcG-mediated repression [187]. In addition, PcG-mediated repression has been involved in neuronal differentiation [276, 298], thus it seems likely that the activation of neuroendocrine-specific transcriptional programs may also be facilitated by increased PcG activity. Supporting this idea, we observed preferential over- expression of PcG genes CBX2 and EZH2 in SCLC, which represents a lung cancer subtype with neuroendocrine differentiation. Taken together, these results suggest that gene expression profiles regulated by PcG genes during normal embryogenesis are re-established in NEPC and possibly other neuroendocrine malignancies.

From a molecular standpoint, it is important to note that CBX2 up-regulation observed in NEPC occurs in the context of over-expressed EZH2 and other PRC2 members, which likely alters the genomic distribution of the H3K27me3 mark. This has considerable mechanistic implications since CBX2 can directly bind H3K27me3 and recruit PRC1 to H3K27me3 sites, which solidifies transcriptional repression at target loci [173]. Thus, CBX2 over-expression might represent an alteration necessary to mediate the downstream epigenetic effect of PRC2 gain of function. 75

Biochemical evidence supports this hypothesis since, although PRC1 composition varies in a context-dependent manner, PRC1 complexes found at H3K27me3 sites are preferentially enriched in CBX2 and not other CBX proteins [181]. While the relative contribution of CBX2 compared to other CBX members remains under investigation, the strong up-regulation of CBX2 in NEPC, in addition to its critical role in cellular differentiation, suggest that CBX2 is functionally involved in PcG-mediated progression of NEPC.

Despite playing imperative roles during embryonic development [193], CBX2 has been overlooked for many years in the cancer field. In chapter 2, we have demonstrated that CBX2 represents a potential therapeutic target in CRPC. In this chapter, we report that CBX2 was consistently the most highly up-regulated EpR in NEPC compared to PCa in our analyzed datasets. More specifically, we have shown that high CBX2 expression, as well as silencing of PcG targets, correlates with metastasis and poor patient outcome. In both CRPC and NEPC, we have provided evidence that CBX2 activity is intricately related to the status of the E2F/RB1 axis, in line with its cell-cycle regulated function in embryonic development. Since the role of CBX2 has yet to be addressed in human cancers, our results in advanced PCa argue that CBX2 is functionally required for optimal tumor growth and therefore may be implicated in the pathogenesis of other malignancies.

4. CBX2 meta-analysis in human cancers: evidence of a widespread oncogenic role 4.1. Introduction In the past decade, a number of chromatin regulators known to play key molecular functions in embryonic development have been identified as novel cancer-driving genes [140, 299]. These epigenetic factors are thought to retain cancer cells in an undifferentiated state, thus enhancing their metastatic potential and resistance to treatment [300, 301]. Accordingly, epigenetic therapies have been developed and have made their way into the clinic [302]. While for many years pharmacological targeting of EpRs has focused around blocking the enzymatic activity of epigenetic writers and erasers [285, 303], more and more efforts are devoted to inhibiting the function of chromatin readers. For example, inhibitors of the H3K4me3 reader BRD4 have been successfully employed in pre-clinical models of advanced cancers [251]. More recently, Tabet et al. developed small molecules blocking the chromodomain of CBX7, which represent the first antagonist of any CBX chromodomain [223, 224]. These data indicate that pharmacologically targeting CBX2 is possible at least in principle and therefore its therapeutic value should be assessed in a number of tumor types other than PCa.

CBX2 is part of the Polycomb Group (PcG) family, a group of epigenetic repressors that assemble in a complex combinatorial fashion to form two main Polycomb Repressive Complexes (PRC1 and PRC2) which function silence lineage-specific genes both in embryonic stem cells and multiple cancer types [187, 304, 305]. In solid malignancies, functional activation of PRCs is almost universally associated with aggressive disease progression across a number of tumor types [187]. However, the exact mechanisms leading to this gain of function remain ill-defined. One emerging possibility is that over-expression of one subunit can induce alterations in PRC activity and sequence specificity. While EZH2 and BMI-1 are the two PcG genes that have garnered most of the attention in the context of human cancer [306, 307], emerging evidence supports a critical role in malignancy for all CBX proteins (CBX4, CBX6, CBX7, CBX8) except for CBX2.

The data presented in chapters 2 and 3 supports an important function for CBX2 in aggressive variants of hormone-independent PCa. These results represent the first clear evidence that CBX2 is functionally involved in any human malignancy. Since we demonstrated that CBX2 represents 77

a potential therapeutic target in advanced prostate malignancies, we set out to investigate whether alterations in CBX2 were also present in other cancer types. Given the lack of literature addressing the implications of CBX2 in a neoplastic context, we conducted a meta-analysis of CBX2 in human cancers at the genomic and transcriptomic level using publically available databases. We report that CBX2 down-regulation and inactivating genetic mutations represent extremely rare events in human tumors. Furthermore, we also provide the first evidence that CBX2 genomic amplification and mRNA up-regulation predicts metastatic progression and poor overall survival (OS) in multiple cancer types, including those of the breast, lung, and prostate.

4.2. Materials and methods 4.2.1. COSMIC database analysis

The COSMIC database (http://cancer.sanger.ac.uk/cancergenome/projects/cosmic/) [308] was used to extract all data relevant to our analysis of genetic alterations at the CBX2 locus. All data originating from the COSMIC database is specifically mentioned either in the main text, the figure legend, or both.

4.2.2. Oncomine database analysis

All transcriptomic data used in this chapter was extracted from the Oncomine platform (www.oncomine.com) [279]. Data was acquired in an unbiased fashion by compiling all the Oncomine studies which showed significantly altered CBX2 expression at the threshold set for each individual analysis. Significant studies in which at least one analyzed group was comprised of 3 patients or less were excluded. All data originating from the Oncomine database is specifically mentioned either in the main text, the figure legend, or both.

4.2.3. Human protein atlas database analysis

The image of CBX2 protein sequence and domains used in Figure 4.3 was obtained from the Protein Atlas database (http://www.proteinatlas.org/) [309].

4.2.4. Analysis of the minimal common region of CBX2 amplification

TCGA breast cancer copy number data was downloaded from Cancer Genomics Browser (https://genome-cancer.ucsc.edu/) and all tumors containing CBX2 gains were determined. We filtered to segments with gain (logR>0.3) and at least 100 kilobases (kb) in size and mapped the minimal common region, defined by the sample with the smallest region containing CBX2. We then determined the Spearman copy number-expression correlation for genes mapping to the region assessed. Those with significant correlation (p<0.05 Spearman) were further analyzed for survival associations using the Curtis Breast dataset on Oncomine and all breast cancer samples from the KMplotter resource (http://kmplot.com/analysis/) [310].

4.2.5. Statistical analysis

Unless otherwise mentioned, all analyses were done using p≤0.05 as the significance threshold. GraphPad Prism software (Version 6) was used for all statistical analyses except for the multivariate analysis of variance (MANOVA) and the Cox proportional hazards regression, which were done using R statistical software.

4.3. Results 4.3.1. Genomic analysis of the CBX2 locus

As the first step of our systematic meta-analysis to study CBX2 molecular profile in human cancers, we analyzed genomic alterations at the CBX2 locus using the COSMIC database, a resource designed to store and display somatic mutation information and related details [308]. Strikingly, there was an extremely low frequency of alterations disrupting CBX2 function. In 8013 tumor samples spanning 29 tissue types, we did not find a single translocation, homozygous loss, insertion/deletion, or any other inactivating large-scale chromosomal abnormality (Figure 4.1A). In total, only 40 point mutations were recorded at the CBX2 gene in these tumors (overall frequency=0.5%). The CBX2 mutation frequency is thus considered very low, as high frequency mutations are typically described as being over 20% and intermediate frequency as being between 2% and 20% [311].

Figure 4.1: Extremely rare occurrence of genetic mutations disrupting CBX2 function in human cancers A) Percentage of specific inactivating genetic alterations at the CBX2 locus (8013 patients, COSMIC database) B) Distribution and frequency of CBX2 point mutations (8013 patients, COSMIC database).

Further analysis revealed that, of those 40 mutations, 2 were nonsense and 22 were missense, while 17 were silent mutations which likely did not have any effect on CBX2 protein sequence (Figure 4.1B). These point mutations were distributed evenly across all tissues, with no tumor type having a marked increase in CBX2 mutation frequency (Figure 4.2). The mutations were concentrated in regions where the CBX2 protein is predicted to be highly disordered and hydrophilic, and within these regions their distribution was relatively homogenous (Figure 4.3). Additionally, no single residue within the CBX2 protein was mutated more than twice. Interestingly, there was a complete absence of point mutations in the CBX2 chromodomain, the region responsible for H3K27me3 binding (Figure 4.3). Overall, the lack of large-scale genomic aberration and the non-synonymous mutation frequency of 0.3% make it unlikely that alterations at the CBX2 locus impact cancer cell phenotype.

Figure 4.2: Percentage of CBX2 point mutations in different human cancer types (data from COSMIC database).

Figure 4.3: Mutational analysis of CBX2 protein in human cancers. Distribution of non-silent point mutations within CBX2 protein sequence (figure exported from the Human Protein Atlas, mutational data from COSMIC database)

The low frequency of genomic CBX2 disruption led us to investigate whether CBX2 copy number increases are favored during oncogenesis. To address this question, we first analyzed the amplification profile of the CBX2 locus in human neoplasms using the COSMIC database. In contrast with the rare nature of CBX2 inactivating mutations, we uncovered that CBX2 gene amplifications occur frequently in a number of tumor types. Overall, 714 out of 8013 samples from the COSMIC database had undergone copy number gain (CNG) at the CBX2 locus (overall frequency: 8.9%; see Figure 4.4 and Table 4.1). Interestingly, the distribution of these amplifications was not homogenous across all tumor types. We observed 5 neoplasms in which the CBX2 CNG frequency ranged between 3% and 15%: those originating from the central nervous system, colon, endometrium, pancreas, and kidneys (Figure 4.4 and Table 4.1). Furthermore, three cancer types harbored a frequency of CBX2 amplification greater than 30%: tumors of the ovaries (34.0%), breast (34.5%), and lungs (35.5 %), suggesting that CBX2 copy number increases may provide a selective advantage to cancer cells.

Figure 4.4: Frequency of CBX2 amplification across different tumor types (data from COSMIC database, see Table 4.1 for patient number)

Table 4.1: List of tumor types harboring a frequency of CBX2 amplification higher than 3% (data from COSMIC database).

# of CBX2 # of Total Amplification Tissue Type Amplified Samples Percentage Pancreas 1 32 3.1 Breast 270 782 34.5 CNS 7 140 5.0 Endometrium 28 246 11.4 Kidney 22 300 7.3 Colon 54 486 11.1 Lung 169 476 35.5 Ovary 157 462 34.0

4.3.2. Transcriptomic analysis of CBX2 expression in human cancers

Since our genomic analysis revealed recurrent CNGs and very rare inactivating mutations at the CBX2 locus, we next investigated whether this trend would also be reflected at the mRNA level. Using the Oncomine database [279], we identified a total of 25 studies which showed significant up-regulation (FC>2, p value<0.001, top 10% over/under-expressed) in cancer compared to normal tissue (Figure 4.5, Table 4.2). Strikingly, not a single study reported down-regulation of CBX2 using the same inclusion criteria (Figure 4.5), once again implying an important functional role in cancer cells. The total number of patients in the 25 studies showing CBX2 up- regulation in cancer tissues is 3848, compared to zero for CBX2 down-regulation (Figure 4.5, Table 4.2). In the studies harboring CBX2 over-expression, fold changes varied between 2.1 and 15, and the p values between 4.0E-3 and 3.6E-73 (Figure 4.5). The most represented cancer types in the CBX2-over-expressed studies were those originating from the colon (29.6%), breast (18.5%), stomach (14.8%), and lung (11.1%). These results demonstrate a clear bias towards CBX2 up-regulation and complement the genomic analysis which hinted towards a selective pressure to maintain CBX2 function.

PcG complexes are known to repress the tumor suppressive loci CDKN2A (encoding p14 and p16) and CDKN2B (encoding p15) in many human cancers [312]. We thus sought to determine whether CBX2 over-expression correlated with silencing of the CDKN2A/B genes. Interestingly, we found that neither p14 nor p16 were down-regulated in any of the 25 studies with CBX2 over-expression (Figure 4.6). However, p15 was found to be down-regulated in 10 of those 25 studies (40%) using the same cutoff as for CBX2 (FC>2, p value<0.01, top 10% under- expressed). However, when using Spearman correlations to investigate the direct relationship between CBX2 and CDKN2A/B, not a single study showed a statistically significant correlation (Table 4.3). Taken together, analysis of CBX2 mRNA levels revealed lack of CBX2 down- regulation contrasted by frequent CBX2 over-expression, an event which occurred independently of CDKN2A/B silencing.

Figure 4.5: Marked up-regulation of CBX2 in cancerous compared to normal tissues (Oncomine database). A) Number of studies displaying significant CBX2 up-regulation or down-regulation in cancer vs normal tissues at different p values. The total number of patients in the significant studies is shown in brackets. (inclusion criteria: FC≥2, top 10% under/over-expressed, student's t-test) B) Tissue distribution of the 25 studies harboring significant CBX2 up-regulation at p≤0.001 (student's t-test).

Table 4.2: Studies with differential CBX2 expression between cancerous and normal tissues (Oncomine database; inclusion criteria: FC≥2, p≤0.001, top 10% under/over-expressed, student's t- test).

# # # Fold Alteration Rank Study Name Cancer Type p-value Normal Cancer Total Change

Invasive ductal breast 1 Curtis Breast 144 1556 1700 2.1 3.6E-73 carcinoma Invasive ductal breast 2 TCGA Breast 61 389 450 3.3 2.5E-32 carcinoma 3 Sun Brain Glioblastoma 23 81 104 3.8 1.9E-16 Richardson Ductal breast 4 7 40 47 9.4 1.3E-15 Breast 2 carcinoma TCGA 5 Rectal adenocarcinoma 22 60 82 3.6 1.9E-15 Colorectal TCGA Rectal mucinous 6 22 6 28 4.8 1.0E-14 Colorectal adenocarcinoma TCGA 7 Colon adenocarcinoma 22 101 123 3.1 5.8E-14 Colorectal Medullary breast 8 Curtis Breast 144 32 176 4.7 3.1E-13 carcinoma Invasive breast 9 TCGA Breast 61 76 137 2.6 4.7E-12 carcinoma Up- Regulation Crabtree Uterine corpus 10 27 50 77 2.9 9.9E-12 Uterus leiomyoma TCGA Colon mucinous 11 22 22 44 3.3 7.4E-11 Colorectal adenocarcinoma Infiltrating bladder 12 Lee Bladder 68 62 130 2.7 1.0E-10 urothelial carcinoma TCGA 13 Cecum adenocarcinoma 22 22 44 4.3 2.5E-10 Colorectal Skrzypczak 14 Colorectal carcinoma 24 36 60 2.2 3.5E-10 Colorectal Squamous cell lung 15 Hou Lung 65 27 92 3.7 5.3E-10 carcinoma Hong 16 Colorectal carcinoma 12 70 82 15 8.0E-10 Colorectal Gastric intestinal type 17 Derrico Gastric 31 26 57 4.5 1.7E-09 carcinoma

18 Cho Gastric Diffuse gastric 19 31 50 2.3 6.0E-09

adenocarcinoma

Skrzypczak 19 Colorectal carcinoma 10 5 15 2.9 6.2E-09 Colorectal 2 20 Hou Lung Lung adenocarcinoma 65 45 110 2.3 1.4E-08 21 Sun Brain Oligodendroglioma 23 50 73 2.7 2.1E-08 Large cell lung 22 Hou Lung 65 19 84 5.4 7.6E-07 carcinoma Korkola 23 Yolk sac tumor 6 9 15 2.2 8.7E-05 Seminoma Brune 24 Burkitt's lymphoma 25 5 30 2.6 1.7E-04 Lymphoma Gastric mixed 25 Cho Gastric 19 10 29 2.1 3.8E-04 adenocarcinoma

Figure 4.6: CBX2 over-expression is not associated with CDKN2A silencing but co-occurs with CDKN2B down-regulation in colorectal cancer only. A) Graph showing the number of studies with down-regulation of p14, p15, and p16 in studies with CBX2 up-regulation at different p values (inclusion criteria: FC≥2, top 10% under/over-expressed, student's t-test) B) Distribution of studies with CBX2 up- regulation and p15 down-regulation in colorectal and non-colorectal cancers.

Table 4.3: Spearman correlation between CBX2 and CDKN2A/B in the datasets displaying CBX2 up-regulation in malignant compared to normal tissues (data from cBIO Portal, MSKCC Prostate dataset)

Dataset # Patients R² CDKN2A R² CDKN2B Brune Lymphoma 42 0.00 0.05 Cho Gastric 71 0.02 0.07 Crabtree Uterus 50 0.05 0.01 Curtis Breast 1992 0.01 0.00 Derrico Gastric 38 0.02 0.00 Hong Colorectal 70 0.04 0.00 Hou Lung 91 0.00 0.01 Korkola Seminoma 101 0.02 0.01 Lee Bladder 188 0.01 0.04 Richardson Breast 2 40 0.12 0.02 Skrzypczak Colorectal 81 0.01 0.03 Skrzypczak Colorectal 2 20 0.03 0.01 Sun Brain 157 0.01 0.00 TCGA Breast 532 0.11 0.05 TCGA Colorectal 215 0.00 0.00

4.3.3. Clinical correlations of differential CBX2 expression

Given the widespread CBX2 up-regulation in cancerous compared to normal tissues, we investigated whether CBX2 expression was also correlated with clinical indicators of tumor progression. Metastasis was the first parameter we assessed. Using the Oncomine database [279], we found 9 studies exhibiting significantly (FC≥1.5, p≤0.05) increased CBX2 mRNA levels in metastatic compared to primary tumors (Table 4.4). Prostate cancer was the most highly represented tumor type, accounting for 3 of the 5 most significant studies. Sarcoma and breast cancer followed with two studies each displaying CBX2 up-regulation in metastatic disease. In

contrast with the strong CBX2 up-regulation observed in disseminated tumors, not a single study with CBX2 under-expression could be detected using the same parameters (Table 4.4), consistent with a selection against CBX2 loss of function in metastatic cells.

Table 4.4: List of studies with differential CBX2 expression between metastatic and primary tumors (Oncomine database; inclusion criteria: FC≥1.5, p≤0.05, top 10% under/over-expressed, student's t-test)

Fold Alteration Rank Study Name #Primary #Metastatic #Total P Value Change

1 Chandran Prostate 10 21 31 7.8 1.7E-12 2 Bittner Breast 327 9 336 1.9 3.2E-05 3 Varambally Prostate 7 6 13 9.2 1.3E-04 4 Segal Sarcoma 2 46 4 50 13 4.4E-04 Up- 5 Grasso Prostate 13 17 30 1.9 4.0E-03 Regulation 6 Riker Melanoma 16 40 56 1.6 6.0E-03 7 Radvanyi Breast 41 4 45 3.1 9.0E-03 8 Bittner Sarcoma 42 10 52 1.7 1.1E-02 9 Liao Liver 4 6 10 4.9 2.3E-02 Down- - None found - - - - - Regulation

To assess the relationship between CBX2 and clinico-pathologic variables, we conducted MANOVA in one cohort for each of breast, lung, and colorectal cancers from the Oncomine platform. All significant covariates (p<0.05) in the MANOVA were further assessed using univariate analyses. We found that CBX2 was significantly associated with sex in colon cancer, with higher expression observed in females (Figure 4.7A, Table 4.5, TCGA Colorectal, p<0.01, student’s test). In the analyzed lung cancer dataset, subtype was the only clinical covariate associating with CBX2, with higher CBX2 expression in squamous cell carcinoma compared to lung adenocarcinoma (Figure 4.7B, Table 4.6, Bild Lung, p<0.05, student’s test). Finally, MANOVA conducted on the Curtis Breast dataset revealed a highly significant association between CBX2 and age, subtype, and grade that was confirmed via Kruskal-Wallis test (Figure

4.7C, D, Table 4.7, all p<10-15, Kruskal-Wallis). More specifically, higher CBX2 expression correlated with younger age, basal-like subtype and higher grade, all of which are linked to poor patient prognosis. We therefore conducted a COX proportional hazards (COXPH) regression on the Curtis Breast dataset to further explore the relationship between CBX2 and clinical outcome. Interestingly, we found that behind age CBX2 was the second covariate most significantly associated with patient survival (Table 4.8, COXPH, p<0.001), suggesting a role for CBX2 in promoting aggressive disease progression.

Figure 4.7: Differential CBX2 expression associates with clinico-pathologic features. A) Sex-specific CBX2 expression in TCGA colon (p=0.0015, student’s test) B) Subtype-specific CBX2 expression in Bild lung (p=0.03, student’s test) C) Subtype-specific CBX2 expression in Curtis breast (p<0.0001, Kruskal-Wallis test) D) Grade-specific CBX2 expression in Curtis breast (p<0.0001, Kruskal-Wallis test)

Table 4.5: MANOVA of CBX2 in TCGA colon dataset (***p≤0.001; **p≤0.01; *p≤0.05).

Factor Df Sum Sq Mean Sq F value P value Significance Age 1 1.389 1.389 1.6459 0.20096 Nodal 1 0.196 0.1965 0.2328 0.629982 Gender 1 9.501 9.5013 11.2584 0.000944 *** Stage 3 1.566 0.5219 0.6184 0.603827

Table 4.6: MANOVA of CBX2 in Bild lung dataset (***p≤0.001; **p≤0.01; *p≤0.05).

Factor Df Sum Sq Mean Sq F value P value Significance Age 1 0.425 0.4248 0.5669 0.45319 Gender 1 0.028 0.0284 0.0379 0.84594 Stage 3 0.398 0.1325 0.1768 0.91188 Histology 1 4.099 4.0988 5.4695 0.02126 *

Table 4.7: MANOVA of CBX2 in Curtis breast dataset (***p≤0.001; **p≤0.01; *p≤0.05).

Factor Df Sum Sq Mean Sq F value P value Significance Age 1 116.24 116.236 123.2413 < 10-15 *** Subtype 4 1095.55 273.888 290.3944 < 10-15 *** Grade 2 124.78 62.39 66.15 < 10-15 *** Nodal 1 2.72 2.723 2.8872 0.0894

Table 4.8: Cox proportional hazards regression for CBX2 in Curtis breast dataset.

Curtis Breast coef exp(coef) se(coef) z p CBX2 0.108 1.11 0.03369 3.205 0.0013 Age 0.0316 1.03 0.00312 10.123 0 Grade 2 0.1439 1.15 0.15029 0.957 0.34 Grade 3 0.3413 1.41 0.15505 2.201 0.028 Basal Subtype 0.2722 1.31 0.12925 2.106 0.035 Luminal B Subtype 0.2495 1.28 0.09536 2.617 0.0089 Her2 Subtype 0.3634 1.44 0.12174 2.985 0.0028 Normal Subtype 0.2839 1.33 0.13662 2.078 0.038

Finally, we investigated whether CBX2 is a potential driver of the recurrent 17q25.3 copy gains we observed. We used multiple criteria to investigate genes mapping to the minimal common region of amplification containing CBX2, including copy number expression association, expression in tumours compared to normals, and survival association, as these features could highlight the gene that is most likely to be driving the 17q25.3 amplicon. Since CBX2 was recurrently amplified and differentially expressed in breast cancer, we investigate the clinical implications of the CBX2 amplicon in breast cancer. We first used breast cancer TCGA copy number data from the Cancer Genomics Browser [313] (https://genome-cancer.ucsc.edu/) to identify all gained genomic segments containing CBX2. We then filtered to segments with gain (defined by having a log ratio > 0.3) and at least 100kb in size and mapped the minimal common region (MCR), which is defined by the sample with the smallest region containing CBX2. We found that the MCR contained 3 genes: ENPP7, CBX2, and CBX8, present at 17q25.3 (Figure 4.8A, B).

We next assessed the correlation between copy number and the expression of genes mapping to the region using Spearman correlation. ENPP7 had a negative correlation, indicating that its amplification did not result in increased expression. In contrast, a weak but significant positive correlation was observed for both CBX2 and CBX8 (p<0.0001, Spearman test). To determine

whether either of the two genes was associated with patient survival, we performed a Mantel- Cox logrank test for both CBX2 and CBX8, comparing survival of patients with expression ranking in the top and bottom quartile of expression. We report that elevated CBX2 levels, but not CBX8, significantly predicts lower overall survival in breast cancer (Figure 4.8C, D, CBX2 p<0.0001; CBX8 p=0.49, Mantel-Cox logrank test). To ensure that these data are reproducible, we also performed logrank test on CBX2 and CBX8 using the KMplotter (http://kmplot.com/analysis/) [310]. Once again, we found that CBX2 was the only gene within the MCR which was able to significantly predict lower overall survival (Figure 4.8E, F, p<0.05, logrank test). Overall, we demonstrate that CBX2 is frequently gained in breast cancer, which leads to its increased expression and associates with poor patient prognosis, suggesting that it represents a candidate driver of the 17q25.3 amplicon.

In summary, genomic analysis has revealed that genetic alterations resulting in loss of function represent extremely infrequent events at the CBX2 locus. In contrast, recurrent amplifications were observed in multiple cancer types. Additionally, transcriptomic analysis revealed a propensity for CBX2 up-regulation in four independent scenarios 1) cancer vs normal tissues 2) metastatic vs primary tumors 3) dead vs alive at 3 years 4) dead vs alive at 5 years. Surprisingly, with the exception of CDKN2B in colorectal cancer, CBX2 over-expression was not correlated with silencing of the tumor suppressive CDKN2A or CDKN2B loci.

Figure 4.8: CBX2 as the driver within the minimal common region of its amplicon. A) Identification of the CBX2 minimal common region on 17q25.3 B) Genes present within the CBX2 minimal common

region. Logrank test assessing link with overall survival for C, E) CBX2 (p<0.05) and D, F) CBX8 (p>0.05)

4.4. Discussion Our results demonstrate a clinically-relevant increase in CBX2 copy number and expression in human cancer. Given the very low frequency of CBX homozygous loss, point mutation, and under-expression, we believe CBX2 may play an important functional role in tumor cells. In parallel, transcriptomic analysis of CBX2 expression revealed a strong bias towards CBX2 up- regulation, although this alteration was observed at different stages among the various tumor types, reflecting some context-specificity in CBX2 activity and binding partners. Overall, breast cancer displayed the most significant associations with CBX2 alterations, exhibiting an overall frequency of CBX2 amplification exceeding 30%. It is interesting to note that we did not observe CBX2 amplification in PCa (chapter 2), indicating that the molecular mechanisms underlying CBX2 over-expression may be tumor-specific. Of all the genes present in the breast cancer MCR containing CBX2, only CBX2 mRNA expression exhibited prognostic significance, suggesting that CBX2 could likely be implicated in the progression of breast cancer. Since CBX2 and CBX8 proteins share a conserved chromodomain but largely differ in their C-terminus [214], we believe further investigation is required to elucidate the structural differences that may underlie the specific cancer-promoting mechanism of CBX2.

Given that many breast cancers also harbor EZH2 over-expression [314, 315], an interesting possibility is that CBX2 up-regulation represents a key adaptation necessary to perpetuate EZH2’s oncogenic activity. This therefore implies an important role for the ability of CBX2 to interact with H3K27me3 and is consistent with our observation that missense mutations are excluded from the chromodomain of CBX2. Interestingly, EZH2 and H3K27me3 also regulate the expression of several tissue-specific lncRNAs [316, 317]. Since many loci encoding lncRNA are rich in AT repeats and CBX2 contains an AT hook domain which can interact with those regions [165], it is conceivable that EZH2 and CBX2 directly regulate the lncRNAs which themselves determine PcG target specificity.

PCa is another neoplasm in which EZH2 over-expression represents a key hallmark [140, 141], and interestingly it was the cancer type in which CBX2 up-regulation most significantly correlated with metastatic progression, consistent with our findings from chapter 2. As opposed to breast cancer, we could not find any significant differential CBX2 expression in neoplastic tissues compared to normal ones, indicating that CBX2 up-regulation likely represents a late event in PCa progression, in accordance with our findings from chapters 2 and 3. However, we did not observe any correlations between CBX2 expression and overall survival. We attribute this result to a technical limitation of the Oncomine database, which can only calculate survival at 1, 3, and 5 years. Since PCa is generally a slow-growing disease and that more than 98% of patients are alive five years post-diagnosis [318], it is likely that analyzing differences in OS strictly below 5 years post-diagnosis is simply too early to achieve statistical significance. Nonetheless, the fact that PCa was the tumor type with the most studies exhibiting elevated CBX2 levels in metastatic tissues solidifies the important role of CBX2 in advanced prostate malignancies described in chapters 2 and 3.

Overall, our results conclusively demonstrate a recurrent CBX2 up-regulation frequently correlating with metastasis and lower OS in numerous cancer types, consistent with our studies in advanced PCa. However, one limitation of our meta-analysis is that it does not allow for an in- depth characterization of subtype-specific CBX2 expression. For example, it is possible that CBX2 expression is markedly high in one molecularly- or histologically-characterized subtype while being low in another from a given neoplasm, something which could not be determined from our analysis. Nonetheless, by using the COSMIC and Oncomine databases we were able to analyze over 40 000 patient samples in a fully unbiased fashion, which allowed us to observe a genotranscriptomic profile for CBX2 that was consistent with that of an oncogene.

Finally, the very low frequency of CBX2 mutation and under-expression suggests that the therapeutic inhibition of CBX2 might represent a valuable clinical strategy for the treatment of many human cancers. We report aberrant regulation of CBX2 in breast, lung, colorectal, prostate, brain, and hematopoietic tumors, all of which rank among the 10 deadliest neoplastic diseases worldwide and are in dire need of novel therapeutics. Furthermore, epigenetic alterations are reversible and recent studies have highlighted the possibility of targeting histone readers with 96

small molecule inhibitors [223, 319]. Taken together, our work has identified a putative oncogenic role for CBX2 in numerous tumor types and has provided the rationale for the design of novel CBX2-targeting therapies.

5. Conclusions 5.1. Summary of findings As the role of CBX2 has never been described in the context of cancer, this thesis provides the very first report that CBX2 is involved in cancer progression, notably in advanced prostate and breast tumors. In Chapter 2, we started by analyzing the expression of PcG genes in PDX models of advanced PCa. We first identified that CBX2 was the PcG member most significantly up- regulated in xenograft models representative of aggressive disease. This was confirmed at both the mRNA and protein levels, and further validated in numerous independent clinical cohorts. Furthermore, elevated CBX2 levels predicted high grade, increased metastatic ability, and lower overall survival in PCa patients. Taken together, these results hinted that CBX2 might represent a driver of PCa progression and prompted us to investigate its phenotypic impact in advanced PCa cell lines.

Using the validated LNCaP/C4-2 model of PCa progression [136], we silenced the expression of CBX2 through siRNA in order to assess the functions regulated by CBX2. Strikingly, CBX2 had a major effect on PCa cell phenotype, inducing significant proliferation arrest and widespread apoptosis, consistent with the hypothesis that CBX2 was required for advanced PCa cell survival. In order to dissect the molecular mechanisms underlying CBX2’s cancer-promoting activity, we conducted microarray analysis on C4-2 cells that were depleted of CBX2. As expected, numerous CRGs were involved in pathways controlling cellular proliferation and differentiation. Notably, many CRGs played pivotal roles in the PI3K/AKT pathway, which is known to be aberrantly activated in more than 80% of CRPC patients. In addition, CBX2 depletion significantly reduced the expression of a number of genes involved in mitotic spindle formation and chromosome segregation, implying a fundamental role in mitosis. Overall, the clinically-relevant up-regulation of CBX2 in CRPC and the striking proliferation arrest induced upon CBX2 depletion in advanced PCa cells suggest that CBX2 functionally contributes to PCa progression.

Since potent AR antagonists have recently been introduced to treat CRPC [14], it is anticipated that the incidence of NEPC will rise quickly in the future as an adaptive means to achieve castration resistance [106]. Thus, NEPC can be seen as a “next-generation” clinical problem in 98

the continuum of treatment resistance observed in late-stage disease [106]. In Chapter 2, we demonstrated that CBX2 expression was elevated in tumors lacking ligand-dependent AR activity so in Chapter 3 we analyzed CBX2 in the context of AR-negative NEPC. By profiling a number of potentially druggable EpRs in patient tumor tissues, we demonstrate that CBX2 is the most highly up-regulated EpR in NEPC compared to prostate adenocarcinoma. Interestingly, CBX2 was closely followed by EZH2, as well as other PcG proteins and epigenetic repressors, supporting a role for altered transcriptional regulation by PcG complexes. In line with this idea, we also found that a set of genes down-regulated in different in vivo NEPC models (NEARS) were enriched in PcG targets and that silencing of NEARS was able to predict poor prognosis in primary PCa patients. Overall, these results indicate that CBX2 is likely implicated in NEPC pathogenesis and complement the data from Chapter 2, thereby supporting the important nature of CBX2 in advanced prostate malignancies.

Given that the role of CBX2 has never been addressed in the cancer literature, we were interested to know whether CBX2 was also relevant to other types of human cancers. Thus, we conducted genomic and transcriptomic analyses of CBX2 in a multitude of tumor types using previously published clinical data. Strikingly, these studies demonstrated that the CBX2 locus is rarely inactivated and that CBX2 mRNA down-regulation is extremely rare. In contrast, CBX2 is frequently amplified and over-expressed in many solid tumors independently of CDKN2A and CDKN2B silencing, both of which are classical PcG targets [312]. Moreover, elevated CBX2 levels were associated with clinical features indicative of poor prognosis, in line with the results presented in Chapters 2 and 3. In summary, the elevated expression of CBX2 coupled to its lack of inactivating mutations suggests a critical role for CBX2 in human tumors and is consistent with the molecular profile of an oncogene.

5.2. Main conclusions The goal of this research project was to identify novel drivers of advanced PCa, a disease for which there is presently no cure. Based on the critical role played by CBX2 in the male urogenital system development, we initially hypothesized that CBX2 may become differentially expressed during PCa progression. Furthermore, we also postulated that elevated CBX2 activity may result in aberrant transcriptional regulation of key genes, which could be reversed through 99

CBX2 inhibition. Addressing the first hypothesis, we demonstrate that CBX2 is not only up- regulated in metastatic CRPC and NEPC, but that elevated CBX2 levels are also observed across a number of tumor types and almost exclusively correlates with poor patient prognosis. Secondly, we demonstrate that, at least in advanced PCa cell lines, CBX2 depletion induces striking cell death accompanied by differential expression of important CRGs preferentially involved in cell proliferation and differentiation. These results are consistent with the phenotypes observed in CBX2-deficient animal models and indicate that CBX2 may represent a critical driver of cancer progression.

Within the context of PCa, we observed highest CBX2 expression in the most metastatic and aggressive tumors. In addition, since elevated CBX2 was observed within metastatic but not in non-metastatic foci of the same patient tumor, it suggests that in primary PCa the up-regulation of CBX2 in a subpopulation of cells may predispose the tumor to follow an aggressive clinical course. Since CBX2 depletion induced cell death in the metastatic cell lines, these data cohesively argue for a critical CBX2 function in driving metastatic CRPC. Mechanistically, microarray analysis demonstrated that CBX2 regulates genes in critical pathways known to contribute to CRPC progression such as the PI3K/AKT pathway [131], which may explain the phenotypes observed upon CBX2 silencing. Therefore, we conclude that CBX2 promotes CRPC survival by reversibly controlling molecular pathways defining cellular proliferation and differentiation.

Since characterization of our PDX model of neuroendocrine transdifferentiation revealed few genetic changes between PCa and NEPC [139], we posited that NEPC was likely driven by epigenetic alterations. By exploring the epigenetic landscape of NEPC, we identified CBX2 and EZH2 as the two most highly up-regulated chromatin regulators, which we also found to be over- expressed in SCLC, another neuroendocrine malignancy [320]. Substantiating a role for altered PcG-mediated silencing, down-regulation of a NEPC-derived repression signature associated with metastasis and poor clinical outcome in primary PCa. Thus, CBX2 and EZH2 represent likely mediators of NEPC aggressiveness and therefore carry some potential as therapeutic targets for NEPC, a malignancy that is typically fatal within a year of diagnosis [111].

100

While our initial focus was on prostate malignancies, our results combined with the role of CBX2 in embryonic development led to the hypothesis that CBX2 might also be involved in the pathogenesis of other tumor types. Based on our genotranscriptomic profiling, there appears to be a selective pressure to maintain functional CBX2 as we found very few tumors displaying inactivating mutations or down-regulation of CBX2. These findings are analogous to what was observed in advanced PCa, where there was a propensity for CBX2 over-expression. As observed in PCa, higher CBX2 levels significantly associated with aggressive behavior and poor clinical outcome in many tumor types. Taken together, these findings support the conclusion that CBX2 represents a critical regulator of human tumorigenesis and metastatic spreading.

5.3. Strengths and limitations Overall, this work represents the first analysis functionally linking CBX2 and human cancers, notably PCa. A critical factor in the discovery of CBX2 as a potential therapeutic target was the clinical relevance of the in vivo and the in vitro models that were employed. As previously mentioned, PDX models of PCa are highly reflective of clinical disease, accurately recapitulating the tumor microenvironment and architecture [132]. Moreover, the LTL313B/LTL313H model is a one-of-a-kind xenograft pair originating from the same patient tumor, thereby reproducing the intratumoral heterogeneity observed in patients [139]. This allowed us to identify CBX2 up- regulation in a subpopulation of tumor cells with high metastatic potential within the primary tumor, at a stage where biopsies are typically conducted. Likewise, LTL331R/LTL331 and LTL313B/313BR represent unique PDX models for advanced CRPC, which are particularly useful since tumor tissue originating from late stage PCa is rare given the lack of biopsy in advanced disease [139]. Additionally, the LNCaP and C4-2 cell lines employed in this study are validated models of PCa progression that we used to complement our in vivo results [136]. Finally, experimental findings have also been confirmed in numerous independent gene expression datasets derived from patient tumors, supporting the clinical significance of our study.

Another important aspect of the study is that we used an epigenetic approach to identify novel therapeutic targets for advanced PCa. Epigenetic alterations are reversible and therefore can be reversed pharmacologically [321]. A number of epigenetic therapies have been approved for clinical use, with many more currently under investigation in clinical trials [302]. In parallel, 101

there is mounting evidence that the PcG complexes induce abnormal chromatin regulation that drives PCa progression [211]. The recent development of antagonists of epigenetic readers, notably targeting the CBX7 chromodomain, makes CBX2 a particularly attractive drug target in the evolving landscape of epigenetic therapies [224]. Given the potential for CBX2 targeting, the translational application of this study therefore represents a considerable advantage. Consequently, the finding that CBX2 is involved in a number of tumor types also potentiates the clinical and pharmaceutical value of CBX2 as a drug target. In summary, we have used the optimal models to address an important clinical need and identified a key epigenetic target with the potential to improve the quality of care for many patients.

Despite these interesting findings, our work also carries some limitations that warrant further clarification. In particular, we did not confirm whether the CRGs identified in chapter 2 were directly bound by CBX2. Confirming the genomic localization of CBX2 could determine which of these direct CBX2 targets are most likely to indirectly drive the expression of other CRGs. Since CBX2 is a transcriptional repressor and a number of CRGs were down-regulated upon CBX2 silencing, this implies that these CRGs are indirectly regulated by CBX2. In addition, it is possible that CBX2 controls specific enhancers that in turn define the transcriptional status of CRGs, which could not be detected by our microarray analysis [322]. Despite these limitations, analysis of CRGs demonstrated a profound link with cellular proliferation and this feature was consistent across all 3 chapters as well as the existing literature [228], suggesting that CRGs were functionally linked to CBX2 activity.

While silencing of CBX2 strikingly blocks the growth of metastatic PCa cell lines in vitro, our analysis would benefit from assessing the in vivo effect of CBX2 inhibition. In this study, we used siRNA to silence CBX2 expression, which does not provide sufficient knockdown in vivo. However, in vivo experiments are conditional to the development of CBX2 inhibitors able to permeate the cell membrane, which are not currently available. Additionally, this work would have benefited from analyzing the phenotypes induced by siRNA-mediated CBX2 silencing in NEPC cell lines. However, there is only a single cell line derived from NEPC cells (H660) and it does not represent a good model of NEPC either molecularly or phenotypically [323]. In contrast to the high proliferation rate observed in clinical NEPC, H660 cells have a doubling time of 102

more than two weeks in culture, making it impossible to conduct siRNA-mediated knockdown [324].

5.4. Overall significance and impact This thesis provides the first evidence that CBX2 represents a novel player in human cancers, particularly PCa. In men afflicted with PCa, there currently exist two main problems. The first major issue is the lack of available biomarkers to accurately distinguish between low-risk and high risk disease in the setting of localized PCa [11]. This creates a major dilemma for clinicians, who have to decide whether to give aggressive treatment that may unnecessarily harm the patient or whether to go forward with less radical options, at the risk of letting the cancer progress faster than if intense therapy was offered [11].

In this project, we have demonstrated that elevated expression of CBX2 was preferentially observed in tumors that were highly aggressive and more likely to disseminate, suggesting it may represent a novel biomarker for aggressive PCa. Using PDX models such as LTL313B/LTL313H which recapitulate the heterogeneity observed in patients, we were able to extract key information about PCa metastasis [139]. Importantly, we have shown that CBX2 is up-regulated in specific pre-metastatic foci within the primary tumor. In line with this idea, we also demonstrated that CBX2 mRNA and protein levels are significantly higher in metastatic compared to primary PCa tumor tissue, which argues that a subpopulation of CBX2-expressing cells may predispose the disease to take an aggressive clinical course. Furthermore, elevated CBX2 expression in primary PCa patients significantly correlated with poor prognostic factors and lower overall survival. Taken together, these results demonstrate that CBX2 over-expression may be a clinically-relevant biomarker to distinguish between low-risk and high-risk disease, and that further efforts should be made to investigate its potential.

The second major problem, and perhaps the most important one, is that the therapies available for patients with CRPC remain palliative in nature and thus there is a profound need for novel treatments against advanced disease [14]. As mentioned in the previous paragraphs elevated CBX2 expression predicts poor clinical outcome, thus it is likely that CBX2 plays a functional role. This idea is substantiated by the fact that, for the first time, we showed that CBX2 depletion

103

induced proliferation arrest and apoptosis in cell line models of advanced PCa. These results position CBX2 as a drug target in CRPC, and provide a novel strategy to therapeutically intervene through epigenetic modulation. In addition, we have identified CBX2 as a potential driver of NEPC, which means that a CBX2 targeting agent would likely reduce NEPC incidence and progression in addition to its effect on CRPC cells. Taken together, this knowledge will provide the foundation for novel therapeutic strategies designed to inhibit CBX2 function. Given the widespread implications of CBX2 in numerous human neoplasms, CBX2-targeting therapies may be used to treat a number of advanced solid tumors, in particular for the treatment of advanced PCa and lung cancer. Finally, given the recent interest in targeting chromatin modulators, the proposed work may provide a paradigm for future efforts aimed at targeting the cancer epigenome.

In addition to demonstrating the potential of CBX2 as a biomarker and therapeutic target, we have also analyzed the molecular mechanisms through which CBX2’s tumor-promoting activities were propagated. As would be expected from an oncogene, microarray analysis revealed that CBX2 was silencing key tumor suppressor genes in PCa cell lines, which lead to the up-regulation of important genes promoting PCa cell survival and proliferation. Notably, CBX2-regulated genes were intricately linked to the PI3K/AKT pathway, which is overactive in more than 80% of CRPC patients [84]. These findings are the first to shed light on the genes and pathways regulated by CBX2 in any human cancer, and substantiate the functional assays demonstrating that CBX2 is a critical component of tumor biology that may emerge as a widespread drug target.

5.5. Future research directions While we have identified a group of CRGs whose expression is modulated upon CBX2 silencing, we have not yet determined whether CBX2 directly binds and represses the CRG loci. Furthermore, it is still unclear whether CBX2 is localized at sites of PRC2 or PRC1 activity. To address these questions, the optimal method of investigation would be a ChIP experiment using antibodies specific to CBX2, H3K27me3 (PRC2), and Ring1B (PRC1). This would provide a map of CBX2 binding in PCa cells that could be analyzed in the context of PRC1 and PRC2 localization. Additionally, the presence of an AT-hook domain right next to the chomodomain 104

suggests that these two domains may work in concert to repress a specific subset of genes [235]. Another mechanistic aspect to explore is the influence of lncRNAs on CBX2 activity. Since lncRNAs are important regulators of PRC1 and PRC2 recruitment to chromatin, understanding the contribution of lncRNAs to aberrant CBX2-mediated transcription should become an active area of research [176, 206]. However, given that CBX2 does not have the ability to bind RNA [215], the effects of RNA on CBX2 activity are likely to be indirect. Taken together, these studies would complement our investigation of the molecular mechanisms through which CBX2 regulates PCa and would allow further insights into mechanistic aspects of PcG-mediated silencing.

Our analysis of CRGs has clearly demonstrated that CBX2 regulates genes involved in cellular proliferation, particularly in the G2/M phase. This is in accordance with our in vitro results showing strong proliferation arrest, as well as with previous studies demonstrating that germline CBX2 deletion induces severe proliferative defects in animal models [228]. While there is evidence that CBX2 nuclear translocation and activity are regulated by phosphorylation [238, 239], the kinases which phosphorylate CBX2 remain unknown. In the future, identifying these CBX2-regulating kinases harbors significant clinical relevance since their pharmacologic inhibition could block CBX2 phosphorylation and therefore suppress its tumor-promoting activity. Since the AKT pathway strongly involves phosphorylation and is activated in a majority of CRPC and NEPC patients, it would be interesting to investigate the link between CBX2 and the PI3K/AKT pathway [325]. We have observed transcriptional changes thought to activate the PI3K/AKT pathway in our microarray analysis, thus it is conceivable that CBX2 and AKT are part of an auto-stimulatory loop that accounts for the aggressive properties of tumor cells observed in late stage disease. To investigate this hypothesis, western blots looking at the phosphorylation state of AKT should be conducted in the presence and absence of CBX2. Conversely, CBX2 activity should also be assessed in the presence or absence of AKT antagonism.

In addition, CBX2 also distinguishes itself from other CBX proteins by containing a positively- charged domain necessary for PRC1-dependent chromatin compaction [183]. The profound phenotypic changes induced by CBX2 inhibition suggest the presence of widespread 105

transcriptomic alterations, which could be explained by large-scale chromatin decompaction at PcG target sites. This model is also consistent with the widely accepted notion that PcG complexes act not to initiate but to maintain mitotically-heritable chromatin states [326]. The strong positive charge found within this domain provides the electrochemical basis for interaction with DNA, thereby facilitating heterochromatin formation [183]. Altering the function of this domain could therefore inhibit these crucial interactions and suppress the tumor- promoting properties of CBX2. However, only structures for the CBX2 chromodomain are currently available [173], such that developing modulators of the positively-charged CBX2 region would be challenging.

In chapter 4, we presented the first evidence that CBX2 is up-regulated and that its expression correlates with poor clinical outcome in a number of tumor types [327]. Notably, CBX2 was over-expressed in tumors of the breast, colon, lung, and prostate, which represent the four deadliest cancers in the western world. This result suggests that CBX2 may be a potential drug target for the treatment of numerous cancer types, which confers CBX2 great clinical and pharmaceutical value. However, further investigations should address whether CBX2 is functionally required in these cancers. Additionally, it is also possible that CBX2 is involved at different stages of tumor development in different tumor types and subtypes, which also requires further analysis. Moreover, it would be interesting to determine whether CBX2 promotes cancer progression through different molecular mechanisms in different tumor types. This could generate important knowledge that can be applied in identifying compensating pathways which can be targeted to prevent resistance to CBX2-targeted therapies.

Finally, we believe that the reported aberrations in CBX2 and PcG-mediated silencing have clear therapeutic implications, particularly given the emerging improvements in targeting the cancer epigenome [251, 321]. In particular, we believe the interaction between CBX2 and H3K27me3 bridges the function of PRC2 and PRC1, thus representing a critical junction in this altered epigenetic pathway [290]. A few strategies can be put forward to therapeutically target this axis in the context of NEPC. First, small molecule inhibitors interfering with the methyltransferase activity of EZH2 have already been developed and warrant further investigation in NEPC [164]. Second, antagonists of the CBX2 chromodomain represent another promising path, as they 106

would disrupt the binding between CBX2 and H3K27me3. At present, there are no small molecules directly targeting CBX2, although the development of CBX7 antagonists hints that a similar strategy could also be employed for CBX2 [223, 224]. Third, antisense oligonucleotides (ASOs) or encapsulated siRNAs may be used to reduce the expression of key PcG genes such as CBX2 and EZH2. An exponentially increasing number of ASOs have entered clinical testing, highlighting the potential of ASOs as therapeutic agents [328]. Taken together, our results highlight clinically-relevant alterations in Polycomb-mediated silencing that may be targetable in a number of tumor types, thereby adding to the growing landscape of cancer epigenetics.

107

Bibliography [1] Ferlay J, Soerjomataram I, Dikshit R, Eser S, Mathers C, Rebelo M, Parkin DM, Forman D, Bray F. Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer, 2015; 136: E359-86. [2] Hanahan D, Weinberg RA. The hallmarks of cancer. Cell, 2000; 100: 57-70. [3] Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell, 2011; 144: 646-74. [4] Siegel RL, Miller KD, Jemal A. Cancer statistics, 2015. CA Cancer J Clin, 2015; 65: 5- 29. [5] Siegel R, Ma J, Zou Z, Jemal A. Cancer statistics, 2014. CA Cancer J Clin, 2014; 64: 9- 29. [6] Eggener S. Commentary on: "Long-term functional outcomes after treatment for localized prostate cancer." Resnick MJ, Koyama T, Fan KH, Albertsen PC, Goodman M, Hamilton AS, Hoffman RM, Potosky AL, Stanford JL, Stroup AM, Van Horn RL, Penson DF. Department of Urologic Surgery and the Center for Surgical Quality and Outcomes Research, Vanderbilt University, Nashville, TN.: N Engl J Med 2013;368(5):436-45. doi: 10.1056/NEJMoa1209978. Urol Oncol, 2014; 32: 513-4. [7] Bubendorf L, Schöpfer A, Wagner U, Sauter G, Moch H, Willi N, Gasser TC, Mihatsch MJ. Metastatic patterns of prostate cancer: an autopsy study of 1,589 patients. Hum Pathol, 2000; 31: 578-83. [8] Schröder F, Crawford ED, Axcrona K, Payne H, Keane TE. Androgen deprivation therapy: past, present and future. BJU Int, 2012; 109 Suppl 6: 1-12. [9] Feldman BJ, Feldman D. The development of androgen-independent prostate cancer. Nat Rev Cancer, 2001; 1: 34-45. [10] Terry S, Beltran H. The many faces of neuroendocrine differentiation in prostate cancer progression. Front Oncol, 2014; 4: 60. [11] Scher HI, Morris MJ, Larson S, Heller G. Validation and clinical utility of prostate cancer biomarkers. Nat Rev Clin Oncol, 2013; 10: 225-34. [12] Loeb S, Bjurlin MA, Nicholson J, Tammela TL, Penson DF, Carter HB, Carroll P, Etzioni R. Overdiagnosis and overtreatment of prostate cancer. Eur Urol, 2014; 65: 1046- 55. [13] Pelzer AE, Bektic J, Akkad T, Ongarello S, Schaefer G, Schwentner C, Frauscher F, Bartsch G, Horninger W. Under diagnosis and over diagnosis of prostate cancer in a screening population with serum PSA 2 to 10 ng/ml. J Urol, 2007; 178: 93-7; discussion 97. [14] Saad F, Hotte S, Catton C, Drachenberg D, Finelli A, Fleshner N, Gleave M, Kapoor A, Kassouf W, Loblaw A, North S, Usmani N, Chi KN. CUA-CUOG guidelines for the management of castration-resistant prostate cancer (CRPC): 2013 update. Can Urol Assoc J, 2013; 7: 231-7. [15] Kirby M, Hirst C, Crawford ED. Characterising the castration-resistant prostate cancer population: a systematic review. Int J Clin Pract, 2011; 65: 1180-92. [16] Wang HT, Yao YH, Li BG, Tang Y, Chang JW, Zhang J. Neuroendocrine Prostate Cancer (NEPC) progressing from conventional prostatic adenocarcinoma: factors

108

associated with time to development of NEPC and survival from NEPC diagnosis-a systematic review and pooled analysis. J Clin Oncol, 2014; 32: 3383-90. [17] Beltran H, Tomlins S, Aparicio A, Arora V, Rickman D, Ayala G, Huang J, True L, Gleave ME, Soule H, Logothetis C, Rubin MA. Aggressive variants of castration- resistant prostate cancer. Clin Cancer Res, 2014; 20: 2846-50. [18] Marker PC, Donjacour AA, Dahiya R, Cunha GR. Hormonal, cellular, and molecular control of prostatic development. Dev Biol, 2003; 253: 165-74. [19] Hayward SW, Cunha GR. The prostate: development and physiology. Radiol Clin North Am, 2000; 38: 1-14. [20] Hayward SW, Cunha GR, Dahiya R. Normal development and carcinogenesis of the prostate. A unifying hypothesis. Ann N Y Acad Sci, 1996; 784: 50-62. [21] Hayward SW, Baskin LS, Haughney PC, Cunha AR, Foster BA, Dahiya R, Prins GS, Cunha GR. Epithelial development in the rat ventral prostate, anterior prostate and seminal vesicle. Acta Anat (Basel), 1996; 155: 81-93. [22] Cunha GR, Chung LW, Shannon JM, Reese BA. Stromal-epithelial interactions in sex differentiation. Biol Reprod, 1980; 22: 19-42. [23] Wang Y, Hayward S, Cao M, Thayer K, Cunha G. Cell differentiation lineage in the prostate. Differentiation, 2001; 68: 270-9. [24] Hayward SW, Baskin LS, Haughney PC, Foster BA, Cunha AR, Dahiya R, Prins GS, Cunha GR. Stromal development in the ventral prostate, anterior prostate and seminal vesicle of the rat. Acta Anat (Basel), 1996; 155: 94-103. [25] Cunha GR, Battle E, Young P, Brody J, Donjacour A, Hayashi N, Kinbara H. Role of epithelial-mesenchymal interactions in the differentiation and spatial organization of visceral smooth muscle. Epithelial Cell Biol, 1992; 1: 76-83. [26] Noordzij MA, van Steenbrugge GJ, van der Kwast TH, Schröder FH. Neuroendocrine cells in the normal, hyperplastic and neoplastic prostate. Urol Res, 1995; 22: 333-41. [27] Lee CH, Akin-Olugbade O, Kirschenbaum A. Overview of prostate anatomy, histology, and pathology. Endocrinol Metab Clin North Am, 2011; 40: 565-75, viii-ix. [28] McNeal JE. The zonal anatomy of the prostate. Prostate, 1981; 2: 35-49. [29] McNeal JE. Normal histology of the prostate. Am J Surg Pathol, 1988; 12: 619-33. [30] Aumüller G, Seitz J. Protein secretion and secretory processes in male accessory sex glands. Int Rev Cytol, 1990; 121: 127-231. [31] Hayward SW, Del Buono R, Deshpande N, Hall PA. A functional model of adult human prostate epithelium. The role of androgens and stroma in architectural organisation and the maintenance of differentiated secretory function. J Cell Sci, 1992; 102 ( Pt 2): 361-72. [32] Sugimura Y, Cunha GR, Donjacour AA. Morphological and histological study of castration-induced degeneration and androgen-induced regeneration in the mouse prostate. Biol Reprod, 1986; 34: 973-83. [33] Kumar B, Koul S, Khandrika L, Meacham RB, Koul HK. Oxidative stress is inherent in prostate cancer cells and is required for aggressive phenotype. Cancer Res, 2008; 68: 1777-85. [34] Elkahwaji JE, Hauke RJ, Brawner CM. Chronic bacterial inflammation induces prostatic intraepithelial neoplasia in mouse prostate. Br J Cancer, 2009; 101: 1740-8.

109

[35] Cano P, Godoy A, Escamilla R, Dhir R, Onate SA. Stromal-epithelial cell interactions and androgen receptor-coregulator recruitment is altered in the tissue microenvironment of prostate cancer. Cancer Res, 2007; 67: 511-9. [36] Brawer MK. Prostatic intraepithelial neoplasia: an overview. Rev Urol, 2005; 7 Suppl 3: S11-8. [37] Bostwick DG, Liu L, Brawer MK, Qian J. High-grade prostatic intraepithelial neoplasia. Rev Urol, 2004; 6: 171-9. [38] De Marzo AM, Platz EA, Sutcliffe S, Xu J, Grönberg H, Drake CG, Nakai Y, Isaacs WB, Nelson WG. Inflammation in prostate carcinogenesis. Nat Rev Cancer, 2007; 7: 256-69. [39] Gann PH. Risk factors for prostate cancer. Rev Urol, 2002; 4 Suppl 5: S3-S10. [40] Patel AR, Klein EA. Risk factors for prostate cancer. Nat Clin Pract Urol, 2009; 6: 87-95. [41] Fournier G, Valeri A, Mangin P, Cussenot O. [Prostate cancer. Epidemiology. Risk factors. Pathology]. Ann Urol (Paris), 2004; 38: 187-206. [42] Steinberg GD, Carter BS, Beaty TH, Childs B, Walsh PC. Family history and the risk of prostate cancer. Prostate, 1990; 17: 337-47. [43] Salem S, Salahi M, Mohseni M, Ahmadi H, Mehrsai A, Jahani Y, Pourmand G. Major dietary factors and prostate cancer risk: a prospective multicenter case-control study. Nutr Cancer, 2011; 63: 21-7. [44] Prins GS. Endocrine disruptors and prostate cancer risk. Endocr Relat Cancer, 2008; 15: 649-56. [45] Julin B, Wolk A, Johansson JE, Andersson SO, Andrén O, Akesson A. Dietary cadmium exposure and prostate cancer incidence: a population-based prospective cohort study. Br J Cancer, 2012; 107: 895-900. [46] [47] Wang Q, Li W, Zhang Y, Yuan X, Xu K, Yu J, Chen Z, Beroukhim R, Wang H, Lupien M, Wu T, Regan MM, Meyer CA, Carroll JS, Manrai AK, Jänne OA, Balk SP, Mehra R, Han B, Chinnaiyan AM, Rubin MA, True L, Fiorentino M, Fiore C, Loda M, Kantoff PW, Liu XS, Brown M. Androgen receptor regulates a distinct transcription program in androgen-independent prostate cancer. Cell, 2009; 138: 245-56. [48] Wyatt AW, Mo F, Wang Y, Collins CC. The diverse heterogeneity of molecular alterations in prostate cancer identified through next-generation sequencing. Asian J Androl, 2013; 15: 301-8. [49] Tomlins SA, Rhodes DR, Perner S, Dhanasekaran SM, Mehra R, Sun XW, Varambally S, Cao X, Tchinda J, Kuefer R, Lee C, Montie JE, Shah RB, Pienta KJ, Rubin MA, Chinnaiyan AM. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science, 2005; 310: 644-8. [50] Gudmundsson J, Sulem P, Manolescu A, Amundadottir LT, Gudbjartsson D, Helgason A, Rafnar T, Bergthorsson JT, Agnarsson BA, Baker A, Sigurdsson A, Benediktsdottir KR, Jakobsdottir M, Xu J, Blondal T, Kostic J, Sun J, Ghosh S, Stacey SN, Mouy M, Saemundsdottir J, Backman VM, Kristjansson K, Tres A, Partin AW, Albers-Akkers MT, Godino-Ivan Marcos J, Walsh PC, Swinkels DW, Navarrete S, Isaacs SD, Aben KK, Graif T, Cashy J, Ruiz-Echarri M, Wiley KE, Suarez BK, Witjes JA, Frigge M, Ober C, Jonsson E, Einarsson GV, Mayordomo JI, Kiemeney LA, Isaacs WB, Catalona WJ, Barkardottir RB, Gulcher JR, Thorsteinsdottir U, Kong A, Stefansson K. Genome-wide

110

association study identifies a second prostate cancer susceptibility variant at 8q24. Nat Genet, 2007; 39: 631-7. [51] McMenamin ME, Soung P, Perera S, Kaplan I, Loda M, Sellers WR. Loss of PTEN expression in paraffin-embedded primary prostate cancer correlates with high Gleason score and advanced stage. Cancer Res, 1999; 59: 4291-6. [52] Bowen C, Bubendorf L, Voeller HJ, Slack R, Willi N, Sauter G, Gasser TC, Koivisto P, Lack EE, Kononen J, Kallioniemi OP, Gelmann EP. Loss of NKX3.1 expression in human prostate cancers correlates with tumor progression. Cancer Res, 2000; 60: 6111-5. [53] Millar DS, Ow KK, Paul CL, Russell PJ, Molloy PL, Clark SJ. Detailed methylation analysis of the glutathione S-transferase pi (GSTP1) gene in prostate cancer. Oncogene, 1999; 18: 1313-24. [54] Ishikawa T, Ali-Osman F. Glutathione-associated cis-diamminedichloroplatinum(II) metabolism and ATP-dependent efflux from leukemia cells. Molecular characterization of glutathione-platinum complex and its biological significance. J Biol Chem, 1993; 268: 20116-25. [55] Schnekenburger M, Karius T, Diederich M. Regulation of epigenetic traits of the glutathione S-transferase P1 gene: from detoxification toward cancer prevention and diagnosis. Front Pharmacol, 2014; 5: 170. [56] Alva A, Hussain M. The changing natural history of metastatic prostate cancer. Cancer J, 2013; 19: 19-24. [57] Banyard J, Chung I, Wilson AM, Vetter G, Le Béchec A, Bielenberg DR, Zetter BR. Regulation of epithelial plasticity by miR-424 and miR-200 in a new prostate cancer metastasis model. Sci Rep, 2013; 3: 3151. [58] Bitting RL, Schaeffer D, Somarelli JA, Garcia-Blanco MA, Armstrong AJ. The role of epithelial plasticity in prostate cancer dissemination and treatment resistance. Cancer Metastasis Rev, 2014; 33: 441-68. [59] Maccauro G, Spinelli MS, Mauro S, Perisano C, Graci C, Rosa MA. Physiopathology of spine metastasis. Int J Surg Oncol, 2011; 2011: 107969. [60] Gartrell BA, Saad F. Managing bone metastases and reducing skeletal related events in prostate cancer. Nat Rev Clin Oncol, 2014; 11: 335-45. [61] Nguyen DX, Bos PD, Massagué J. Metastasis: from dissemination to organ-specific colonization. Nat Rev Cancer, 2009; 9: 274-84. [62] Lilja H, Ulmert D, Vickers AJ. Prostate-specific antigen and prostate cancer: prediction, detection and monitoring. Nat Rev Cancer, 2008; 8: 268-78. [63] Balk SP, Ko YJ, Bubley GJ. Biology of prostate-specific antigen. J Clin Oncol, 2003; 21: 383-91. [64] Wang TJ, Rittenhouse HG, Wolfert RL, Lynne CM, Brackett NL. PSA concentrations in seminal plasma. Clin Chem, 1998; 44: 895-6. [65] Bell N, Connor Gorber S, Shane A, Joffres M, Singh H, Dickinson J, Shaw E, Dunfield L, Tonelli M, Care CTFoPH. Recommendations on screening for prostate cancer with the prostate-specific antigen test. CMAJ, 2014; 186: 1225-34. [66] Birtle AJ, Freeman A, Masters JR, Payne HA, Harland SJ, Registry BSoOC. Clinical features of patients who present with metastatic prostate carcinoma and serum prostate- specific antigen (PSA) levels < 10 ng/mL: the "PSA negative" patients. Cancer, 2003; 98: 2362-7. 111

[67] Carter HB, Albertsen PC, Barry MJ, Etzioni R, Freedland SJ, Greene KL, Holmberg L, Kantoff P, Konety BR, Murad MH, Penson DF, Zietman AL. Early detection of prostate cancer: AUA Guideline. J Urol, 2013; 190: 419-26. [68] Humphrey PA. Diagnosis of adenocarcinoma in prostate needle biopsy tissue. J Clin Pathol, 2007; 60: 35-42. [69] Deliveliotis C, John V, Louras G, Andreas S, Alargof E, Sofras F, Goulandris N. Multiple transrectal ultrasound guided prostatic biopsies: morbidity and tolerance. Int Urol Nephrol, 1999; 31: 681-6. [70] Dickinson L, Ahmed HU, Allen C, Barentsz JO, Carey B, Futterer JJ, Heijmink SW, Hoskin PJ, Kirkham A, Padhani AR, Persad R, Puech P, Punwani S, Sohaib AS, Tombal B, Villers A, van der Meulen J, Emberton M. Magnetic resonance imaging for the detection, localisation, and characterisation of prostate cancer: recommendations from a European consensus meeting. Eur Urol, 2011; 59: 477-94. [71] van der Kwast TH, Lopes C, Santonja C, Pihl CG, Neetens I, Martikainen P, Di Lollo S, Bubendorf L, Hoedemaeker RF, Cancer MotpcotERSoSfP. Guidelines for processing and reporting of prostatic needle biopsies. J Clin Pathol, 2003; 56: 336-40. [72] Humphrey PA. Gleason grading and prognostic factors in carcinoma of the prostate. Mod Pathol, 2004; 17: 292-306. [73] Gleason DF. Classification of prostatic carcinomas. Cancer Chemother Rep, 1966; 50: 125-8. [74] Schröder FH, Hermanek P, Denis L, Fair WR, Gospodarowicz MK, Pavone-Macaluso M. The TNM classification of prostate cancer. Prostate Suppl, 1992; 4: 129-38. [75] Grossfeld GD, Chang JJ, Broering JM, Miller DP, Yu J, Flanders SC, Henning JM, Stier DM, Carroll PR. Impact of positive surgical margins on prostate cancer recurrence and the use of secondary cancer treatment: data from the CaPSURE database. J Urol, 2000; 163: 1171-7; quiz 1295. [76] Cooper EH. Prostate specific antigen in diagnosis, staging, and follow-up of prostate cancer. Prostate Suppl, 1992; 4: 125-8. [77] Troyer DA, Mubiru J, Leach RJ, Naylor SL. Promise and challenge: Markers of prostate cancer detection, diagnosis and prognosis. Dis Markers, 2004; 20: 117-28. [78] Akduman B, Crawford ED. Treatment of localized prostate cancer. Rev Urol, 2006; 8 Suppl 2: S15-21. [79] Wilt TJ, Brawer MK, Jones KM, Barry MJ, Aronson WJ, Fox S, Gingrich JR, Wei JT, Gilhooly P, Grob BM, Nsouli I, Iyer P, Cartagena R, Snider G, Roehrborn C, Sharifi R, Blank W, Pandya P, Andriole GL, Culkin D, Wheeler T, Group PCIvOTPS. Radical prostatectomy versus observation for localized prostate cancer. N Engl J Med, 2012; 367: 203-13. [80] Gomella LG. Effective testosterone suppression for prostate cancer: is there a best castration therapy? Rev Urol, 2009; 11: 52-60. [81] Colombel MC, Buttyan R. Hormonal control of apoptosis: the rat prostate gland as a model system. Methods Cell Biol, 1995; 46: 369-85. [82] Attar RM, Takimoto CH, Gottardis MM. Castration-resistant prostate cancer: locking up the molecular escape routes. Clin Cancer Res, 2009; 15: 3251-5.

112

[83] Harris WP, Mostaghel EA, Nelson PS, Montgomery B. Androgen deprivation therapy: progress in understanding mechanisms of resistance and optimizing androgen depletion. Nat Clin Pract Urol, 2009; 6: 76-85. [84] Karantanos T, Corn PG, Thompson TC. Prostate cancer progression after androgen deprivation therapy: mechanisms of castrate resistance and novel therapeutic approaches. Oncogene, 2013; 32: 5501-11. [85] Ryan CJ, Smith MR, de Bono JS, Molina A, Logothetis CJ, de Souza P, Fizazi K, Mainwaring P, Piulats JM, Ng S, Carles J, Mulders PF, Basch E, Small EJ, Saad F, Schrijvers D, Van Poppel H, Mukherjee SD, Suttmann H, Gerritsen WR, Flaig TW, George DJ, Yu EY, Efstathiou E, Pantuck A, Winquist E, Higano CS, Taplin ME, Park Y, Kheoh T, Griffin T, Scher HI, Rathkopf DE, Investigators C-A-. Abiraterone in metastatic prostate cancer without previous chemotherapy. N Engl J Med, 2013; 368: 138-48. [86] Tannock IF, de Wit R, Berry WR, Horti J, Pluzanska A, Chi KN, Oudard S, Théodore C, James ND, Turesson I, Rosenthal MA, Eisenberger MA, Investigators T. Docetaxel plus prednisone or mitoxantrone plus prednisone for advanced prostate cancer. N Engl J Med, 2004; 351: 1502-12. [87] de Bono JS, Logothetis CJ, Molina A, Fizazi K, North S, Chu L, Chi KN, Jones RJ, Goodman OB, Saad F, Staffurth JN, Mainwaring P, Harland S, Flaig TW, Hutson TE, Cheng T, Patterson H, Hainsworth JD, Ryan CJ, Sternberg CN, Ellard SL, Fléchon A, Saleh M, Scholz M, Efstathiou E, Zivi A, Bianchini D, Loriot Y, Chieffo N, Kheoh T, Haqq CM, Scher HI, Investigators C-A-. Abiraterone and increased survival in metastatic prostate cancer. N Engl J Med, 2011; 364: 1995-2005. [88] Bianchini D, Lorente D, Rodriguez-Vida A, Omlin A, Pezaro C, Ferraldeschi R, Zivi A, Attard G, Chowdhury S, de Bono JS. Antitumour activity of enzalutamide (MDV3100) in patients with metastatic castration-resistant prostate cancer (CRPC) pre-treated with docetaxel and abiraterone. Eur J Cancer, 2014; 50: 78-84. [89] de Bono JS, Oudard S, Ozguroglu M, Hansen S, Machiels JP, Kocak I, Gravis G, Bodrogi I, Mackenzie MJ, Shen L, Roessner M, Gupta S, Sartor AO, Investigators T. Prednisone plus cabazitaxel or mitoxantrone for metastatic castration-resistant prostate cancer progressing after docetaxel treatment: a randomised open-label trial. Lancet, 2010; 376: 1147-54. [90] Shahinian VB, Kuo YF, Freeman JL, Goodwin JS. Risk of fracture after androgen deprivation for prostate cancer. N Engl J Med, 2005; 352: 154-64. [91] Yuan X, Cai C, Chen S, Yu Z, Balk SP. Androgen receptor functions in castration- resistant prostate cancer and mechanisms of resistance to new agents targeting the androgen axis. Oncogene, 2014; 33: 2815-25. [92] Visakorpi T, Hyytinen E, Koivisto P, Tanner M, Keinänen R, Palmberg C, Palotie A, Tammela T, Isola J, Kallioniemi OP. In vivo amplification of the androgen receptor gene and progression of human prostate cancer. Nat Genet, 1995; 9: 401-6. [93] Gregory CW, He B, Johnson RT, Ford OH, Mohler JL, French FS, Wilson EM. A mechanism for androgen receptor-mediated prostate cancer recurrence after androgen deprivation therapy. Cancer Res, 2001; 61: 4315-9.

113

[94] Taplin ME, Bubley GJ, Ko YJ, Small EJ, Upton M, Rajeshkumar B, Balk SP. Selection for androgen receptor mutations in prostate cancers treated with androgen antagonist. Cancer Res, 1999; 59: 2511-5. [95] Sanford EJ, Paulson DF, Rohner TJ, Santen RJ, Bardin CW. The effects of castration on adrenal testosterone secretion in men with prostatic carcinoma. J Urol, 1977; 118: 1019- 21. [96] Locke JA, Guns ES, Lubik AA, Adomat HH, Hendy SC, Wood CA, Ettinger SL, Gleave ME, Nelson CC. Androgen levels increase by intratumoral de novo steroidogenesis during progression of castration-resistant prostate cancer. Cancer Res, 2008; 68: 6407-15. [97] Li Y, Chan SC, Brand LJ, Hwang TH, Silverstein KA, Dehm SM. Androgen receptor splice variants mediate enzalutamide resistance in castration-resistant prostate cancer cell lines. Cancer Res, 2013; 73: 483-9. [98] Dai B, Chen H, Guo S, Yang X, Linn DE, Sun F, Li W, Guo Z, Xu K, Kim O, Kong X, Melamed J, Qiu S, Qiu Y. Compensatory upregulation of tyrosine kinase Etk/BMX in response to androgen deprivation promotes castration-resistant growth of prostate cancer cells. Cancer Res, 2010; 70: 5587-96. [99] Guo Z, Dai B, Jiang T, Xu K, Xie Y, Kim O, Nesheiwat I, Kong X, Melamed J, Handratta VD, Njar VC, Brodie AM, Yu LR, Veenstra TD, Chen H, Qiu Y. Regulation of androgen receptor activity by tyrosine phosphorylation. Cancer Cell, 2006; 10: 309-19. [100] Köhler A, Demir U, Kickstein E, Krauss S, Aigner J, Aranda-Orgillés B, Karagiannidis AI, Achmüller C, Bu H, Wunderlich A, Schweiger MR, Schaefer G, Schweiger S, Klocker H, Schneider R. A hormone-dependent feedback-loop controls androgen receptor levels by limiting MID1, a novel translation enhancer and promoter of oncogenic signaling. Mol Cancer, 2014; 13: 146. [101] Augello MA, Burd CJ, Birbe R, McNair C, Ertel A, Magee MS, Frigo DE, Wilder- Romans K, Shilkrut M, Han S, Jernigan DL, Dean JL, Fatatis A, McDonnell DP, Visakorpi T, Feng FY, Knudsen KE. Convergence of oncogenic and hormone receptor pathways promotes metastatic phenotypes. J Clin Invest, 2013; 123: 493-508. [102] Colombel M, Symmans F, Gil S, O'Toole KM, Chopin D, Benson M, Olsson CA, Korsmeyer S, Buttyan R. Detection of the apoptosis-suppressing oncoprotein bc1-2 in hormone-refractory human prostate cancers. Am J Pathol, 1993; 143: 390-400. [103] Gleave M, Tolcher A, Miyake H, Nelson C, Brown B, Beraldi E, Goldie J. Progression to androgen independence is delayed by adjuvant treatment with antisense Bcl-2 oligodeoxynucleotides after castration in the LNCaP prostate tumor model. Clin Cancer Res, 1999; 5: 2891-8. [104] Thomas C, Lamoureux F, Crafter C, Davies BR, Beraldi E, Fazli L, Kim S, Thaper D, Gleave ME, Zoubeidi A. Synergistic targeting of PI3K/AKT pathway and androgen receptor axis significantly delays castration-resistant prostate cancer progression in vivo. Mol Cancer Ther, 2013; 12: 2342-55. [105] Wyatt AW, Gleave ME. Targeting the adaptive molecular landscape of castration- resistant prostate cancer. EMBO Mol Med, 2015; 7: 878-94. [106] Beltran H, Tagawa ST, Park K, MacDonald T, Milowsky MI, Mosquera JM, Rubin MA, Nanus DM. Challenges in recognizing treatment-related neuroendocrine prostate cancer. J Clin Oncol, 2012; 30: e386-9.

114

[107] Lotan TL, Gupta NS, Wang W, Toubaji A, Haffner MC, Chaux A, Hicks JL, Meeker AK, Bieberich CJ, De Marzo AM, Epstein JI, Netto GJ. ERG gene rearrangements are common in prostatic small cell carcinomas. Mod Pathol, 2011; 24: 820-8. [108] Shen R, Dorai T, Szaboles M, Katz AE, Olsson CA, Buttyan R. Transdifferentiation of cultured human prostate cancer cells to a neuroendocrine cell phenotype in a hormone- depleted medium. Urol Oncol, 1997; 3: 67-75. [109] Humphrey PA. Histological variants of prostatic carcinoma and their significance. Histopathology, 2012; 60: 59-74. [110] Papandreou CN, Daliani DD, Thall PF, Tu SM, Wang X, Reyes A, Troncoso P, Logothetis CJ. Results of a phase II study with doxorubicin, etoposide, and cisplatin in patients with fully characterized small-cell carcinoma of the prostate. J Clin Oncol, 2002; 20: 3072-80. [111] Sargos P, Ferretti L, Gross-Goupil M, Orre M, Cornelis F, Henriques de Figueiredo B, Houédé N, Merino C, Roubaud G, Dallaudiére B, Richaud P, Fléchon A. Characterization of prostate neuroendocrine cancers and therapeutic management: a literature review. Prostate Cancer Prostatic Dis, 2014; 17: 220-6. [112] Rubenstein JH, Katin MJ, Mangano MM, Dauphin J, Salenius SA, Dosoretz DE, Blitzer PH. Small cell anaplastic carcinoma of the prostate: seven new cases, review of the literature, and discussion of a therapeutic strategy. Am J Clin Oncol, 1997; 20: 376-80. [113] Leibovici D, Spiess PE, Agarwal PK, Tu SM, Pettaway CA, Hitzhusen K, Millikan RE, Pisters LL. Prostate cancer progression in the presence of undetectable or low serum prostate-specific antigen level. Cancer, 2007; 109: 198-204. [114] Furtado P, Lima MV, Nogueira C, Franco M, Tavora F. Review of small cell carcinomas of the prostate. Prostate Cancer, 2011; 2011: 543272. [115] Abrahamsson PA. Neuroendocrine differentiation in prostatic carcinoma. Prostate, 1999; 39: 135-48. [116] Wang W, Epstein JI. Small cell carcinoma of the prostate. A morphologic and immunohistochemical study of 95 cases. Am J Surg Pathol, 2008; 32: 65-71. [117] Ather MH, Abbas F, Faruqui N, Israr M, Pervez S. Correlation of three immunohistochemically detected markers of neuroendocrine differentiation with clinical predictors of disease progression in prostate cancer. BMC Urol, 2008; 8: 21. [118] Sagnak L, Topaloglu H, Ozok U, Ersoy H. Prognostic significance of neuroendocrine differentiation in prostate adenocarcinoma. Clin Genitourin Cancer, 2011; 9: 73-80. [119] Lapuk AV, Wu C, Wyatt AW, McPherson A, McConeghy BJ, Brahmbhatt S, Mo F, Zoubeidi A, Anderson S, Bell RH, Haegert A, Shukin R, Wang Y, Fazli L, Hurtado-Coll A, Jones EC, Hach F, Hormozdiari F, Hajirasouliha I, Boutros PC, Bristow RG, Zhao Y, Marra MA, Fanjul A, Maher CA, Chinnaiyan AM, Rubin MA, Beltran H, Sahinalp SC, Gleave ME, Volik SV, Collins CC. From sequence to molecular pathology, and a mechanism driving the neuroendocrine phenotype in prostate cancer. J Pathol, 2012; 227: 286-97. [120] Tan HL, Sood A, Rahimi HA, Wang W, Gupta N, Hicks J, Mosier S, Gocke CD, Epstein JI, Netto GJ, Liu W, Isaacs WB, De Marzo AM, Lotan TL. Rb loss is characteristic of prostatic small cell neuroendocrine carcinoma. Clin Cancer Res, 2014; 20: 890-903.

115

[121] Chen H, Sun Y, Wu C, Magyar CE, Li X, Cheng L, Yao JL, Shen S, Osunkoya AO, Liang C, Huang J. Pathogenesis of prostatic small cell carcinoma involves the inactivation of the P53 pathway. Endocr Relat Cancer, 2012; 19: 321-31. [122] Mosquera JM, Beltran H, Park K, MacDonald TY, Robinson BD, Tagawa ST, Perner S, Bismar TA, Erbersdobler A, Dhir R, Nelson JB, Nanus DM, Rubin MA. Concurrent AURKA and MYCN gene amplifications are harbingers of lethal treatment-related neuroendocrine prostate cancer. Neoplasia, 2013; 15: 1-10. [123] Beltran H, Rickman DS, Park K, Chae SS, Sboner A, MacDonald TY, Wang Y, Sheikh KL, Terry S, Tagawa ST, Dhir R, Nelson JB, de la Taille A, Allory Y, Gerstein MB, Perner S, Pienta KJ, Chinnaiyan AM, Collins CC, Gleave ME, Demichelis F, Nanus DM, Rubin MA. Molecular characterization of neuroendocrine prostate cancer and identification of new drug targets. Cancer Discov, 2011; 1: 487-95. [124] Lappalainen T, Sammeth M, Friedländer MR, 't Hoen PA, Monlong J, Rivas MA, Gonzàlez-Porta M, Kurbatova N, Griebel T, Ferreira PG, Barann M, Wieland T, Greger L, van Iterson M, Almlöf J, Ribeca P, Pulyakhina I, Esser D, Giger T, Tikhonov A, Sultan M, Bertier G, MacArthur DG, Lek M, Lizano E, Buermans HP, Padioleau I, Schwarzmayr T, Karlberg O, Ongen H, Kilpinen H, Beltran S, Gut M, Kahlem K, Amstislavskiy V, Stegle O, Pirinen M, Montgomery SB, Donnelly P, McCarthy MI, Flicek P, Strom TM, Lehrach H, Schreiber S, Sudbrak R, Carracedo A, Antonarakis SE, Häsler R, Syvänen AC, van Ommen GJ, Brazma A, Meitinger T, Rosenstiel P, Guigó R, Gut IG, Estivill X, Dermitzakis ET, Consortium G. Transcriptome and genome sequencing uncovers functional variation in humans. Nature, 2013; 501: 506-11. [125] Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, Jacobsen A, Byrne CJ, Heuer ML, Larsson E, Antipin Y, Reva B, Goldberg AP, Sander C, Schultz N. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov, 2012; 2: 401-4. [126] Rhodes DR, Kalyana-Sundaram S, Mahavisno V, Varambally R, Yu J, Briggs BB, Barrette TR, Anstet MJ, Kincead-Beal C, Kulkarni P, Varambally S, Ghosh D, Chinnaiyan AM. Oncomine 3.0: genes, pathways, and networks in a collection of 18,000 cancer gene expression profiles. Neoplasia, 2007; 9: 166-80. [127] Győrffy B, Surowiak P, Budczies J, Lánczky A. Online survival analysis software to assess the prognostic value of biomarkers using transcriptomic data in non-small-cell lung cancer. PLoS One, 2013; 8: e82241. [128] Rosol TJ, Tannehill-Gregg SH, Corn S, Schneider A, McCauley LK. Animal models of bone metastasis. Cancer Treat Res, 2004; 118: 47-81. [129] Gingrich JR, Barrios RJ, Morton RA, Boyce BF, DeMayo FJ, Finegold MJ, Angelopoulou R, Rosen JM, Greenberg NM. Metastatic prostate cancer in a transgenic mouse. Cancer Res, 1996; 56: 4096-102. [130] Hurwitz AA, Foster BA, Allison JP, Greenberg NM, Kwon ED. The TRAMP mouse as a model for prostate cancer. Curr Protoc Immunol, 2001; Chapter 20: Unit 20.5. [131] Wang S, Gao J, Lei Q, Rozengurt N, Pritchard C, Jiao J, Thomas GV, Li G, Roy-Burman P, Nelson PS, Liu X, Wu H. Prostate-specific deletion of the murine Pten tumor suppressor gene leads to metastatic prostate cancer. Cancer Cell, 2003; 4: 209-21.

116

[132] Lin D, Xue H, Wang Y, Wu R, Watahiki A, Dong X, Cheng H, Wyatt AW, Collins CC, Gout PW. Next generation patient-derived prostate cancer xenograft models. Asian J Androl, 2014; 16: 407-12. [133] Russell PJ, Kingsley EA. Human prostate cancer cell lines. Methods Mol Med, 2003; 81: 21-39. [134] Thalmann GN, Sikes RA, Wu TT, Degeorges A, Chang SM, Ozen M, Pathak S, Chung LW. LNCaP progression model of human prostate cancer: androgen-independence and osseous metastasis. Prostate, 2000; 44: 91-103 Jul 1;44(2). [135] Liu AY, Brubaker KD, Goo YA, Quinn JE, Kral S, Sorensen CM, Vessella RL, Belldegrun AS, Hood LE. Lineage relationship between LNCaP and LNCaP-derived prostate cancer cell lines. Prostate, 2004; 60: 98-108. [136] Chen Q, Watson JT, Marengo SR, Decker KS, Coleman I, Nelson PS, Sikes RA. Gene expression in the LNCaP human prostate cancer progression model: progression associated expression in vitro corresponds to expression changes associated with prostate cancer progression in vivo. Cancer Lett, 2006; 244: 274-88. [137] Sobel RE, Sadar MD. Cell lines used in prostate cancer research: a compendium of old and new lines--part 1. J Urol, 2005; 173: 342-59. [138] Choi SY, Lin D, Gout PW, Collins CC, Xu Y, Wang Y. Lessons from patient-derived xenografts for better in vitro modeling of human cancer. Adv Drug Deliv Rev, 2014; 79- 80: 222-37. [139] Lin D, Wyatt AW, Xue H, Wang Y, Dong X, Haegert A, Wu R, Brahmbhatt S, Mo F, Jong L, Bell RH, Anderson S, Hurtado-Coll A, Fazli L, Sharma M, Beltran H, Rubin M, Cox M, Gout PW, Morris J, Goldenberg L, Volik SV, Gleave ME, Collins CC. High fidelity patient-derived xenografts for accelerating prostate cancer discovery and drug development. Cancer Res, 2014; 74: 1272-83. [140] Varambally S, Dhanasekaran SM, Zhou M, Barrette TR, Kumar-Sinha C, Sanda MG, Ghosh D, Pienta KJ, Sewalt RG, Otte AP, Rubin MA, Chinnaiyan AM. The polycomb group protein EZH2 is involved in progression of prostate cancer. Nature, 2002; 419: 624-9. [141] van Leenders GJ, Dukers D, Hessels D, van den Kieboom SW, Hulsbergen CA, Witjes JA, Otte AP, Meijer CJ, Raaphorst FM. Polycomb-group oncogenes EZH2, BMI1, and RING1 are overexpressed in prostate cancer with adverse pathologic and clinical features. Eur Urol, 2007; 52: 455-63. [142] Berger SL, Kouzarides T, Shiekhattar R, Shilatifard A. An operational definition of epigenetics. Genes Dev, 2009; 23: 781-3. [143] Bonasio R, Tu S, Reinberg D. Molecular signals of epigenetic states. Science, 2010; 330: 612-6. [144] Feinberg AP. Phenotypic plasticity and the epigenetics of human disease. Nature, 2007; 447: 433-40. [145] Murrell A, Hurd PJ, Wood IC. Epigenetic mechanisms in development and disease. Biochem Soc Trans, 2013; 41: 697-9. [146] Hashimshony T, Zhang J, Keshet I, Bustin M, Cedar H. The role of DNA methylation in setting up chromatin structure during development. Nat Genet, 2003; 34: 187-92. [147] Jenuwein T, Allis CD. Translating the histone code. Science, 2001; 293: 1074-80.

117

[148] Kornberg RD. Chromatin structure: a repeating unit of histones and DNA. Science, 1974; 184: 868-71. [149] Kornberg RD, Lorch Y. Twenty-five years of the nucleosome, fundamental particle of the eukaryote chromosome. Cell, 1999; 98: 285-94. [150] Holliday R, Pugh JE. DNA modification mechanisms and gene activity during development. Science, 1975; 187: 226-32. [151] Cedar H, Bergman Y. Linking DNA methylation and histone modification: patterns and paradigms. Nat Rev Genet, 2009; 10: 295-304. [152] Arnaudo AM, Garcia BA. Proteomic characterization of novel histone post-translational modifications. Epigenetics Chromatin, 2013; 6: 24. [153] Musselman CA, Lalonde ME, Côté J, Kutateladze TG. Perceiving the epigenetic landscape through histone readers. Nat Struct Mol Biol, 2012; 19: 1218-27. [154] Taunton J, Hassig CA, Schreiber SL. A mammalian histone deacetylase related to the yeast transcriptional regulator Rpd3p. Science, 1996; 272: 408-11. [155] Strahl BD, Allis CD. The language of covalent histone modifications. Nature, 2000; 403: 41-5. [156] Meyerson M, Gabriel S, Getz G. Advances in understanding cancer genomes through second-generation sequencing. Nat Rev Genet, 2010; 11: 685-96. [157] Swanton C. Intratumor heterogeneity: evolution through space and time. Cancer Res, 2012; 72: 4875-82. [158] Gerlinger M, Rowan AJ, Horswell S, Larkin J, Endesfelder D, Gronroos E, Martinez P, Matthews N, Stewart A, Tarpey P, Varela I, Phillimore B, Begum S, McDonald NQ, Butler A, Jones D, Raine K, Latimer C, Santos CR, Nohadani M, Eklund AC, Spencer- Dene B, Clark G, Pickering L, Stamp G, Gore M, Szallasi Z, Downward J, Futreal PA, Swanton C. Intratumor heterogeneity and branched evolution revealed by multiregion sequencing. N Engl J Med, 2012; 366: 883-92. [159] McDonald OG, Wu H, Timp W, Doi A, Feinberg AP. Genome-scale epigenetic reprogramming during epithelial-to-mesenchymal transition. Nat Struct Mol Biol, 2011; 18: 867-74. [160] Timp W, Feinberg AP. Cancer as a dysregulated epigenome allowing cellular growth advantage at the expense of the host. Nat Rev Cancer, 2013; 13: 497-510. [161] Perry AS, Watson RW, Lawler M, Hollywood D. The epigenome as a therapeutic target in prostate cancer. Nat Rev Urol, 2010; 7: 668-80. [162] Seligson DB, Horvath S, Shi T, Yu H, Tze S, Grunstein M, Kurdistani SK. Global histone modification patterns predict risk of prostate cancer recurrence. Nature, 2005; 435: 1262-6. [163] Seligson DB, Horvath S, McBrian MA, Mah V, Yu H, Tze S, Wang Q, Chia D, Goodglick L, Kurdistani SK. Global levels of histone modifications predict prognosis in different cancers. Am J Pathol, 2009; 174: 1619-28. [164] Crea F, Hurt EM, Mathews LA, Cabarcas SM, Sun L, Marquez VE, Danesi R, Farrar WL. Pharmacologic disruption of Polycomb Repressive Complex 2 inhibits tumorigenicity and tumor progression in prostate cancer. Mol Cancer, 2011; 10: 40. [165] Senthilkumar R, Mishra RK. Novel motifs distinguish multiple homologues of Polycomb in vertebrates: expansion and diversification of the epigenetic toolkit. BMC Genomics, 2009; 10: 549. 118

[166] Lewis EB. A gene complex controlling segmentation in Drosophila. Nature, 1978; 276: 565-70. [167] Margueron R, Reinberg D. The Polycomb complex PRC2 and its mark in life. Nature, 2011; 469: 343-9. [168] Schwartz YB, Pirrotta V. A new world of Polycombs: unexpected partnerships and emerging functions. Nat Rev Genet, 2013; 14: 853-64. [169] Francis NJ, Saurin AJ, Shao Z, Kingston RE. Reconstitution of a functional core polycomb repressive complex. Mol Cell, 2001; 8: 545-56. [170] Whitcomb SJ, Basu A, Allis CD, Bernstein E. Polycomb Group proteins: an evolutionary perspective. Trends Genet, 2007; 23: 494-502. [171] Cao R, Wang L, Wang H, Xia L, Erdjument-Bromage H, Tempst P, Jones RS, Zhang Y. Role of histone H3 lysine 27 methylation in Polycomb-group silencing. Science, 2002; 298: 1039-43. [172] Margueron R, Justin N, Ohno K, Sharpe ML, Son J, Drury WJ, Voigt P, Martin SR, Taylor WR, De Marco V, Pirrotta V, Reinberg D, Gamblin SJ. Role of the polycomb protein EED in the propagation of repressive histone marks. Nature, 2009; 461: 762-7. [173] Kaustov L, Ouyang H, Amaya M, Lemak A, Nady N, Duan S, Wasney GA, Li Z, Vedadi M, Schapira M, Min J, Arrowsmith CH. Recognition and specificity determinants of the human cbx chromodomains. J Biol Chem, 2011; 286: 521-9. [174] Wang H, Wang L, Erdjument-Bromage H, Vidal M, Tempst P, Jones RS, Zhang Y. Role of histone H2A ubiquitination in Polycomb silencing. Nature, 2004; 431: 873-8. [175] van der Stoop P, Boutsma EA, Hulsman D, Noback S, Heimerikx M, Kerkhoven RM, Voncken JW, Wessels LF, van Lohuizen M. Ubiquitin E3 ligase Ring1b/Rnf2 of polycomb repressive complex 1 contributes to stable maintenance of mouse embryonic stem cells. PLoS One, 2008; 3: e2235. [176] Gupta RA, Shah N, Wang KC, Kim J, Horlings HM, Wong DJ, Tsai MC, Hung T, Argani P, Rinn JL, Wang Y, Brzoska P, Kong B, Li R, West RB, van de Vijver MJ, Sukumar S, Chang HY. Long non-coding RNA HOTAIR reprograms chromatin state to promote cancer metastasis. Nature, 2010; 464: 1071-6. [177] Yap KL, Li S, Muñoz-Cabello AM, Raguz S, Zeng L, Mujtaba S, Gil J, Walsh MJ, Zhou MM. Molecular interplay of the noncoding RNA ANRIL and methylated histone H3 lysine 27 by polycomb CBX7 in transcriptional silencing of INK4a. Mol Cell, 2010; 38: 662-74. [178] Klauke K, Radulović V, Broekhuis M, Weersing E, Zwart E, Olthof S, Ritsema M, Bruggeman S, Wu X, Helin K, Bystrykh L, de Haan G. Polycomb Cbx family members mediate the balance between haematopoietic stem cell self-renewal and differentiation. Nat Cell Biol, 2013; 15: 353-62. [179] Eskeland R, Leeb M, Grimes GR, Kress C, Boyle S, Sproul D, Gilbert N, Fan Y, Skoultchi AI, Wutz A, Bickmore WA. Ring1B compacts chromatin structure and represses gene expression independent of histone ubiquitination. Mol Cell, 2010; 38: 452-64. [180] Tavares L, Dimitrova E, Oxley D, Webster J, Poot R, Demmers J, Bezstarosti K, Taylor S, Ura H, Koide H, Wutz A, Vidal M, Elderkin S, Brockdorff N. RYBP-PRC1 complexes mediate H2A ubiquitylation at polycomb target sites independently of PRC2 and H3K27me3. Cell, 2012; 148: 664-78. 119

[181] Gao Z, Zhang J, Bonasio R, Strino F, Sawai A, Parisi F, Kluger Y, Reinberg D. PCGF homologs, CBX proteins, and RYBP define functionally distinct PRC1 family complexes. Mol Cell, 2012; 45: 344-56. [182] Lehmann L, Ferrari R, Vashisht AA, Wohlschlegel JA, Kurdistani SK, Carey M. Polycomb repressive complex 1 (PRC1) disassembles RNA polymerase II preinitiation complexes. J Biol Chem, 2012; 287: 35784-94. [183] Grau DJ, Chapman BA, Garlick JD, Borowsky M, Francis NJ, Kingston RE. Compaction of chromatin by diverse Polycomb group proteins requires localized regions of high charge. Genes Dev, 2011; 25: 2210-21. [184] Francis NJ, Kingston RE, Woodcock CL. Chromatin compaction by a polycomb group protein complex. Science, 2004; 306: 1574-7. [185] Shao Z, Raible F, Mollaaghababa R, Guyon JR, Wu CT, Bender W, Kingston RE. Stabilization of chromatin structure by PRC1, a Polycomb complex. Cell, 1999; 98: 37- 46. [186] Blackledge NP, Farcas AM, Kondo T, King HW, McGouran JF, Hanssen LL, Ito S, Cooper S, Kondo K, Koseki Y, Ishikura T, Long HK, Sheahan TW, Brockdorff N, Kessler BM, Koseki H, Klose RJ. Variant PRC1 complex-dependent H2A ubiquitylation drives PRC2 recruitment and polycomb domain formation. Cell, 2014; 157: 1445-59. [187] Bracken AP, Helin K. Polycomb group proteins: navigators of lineage pathways led astray in cancer. Nat Rev Cancer, 2009; 9: 773-84. [188] Sauvageau M, Sauvageau G. Polycomb group proteins: multi-faceted regulators of somatic stem cells and cancer. Cell Stem Cell, 2010; 7: 299-313. [189] Lee TI, Jenner RG, Boyer LA, Guenther MG, Levine SS, Kumar RM, Chevalier B, Johnstone SE, Cole MF, Isono K, Koseki H, Fuchikami T, Abe K, Murray HL, Zucker JP, Yuan B, Bell GW, Herbolsheimer E, Hannett NM, Sun K, Odom DT, Otte AP, Volkert TL, Bartel DP, Melton DA, Gifford DK, Jaenisch R, Young RA. Control of developmental regulators by Polycomb in human embryonic stem cells. Cell, 2006; 125: 301-13. [190] Bantignies F, Roure V, Comet I, Leblanc B, Schuettengruber B, Bonnet J, Tixier V, Mas A, Cavalli G. Polycomb-dependent regulatory contacts between distant Hox loci in Drosophila. Cell, 2011; 144: 214-26. [191] Beuchle D, Struhl G, Müller J. Polycomb group proteins and heritable silencing of Drosophila Hox genes. Development, 2001; 128: 993-1004. [192] Yu BD, Hess JL, Horning SE, Brown GA, Korsmeyer SJ. Altered Hox expression and segmental identity in Mll-mutant mice. Nature, 1995; 378: 505-8. [193] Katoh-Fukui Y, Tsuchiya R, Shiroishi T, Nakahara Y, Hashimoto N, Noguchi K, Higashinakagawa T. Male-to-female sex reversal in M33 mutant mice. Nature, 1998; 393: 688-92. [194] Lukacs RU, Memarzadeh S, Wu H, Witte ON. Bmi-1 is a crucial regulator of prostate stem cell self-renewal and malignant transformation. Cell Stem Cell, 2010; 7: 682-93. [195] Yu J, Rhodes DR, Tomlins SA, Cao X, Chen G, Mehra R, Wang X, Ghosh D, Shah RB, Varambally S, Pienta KJ, Chinnaiyan AM. A polycomb repression signature in metastatic prostate cancer predicts cancer outcome. Cancer Res, 2007; 67: 10657-63. [196] Morin RD, Johnson NA, Severson TM, Mungall AJ, An J, Goya R, Paul JE, Boyle M, Woolcock BW, Kuchenbauer F, Yap D, Humphries RK, Griffith OL, Shah S, Zhu H, 120

Kimbara M, Shashkin P, Charlot JF, Tcherpakov M, Corbett R, Tam A, Varhol R, Smailus D, Moksa M, Zhao Y, Delaney A, Qian H, Birol I, Schein J, Moore R, Holt R, Horsman DE, Connors JM, Jones S, Aparicio S, Hirst M, Gascoyne RD, Marra MA. Somatic mutations altering EZH2 (Tyr641) in follicular and diffuse large B-cell lymphomas of germinal-center origin. Nat Genet, 2010; 42: 181-5. [197] Ernst T, Chase AJ, Score J, Hidalgo-Curtis CE, Bryant C, Jones AV, Waghorn K, Zoi K, Ross FM, Reiter A, Hochhaus A, Drexler HG, Duncombe A, Cervantes F, Oscier D, Boultwood J, Grand FH, Cross NC. Inactivating mutations of the histone methyltransferase gene EZH2 in myeloid disorders. Nat Genet, 2010; 42: 722-6. [198] Jene-Sanz A, Váraljai R, Vilkova AV, Khramtsova GF, Khramtsov AI, Olopade OI, Lopez-Bigas N, Benevolenskaya EV. Expression of polycomb targets predicts breast cancer prognosis. Mol Cell Biol, 2013; 33: 3951-61. [199] Cao Q, Mani RS, Ateeq B, Dhanasekaran SM, Asangani IA, Prensner JR, Kim JH, Brenner JC, Jing X, Cao X, Wang R, Li Y, Dahiya A, Wang L, Pandhi M, Lonigro RJ, Wu YM, Tomlins SA, Palanisamy N, Qin Z, Yu J, Maher CA, Varambally S, Chinnaiyan AM. Coordinated regulation of polycomb group complexes through microRNAs in cancer. Cancer Cell, 2011; 20: 187-99. [200] Yu J, Cao Q, Mehra R, Laxman B, Tomlins SA, Creighton CJ, Dhanasekaran SM, Shen R, Chen G, Morris DS, Marquez VE, Shah RB, Ghosh D, Varambally S, Chinnaiyan AM. Integrative genomics analysis reveals silencing of beta-adrenergic signaling by polycomb in prostate cancer. Cancer Cell, 2007; 12: 419-31. [201] Cao Q, Yu J, Dhanasekaran SM, Kim JH, Mani RS, Tomlins SA, Mehra R, Laxman B, Cao X, Kleer CG, Varambally S, Chinnaiyan AM. Repression of E-cadherin by the polycomb group protein EZH2 in cancer. Oncogene, 2008; 27: 7274-84. [202] Varambally S, Cao Q, Mani RS, Shankar S, Wang X, Ateeq B, Laxman B, Cao X, Jing X, Ramnarayanan K, Brenner JC, Yu J, Kim JH, Han B, Tan P, Kumar-Sinha C, Lonigro RJ, Palanisamy N, Maher CA, Chinnaiyan AM. Genomic loss of microRNA-101 leads to overexpression of histone methyltransferase EZH2 in cancer. Science, 2008; 322: 1695-9. [203] van der Vlag J, Otte AP. Transcriptional repression mediated by the human polycomb- group protein EED involves histone deacetylation. Nat Genet, 1999; 23: 474-8. [204] Viré E, Brenner C, Deplus R, Blanchon L, Fraga M, Didelot C, Morey L, Van Eynde A, Bernard D, Vanderwinden JM, Bollen M, Esteller M, Di Croce L, de Launoit Y, Fuks F. The Polycomb group protein EZH2 directly controls DNA methylation. Nature, 2006; 439: 871-4. [205] Tsai MC, Manor O, Wan Y, Mosammaparast N, Wang JK, Lan F, Shi Y, Segal E, Chang HY. Long noncoding RNA as modular scaffold of histone modification complexes. Science, 2010; 329: 689-93. [206] Chu C, Qu K, Zhong FL, Artandi SE, Chang HY. Genomic maps of long noncoding RNA occupancy reveal principles of RNA-chromatin interactions. Mol Cell, 2011; 44: 667-78. [207] Bracken AP, Pasini D, Capra M, Prosperini E, Colli E, Helin K. EZH2 is downstream of the pRB-E2F pathway, essential for proliferation and amplified in cancer. EMBO J, 2003; 22: 5323-35.

121

[208] Coe BP, Thu KL, Aviel-Ronen S, Vucic EA, Gazdar AF, Lam S, Tsao MS, Lam WL. Genomic deregulation of the E2F/Rb pathway leads to activation of the oncogene EZH2 in small cell lung cancer. PLoS One, 2013; 8: e71670. [209] Siddique HR, Saleem M. Role of BMI1, a stem cell factor, in cancer recurrence and chemoresistance: preclinical and clinical evidences. Stem Cells, 2012; 30: 372-8. [210] McGinty RK, Henrici RC, Tan S. Crystal structure of the PRC1 ubiquitylation module bound to the nucleosome. Nature, 2014; 514: 591-6. [211] Berezovska OP, Glinskii AB, Yang Z, Li XM, Hoffman RM, Glinsky GV. Essential role for activation of the Polycomb group (PcG) protein chromatin silencing pathway in metastatic prostate cancer. Cell Cycle, 2006; 5: 1886-901. [212] Fan C, He L, Kapoor A, Rybak AP, De Melo J, Cutz JC, Tang D. PTEN inhibits BMI1 function independently of its phosphatase activity. Mol Cancer, 2009; 8: 98. [213] Crea F, Duhagon Serrat MA, Hurt EM, Thomas SB, Danesi R, Farrar WL. BMI1 silencing enhances docetaxel activity and impairs antioxidant response in prostate cancer. Int J Cancer, 2011; 128: 1946-54. [214] Vincenz C, Kerppola TK. Different polycomb group CBX family proteins associate with distinct regions of chromatin using nonhomologous protein sequences. Proc Natl Acad Sci U S A, 2008; 105: 16572-7. [215] Bernstein E, Duncan EM, Masui O, Gil J, Heard E, Allis CD. Mouse polycomb proteins bind differentially to methylated histone H3 and RNA and are enriched in facultative heterochromatin. Mol Cell Biol, 2006; 26: 2560-9. [216] Vandamme J, Völkel P, Rosnoblet C, Le Faou P, Angrand PO. Interaction proteomics analysis of polycomb proteins defines distinct PRC1 complexes in mammalian cells. Mol Cell Proteomics, 2011; 10: M110.002642. [217] O'Loghlen A, Muñoz-Cabello AM, Gaspar-Maia A, Wu HA, Banito A, Kunowska N, Racek T, Pemberton HN, Beolchi P, Lavial F, Masui O, Vermeulen M, Carroll T, Graumann J, Heard E, Dillon N, Azuara V, Snijders AP, Peters G, Bernstein E, Gil J. MicroRNA regulation of Cbx7 mediates a switch of Polycomb orthologs during ESC differentiation. Cell Stem Cell, 2012; 10: 33-46. [218] Morey L, Pascual G, Cozzuto L, Roma G, Wutz A, Benitah SA, Di Croce L. Nonoverlapping functions of the Polycomb group Cbx family of proteins in embryonic stem cells. Cell Stem Cell, 2012; 10: 47-62. [219] Wang B, Tang J, Liao D, Wang G, Zhang M, Sang Y, Cao J, Wu Y, Zhang R, Li S, Ding W, Zhang G, Kang T. Chromobox homolog 4 is correlated with prognosis and tumor cell growth in hepatocellular carcinoma. Ann Surg Oncol, 2013; 20 Suppl 3: 684-92. [220] Rothman N, Garcia-Closas M, Chatterjee N, Malats N, Wu X, Figueroa JD, Real FX, Van Den Berg D, Matullo G, Baris D, Thun M, Kiemeney LA, Vineis P, De Vivo I, Albanes D, Purdue MP, Rafnar T, Hildebrandt MA, Kiltie AE, Cussenot O, Golka K, Kumar R, Taylor JA, Mayordomo JI, Jacobs KB, Kogevinas M, Hutchinson A, Wang Z, Fu YP, Prokunina-Olsson L, Burdett L, Yeager M, Wheeler W, Tardón A, Serra C, Carrato A, García-Closas R, Lloreta J, Johnson A, Schwenn M, Karagas MR, Schned A, Andriole G, Grubb R, Black A, Jacobs EJ, Diver WR, Gapstur SM, Weinstein SJ, Virtamo J, Cortessis VK, Gago-Dominguez M, Pike MC, Stern MC, Yuan JM, Hunter DJ, McGrath M, Dinney CP, Czerniak B, Chen M, Yang H, Vermeulen SH, Aben KK, Witjes JA, Makkinje RR, Sulem P, Besenbacher S, Stefansson K, Riboli E, Brennan P, 122

Panico S, Navarro C, Allen NE, Bueno-de-Mesquita HB, Trichopoulos D, Caporaso N, Landi MT, Canzian F, Ljungberg B, Tjonneland A, Clavel-Chapelon F, Bishop DT, Teo MT, Knowles MA, Guarrera S, Polidoro S, Ricceri F, Sacerdote C, Allione A, Cancel- Tassin G, Selinski S, Hengstler JG, Dietrich H, Fletcher T, Rudnai P, Gurzau E, Koppova K, Bolick SC, Godfrey A, Xu Z, Sanz-Velez JI, D García-Prats M, Sanchez M, Valdivia G, Porru S, Benhamou S, Hoover RN, Fraumeni JF, Silverman DT, Chanock SJ. A multi- stage genome-wide association study of bladder cancer identifies multiple susceptibility loci. Nat Genet, 2010; 42: 978-84. [221] Forzati F, Federico A, Pallante P, Abbate A, Esposito F, Malapelle U, Sepe R, Palma G, Troncone G, Scarfò M, Arra C, Fedele M, Fusco A. CBX7 is a tumor suppressor in mice and humans. J Clin Invest, 2012; 122: 612-23. [222] Mohammad HP, Cai Y, McGarvey KM, Easwaran H, Van Neste L, Ohm JE, O'Hagan HM, Baylin SB. Polycomb CBX7 promotes initiation of heritable repression of genes frequently silenced with cancer-specific DNA hypermethylation. Cancer Res, 2009; 69: 6322-30. [223] Tabet S, Douglas SF, Daze KD, Garnett GA, Allen KJ, Abrioux EM, Quon TT, Wulff JE, Hof F. Synthetic trimethyllysine receptors that bind histone 3, trimethyllysine 27 (H3K27me3) and disrupt its interaction with the epigenetic reader protein CBX7. Bioorg Med Chem, 2013; 21: 7004-10. [224] Simhadri C, Daze KD, Douglas SF, Quon TT, Dev A, Gignac MC, Peng F, Heller M, Boulanger MJ, Wulff JE, Hof F. Chromodomain antagonists that target the polycomb- group methyllysine reader protein chromobox homolog 7 (CBX7). J Med Chem, 2014; 57: 2874-83. [225] Tan J, Jones M, Koseki H, Nakayama M, Muntean AG, Maillard I, Hess JL. CBX8, a polycomb group protein, is essential for MLL-AF9-induced leukemogenesis. Cancer Cell, 2011; 20: 563-75. [226] Dietrich N, Bracken AP, Trinh E, Schjerling CK, Koseki H, Rappsilber J, Helin K, Hansen KH. Bypass of senescence by the polycomb group protein CBX8 through direct binding to the INK4A-ARF locus. EMBO J, 2007; 26: 1637-48. [227] Gecz J, Gaunt SJ, Passage E, Burton RD, Cudrey C, Pearce JJ, Fontes M. Assignment of a Polycomb-like chromobox gene (CBX2) to human chromosome 17q25. Genomics, 1995; 26: 130-3. [228] Coré N, Bel S, Gaunt SJ, Aurrand-Lions M, Pearce J, Fisher A, Djabali M. Altered cellular proliferation and mesoderm patterning in Polycomb-M33-deficient mice. Development, 1997; 124: 721-9. [229] Pearce JJ, Singh PB, Gaunt SJ. The mouse has a Polycomb-like chromobox gene. Development, 1992; 114: 921-9. [230] Müller J, Gaunt S, Lawrence PA. Function of the Polycomb protein is conserved in mice and flies. Development, 1995; 121: 2847-52. [231] Völkel P, Le Faou P, Vandamme J, Pira D, Angrand PO. A human Polycomb isoform lacking the Pc box does not participate to PRC1 complexes but forms protein assemblies and represses transcription. Epigenetics, 2012; 7: 482-91. [232] Zhen CY, Duc HN, Kokotovic M, Phiel CJ, Ren X. Cbx2 stably associates with mitotic chromosomes via a PRC2- or PRC1-independent mechanism and is needed for recruiting PRC1 complex to mitotic chromosomes. Mol Biol Cell, 2014; 25: 3726-39. 123

[233] Alkema MJ, Bronk M, Verhoeven E, Otte A, van 't Veer LJ, Berns A, van Lohuizen M. Identification of Bmi1-interacting proteins as constituents of a multimeric mammalian polycomb complex. Genes Dev, 1997; 11: 226-40. [234] Bezsonova I, Walker JR, Bacik JP, Duan S, Dhe-Paganon S, Arrowsmith CH. Ring1B contains a ubiquitin-like docking module for interaction with Cbx proteins. Biochemistry, 2009; 48: 10542-8. [235] Tardat M, Albert M, Kunzmann R, Liu Z, Kaustov L, Thierry R, Duan S, Brykczynska U, Arrowsmith CH, Peters AH. Cbx2 Targets PRC1 to Constitutive Heterochromatin in Mouse Zygotes in a Parent-of-Origin-Dependent Manner. Mol Cell, 2015; 58: 157-71. [236] Wang G, Horsley D, Ma A, Otte AP, Hutchings A, Butcher GW, Singh PB. M33, a mammalian homologue of Drosophila Polycomb localises to euchromatin within interphase nuclei but is enriched within the centromeric heterochromatin of metaphase chromosomes. Cytogenet Cell Genet, 1997; 78: 50-5. [237] Plath K, Talbot D, Hamer KM, Otte AP, Yang TP, Jaenisch R, Panning B. Developmentally regulated alterations in Polycomb repressive complex 1 proteins on the inactive X chromosome. J Cell Biol, 2004; 167: 1025-35. [238] Noguchi K, Shiurba R, Higashinakagawa T. Nuclear translocation of mouse polycomb m33 protein in regenerating liver. Biochem Biophys Res Commun, 2002; 291: 508-15. [239] Hatano A, Matsumoto M, Higashinakagawa T, Nakayama KI. Phosphorylation of the chromodomain changes the binding specificity of Cbx2 for methylated histone H3. Biochem Biophys Res Commun, 2010; 397: 93-9. [240] Pelegri F, Lehmann R. A role of polycomb group genes in the regulation of gap gene expression in Drosophila. Genetics, 1994; 136: 1341-53. [241] Bel S, Coré N, Djabali M, Kieboom K, Van der Lugt N, Alkema MJ, Van Lohuizen M. Genetic interactions and dosage effects of Polycomb group genes in mice. Development, 1998; 125: 3543-51. [242] Bel-Vialar S, Coré N, Terranova R, Goudot V, Boned A, Djabali M. Altered retinoic acid sensitivity and temporal expression of Hox genes in polycomb-M33-deficient mice. Dev Biol, 2000; 224: 238-49. [243] Biason-Lauber A, Konrad D, Meyer M, DeBeaufort C, Schoenle EJ. Ovaries and female phenotype in a girl with 46,XY karyotype and mutations in the CBX2 gene. Am J Hum Genet, 2009; 84: 658-63. [244] Zhang X, Ho SM. Epigenetics meets endocrinology. J Mol Endocrinol, 2011; 46: R11- 32. [245] Nef S, Parada LF. Hormones in male sexual development. Genes Dev, 2000; 14: 3075- 86. [246] Katoh-Fukui Y, Owaki A, Toyama Y, Kusaka M, Shinohara Y, Maekawa M, Toshimori K, Morohashi K. Mouse Polycomb M33 is required for splenic vascular and adrenal gland formation through regulating Ad4BP/SF1 expression. Blood, 2005; 106: 1612-20. [247] Ogawa H, Ishiguro K, Gaubatz S, Livingston DM, Nakatani Y. A complex with chromatin modifiers that occupies E2F- and Myc-responsive genes in G0 cells. Science, 2002; 296: 1132-6. [248] Coré N, Joly F, Boned A, Djabali M. Disruption of E2F signaling suppresses the INK4a- induced proliferative defect in M33-deficient mice. Oncogene, 2004; 23: 7660-8.

124

[249] Gu L, Frommel SC, Oakes CC, Simon R, Grupp K, Gerig CY, Bär D, Robinson MD, Baer C, Weiss M, Gu Z, Schapira M, Kuner R, Sültmann H, Provenzano M, Yaspo ML, Brors B, Korbel J, Schlomm T, Sauter G, Eils R, Plass C, Santoro R, Cancer IPoEOP. BAZ2A (TIP5) is involved in epigenetic alterations in prostate cancer and its overexpression predicts disease recurrence. Nat Genet, 2015; 47: 22-30. [250] Richly H, Aloia L, Di Croce L. Roles of the Polycomb group proteins in stem cells and cancer. Cell Death Dis, 2011; 2: e204. [251] Filippakopoulos P, Qi J, Picaud S, Shen Y, Smith WB, Fedorov O, Morse EM, Keates T, Hickman TT, Felletar I, Philpott M, Munro S, McKeown MR, Wang Y, Christie AL, West N, Cameron MJ, Schwartz B, Heightman TD, La Thangue N, French CA, Wiest O, Kung AL, Knapp S, Bradner JE. Selective inhibition of BET bromodomains. Nature, 2010; 468: 1067-73. [252] Li LC, Carroll PR, Dahiya R. Epigenetic changes in prostate cancer: implication for diagnosis and treatment. J Natl Cancer Inst, 2005; 97: 103-15. [253] Horwich A, Hugosson J, de Reijke T, Wiegel T, Fizazi K, Kataja V, Members P, Oncology ESfM. Prostate cancer: ESMO Consensus Conference Guidelines 2012. Ann Oncol, 2013; 24: 1141-62. [254] Beer TM, Armstrong AJ, Rathkopf DE, Loriot Y, Sternberg CN, Higano CS, Iversen P, Bhattacharya S, Carles J, Chowdhury S, Davis ID, de Bono JS, Evans CP, Fizazi K, Joshua AM, Kim CS, Kimura G, Mainwaring P, Mansbach H, Miller K, Noonberg SB, Perabo F, Phung D, Saad F, Scher HI, Taplin ME, Venner PM, Tombal B, Investigators P. Enzalutamide in metastatic prostate cancer before chemotherapy. N Engl J Med, 2014; 371: 424-33. [255] Colwill K, Gräslund S, Group RPBW. A roadmap to generate renewable protein binders to the human proteome. Nat Methods, 2011; 8: 551-8. [256] Lin D, Watahiki A, Bayani J, Zhang F, Liu L, Ling V, Sadar MD, English J, Fazli L, So A, Gout PW, Gleave M, Squire JA, Wang YZ. ASAP1, a gene at 8q24, is associated with prostate cancer metastasis. Cancer Res, 2008; 68: 4352-9. [257] Gashaw I, Ellinghaus P, Sommer A, Asadullah K. What makes a good drug target? Drug Discov Today, 2012; 17 Suppl: S24-30. [258] Hayward SW, Dahiya R, Cunha GR, Bartek J, Deshpande N, Narayan P. Establishment and characterization of an immortalized but non-transformed human prostate epithelial cell line: BPH-1. In Vitro Cell Dev Biol Anim, 1995; 31: 14-24. [259] Grasso CS, Wu YM, Robinson DR, Cao X, Dhanasekaran SM, Khan AP, Quist MJ, Jing X, Lonigro RJ, Brenner JC, Asangani IA, Ateeq B, Chun SY, Siddiqui J, Sam L, Anstett M, Mehra R, Prensner JR, Palanisamy N, Ryslik GA, Vandin F, Raphael BJ, Kunju LP, Rhodes DR, Pienta KJ, Chinnaiyan AM, Tomlins SA. The mutational landscape of lethal castration-resistant prostate cancer. Nature, 2012; 487: 239-43. [260] Crea F, Clermont PL, Mai A, Helgason CD. Histone modifications, stem cells and prostate cancer. Curr Pharm Des, 2014; 20: 1687-97. [261] Collins AT, Berry PA, Hyde C, Stower MJ, Maitland NJ. Prospective identification of tumorigenic prostate cancer stem cells. Cancer Res, 2005; 65: 10946-51. [262] Chang AJ, Autio KA, Roach M, Scher HI. High-risk prostate cancer-classification and therapy. Nat Rev Clin Oncol, 2014; 11: 308-23.

125

[263] Cheng L, Pisansky TM, Sebo TJ, Leibovich BC, Ramnani DM, Weaver AL, Scherer BG, Blute ML, Zincke H, Bostwick DG. Cell proliferation in prostate cancer patients with lymph node metastasis: a marker for progression. Clin Cancer Res, 1999; 5: 2820-3. [264] Fisher G, Yang ZH, Kudahetti S, Møller H, Scardino P, Cuzick J, Berney DM, Group TP. Prognostic value of Ki-67 for prostate cancer death in a conservatively managed cohort. Br J Cancer, 2013; 108: 271-7. [265] Bettencourt MC, Bauer JJ, Sesterhenn IA, Mostofi FK, McLeod DG, Moul JW. Ki-67 expression is a prognostic marker of prostate cancer recurrence after radical prostatectomy. J Urol, 1996; 156: 1064-8. [266] Smith MR, Kabbinavar F, Saad F, Hussain A, Gittelman MC, Bilhartz DL, Wynne C, Murray R, Zinner NR, Schulman C, Linnartz R, Zheng M, Goessl C, Hei YJ, Small EJ, Cook R, Higano CS. Natural history of rising serum prostate-specific antigen in men with castrate nonmetastatic prostate cancer. J Clin Oncol, 2005; 23: 2918-25. [267] Bitting RL, Armstrong AJ. Targeting the PI3K/Akt/mTOR pathway in castration- resistant prostate cancer. Endocr Relat Cancer, 2013; 20: R83-99. [268] Palmgren JS, Karavadia SS, Wakefield MR. Unusual and underappreciated: small cell carcinoma of the prostate. Semin Oncol, 2007; 34: 22-9. [269] Hubaux R, Thu KL, Coe BP, MacAulay C, Lam S, Lam WL. EZH2 promotes E2F-driven SCLC tumorigenesis through modulation of apoptosis and cell-cycle regulation. J Thorac Oncol, 2013; 8: 1102-6. [270] Sato T, Kaneda A, Tsuji S, Isagawa T, Yamamoto S, Fujita T, Yamanaka R, Tanaka Y, Nukiwa T, Marquez VE, Ishikawa Y, Ichinose M, Aburatani H. PRC2 overexpression and PRC2-target gene repression relating to poorer prognosis in small cell lung cancer. Sci Rep, 2013; 3: 1911. [271] Hirabayashi Y, Suzki N, Tsuboi M, Endo TA, Toyoda T, Shinga J, Koseki H, Vidal M, Gotoh Y. Polycomb limits the neurogenic competence of neural precursor cells to promote astrogenic fate transition. Neuron, 2009; 63: 600-13. [272] Golden MG, Dasen JS. Polycomb repressive complex 1 activities determine the columnar organization of motor neurons. Genes Dev, 2012; 26: 2236-50. [273] Parrish JZ, Emoto K, Jan LY, Jan YN. Polycomb genes interact with the tumor suppressor genes hippo and warts in the maintenance of Drosophila sensory neuron dendrites. Genes Dev, 2007; 21: 956-72. [274] Pereira JD, Sansom SN, Smith J, Dobenecker MW, Tarakhovsky A, Livesey FJ. Ezh2, the histone methyltransferase of PRC2, regulates the balance between self-renewal and differentiation in the cerebral cortex. Proc Natl Acad Sci U S A, 2010; 107: 15957-62. [275] Egan CM, Nyman U, Skotte J, Streubel G, Turner S, O'Connell DJ, Rraklli V, Dolan MJ, Chadderton N, Hansen K, Farrar GJ, Helin K, Holmberg J, Bracken AP. CHD5 is required for neurogenesis and has a dual role in facilitating gene expression and polycomb gene repression. Dev Cell, 2013; 26: 223-36. [276] Bello B, Holbro N, Reichert H. Polycomb group genes are required for neural stem cell survival in postembryonic neurogenesis of Drosophila. Development, 2007; 134: 1091-9. [277] Peruzzi P, Bronisz A, Nowicki MO, Wang Y, Ogawa D, Price R, Nakano I, Kwon CH, Hayes J, Lawler SE, Ostrowski MC, Chiocca EA, Godlewski J. MicroRNA-128 coordinately targets Polycomb Repressor Complexes in glioma stem cells. Neuro Oncol, 2013; 15: 1212-24. 126

[278] Li G, Warden C, Zou Z, Neman J, Krueger JS, Jain A, Jandial R, Chen M. Altered expression of polycomb group genes in glioblastoma multiforme. PLoS One, 2013; 8: e80970. [279] Rhodes DR, Yu J, Shanker K, Deshpande N, Varambally R, Ghosh D, Barrette T, Pandey A, Chinnaiyan AM. ONCOMINE: a cancer microarray database and integrated data- mining platform. Neoplasia, 2004; 6: 1-6. [280] (CLCGP) CLCGP, (NGM) NGM. A genomics-based classification of human lung tumors. Sci Transl Med, 2013; 5: 209ra153. [281] Bestor TH. The DNA methyltransferases of mammals. Hum Mol Genet, 2000; 9: 2395- 402. [282] Allis CD, Berger SL, Cote J, Dent S, Jenuwien T, Kouzarides T, Pillus L, Reinberg D, Shi Y, Shiekhattar R, Shilatifard A, Workman J, Zhang Y. New nomenclature for chromatin-modifying enzymes. Cell, 2007; 131: 633-6. [283] Kinney SR, Pradhan S. Ten eleven translocation enzymes and 5-hydroxymethylation in mammalian development and cancer. Adv Exp Med Biol, 2013; 754: 57-79. [284] Kohli RM, Zhang Y. TET enzymes, TDG and the dynamics of DNA demethylation. Nature, 2013; 502: 472-9. [285] Dokmanovic M, Clarke C, Marks PA. Histone deacetylase inhibitors: overview and perspectives. Mol Cancer Res, 2007; 5: 981-9. [286] North BJ, Verdin E. Sirtuins: Sir2-related NAD-dependent protein deacetylases. Genome Biol, 2004; 5: 224. [287] Parry L, Clarke AR. The Roles of the Methyl-CpG Binding Proteins in Cancer. Genes Cancer, 2011; 2: 618-30. [288] Filippakopoulos P, Knapp S. The bromodomain interaction module. FEBS Lett, 2012; 586: 2692-704. [289] Tajul-Arifin K, Teasdale R, Ravasi T, Hume DA, Mattick JS, Group RG, Members G. Identification and analysis of chromodomain-containing proteins encoded in the mouse transcriptome. Genome Res, 2003; 13: 1416-29. [290] Di Croce L, Helin K. Transcriptional regulation by Polycomb group proteins. Nat Struct Mol Biol, 2013; 20: 1147-55. [291] Simon JA, Kingston RE. Occupying chromatin: Polycomb mechanisms for getting to genomic targets, stopping transcriptional traffic, and staying put. Mol Cell, 2013; 49: 808-24. [292] Kouzarides T. Chromatin modifications and their function. Cell, 2007; 128: 693-705. [293] Oesterling JE, Hauzeur CG, Farrow GM. Small cell anaplastic carcinoma of the prostate: a clinical, pathological and immunohistological study of 27 patients. J Urol, 1992; 147: 804-7. [294] Shah RB, Mehra R, Chinnaiyan AM, Shen R, Ghosh D, Zhou M, Macvicar GR, Varambally S, Harwood J, Bismar TA, Kim R, Rubin MA, Pienta KJ. Androgen- independent prostate cancer is a heterogeneous group of diseases: lessons from a rapid autopsy program. Cancer Res, 2004; 64: 9209-16. [295] Molina JR, Yang P, Cassivi SD, Schild SE, Adjei AA. Non-small cell lung cancer: epidemiology, risk factors, treatment, and survivorship. Mayo Clin Proc, 2008; 83: 584- 94.

127

[296] Sutherland KD, Proost N, Brouns I, Adriaensen D, Song JY, Berns A. Cell of origin of small cell lung cancer: inactivation of Trp53 and Rb1 in distinct cell types of adult mouse lung. Cancer Cell, 2011; 19: 754-64. [297] Bolton EC, So AY, Chaivorapol C, Haqq CM, Li H, Yamamoto KR. Cell- and gene- specific regulation of primary target genes by the androgen receptor. Genes Dev, 2007; 21: 2005-17. [298] Testa G. The time of timing: how Polycomb proteins regulate neurogenesis. Bioessays, 2011; 33: 519-28. [299] You JS, Kang JK, Seo DW, Park JH, Park JW, Lee JC, Jeon YJ, Cho EJ, Han JW. Depletion of embryonic stem cell signature by histone deacetylase inhibitor in NCCIT cells: involvement of Nanog suppression. Cancer Res, 2009; 69: 5716-25. [300] Ben-Porath I, Thomson MW, Carey VJ, Ge R, Bell GW, Regev A, Weinberg RA. An embryonic stem cell-like gene expression signature in poorly differentiated aggressive human tumors. Nat Genet, 2008; 40: 499-507. [301] Ohm JE, McGarvey KM, Yu X, Cheng L, Schuebel KE, Cope L, Mohammad HP, Chen W, Daniel VC, Yu W, Berman DM, Jenuwein T, Pruitt K, Sharkis SJ, Watkins DN, Herman JG, Baylin SB. A stem cell-like chromatin pattern may predispose tumor suppressor genes to DNA hypermethylation and heritable silencing. Nat Genet, 2007; 39: 237-42. [302] Jones PA. At the tipping point for epigenetic therapies in cancer. J Clin Invest, 2014; 124: 14-6. [303] Jones PA, Taylor SM. Cellular differentiation, cytidine analogs and DNA methylation. Cell, 1980; 20: 85-93. [304] Kuzmichev A, Nishioka K, Erdjument-Bromage H, Tempst P, Reinberg D. Histone methyltransferase activity associated with a human multiprotein complex containing the Enhancer of Zeste protein. Genes Dev, 2002; 16: 2893-905. [305] Satijn DP, Gunster MJ, van der Vlag J, Hamer KM, Schul W, Alkema MJ, Saurin AJ, Freemont PS, van Driel R, Otte AP. RING1 is associated with the polycomb group protein complex and acts as a transcriptional repressor. Mol Cell Biol, 1997; 17: 4105-13. [306] Bachmann IM, Halvorsen OJ, Collett K, Stefansson IM, Straume O, Haukaas SA, Salvesen HB, Otte AP, Akslen LA. EZH2 expression is associated with high proliferation rate and aggressive tumor subgroups in cutaneous melanoma and cancers of the endometrium, prostate, and breast. J Clin Oncol, 2006; 24: 268-73. [307] Cao L, Bombard J, Cintron K, Sheedy J, Weetall ML, Davis TW. BMI1 as a novel target for drug discovery in cancer. J Cell Biochem, 2011; 112: 2729-41. [308] Forbes SA, Bhamra G, Bamford S, Dawson E, Kok C, Clements J, Menzies A, Teague JW, Futreal PA, Stratton MR. The Catalogue of Somatic Mutations in Cancer (COSMIC). Curr Protoc Hum Genet, 2008; Chapter 10: Unit 10.11. [309] Uhlen M, Oksvold P, Fagerberg L, Lundberg E, Jonasson K, Forsberg M, Zwahlen M, Kampf C, Wester K, Hober S, Wernerus H, Björling L, Ponten F. Towards a knowledge- based Human Protein Atlas. Nat Biotechnol, 2010; 28: 1248-50. [310] Mihaly Z, Kormos M, Lanczky A, Dank M, Budczies J, Szasz MA, Gyorffy B. A meta- analysis of gene expression-based biomarkers predicting outcome after tamoxifen treatment in breast cancer. Breast Cancer Res Treat, 2013; 140: 219-32.

128

[311] Lawrence MS, Stojanov P, Mermel CH, Robinson JT, Garraway LA, Golub TR, Meyerson M, Gabriel SB, Lander ES, Getz G. Discovery and saturation analysis of cancer genes across 21 tumour types. Nature, 2014; 505: 495-501. [312] Aguilo F, Zhou MM, Walsh MJ. Long noncoding RNA, polycomb, and the ghosts haunting INK4b-ARF-INK4a expression. Cancer Res, 2011; 71: 5365-9. [313] Zack TI, Schumacher SE, Carter SL, Cherniack AD, Saksena G, Tabak B, Lawrence MS, Zhang CZ, Wala J, Mermel CH, Sougnez C, Gabriel SB, Hernandez B, Shen H, Laird PW, Getz G, Meyerson M, Beroukhim R. Pan-cancer patterns of somatic copy number alteration. Nat Genet, 2013; 45: 1134-1140. [314] Chase A, Cross NC. Aberrations of EZH2 in cancer. Clin Cancer Res. 2011 May 1;17(9):2613-8. [315] Granit RZ, Gabai Y, Hadar T, Karamansha Y, Liberman L, Waldhorn I, Gat-Viks I, Regev A, Maly B, Darash-Yahana M, Peretz T, Ben-Porath I. EZH2 promotes a bi- lineage identity in basal-like breast cancer cells. Oncogene, 2013; 32: 3886-95. [316] Guil S, Soler M, Portela A, Carrère J, Fonalleras E, Gómez A, Villanueva A, Esteller M. Intronic RNAs mediate EZH2 regulation of epigenetic targets. Nat Struct Mol Biol, 2012; 19: 664-70. [317] Wu SC, Kallin EM, Zhang Y. Role of H3K27 methylation in the regulation of lncRNA expression. Cell Res, 2010; 20: 1109-16. [318] Simmons MN, Berglund RK, Jones JS. A practical guide to prostate cancer diagnosis and management. Cleve Clin J Med, 2011; 78: 321-31. [319] Dawson MA, Kouzarides T, Huntly BJ. Targeting epigenetic readers in cancer. N Engl J Med, 2012; 367: 647-57. [320] Travis WD. Advances in neuroendocrine lung tumors. Ann Oncol, 2010; 21 Suppl 7: vii65-71. [321] Dawson MA, Kouzarides T. Cancer epigenetics: from mechanism to therapy. Cell, 2012; 150: 12-27. [322] Ferrari KJ, Scelfo A, Jammula S, Cuomo A, Barozzi I, Stützer A, Fischle W, Bonaldi T, Pasini D. Polycomb-dependent H3K27me1 and H3K27me2 regulate active transcription and enhancer fidelity. Mol Cell, 2014; 53: 49-62. [323] Johnson BE, Whang-Peng J, Naylor SL, Zbar B, Brauch H, Lee E, Simmons A, Russell E, Nam MH, Gazdar AF. Retention of chromosome 3 in extrapulmonary small cell cancer shown by molecular and cytogenetic studies. J Natl Cancer Inst, 1989; 81: 1223-8. [324] Aggarwal R, Zhang T, Small EJ, Armstrong AJ. Neuroendocrine prostate cancer: subtypes, biology, and clinical outcomes. J Natl Compr Canc Netw, 2014; 12: 719-26. [325] Majumder PK, Sellers WR. Akt-regulated pathways in prostate cancer. Oncogene, 2005; 24: 7465-74. [326] Akasaka T, van Lohuizen M, van der Lugt N, Mizutani-Koseki Y, Kanno M, Taniguchi M, Vidal M, Alkema M, Berns A, Koseki H. Mice doubly deficient for the Polycomb Group genes Mel18 and Bmi1 reveal synergy and requirement for maintenance but not initiation of Hox gene expression. Development, 2001; 128: 1587-97. [327] Siegel R, Desantis C, Jemal A. Colorectal cancer statistics, 2014. CA Cancer J Clin, 2014; 64: 104-17. [328] Dean NM, Bennett CF. Antisense oligonucleotide-based therapeutics for cancer. Oncogene, 2003; 22: 9087-96. 129

Appendix This appendix lists all of the peer-reviewed publications to which I have participated, including manuscripts that were either published, accepted, currently in submission, or prepared for submission during my degree. 1- Vadnais C, Davoudi S, Afshin M, Harada R, Dudley R, Clermont PL, Drobetsky E, Nepveu A. CUX1 transcription factor is required for optimal ATM/ATR-mediated responses to DNA damage. Nucleic Acids Res. 2012 May;40(10):4483-95. 2- Vadnais C, Awan AA, Harada R, Clermont PL, Leduy L, Bérubé G, Nepveu A. Long- range transcriptional regulation by the p110 CUX1 homeodomain protein on the ENCODE array. BMC Genomics. 2013 Apr 16;14:258. 3- Crea F, Clermont PL, Mai A, Helgason CD. Histone Modifications, Stem Cells, and Prostate Cancer. Curr Pharm Des. 2014;20(11):1687-97. 4- Clermont PL, Crea F, Helgason CD. Chapter 22: Trithorax Genes in Prostate Cancer. In Advances in Prostate Cancer (ISBN 978-953-51-0932-7; edited by Gerhard Hamilton), InTech, 2013. 5- Crea F, Clermont PL, Parolia A, Wang Y, Helgason CD. The Non-Coding Transcriptome as a Dynamic Regulator of Cancer Metastasis. Cancer Met Rev. 2014 Mar;33(1):1-16. 6- Clermont PL, Sun L, Crea F, Thu KL, Zhang A, Parolia A, Lam WL, Helgason CD. Genotranscriptomic meta-analysis of the Polycomb gene CBX2 in human cancers: initial evidence of an oncogenic role. Br J Cancer. 2014 Oct 14;111(8):1663-72. 7- Parolia A, Crea F, Xue H, Wang Y, Mo F, Ramnarine V, Liu H, Lin D, Saidy N, Clermont PL, Cheng H, Collins C, Wang Y, Helgason CD. The long non-coding RNA PCGEM1 is regulated by androgen receptor activity in vivo. Mol Cancer. 2015 Feb 21;14(1):46. 8- Clermont PL, Lin D, Crea F, Wu R, Xue H, Wang Y, Thu KL, Lam WL, Gout PW, Wang Y, Helgason CD. Polycomb-Mediated Silencing in Neuroendocrine Prostate Cancer. Clinical Epigenetics (in press) 9- Clermont PL, Parolia A, Liu HH, Helgason CD. DNA methylation at enhancer regions: novel avenues for epigenetic biomarker development. Frontiers in Bioscience (in press)

130

10- Crea F, Di Paolo A, Liu HH, Polillo M, Clermont PL, Guerrini F, Ciabatti E, Ricci F, Baratè C, Fontanelli G, Barsotti S, Morganti R, Danesi R, Wang Y, Petrini M, Galimberti S, Helgason CD. Polycomb genes are associated with response to imatinib in chronic myeloid leukemia. Epigenomics (in press) 11- Clermont PL, Crea F, Chiang YT, Lin D, Zhang A, Parolia A, Wu R, Xue H, Wang Y, Sun L, Wang Y, Helgason CD. Identification of the Polycomb gene CBX2 as a novel drug target in advanced prostate cancer. (in preparation) 12- Sun X, Qu S, Clermont PL, Helgason CD, Jiao W, Wang Y. The molecular profile of centromere protein A (CENPA) is associated with human cancer progression and aggressiveness (in preparation) 13- Clermont PL, Thu KL, Lam WL, Helgason CD. Elevated CBX2 expression correlates with prognosis and smoking status in non-small cell lung cancer (in preparation) 14- Clermont PL, Crea F, Helgason CD. Targeting CBX proteins in human cancers. (in preparation)

131