Reversible inhibition of lysine-specific demethylase 1 is a novel therapeutic strategy for solid tumors

Cancer is a genomic disease driven by interplay between genetic and epigenetic factors. While genetic mutations are irreversible events, epigenetic regulation is dynamic and reversible, and small molecule blockade of the epigenetic machinery has shown clinical benefit in hematological malignancies. However, the promise of epigenetic therapy has yet to be realized in solid tumors due do limited efficacy and elevated risk of toxicity. Development of potent and specific inhibitors targeting the histone methylation machinery shows promise in tailoring epigenetic therapy for a specific malignancy and decreasing the risk of off-target effects. One such target of interest is the histone lysine-specific demethylase 1 (LSD1). Several solid malignancies show upregulation of LSD1 associated with an aggressive clinical course. Validation of LSD1 as a target has been limited by poorly potent and nonspecific tool compounds, hindering evaluation in in vivo models of disease. This work describes the discovery of a novel potent, specific, and reversible series of LSD1 inhibitors. The identified lead compound, HCI2509, is a noncompetitive inhibitor with nanomolar affinity for LSD1. HCI2509 impaired cell viability across several human cancer cell lines, with both Ewing sarcoma and endometrial cancers showing particularly potent responses. Ewing sarcoma is a rare and aggressive pediatric malignancy characterized by by the chromosomal translocation-derived EWS/ETS fusion proteins. EWS/ETS fusions act iv as oncogenic transcription factors and facilitate cellular reprogramming through the activation of oncogenes and repression of tumor suppressors. Treatment with HCI2509 reverses both EWS/ETS-mediated transcriptional activation and transcriptional repression, and leads to apoptotic cell death in Ewing sarcoma cells. Notably, HCI2509 shows single-agent efficacy in xenograft models of Ewing sarcoma and represents a new therapeutic strategy for this devastating disease. HCI2509 also shows single-agent efficacy in a xenograft model of Type II endometrial carcinoma. Cases of Type II endometrial carcinoma comprise 11% of the incidence and 48% of the deaths due to endometrial cancer annually, such that new therapies are needed for this aggressive subtype. Reversible LSD1 inhibition was associated with tumor regression in an orthotopic model of this disease. These results demonstrate the promise of targeting the histone methylation machinery, specifically LSD1, as a therapeutic strategy for solid tumors.

REVERSIBLE INHIBITION OF LYSINE-SPECIFIC DEMETHYLASE 1 IS A NOVEL THERAPEUTIC STRATEGY FOR SOLID TUMORS by Emily Rose Theisen A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Pharmaceutics and Pharmacuetical Chemistry The University of Utah May 2015 Copyright © Emily Rose Theisen 2015 All Rights Reserved The University of Utah Graduate School STATEMENT OF DISSERTATION APPROVAL The dissertation of Emily Rose Theisen has been approved by the following supervisory committee members: , Chair Sunil Sharma 10/22/2014 Date Approved , Member Margit Janat-Amsbury 10/23/2014 Date Approved , Member James Herron 10/30/2014 Date Approved , Member David W. Grainger 10/25/2014 Date Approved , Member Hamid Ghandehari 10/23/2014 Date Approved and by David W. Grainger the Department/College/School of , Chair/Dean of Pharmaceutics and Pharmaceutical Chemistry and by David B. Kieda, Dean of The Graduate School. ii ABSTRACT Cancer is a genomic disease driven by interplay between genetic and epigenetic factors. While genetic mutations are irreversible events, epigenetic regulation is dynamic and reversible, and small molecule blockade of the epigenetic machinery has shown clinical benefit in hematological malignancies. However, the promise of epigenetic therapy has yet to be realized in solid tumors due do limited efficacy and elevated risk of toxicity. Development of potent and specific inhibitors targeting the histone methylation machinery shows promise in tailoring epigenetic therapy for a specific malignancy and decreasing the risk of off-target effects. One such target of interest is the histone lysine-specific demethylase 1 (LSD1). Several solid malignancies show upregulation of LSD1 associated with an aggressive clinical course. Validation of LSD1 as a target has been limited by poorly potent and nonspecific tool compounds, hindering evaluation in in vivo models of disease. This work describes the discovery of a novel potent, specific, and reversible series of LSD1 inhibitors. The identified lead compound, HCI2509, is a noncompetitive inhibitor with nanomolar affinity for LSD1. HCI2509 impaired cell viability across several human cancer cell lines, with both Ewing sarcoma and endometrial cancers showing particularly potent responses. Ewing sarcoma is a rare and aggressive pediatric malignancy characterized by by the chromosomal translocation-derived EWS/ETS fusion proteins. EWS/ETS fusions act iii as oncogenic transcription factors and facilitate cellular reprogramming through the activation of oncogenes and repression of tumor suppressors. Treatment with HCI2509 reverses both EWS/ETS-mediated transcriptional activation and transcriptional repression, and leads to apoptotic cell death in Ewing sarcoma cells. Notably, HCI2509 shows single-agent efficacy in xenograft models of Ewing sarcoma and represents a new therapeutic strategy for this devastating disease. HCI2509 also shows single-agent efficacy in a xenograft model of Type II endometrial carcinoma. Cases of Type II endometrial carcinoma comprise 11% of the incidence and 48% of the deaths due to endometrial cancer annually, such that new therapies are needed for this aggressive subtype. Reversible LSD1 inhibition was associated with tumor regression in an orthotopic model of this disease. These results demonstrate the promise of targeting the histone methylation machinery, specifically LSD1, as a therapeutic strategy for solid tumors. iv This work is dedicated to my parents, Timothy O. Theisen and Barbara A. Filip, my siblings, Jeffrey Martin and Frances Claire, and my grandparents, Gertrude and Robert Theisen and Elaine and Lambert Lasecki, for everything they have given and continue to give in love. v "Nothing in life is to be feared, it is only to be understood. Now is the time to understand more, so that we may fear less." Marie Curie as quoted in Our Precarious Habitat, Melvin A Bernarde, 1973 vi TABLE OF CONTENTS ABSTRACT………………………………………………………………………. iii LIST OF FIGURES………………………………………………………………. ix LIST OF TABLES………………………………………………………………… xii LIST OF ABBREVIATIONS...…………………………………………………… xiv ACKNOWLEDGEMENTS………………………………………………………. xvii Chapters 1. INTRODUCTION AND BACKGROUND…………………………………... 1 1.1 Introduction……...……………………………………………………. 1.2 Summary of Dissertation......…………………………………………. 1.3 Background………...…………………………………………………. 1.4 Study Rationale and Objectives………….…………………………… 1.5 References…………….………………………………………………. 1 2 4 29 32 2. HIGH THROUGHPUT VIRTUAL SCREENING IDENTIFIES NOVEL N'(1-PHENYLETHYLIDENE)-BENZOHYDRAZIDES AS POTENT, SPECIFIC, AND REVERSIBLE INHIBITORS OF LSD1 ..………………... 54 2.1 Abstract…..……...……………………………………………………. 2.2 Introduction……...……………………………………………………. 2.3 Results……………...…………………………………………………. 2.4 Discussion..………...…………………………………………………. 2.5 Conclusions………….………………………...……………………… 2.6 Experimental Section………….………….…………………………… 2.7 References…………….………………………………………………. 2.8 Supplementary Materials...……………………………………………. 55 55 56 62 63 63 66 68 3. REVERSIBLE LSD1 INHIBITION INTERFERES WITH GLOBAL EWS/ETS TRANSCRIPTIONAL ACTIVITY AND IMPEDES EWING SARCOMA TUMOR GROWTH…………………………………... vii 88 3.1 Abstract……...……………………………………….………………. 89 3.2 Introduction……...………………......….………….………………… 89 3.3 Translational Relevance.….………….………………………………. 90 3.4 Materials and Methods………………...…………...…………………. 90 3.5 Results…....………...…………………………………………………. 91 3.6 Discussion...………….………………………...……………………… 98 3.7 References…………….………………………………………………. 101 4. REVERSIBLE INHIBITION OF LYSINE SPECIFIC DEMETHYLASE 1 IS A NOVEL ANTITUMOR STRATEGY FOR POORLY DIFFERENTIATED ENDOMETRIAL CARCINOMA………...................... 4.1 Abstract…..……...……………………………………………………. 4.2 Background……...………………......….………….…………………. 4.3 Methods……………………..….……...…………...…………………. 4.4 Results…....………...…………………………………………………. 4.5 Discussion...………….………………………...……………………… 4.6 Conclusions…………….…………………………………………..…. 4.7 References…………….………………………………………………. 115 116 116 117 118 122 125 126 5. CONCLUSIONS AND OUTLOOK……...…………………………………... 136 5.1 Conclusions……...……………………………………………………. 5.2 Future Studies...…………………………………………….…………. 5.3 Outlook………...……………………………………….….….………. 5.4 References…………….………………………………………………. 136 142 150 151 Appendices A. PURIFICATION OF LYSINE-SPECIFIC DEMETHYLASE 1...…………... 158 B. IN-HOUSE CELL LINE SCREEN AND XCELLIGENCE PROFILING…... 163 C. PHARMACOKINETIC MEASUREMENTS IN MICE AND RATS..……... viii 169 LIST OF FIGURES 1.1 Histone methylation machinery ...……………………………………………. 50 1.2 The crystal structure of lysine-specific demethylase 1..……………………… 51 1.3 Catalytic oxidative demethylation of histone H3 lysine 4 by LSD1…………. 52 1.4 Other proteins bind LSD1 through molecular mimicry …………….……..…. 53 2.1 Mode of binding of compound 12 in complex with LSD1...………….……… 58 2.2 MAO activity of select compounds……….….…………………………….…. 59 2.3 Compound 12 reversibly inhibits the activity of LSD1..……………………… 59 2.4 Derivative melt curves of LSD1 in the presence of DMSO, compound 12, compound 13, and TCP………………………………………………………... 60 2.5 LSD1 kinetics with multiple concentrations of compound 12………………… 60 2.6 Compounds with biochemical activity against LSD1 show in vitro EC50s clustered near 1 µM…………………………………………………………… 61 2.7 VCaP cells treated with compound 12 show a dose-dependent increase in H3K9 dimethylation…………………………………………….…….….…… 61 2.8 Scheme 1. General Procedure for the Synthesis..……..………………………. 63 S2.1 Binding site model and definition of active site of LSD1 structure generated from PDB ID 2Z5U.………………….…………..…………………………… 85 S2.2 Flow diagram for the virtual ligand screening (VLS) using ICM-VLS, Schrodinger workflow GOLD programs……………………………………… 85 S2.3 Complete reaction schemes for compounds 11-22.…………………………… 86 S2.4 LC-MS data for compound 12 (96% purity)…………………………………... 87 3.1 Global EWS/FLI transcriptional activity is disrupted by HCI2509…..….……. 92 ix 3.2 Morphological changes in A673 with HCI2509 treatment .………………..…. 94 3.3 Mechanism of action of HCI2509 in vitro. .…………………………………. 95 3.4 Regulation of HMOX1………………….…………………………………… 97 3.5 HCI2509 activity in vivo.………….………………………………………… 99 3.6 Model for HCI2509 mechanism of action in Ewing sarcoma…..….…..……. 100 S3.1 Transcriptional profiling of HCI2509 in A673 and TTC-466..……………… 104 S3.2 Morphological changes with HCI2509 treatment…………………………… 109 S3.3 Effects of HCI2509 on transformation, methylation, and apoptosis …....…... 111 S3.4 Regulation of HMOX1 in Ewing sarcoma…………………………………... 112 S3.5 Tumor volume, body weight and blood counts ……………………………… 114 4.1 HCI2509 impairs cell viability, proliferation and transformation in Type II EC cell lines.…………..................................................................................... 119 4.2 Treatment with HCI2509 causes changes in global histone methylation and induces LSD1 target genes …........................................................................... 120 4.3 Dose-dependent cell cycle perturbation in Type II EC cell lines with HCI2509 treatment..…………………………………………………………. 121 4.4 HCI2509 induces apoptotic cell death…..….….……………………………. 123 4.5 HCI2509 treatment causes tumor regression in vivo.….……………………. 124 S4.1 Time course evaluation of cell cycle perturbations caused by HCI2509 treatment…………………………………………………………………….. 128 S4.2 TUNEL assay replicates and controls…..…………………………………… 131 S4.3 In depth xenograft model analysis……….…….……………..….………….. 134 5.1 The effects of different classes of LSD1 inhibitors on EWS/FLI targets...…. 153 5.2 HCI2509 decreases metastasis in nude rat model of Ewing sarcoma..……… 155 5.3 Potential synergy between HCI2509 and temozolomide in vivo……………. 156 x 5.4 HCI2509 does not sensitize cell to treatment with medroxyprogesterone 17-acetate (MPA)..………………………………………………...………… 157 A.1 Chromatography tracking LSD1 purfication……………………..…………. 162 A.2 Purified protein is active……………………..…………………….……..…. 162 B.1 xCelligence screen of Ewing sarcoma cell lines …………………………..... 167 C.1 Typical standard curves to quantitate HCI2509 by LC-MS/MS..…………… 177 C.2 Plasma concentration-time curves for HCI2509 in mice as determined by LC-MS/MS………………………………………………………….…....….. 178 C.3 Plasma concentration-time curves for HCI2509 in mice as determined by LC-MS/MS - semilog…………………………………………….……..…… 178 C.4 Plasma concentration-time curves for 40 mg/kg HCI2509 in mice as determined by LC-MS/MS………………………………………….…....….. 179 C.5 Plasma concentration-time curves for 40 mg/kg HCI2509 in mice as determined by LC-MS/MS - semilog……….…………………………..…… 179 xi LIST OF TABLES 1.1 FDA-approved epigenetic therapies…………..…………………….………… 48 1.2 Histone methylation is globally misregulated in cancer ….…………….…… 48 1.3 Altered expression of KDMs in cancer.………………………………………. 49 2.1 Commercially available highly-ranked hits from 121 screened compounds.… 57 2.2 Synthesized compounds and their biochemical activity against LSD1………. 58 2.3 Off-target panel for compound 12…………………………………………… 59 2.4 Melting temperatures as determined by DSF.….….…………………………. 60 2.5 Summary of Michaelis-Menten curve fits………………………………….… 61 2.6 Compound 12 inhibits proliferation in several cell lines in vitro.……………. 61 2.7 In vitro growth inhibition of compound panel in T-47D cells………………… 61 S2.1 Docking scores of compounds 1-10...……...………..………………………… 73 S2.2 Commercially available LSD1 hits (111) from the list of 121 compounds selected………...…………..…………………………………………………... 74 S2.3 Tanimoto similarity coefficients comparing compound 12 and known LSD1 inhibitors from Chart 1…………………...………………..…………………... 83 S2.4 Off-target inhibition assay results………...…………..……………………….. 84 S2.5 Different model fits for enzyme kinetics...……..……………………………... 84 S3.1 Primer sequences for qRT-PCR analysis...…..………………………………... 103 B.1 A 96-cell line panel…..…………………..…....………….…..………………. 164 C.1 Pharmacokinetic parameters for 5 mg/kg HCI2509 dosed as solution IV...….. 174 xii C.2 Pharmacokinetic parameters for 20 mg/kg HCI2509 dosed as solution PO…. 175 C.3 Pharmacokinetic parameters for 50 mg/kg HCI2509 dosed as suspension PO…………………………………………………………………………….. 176 xiii LIST OF ABBREVIATIONS ALL Acute lymphocytic leukemia AML Acute myeloid leukemia AOD Amine oxidase domain BHC80 BRAF35-HDAC complex protein 80 CETSA In-cell thermal shift assay ChIP Chromatin immunoprecipitation CK2 Protein kinase 2 Co-REST REST corepressor 1 CTCL Cutaneous T-cell lymphoma CYP Cytochrome P450 DAPI 4',6-diamidino-2-phenylindole DAVID Database for Annotation, Visualization and Integrated Discovery DNMT DNA methyltransferase DOT1L DOT1-like histone H3K79 methyltransferase DSF Differential scanning fluorimetry EC Endometrial carcinoma EC50 Concentration at 50% effect EMT Epithelial-to-mesenchymal transition EZH2 Enhancer of zeste homolog 2 xiv FAD Flavin adenine dinucleotide FAK Focal adhesion kinase Gfi-1 Growth factor independent 1 GSEA Gene set enrichment analysis H2a Histone 2a H2b Histone 2b H2O2 Hydrogen peroxide H3 Histone H3 H3K16 Histone H3 lysine 16 H3K27 Histone H3 lysine 27 H3K36 Histone H3 lysine 36 H3K4 Histone H3 lysine 4 H3K79 Histone H3 lysine 79 H3K9 Histone H3 lysine 9 H3T6 Histone H3 threonine 6 H4 Histone H4 H4K20 Histone H4 lysine 20 HAT Histone acetyltransferase HDAC Histone deacetylase 1 hERG Human ether-à-go-go HMOX1 Heme oxygenase 1 HTVS High throughput virtual screening IGF-1 Insulin-like growth factor 1 xv JmjC Jumonji-C-domain containing KDM Lysine demethylase KMT Lysine methyltransferase LSD1 Lysine specific demethylase 1 MAO Monoamine oxidase MDS Myelodysplastic syndrome me1 Monomethyl me2 Dimethyl me3 Trimethyl MSCV Murine stem cell virus MTA Metastasis-associated 1 NuRD Nucleosome remodeling and deacetylase complex NURF Nucleosome remodeling factor complex PAO Polyamine oxidase PKCα Protein kinase α PRC2 Polycomb repressive complex 2 SAR Structure-activity relationship SNAG Snail/Gfi SWIRM Swi3p, Rsc8p and Moira domain TCP Tranylcypromine TGFβ Tumor growth factor β VS Virtual screen xvi ACKNOWLEDGEMENTS I would also like to thank my dissertation committee: David Grainger, James Herron, Margit Janat-Amsbury, and Hamid Ghandehari. I have been honored to be mentored by Sunil Sharma in the laboratory he has established at the Huntsman Cancer Institute. To Sunil Sharma: Thank you for the opportunities and guidance that you have continued to provide me, and for seeing this project through to completion. I have been fortunate to work with Savita Sankar, Snehal Gajiwala, Jared Bearss, Raffaella Soldi, Barbara Graves, Stephen Lessnick, Michael Engel, and David Bearss, as well as other exceptional members of the CIT, Graves, and Lessnick Labs. I would like to express the deepest gratitude to my brother Jeffrey in the compilation of this manuscript. I must also acknowledge Charles Fehl, Jason Tanner, Holly Grainger, Phil Wilkes, Alana Jonat, Jessica McCombs, Abood Okal, Andrew Dixon, and Carol Lim for tending the light. Heartfelt thanks to all of my family and friends for their enduring support. I would also like to recognize Dr. You Han Bae for his enthusiasm for this area of research and his support in showcasing it to the department. This work was supported in part by the American Foundation for Pharmaceutical Education, Women in Cancer Research, and NCI/NIH Grants P30 CA042014 (to Huntsman Cancer Institute). We also acknowledge the use of the DNA sequencing and genomics core facilities at the Huntsman Cancer Institute. xvii 1 CHAPTER 1 INTRODUCTION AND BACKGROUND 1.1 Introduction While genetic information encoded in DNA contains the program for every cell, cell- and tissue-specific programming required for normal physiological function are regulated by a dynamic array of epigenetic and transcriptional machinery (1). This epigenomic level of regulation allows interaction between one's environment and one's genes and can result in heritable patterns of gene expression in the absence of genetic mutation. Cancer is a disease of the whole genome, characterized by both genetic aberrations and epigenomic misregulation driving the malignant phenotypes comprehensively described by Hanahan and Weinberg (2). Worldwide, the incidence of cancer is projected to double from 12.7 million new cases in 2008 to 21.4 million new cases, and 13.5 million deaths, by 2030 (3). Given that genomes and environments are singular for each patient, each individual malignancy is unique, such that universally efficacious treatment options are nonexistent. However, where genetic mutations are irreversible, the dynamic nature of the epigenetic machinery is susceptible to pharmacological intervention. Epigenetic enzymes which fuel oncogenic misregulation are emerging therapeutic targets. 2 The histone demethylase lysine specific demethylase 1 (LSD1) is one such target and is either upregulated in or critically important to the development and progression of various cancers, including neuroblastoma (4), acute myeloid leukemia (5), and prostate cancer (6). However, having only been discovered in 2004, the complicated biological mechanisms which regulate LSD1 function in healthy and diseased states are not yet fully elucidated. Moreover, the available tool compounds suffer from both poor potency and specificity, complicating interpretation of reported results. Hence, we pursued a drug discovery program to identify potent, specific and reversible LSD1 inhibitors to use as tool compounds to preclinically screen the viability of LSD1 inhibition as a therapeutic strategy for solid tumors. The discovery of such a series will further enable detailed investigation of the biological role of LSD1 in various cancers, and diffentiate mechanisms that are common between malignancies and those that are more diseasespecific. While this work focuses primarily on Ewing sarcoma and endometrial cancer, the compound series identified may provide therapeutic benefit in a diverse array of cancers for which LSD1 overexpression has been reported or LSD1 biology implicated. 1.2 Summary of this Dissertation This chapter will provide an overview of the rapidly evolving field of cancer epigenetics, describing both the clinical challenges encountered to date by FDA-approved epigenetic therapies and the ways in which second generation epigenetic targeted therapies address these, focusing specifically on the challenges and promises of targeting LSD1. Additionally, this chapter will introduce the rationale for the studies described herein and identify the objectives met in Chapters 2-4. Chapter 2 describes the initial 3 discovery, hit-to-lead optimization, and biochemical characterization of the N'-(1phenylethylidene)-benzohydrazide series of LSD1 inhibitors that are the subject of the remainder of the dissertation. Chapters 3 and 4 describe validation of the activity of LSD1 inhibition in two solid tumors of interest, Ewing sarcoma and Type II endometrial carcinoma. Chapter 3 investigates the unique activity of the lead compound, HCI2509, in Ewing sarcoma, focusing both on characterizing the effects of HCI2509 on the molecular drivers of Ewing sarcoma in vitro and validating single-agent efficacy in xenograft models in vivo. Ewing sarcoma is driven solely by the chromosomal translocation leading to expression of an EWS/ETS fusion oncoprotein and transcription factor, lacking additional genomic aberrations (7). Subsequent transcriptional reprogramming relies heavily on misregulation of the transcriptional and epigenetic machinery, presenting an ideal proofof-concept system for in vivo studies. Chapter 4 moves beyond this to Type II endometrial cancer, which primarily occurs in adulthood, is clinically aggressive, and is driven by a more diverse and complex set of genetic, epigenetic, and environmental factors. In this chapter, studies describe both the in vitro anticancer effects of HCI2509 in multiple cell lines and the in vivo antitumor efficacy of HCI2509 in an orthotopic model of Type II endometrial cancer. Chapter 5 provides conclusions while also outlining future work suggested by the results described herein. Additionally, three appendices include data which were critical to the completion of these studies, but not published, including protein purification, cell-based screening results, and pharmacokinetic studies. 4 1.3 Background 1.3.1 Epigenetics and Cancer Pathogenesis Epigenetics broadly refers to four layers of dynamic regulation within the nucleus: DNA methylation, histone posttranslational modifications, nucleosome positioning, and the expression of various noncoding RNAs. Mounting evidence implicates all four levels in the development and maintenance of oncogenic gene expression programs characteristic of cancer. However, the roles that nucleosome positioning and noncoding RNAs play in cancer are outside the scope of this work. 1.3.1.1 DNA Methylation DNA methylation in mammals occurs only on cytosine bases that are 5' linked to guanosine (CpG) (8). Methylation is catalyzed by DNA methyltransferases (DNMT) DNMT1, DNMT3a, and DNMT3b. DNMT1 acts only on hemimethylated DNA and is responsible for the maintenance of DNA methylation patterns during replication, while both DNMT3a and 3b are capable of de novo DNA methylation (9,10). Cytosine is largely underrepresented in the genome with the exception of short regions (0.5-4 kb) called CpG islands, which are GC enriched (8,11). CpG islands are located at the proximal promoter region of roughly 50% of genes in the human genome (11). DNA methylation at the promoter functions to silence the downstream gene. As a cell progresses through normal development, increased promoter methylation at particular loci silences expression of genes which are lineage-inappropriate, reinforcing cellular differentiation (10). In cancer, global genomic demethylation is observed, with increased methylation occurring at the promoter regions of tumor suppressor genes (12,13). Many 5 of the silenced tumor suppressors are known to be frequently mutated, like MGMT, CDKN2A, MLH1, and BRCA1, suggesting both genetic and epigenetic routes can lead to the same oncogenic phenotype (14-18). In addition to the promoter hypermethylation observed across human neoplasms, DNA methylation can itself promote C to T mutations through spontaneous hydrolytic deamination (19). This effect is not insignificant, up to 50% of the inactivating mutations of the tumor suppressor TP53 occur at methylated cytosines (20). Additionally, cancer cells display genomic hypomethylation outside of the proximal promoter regions (21,22). This global loss of methylation is thought to contribute to genomic instability and structural alterations of chromosomes (23). Overall, changes in DNA methylation were the first characterized epigenetic phenomena observed in cancer, and inspired the development of pharmacological agents targeting the DNMTs, the first FDA-approved epigenetic therapies for cancer, discussed in section 1.3.2. 1.3.1.2 Histone Modifications In order to fit the whole genome into the nucleus, eukaryotic cells utilize a packing scheme in which 147 bp of DNA is wrapped around an octameric complex containing two each of histone 2a (H2a), histone 2b (H2b), histone 3 (H3), and histone 4 (H4) (1). This DNA-histone complex comprises the nucleosome, which is further compacted into chromatin. Tightly compacted chromatin is termed heterochromatin, and genes located here are repressed or silenced, whereas euchromatin has an open conformation allowing the transcriptional machinery access to promote active gene expression (1). Conserved residues on histones, often found on the unstructured and 6 lysine-rich N-terminus, are subject to a variety of posttranslational modifications including methylation, acetylation, phosphorylation, SUMOylation, and ubiquitinylation, such that histone modification is more diverse and dynamic than DNA methylation (24). Particular modifications act combinatorially such that various patterns of modifications interact with DNA- and chromatin-binding proteins to define chromatin status and recruit transcriptional machinery. Using the N-terminal tail of H3 as an example, heterochromatin is marked by increased H3K9 and H3K27 trimethylation as well as DNA methylation, whereas euchromatin is characterized by acetylation at H3K9 and H3K16 (25). While acetylation of histone residues directly affects gene accessibility through increasing the strength of the DNA-histone electrostatic repulsion, histone methylation plays an important, but more complex, role in transcriptional regulation. Notably H3K9 and H3K27 methyl marks are repressive, where methylation at H3K4 is permissive and commonly found associated with the proximal promoter of actively transcribed genes (24,25). The suite of histone modifications are written, erased, and read by a diverse complement of biomolecules, including proteins and nucleic acids. Often complexes possessing opposing functions are found co-localized in the nucleus, facilitating contextdependent dynamism (26-28). The existence of bivalent domains, containing both activating and repressive histone marks, further undrescores the importance of epigenomic dynamism (29,30). Using histone acetylation as an example, acetyl marks are written by histone acetyltranferases (HATs), including the p300/CBP, GNAT, and MYST subfamilies, and erased by histone deacetylases (HDACs) (31). Acetylated histone lysines are recognized by proteins which contain a structural motif termed a bromodomain. 7 Bromodomain-containing proteins include chromatin remodelers, HATs, histone methyltransferases, and transcriptional co-activators and often possess another histone reader domain, such as the methyllysine-specific PHD finger, to facilitate combinatorial recognition of the chromatin state (32). Aberrant global histone acetylation patterns are broadly observed in cancer (32), and many studies have implicated HATs, HDACs, and bromodomains in malignant epigenetic misregulation. For example, HATs are present in multiple oncogenic fusion proteins (33), the most well known being MOZ-TIF2 in aggressive leukemia (34,35). Somatic mutations are also documented in HATs, such as those documented in p300/CBP, in various solid tumors and hematological malignancies (36,37). While somatic mutations are not observed as commonly in HDACs, levels of HDAC expression are often altered in cancer, with overexpression correlating with aberrant silencing of tumor suppressor genes and impaired apoptosis (38,39). With respect to histone readers, mutations in bromodomain-containing proteins have also been documented in acute lymphocytic leukemia (ALL), midline carcinoma, renal carcinoma, and breast cancer (31). 1.3.2 Implications of Cancer Epigenetics for Therapy These observations illustrate an emerging paradigm, whereby genetic mutations and epigenetic factors are two sides of the same coin. It should be noted that discoveries analagous to those described for HATs implicate mutations in DNMTs, the histone lysine methylation machinery, nucleosome remodelers, and noncoding RNAs in carcinogenesis. In fact, mutations in epigenetic regulators are now documented in almost all human malignancies (40-51). Alterations in the epigenetic regulatory machinery lead to genomic 8 instability, which further promotes mutations in other tumor suppressors and epigenetic proteins, further compounding epigenetic misregulation, and so on. However, unlike genetic drivers of cancer, epigenetic modifications are often reversible, presenting opportunities to pharmacologically disrupt and reverse malignant programming. Understanding the interplay and intersection between genetic and epigenetic factors is of critical importance to determine the most appropriate and efficacious way to therapeutically target genomic misregulation in cancer. Most importantly, better tools are needed to determine which epigenetic players represent oncogenic drivers in a given malignancy, such that small-molecule blockade disproportionately affects the cancer cell while leaving required epigenetic and genomic regulatory mechanisms intact in normal tissue. To date, DNMT inhibitors and HDAC inhibitors, are FDA-approved and their clinical use over the past decade has enhanced our knowledge about the promises and challenges of epigenetic therapy in the clinic (Table 1.1). 1.3.2.1 FDA-Approved Epigenetic Therapies 1.3.2.1.1 DNA Methyltransferase Inhibitors The DNA demethylating agents, decitabine and 5-azacytidine, were first designed as cytotoxic chemotherapy in the 1960s (52,53), and their activity against DNMTs was only established 20 years later (54). 5-azacytidine gained FDA-approval in 2004 for the treatment of myelodysplastic syndrome (MDS), while decitabine was approved in 2006 for MDS and acute myeloid leukemia (AML) (55-58). Their approval was dependent upon drastic reductions in dose from the maximally tolerated dose, such that dose deescalation improved both tolerability and shifted the mechanism of action from cytotoxic 9 activity to DNMT inhibition (59). At low doses, the nucleotide analogue is incorporated into DNA and acts as a suicide inhibitor for the DNMT, triggering its degradation (6063). Pharmacokinetic studies of the doses used clinically shows nanomolar plasma concentrations, and at these exposures minimal cell death is observed in vitro (64). However, even after 1 exposure, increased expression of immunomodulatory and proapopotic genes, as well as whole-genome demethylation were observed and coincided with decreased clonogenicity and tumorgenicity (64). Importantly, findings of durable cellular reprogramming seem to also apply to tumor stem-like cells (64). This suggests epigenetic therapies may be able to target this population of cells, which is typically resistant to multiple other treatment modalities and drives the metastases and disease recurrence that often prove fatal. These laboratory results are consistent with clinical observations of patients treated with DNMTs. A large proportion of the patient population treated with 5-azacytidine for AML or MDS need months before a response becomes apparent, perhaps due to long-term exhaustion of stem-like cells (65). Additionally, ~48% of high-risk MDS patients who prolonged DNMTi treatment duration beyond their initial response improved the magnitude of response with subsequent therapies (65). The clinical use of DNMT inhibitors has greatly improved therapeutic options for patient with both MDS and AML. 1.3.2.1.2 Histone Deacetylase Inhibitors The second class of FDA-approved epigenetic therapies are the histone deacetylase inhibitors. Both vorinostat and romidepsin were approved for the treatment of cutaneous T-cell lymphoma (CTCL), with romidepsin also indicated for the treatment of 10 relapsed peripheral T-cell lymphoma (66-69). Multiple additional HDAC inhibitors are in development, but are outside the scope of this work and are reviewed comprehensively by Lane, et al. (70). Whereas HDAC inhibitors have shown striking responses in CTCL, their value remains largely unproven elsewhere in the clinic, likely due to analogous but less understood differences in dose-dependent mechanisms of action, the discussion of which is beyond the scope of this work. 1.3.2.2 Clinical Challenges Facing Epigenetics In the indications where epigenetic therapy is approved significant clinical benefits have been observed. However, while epigenetic mechanisms play a central role in cancer pathogenesis across malignancies, clinical benefit in the most common solid tumor remains largely unachieved. The difficulty in translating epigenetic insights to clinical benefit stems from our limited understanding of the basic science through to the design and execution of clinical trials. In the laboratory, prior to the advent of next generation sequencing and ensuing flood of genomic data, epigenetics research was heavily biased towards events occurring at the transcription start site. As these research programs were initiated the most obvious and observable phenomena was DNA methylation at silenced gene promoters and the downstream effects on transcription (51). However, cancer manifests at the level of the whole genome as is visible in the nuclei of cancer cells under a microscope. In the new era of "-omics," our understanding of the global epigenomic events leading to cancer is ceretainly growing, however, the detailed mechanisms by which these events occur and how exactly they drive tumorigenesis remain largely undetermined and unexplored. 11 It is clear, however, that misregulation in the epigenome is far-reaching, representing a sort of software glitch that alters expression programs across hundreds of genes in diverse pathways and promotes tumorigenesis. Currently approved therapies clearly can rise to the challenge of targeting the cellular reprogramming. The Peter A. Jones group has largely demonstrated durable reprogramming of cancer cells following long-term low exposure to DNMT inhibitors in cell culture (64). Additionally, the process of "reprogramming" induced pluripotent stem cells from differentiated adult cells is enhanced by DNMT and HDAC inhibitors (71-73). These results buttress the potential for epigenetic therapy to show sweeping effects in solid tumors. However, the sort of "knockdown-rescue" experiments that are required to prove causality and achieve mechanistic insight in this realm are largely beyond our technical prowess. While comprehensive understanding of the mechanisms by which approved therapies act remains elusive, several empirical observations have informed the current paradigm for dose de-escalation in their clinical use. At high doses, both DNMT and HDAC inhibitors show cytotoxic effects, with the more potent on-target effects dominant at low doses (51). HDAC inhibitors are limited in that HDACs are fairly promiscuous enzymes and show diverse function (74,75). In fact, some HDACs are localized to the cytoplasm, such that an analysis of whether the antitumor efficacy seen with HDAC treatment are on- or off-target is largely confounded (74). These types of observations continue to muddy the water. Early trials of epigenetic therapy in solid tumors followed traditional clinical trial design, using dose-escalation to identify the maximally tolerated dose (MTD). In Phase II efficacy testing, the MTD for both HDAC and DNMT inhibitors showed pronounced off- 12 target cytotoxicity with little effect on the epigenetic pharmacodynamic endpoints. In order to evaluate the epigenetic activity of these classes of drugs, the doses needed to be reduced. This dose de-escalation was accompanied by the observation that cellular reprogramming was not apparent in the short-term, and required long-term pharmacodynamic monitoring (51). The high cytotoxic doses likely preclude the reprogramming required for true epigenetic therapy. Thus, clinical translation has been limited to date by suboptimal trial design which fails to account for the low-dose, longterm efficacy expected with epigenetic drugs. As such, innovative trial designs are required to build on the early data in hematological malignancies and establish a new paradigm for epigenetic treatment in solid tumors. This really is early days, as several fundamental parameters remain undefined. For example, the therapeutic window for reprogramming in malignant versus normal tissues, or the length of time and criteria used to assess response, are unknown and yet unstudied. Clinical evaluation of new epigenetic therapeutic strategies may benefit from trial design used in other fields, like translational immunotherapy, to assess these sorts of parameters (76). Encouragingly, these lessons have been learned and dose de-escalation is being tested clinically in solid tumors, including nonsmall cell lung carcinoma, colorectal, and breast cancers and with promising early results (77). 1.3.2.3 Sensitizing Cancer to Other Treatments Further optimization of DNMT and HDAC inhibitor dosing in solid tumors in the clinic will provide an opportunity to validate observations from the laboratory that cellular reprogramming induced by epigenetic targeted agents confers increased cellular 13 sensitivity to other modalities of treatment. This includes hormone therapy, chemotherapy, and immunotherapy either combined in parallel or implemented sequentially (51). Notably, Sharma, et al. (78) observed in vitro the existence of a drugtolerant population of cells in multiple human tumor cell lines. The drug-tolerant phenotype was transient and reversible, and mediated by both IGF-1 receptor signaling as well as chromatin changes, suggesting a role for dynamic epigenetic regulation in the development of drug resistance (78). Inhibition of IGF-1 receptor signaling, HDACs, and the histone lysine demethylase JARID1A ablated this phenotype, suggesting a potential role for epigenetic therapies to augment sensitivity to other systemic anticancer therapies (78). 1.3.3 Emerging Epigenetic Targets While DNMT and HDAC inhibitors provided proof-of-concept for epigenetic therapies in the clinic, insights from the last decade of cancer epigenetics research has uncovered mutations or aberrations in countless other classes of epigenetic regulators including histone mark readers, histone lysine methylation regulators, nucleosome remodelers, and the noncoding RNA machinery. This has resulted in a wave of target validation and drug discovery efforts across academia and industry. Several of these research tracks are now bearing fruit, with several novel classes of epigenetic targeted agents in Phase I and Phase II studies. For each program described herein, the Phase I studies have focused on or are studying a particular malignancy in which the target is an established driver of the disease, either through direct mutation or as a required player in epigenomic misregulation. This underscores the importance of picking the right patient 14 population with clear pharmacodynamic criteria for proof-of-concept studies. Having optimized dosing and pharmacokinetic/pharmacodynamic relationships in these simpler populations, clinical research can move forward with the process of evaluating these agents in diverse patient populations with more complex disease states. This progress will lean heavily on continued insights from both basic and translational studies validating potential biomarkers to define the patient populations most likely to respond to different classes of epigenetic agents. The ultimate goal is to enable personalized epigenetic treatment for each individual malignancy. To date, the most advanced clinical programs are those targeting histone acetylation readers, or bromodomain inhibitors, and those targeting the histone lysine methylation machinery. 1.3.3.1 Targeting Bromodomains Protein-protein interactions are notoriously difficult to target with small molecules, however, a class of inhibitors, exemplified by the molecules JQ-1 and iBET, have been shown to disrupt the interaction of the BET family (BRD2, BRD3, BRD4, and BRDt) of bromodomain proteins with acetylated histone lysine residues. Bromodomain inhibitors represent the first epigenetic agents to target histone posttranslational readers. The bromodomain of BET proteins is highly conserved, plays a critical role in cell cycle progression and transcriptional elongation, and is involved in translocations which drive the fatal NUT-midline carcinoma. In vitro and in vivo studies of the BET inhibitors consistently showed both downregulation of MYC transcript and disruption of the MYC transcriptional program across a wide variety of hematological and solid malignancies, as well as disruption of superenhancer motifs that reinforce MYC and BLC2 expression (79- 15 88). At the time of writing three BET inhibitors programs had initiated clinical trials, with GSK525672 in two Phase I trials for NUT midline carcinoma and relapse or refractory hematological malignancies (89), TEN-010 for advanced solid malignancies or NUT midline carcinoma (90), and CPI-0610 in previously treated and aggressive lymphomas (91). 1.3.3.3 Targeting Histone Lysine Methylation While histones can be methylated at lysine, arginine, and histidine side chains, lysine methylation is the best characterized and disproportionately represents the therapeutic development by targeting histone methylation, so it will be the focus of discussion. Unlike acetylation and phosphorylation, lysine methylation does not alter the charge of the residue. Of the posttranslational modifications, methylation shows the slowest turnover (92) and was originally thought to be irreversible, until the discovery of the first histone lysine demethylase in 2004 (93). Lysine residues can be either mono(me1) (94), di- (me2) (95), or tri-methylated (me3) (96). Methylation at histone H3 lysine 4 (H3K4), lysine 9 (H3K9), lysine 27 (H3K27), lysine 36 (H3K36), lysine 79 (H3K79), and histone H4 lysine20 (H4K20) are the most studied, and a plethora of methyl mark writers, readers, and erasers have been identified which display diverse substrate specificities and allow for nuanced control of histone methylation status (Figure 1.1). Broadly speaking, methylation at H3K4, H3K36, and H3K79 typically correlates with euchromatin, while that at H3K9, H3K27, and H4K20 corresponds to repressive heterochromatin (31). Even more specifically, some methylation states may require stability through mitosis, such as established silencing within heterochromatin, while 16 others depend upon dynamism, so as to facilitate cell differentiation in response to external stimuli. Additionally, different modifications on the same residue may denote specific chromatin states. For example, H3K9me1 is typically associated with active chromatin, while H3K9me3 is associated with repressed genes (31). Moreover, these marks may distribute to different regions spatially, for example, H3K4me1 is found within enhancer regions of the genome, while H3K4me2/3 is enriched at the transcription start sites of actively-transcribed genes (31). This model of complexity between different marks and within methyl marks on the same lysine is supported by the observation that methyl marks at different lysine residues display different turnover rates (97). The dynamics of histone methylation are regulated by histone lysine methyltransferases (KMTs) and histone lysine demethylases (KDMs). KMTs catalyze the addition of methyl marks to lysine from S-adenosylmethionine (94) and fall into one of two families, either the SET-domain containing proteins (98) or DOT1L-like proteins (99). KDMs likewise fall into two classes, either the amine oxidases (94) or jumonji C (JmjC)-domain containing, iron-dependent dioxygenases (100,101). In addition to the complexity by which histone methylation regulates chromatin, many KMTs and KDMs also act upon nonhistone substrates, challenging the interpretation of the biological role for these enzymes in the cell. Like DNA methylation and histone acetylation, histone lysine methylation has been widely implicated in the development of various malignancies both through alterations in levels of expression as well as through mutation of KMTs and KDMs. The complete details of this are beyond the scope of this work, but are reviewed comprehensively by Albert and Helin (102) and Chi et al. (103). Broadly speaking, 17 various malignancies show aberrations in the global levels of histone lysine methyl marks, most commonly hypomethylation, that are associated with poorer survival, worse clinical outcomes, or higher disease recurrence (Table 1.2) (104). While causality remains unestablished, these observations may lead to the development of future biomarkers. At the interface between genetics and epigenetics, genomic studies have also identified several somatic mutations in the histone lysine methylation machinery or chromosomal translocations which involve KMTs or KDMs, further implicating misregulation of histone methylation in oncogenesis (102-104). To date, two KMTs have proven to be tractable targets for the development of pharmacological inhibitors. The first is DOT1L, a KMT with specificity for H3K79 (99). Roughly 5-10% of acute leukemias, particularly infant and relapsed leukemias, present with a chromosomal translocation involving the KMT MLL at 11q23 (105). Loss of the C-terminus of MLL in rearrangements replaces the SET KMT domain with sequences derived from AF4, AF9, AF10, and ENL (105). These domains interact directly with DOT1L to maintain the MLL-r fusion-driven oncogenic transcriptional activity, such that DOT1L is necessary for transcription of key target genes driving leukemogenesis (105). Epizyme recently concluded the dose-escalation portion of their Phase I study of the DOT1L inhibitor, EPZ-5676, and began enrolling for the expansion phase of the trial in December 2013 (106). This was the first histone methyltransferase inhibitor to enter the clinic. The second KMT with drug development programs entering early clinical studies is enhancer of zeste homolog 2 (EZH2). EZH2 is a KMT with substrate specificity for H3K27 and is the catalytic subunit of the polycomb repressive complex 2 (PRC2), 18 promoting gene silencing. EZH2 is a prime example of target complexity when considering histone methylation. EZH2 is observed in numerous cancers including breast (107), prostate (108), lung (109), skin (109), and colon cancer (109), as well as lymphomas (110). B-cell lymphomas have also been shown to contain somatic activating point mutations in EZH2, supporting its role as an oncogene (110). Further buttressing this model, the histone demethylase with substrate specificity for H3K27, UTX, contains inactivating point mutations in a variety of cancers (111). However, loss-of-function EZH2 mutations in MDS have been reported (112), highlighting the context-dependence of a single epigenetic target in a given disease. The second KMT inhibitor to enter the clinic was Epizyme's EPZ-6438, with the dose escalation study still active for patients with advanced solid tumors and relapsed or refractory B-cell lymphoma (113). GlaxoSmithKline has also initiated clinical trials with their EZH2 inhibitor, GSK2816126, in patients with relapsed or refractory diffuse large B cell and transformed follicular lymphoma (114). In addition to KMTs, KDMs are attractive therapeutic targets in various cancers. Beyond genetic aberrations, expression levels of various KDMs are observed across many human malignancies (Table 1.3) (115). The first KDM inhibitor to reach the clinic is GSK2879552, an irreversible inhibitor of lysine specific demethylase 1 (LSD1/KDM1A) with Phase I studies initiated in early 2014 for patients with relapsed or refractory nonsmall cell lung carcinoma (116). LSD1 is the focus of the remainder of this work. 19 1.3.4 Lysine Specific Demethylase 1 Somatic mutations in KDM1A are not observed in cancer, but LSD1 overexpression has been documented in a number of both hematological and solid malignancies and is typically associated with de-differentiation, aggressive biology, and poorer outcomes. Increased levels of LSD1 are a biomarker for aggressive tumor biology and poor prognosis in breast (117) and prostate cancers (118-121). In prostate cancer, the overexpression of LSD1 promotes ligand-independent androgen-receptor-dependent transcription (119,120). LSD1 expression is inversely correlated with differentiation in neuroblastoma, suggesting a role in repressing differentiation (122). Interaction of the transcription factor TAL1 with LSD1 drives hematopoietic differentiation programs, with aberrant function of this axis observed in ~60% of T-cell acute lymphoblastic leukemia (123,124). In acute myeloid leukemia, LSD1 blocks differentiation and perpetuates the cancer stem-cell compartment (125,126). Upregulation of LSD1 has also been observed in bladder (127,128), lung (127), colorectal tumors (127), high grade sarcomas (129,130), and hepatocellular carcinoma (131,132). 1.3.4.1 Discovery, Structure, and Function LSD1 is the main histone demethylase in the cell and comprises an 852 aminoacid flavin adenine dinucleotide (FAD)-dependent amine oxidase, depicted in Figure 1.2 (94). The discovery of this enzyme, conserved from yeast through humans, was the first concrete evidence for dynamic regulation of histone methylation. The first 171 Nterminal residues are unstructured, but appear to act as a tether for interactions with other proteins in chromatin-remodeling complexes (133,134). The majority of conserved 20 residues reside in close proximity to the amine oxidase domain (AOD) and appear to facilitate ligand packing and binding (133,134). The AOD is conserved and homologous to monoamine oxidases (MAO) A and B, as well as polyamine oxidase (PAO) (133,134). Like the MAOs, LSD1 binds FAD in a conserved Rossman fold, however, unlike the MAOs, LSD1 bind FAD noncovalently (133,134). The FAD cofactor binding pocket is a narrow cavity through the center of the enzyme and within this pocket FAD interacts with LSD1 through salt bridges with Arg310 and Arg316. The isoalloxazine ring system is positioned for catalytic activity at the base of the substrate binding pocket near Lys661 (133). The FAD is reduced during the formation of the imine intermediate. Hydrolysis of the imine leaves the demethylated lysine and releases a molecule of formaldehyde, while the FADH- is oxidized to FAD by oxygen, releasing a molecule of H2O2 (Figure 1.3) (135). Lys661 is critical for enzymatic activity by channeling molecular oxygen for FADH- oxidation and recharging the redox potential at the catalytic site (136). Other domains include a Swi3p, Rsc8p and Moira (SWIRM) domain and a tower domain. In most proteins, SWIRM domains interact directly with DNA, though this is not the case for LSD1. The tower domain, required for enzymatic activity (134), is a coiledcoil sequence inserted within the AOD which prominently protrudes as a docking site for additional protein-protein interactions, such as that with Co-REST, a common binding partner for LSD1. Co-REST contains two SANT domains, which confer DNA-binding in place of LSD1's odd SWIRM domain, and are required for functional demethylation of residues in the native nucleosome (133). LSD1 demethylates both mono- and di-methyl marks on H3K4 and H3K9 (94,118). Methylation of H3K4 is associated with gene activation, while H3K9 21 methylation generally denotes gene repression. Due to the imine intermediate formed during demethylation, removal of the trimethyl mark is chemically inaccessible to LSD1 (134). The substrate N-terminal tail of H3 packs tightly into the asymmetric funnelshaped binding pocket of LSD1. The amine terminus of H3Ala1 is bound in a highly electronegative pocket showing hydrogen bonding and electrostatic interactions with Asp555 of LSD1 (137). Lys4 is oriented toward the isoalloxazine ring system to facilitate oxidative attack on the N-CH3 group by flavin (137,138). Residues 1-5 of histone H3 adopt a helical turn, 6-9 are sharply bent, and residue 10-16 are more extended and partially solvent exposed along the rim of the binding pocket (138). Recent structural studies by Baron, et al. (136) show that multiple proteins contain conserved N-terminal sequences homologous to histone H3. These are often transcription factors that may function to hook LSD1 for recruitment to different genomic sites (136). Some examples include SNAI1 (related to morphogenetic events mediating tumor invasiveness), Ovo-like1 (epidermal proliferation and differentiation factor), SCRATCH1 (nervous system specific), gfi1 (a gene repressor involved in hematopoiesis whose expression is regulated by LSD1-containing complexes), and insm1 (insulinomaassociated 1; associated with differentiation of neural and pancreatic precursors; discovered in an neuroendocrine tumor) (136). The SNAI1-LSD1 interaction has been crystallized and shows a similar binding mode to histone H3 (Figure 1.4) (136). Other binding partners dock on the tower domain; for example, another SANT-domaincontaining protein called MTA2 has also been shown to recruit LSD1 to chromatin as a member of the nuclear remodeling and deacetylase (NuRD) repressive complex, using a mechanism analogous to Co-REST (139). LSD1 has also been shown to be recruited by 22 the noncoding RNA HOTAIR into larger complexes containing PRC2 proteins (140). Combinatorial regulation is commonly used to achieve the complex epigenetic functions that are required for differentiation and maintenance of cell- and tissue-specific gene expression programs. In addition to several interaction partners with H3homologous N-terminal tails, additional posttranslational histone modifications alter binding affinity of LSD1 for the H3 substrate, such that the histone code can drastically help or hinder LSD1 activity. At least 16 amino acids are required for functional demethylation, though the 21 amino acid substrate shows higher binding affinity (141). Many of the H3 residues 10-21 bind along the SWIRM/AOD boundary. Modifications here will affect binding affinity of the histone tail and change enzymatic efficiency. Closer to the histone binding pocket, many residues have been studied in detail. Methylation of Lys9 shows no effect on LSD1 activity, while acetylation of this residue shows a 6-fold decrease in binding affinity (141). LSD1 activity is completely abolished by phosphorylation at Ser10 (141). Additionally, phosphorylation of H3T6 removes access to H3K4 and shifts the substrate specificity of LSD1 to H3K9 (142). In addition to combinatorial regulation on the substrate histone tail, LSD1's function can be regulated by alternative splicing and posttranslational modifications. Two additional splice sites are observed in the KDM1A gene, one at exon 2 and exon 8 (143). The two additional splice sites may be incorporated either separately or together to result in three additional possibilities, either the 8a, 2a, or 8a/2a variants (143). The 2a variant results in a twenty amino acid insertion in the unstructured N-terminal region and may confer additional or altered specificity in recruiting other partners into chromatinremodeling complexes (143). The 8a variant is found only in neuronal tissue and contains 23 a four amino acid insertion at the base of the tower domain. This insertion contains a phosphorylation site at Thr396b that, when phosphorylated, acts as a dominant negative form of LSD1 that cannot bind CoREST or HDACs and fails to repress neuronal differentiation genes (143,144). Levels of 8a accumulate as neuronal development progresses and the loss of this variant results in decreased development of neuronal morphological features in vitro (143). The combined 8a/2a variant is found in the brain and testis (143). LSD1 has also been reported as a substrate for phosphorylation by protein kinase Cα (PKCα) (145) and protein kinase CK2 at Ser131, Ser137, and Ser166 (146), though the function of these posttranslational modifications remains undetermined. 1.3.4.2 An Epigenetic Effector with Context-Dependent Function Based on the complexity of structural mechanisms which regulate LSD1 enzymatic activity and its protein-protein interactions, it is unsurprising that the physiological function of LSD1 depends largely on both the cellular context and proteinprotein interaction partners. A few illustrative examples are described. At the most basic level, many LSD1/CoREST complexes contain BHC80, which recognizes and binds unmethylated H3K4 to prevent reactivation of target genes (147). Knockdown of BHC80 results in de-repression of LSD1-repressed genes, so BHC80 might function to keep LSD1-containing complexes at the unmodified site for continued repression (147). This also builds on data from Forneris, et al. (141) suggesting that LSD1 requires a histone substrate relatively free of other posttranslational modifications and is the last actor during events which effectively switch the local chromatin state. However, while the most commonly studied LSD1-containing complexes are repressive, involve interaction 24 with CoREST and HDACs, and are targeted at H3K4, in complex with the androgen and estrogen receptors LSD1 shows specificity for H3K9 and demethylation activates hormone-receptor-dependent transcription (118,148). Members of the Snai1 family recruit LSD1, through their H3 homologous Snail/Gfi (SNAG) domain, to the promoters of epithelial genes, like E-cadherin (CDH1). This is particularly important during the epithelial-to-mesenchymal transition (EMT) in order to repress the epithelial gene expression program and cellular phenotype (149-151). In malignant cells, the resulting cellular program drives cells to display a more invasive phenotype, and may partially explain the observed association of increased LSD1 expression with aggressive tumor biology (149). While it is relatively easy to envision the context-dependent function of LSD1 on a complex-to-complex basis, the most interesting data have demonstrated that LSD1 exists in opposing complexes that co-localize in the nucleus. These complexes often exist at the boundaries between heterochromatin and euchromatin and are important for normal development (152,153). In drosophila, the LSD1 homolog dLsd1 and the histone demethylase Lid oppose each other at the boundaries of hetero- and euchromatin, with dLsd1 promoting the expansion of heterchromatic regions. Interestingly, both play a pivotal role in modulating Notch-dependent gene expression (154). dLsd1 is present at Notch target gene promoters and facilitates activation of transcription in antagonism of Lid when Notch signaling is on. However, when Notch signaling is off dLsd1 and Lid cooperate to repress Notch target genes (154). Based on the complex structural and biochemical factors that affect LSD1's specificity, this dual role is not unexpected, but the regulatory mechanisms that mediate these phenotypes in vivo remain poorly understood. 25 In another interesting example, LSD1 is critical for pituitary development (155). However, early development requires LSD1-mediated gene activation, while terminal differentiation events require LSD1-mediated gene repression at the same target genes (155); again, the precise mechanisms which determine the spatiotemporal regulation of LSD1 remain undetermined. As a final note on substrate specificity, it should be noted that LSD1 is known to demethylate nonhistone substrates, though how LSD1 is targeted to these substrates is not understood. Demethylation of the p53 protein at Lys370 by LSD1 prevents the binding of p53 with 53BP1 that is required for p53-mediated transcriptional activation (156). Thus, LSD1 can repress p53 tumor suppressive function through methylation status of a single lysine residue. LSD1 is also critical for the maintenance of global DNA methylation levels in vivo through regulation of DNMT1 (157). Demethylation by LSD1 is required for DNMT1 protein stability and comprises a functional link between the histone and DNA methylation apparatus (157). Another interesting nonhistone substrate for LSD1 is metastatic tumor antigen 1 (MTA1), a member of both the NuRD repressive and nucleosome remodeling factor (NURF) coactivator complexes. When methylated, MTA1 promotes formation of the NuRD complex, while demethylation by LSD1 induces a conformational change which promotes assembly of NURF complex components and switching function from transcriptionally repressive to transcriptionally activating (158); thereby neatly demonstrating the ways in which LSD1 can assemble in complexes with opposing function. 26 1.3.4.3 Challenges of Studying LSD1 Biology The complexity of LSD1 biology presents obvious challenges to translate laboratory findings to the clinic. Whole-genome studies investigating TGFβ-induced EMT have shown that LSD1 is the critical regulator of decreased genomic H3K9me2 and increased H3K4me3 and H3K36me3 in a manner that may apply to malignant transformation (151). These types of findings speak to the potential power of specifically inhibiting LSD1 in cancer, but the biggest challenge remains understanding the underpinning of the biological mechanisms by which LSD1 acts. While the association of LSD1 with an aggressive clinical course has been established, LSD1 has also been reported to have some tumor suppressor function (139). Target validation studies have traditionally used RNAi, MAO inhibitors, or polyamines to probe LSD1 biology in cancer. In various cancer models, inhibition of LSD1 or RNAi-mediated knockdown resulted in increased H3K4 methylation, reexpression of aberrantly silenced tumor suppressor genes (127,159), decrease in prosurvival gene expression (127,160), differentiation of dedifferentiated cancer cell (124,128), and decreased cancer cell proliferation and survival (117,121,122,124,125,127-129,160). However, none of these modes of LSD1 inhibition represent ideal positive controls for novel compound development. Knockdown-rescue experiments have shown rescue of complex phenotypes in vivo with an enzymatically dead mutant (personal communication Michael Engel), and no published studies that the author could find attempted knockdown and rescue with both the wild-type and enzymatically dead constructs. LSD1 is present in complexes at thousands of gene promoters throughout the nucleus, but only enzymatically active at a smaller subset 27 depending on external stimuli (161). Global knockdown removes LSD1 and may affect the stability of multiple nuclear complexes independent of LSD1 enzymatic activity, confounding extrapolation of results to enzymatic inhibitors. The monoamine oxidase inhibitor most commonly used for target validation has been tranylcypromine, which is not potent or specific for LSD1, nor are the polyamines. Thus, deconvoluting which effects are on- vs. off-target with these treatments is challenging and holding novel classes of LSD1 inhibitors to display the same biological output as MAOi and polyamines may falsely discredit bonafide LSD1 inhibitors. This is supported by recent molecular modeling work suggesting that LSD1 has multiple binding pockets on its surface (162) coupled with results described in Chapter 2 which show different classes of LSD1 inhibitor displaying different biophysical effects on LSD1 protein in solution. One could envision disrupting both enzymatic activity and potentially protein-protein interactions through direct or allosteric mechanisms, and depending on the interaction disrupted the biological readout could be different. The complexity of LSD1 biology and the factors confounding its exploration also make predicting toxicity difficult. LSD1 knockout is embryonic lethal (163) and it is an important regulator for normal developmental transcriptional programs in hematopoiesis (123), adipogenesis (164), and neurogenesis (165,166). LSD1 expression is highest during early development and has roles in maintaining pluripotency (167,168) and the cell cycle in stem cells (169) and initiating differentiation during development (170). Conditional knockdown of LSD1 in adult mice led to alterations in hematopoiesis characterized by an expanded progenitor compartment and decreased terminal differentiation (171). This phenomenon was recapitulated to varying degrees with 28 tranylcypromine and its derivatives, though small molecules were not tested in vivo (171). It is unknown what the therapeutic window would be for a potent LSD1 inhibitor in vivo, and whether differences would be observed for reversible and irreversible inhibitors. 1.3.5 Pharmaceutics in Translational Science In addressing the complexity of in vivo efficacy and toxicity of a novel epigenetic drug, translational scientists must exercise care in choosing an appropriate preclinical formulation. Primarily, the vehicle used for drug delivery and route of administration should not interfere in any way with evaluation of the biological system used for testing. Importantly, different stages of preclinical work place different constraints on the formulation used. For example, in proof-of-concept studies the formulation scientist should maximize exposure within the limits of tolerability. However, for pharmacokinetics studies, the formulation must provide detectable exposure without altering the properties of the test compound drastically. As a lead compound emerges and progresses toward the clinic, the formulations used must evolve to be more clinically relevant and acceptable to regulatory agencies. While this may seem straightforward, the general solubility of new chemical entities evaluated is declining (172). Solubilizing agents which are acceptable in in vitro settings are poorly tolerated in vivo. Several alternative strategies can be used to overcome poor aqueous solubility and the appropriate solution is highly compound specific (173). For compounds which are weakly acidic or basic, pH adjustment using different buffering systems within the range of pH 2-9 can greatly improve aqueous 29 solubility. Another common strategy involves the use of cosolvents, though these have highly variable tolerability depending on the intended route of administration. Here organic molecules which are miscible with water provide nonpolar regions to interact with the solute. Commonly used cosolvents include polyethylene glycol 400 (PEG400), propylene glycol, dimethylacetamide, ethanol, and dimethyl sulfoxide (DMSO). Other strategies include co-complexation with cyclodextrins, which contain an interior nonpolar region for solute binding, and inclusion of lipids or surfactants, which can lead to micelle formation of other macromolecular complexes. Use of surfactants can also stabilize drug suspensions and facilitate drug uptake. In the studies reported herein, cosolvent systems were the primary strategy for drug formulation, using guidelines previously reported to minimize confounding toxicities (173). 1.4 Study Rationale and Objectives 1.4.1 Rationale Most studies of LSD1 biology to date have utilized either RNAi-mediated knockdown of LSD1, polyamine analogues, or tranylcypromine and its derivatives to investigate the biology of LSD1. These investigations suggest inhibition of LSD1 in cancer may provide benefit to some patients. However, these modes of LSD1 inhibition are neither potent nor specific, and even with well defined biological output, translation from mouse studies to the clinic remain difficult (174). More potent, specific, and reversible inhibitors of LSD1 are needed to carefully evaluate the biological impacts of LSD1 inhibition on global epigenomic regulation and tumorigenic phenotpes in in vitro and in vivo models of malignancy, identify biomarkers to guide clinical translation, and 30 assess the preclinical efficacy and toxicity of LSD1 inhibition.  Rationale: Reversible inhibitors with improved specificity and potency profiles can be used to perform proof-of-concept studies to validate LSD1 inhibition as a prospective therapeutic strategy in solid tumors. 1.4.2 Hypothesis and Objectives The primary objective of this work is to discover and evaluate a novel potent, specific, and reversible series of LSD1 inhibitors with physicochemical properties amenable for translation to the clinic. Our overarching hypothesis is that by targeting the key histone demethylase, potent and specific LSD1 inhibitors will exhibit single-agent efficacy in solid tumor models. In order to test this hypothesis, the first requirement is to discover and biochemically validate a series of novel LSD1 inhibitors, described in Chapter 2. The second requirement is to show that in malignancies sensitive to lead compound HCI2509, this efficacy translates to in vivo models of disease, described in Chapters 3 and 4. Critical aims and approaches are as follows: 1.4.2.1 Chapter 2, Hypothesis 1 A high-throughput virtual screening strategy can identify novel scaffolds which 1) show inhibitory activity in an enzymatic assay (goal IC50 < 1 μM) and 2) can be optimized to low nanomolar potency leads.  Approach: Identify a commercially available hit compound using high throughput virtual screening and followed by subsequent iterative biochemical testing and medicinal chemistry for hit-to-lead optimization. 31 1.4.2.2 Chapter 2, Hypothesis 2 The novel lead compound will 1) show specificity over homologous enzymes, 2) bind LSD1 reversibly, 3) not compete with the histone H3 N-terminal substrate for LSD1, and 4) show decreased cancer cell line viability.  Approach: Biochemically characterize the lead compound HCI2509 in an array of biochemical assays and assess the effect of HCI2509 on cancer cell line viability in vitro. 1.4.2.3 Chapter 3, Hypothesis LSD1 inhibition with HCI2509 impairs function of the NuRD complex, causes global derepression of EWS/FLI repressed target genes, and shows antitumor activity in vivo across Ewing sarcoma cell lines.  Approach 1: Compare the global transcriptional profile of HCI2509 treatment against that of EWS/FLI- and EWS/ERG-knockdown and to validate selected findings across multiple cell lines.  Approach 2: In vitro characterization of HCI2509 treatment against other EWS/FLI-knockdown associated phenotypes, including cell morphology and oncogenic transformation. Additionally characterize global methylation changes caused by HCI2509 treatment.  Approach 3: Characterize LSD1 target engagement in cells through evaluation of the relationship between LSD1 inhibition, HMOX1 induction, and EWS/FLI function.  Approach 4: Evaluate the antitumor efficacy of HCI2509 in multiple Ewing 32 xenograft models. 1.4.2.4 Chapter 4, Hypothesis LSD1 inhibition with HCI2509 is an effective therapeutic strategy for malignancies with a more complex etiology, specifically Type II endometrial carcinoma.  Approach 1: Evaluate the effects of HCI2509 on proliferation, transformation, global histone methylation, and target gene derepression in hormone-resistant Type II endometrial carcinoma cell lines.  Approach 2: Test HCI2509 for in vivo antitumor efficacy using an orthotopic xenograft model of Type II endometrial carcinoma tracking disease with bioluminescence. These hypotheses and approaches, and the resulting data are discussed at length in their respective chapters. The results from Chapter 2 have been peer-reviewed and published in the Journal of Medicinal Chemistry. The studies described in Chapters 3 and 4 have been accepted at Clinical Cancer Research and submitted to BMC Cancer, respectively. 1.5 References 1. Alberts B, et al. Molecular Biology of the Cell, 5th ed. New York: Garland Science, 2008. Print. 2. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 2011;144:646-74. 3. Global Health [Internet]. Atlanta: American Cancer Society; c2014 [cited 2014 May 3]. Available from: http://www.cancer.org/aboutus/globalhealth/. 4. Schulte JH, Lim S, Schramm A, Friedrichs N, Koster J, Versteeg R, et al. Lysinespecific demethylase 1 is strongly expressed in poorly differentiated neuroblastoma: implications for therapy. Cancer Res 2009;69:2065-71. 33 5. Harris WJ, Huang X, Lynch JT, Spencer GJ, Hitchin JR, Li Y, et al. The histone demethylase KDM1A sustains the oncogenic potential of MLL-AF9 leukemia stem cells. Cancer Cell 2012;21:473-87. 6. Lim S, Janzer A, Becker A, Zimmer A, Schüle R, Buettner R, et al. Lysine-specific demethylase 1 (LSD1) is highly expressed in ER-negative breast cancers and a biomarker predicting aggressive biology. Carcinogenesis 2010;31:512-520. 7. Crompton B, Stewart C, Taylor-Weiner A, Alexa G, Kurek K, Calicchio M, et al. The genomic landscape of pediatric Ewing sarcoma. [abstract]. In: Proceedings of the 105 th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Abstract nr 999. 8. Bird, A. DNA methylation patterns and epigenetic memory. Genes Dev 2002;16:6-21. 9. Kumar S, Cheng X, Klimasauskas S, Mi S, Posfai J, Roberts RJ, et al. The DNA (cytosine-5) methyltransferases. Nucleic Acids Res 1994;22:1-10. 10. Okano M, Bell DW, Haber DA, Li E. DNA methyltransferases DNMT3a and DNMT3b are essential for de novo methylation and mammalian development. Cell 1999;99:247-57. 11. Takai D, Jones PA. Comprehensive analysis of CpG islands in human chromosomes 21 and 22. Proc Natl Acad Sci USA 2002;99:3740-5. 12. Baylin SB, Herman JG. DNA hypermethylation in tumorigenesis: epigenetics joins genetics. Trends Genet 2000;16:168-174. 13. Jones PA, Laird PW. Cancer epigenetics comes of age. Nature Genet 1999;21:163- 167. 14. Esteller M, Toyota M, Sanchez-Cespedes M, Capella G, Peinado MA, Watkins DN, et al. Inactivation of the DNA repair gene O6-methylguanine-DNA methyltransferase by promoter hypermethylation is associated with G to A mutations in K-ras in colorectal tumorigenesis. Cancer Res 2000;60:2368-71. 15. Herman JG, Civin CI, Issa JP, Collector MI, Sharkis SJ, Baylin SB. Distinct patterns of inactivation of p15INK4B and p16INK4A characterize the major types of hematological malignancies. Cancer Res 1997;57:837-41. 16. Esteller M, Silva JM, Dominguez G, Bonilla F, Matias-Guiu X, Lerma E. Promoter hypermethylation and BRCA1 inactivation in sporadic breast and ovarian tumors. J Natl Cancer Inst 2000;92:564-569. 34 17. Kane MF, Loda M, Gaida GM, Lipman J, Mishra R, Goldman H, et al. Methylation of the hMLH1 promoter correlates with lack of expression of hMLH1 in sporadic colon tumors and mismatch repair-defective human tumor cell lines. Cancer Res 1997;57:808- 811. 18. Herman JG, Umar A, Polyak K, Graff JR, Ahuja N, Issa JP, et al. Incidence and functional consequences of hMLH1 promoter hypermethylation in colorectal carcinoma. Proc Natl Acad Sci USA 1998;95:6870-6875. 19. Coulondre C, Miller JH, Farabaugh PJ, Gilbert W. Molecular basis of base substitution hotspots in Escherichia coli. Nature 1978;274;775-780. 20. Rideout WM III, Coetzee GA, Olumi AF, Jones PA. 5-Methylcytosine as an endogenous mutagen in the human LDL receptor and p53 genes. Science 1990;249:1288-1290. 21. Feinberg AP, Vogelstein B. Hypomethylation distinguishes genes of some human cancers from their normal counterparts. Nature 1983;301:89-92. 22. Feinberg AP, Gehrke CW, Kuo KC, Ehrlich M. Reduced genomic 5-methylcytosine content in human colonic neoplasia. Cancer Res 1988;48:1159-1161. 23. Qu GZ, Grundy PE, Narayan A, Ehrlich M. Frequent hypomethylation in Wilms tumors of pericentromeric DNA in chromosomes 1 and 16. Cancer Genet Cytogenet 1999;109:34-39. 24. Fullgrabe J Kavanaugh, E, Joseph B. Histone onco-modifications. Oncogene 2011;30:3391-403. 25. Zinner R, Albiez H, Walter J, Peters A, Cremer T, Cremer M. Histone lysine methylation patterns in cell types are arranged in distinct three-dimensional nuclear zones. Histochem Cell Biol 2006;125:3-19 26. Asp P, Blum R, Vethantham V, Parisi F, Micsinai M, Cheng J, et al. Genome-wide remodeling of the epigenetic landscape during myogenic differentiation. Proc Natl Acad Sci USA 2011;108:E149-58 27. Wang J, Scully K, Zhu X, Cai L, Zhang J, Prefontaine GG, et al. Opposing LSD1 complexes function in developmental gene activation and repression programmes. Nature 2007;446:882-7. 28. Gao H, Lukin K, Ramírez J, Fields S, Lopez D, Hagman J. Opposing effects of SWI/SNF and Mi-2/NuRD chromatin remodeling complexes on epigenetic reprogramming by EBF and Pax5. Proc Natl Acad Sci USA 2009;106:11258-63. 35 29. Bernstein BE, Mikkelsen TS, Xie X, Kamal M, Huebert DJ, Cuff J, et al. A bivalent chromatin structure marks key developmental genes in embryonic stem cells. Cell 2006;125:315-26. 30. Ohm JE, McGarvey KM, Yu X, Cheng L, Schuebel KE, Cope L, et al. A stem celllike chromatin pattern may predispose tumor suppressor genes to DNA hypermethylation and heritable silencing. Nat Genet 2007;39:237-42. 31. Dawson MA, Kouzarides T. Cancer epigenetics: from mechanism to therapy. Cell 2012;150:12-27. 32. Chung CW, Witherington J. Progress in the discovery of small-molecule inhibitors of bromodomain--histone interactions. J Biomol Screen 2011;16:1170-85. 33. Wang J, Iwasaki H, Krivtsov A, Febbo PG, Thorner AR, Ernst P, et al. Conditional MLL-CBP targets GMP and models therapy-related myeloproliferative disease. EMBO J 2005;24:368-81. 34. Deguchi K, Ayton PM, Carapeti M, Kutok JL, Snyder CS, Williams IR, et al. MOZTIF2-induced acute myeloid leukemia requires the MOZ nucleosome binding motif and TIF2-mediated recruitment of CBP. Cancer Cell 2003;3:259-7 35. Huntly BJ, Shigematsu H, Deguchi K, Lee BH, Mizuno S, Duclos N, et al. MOZTIF2, but not BCR-ABL, confers properties of leukemic stem cells to committed murine hematopoietic progenitors. Cancer Cell 2004;6:587-596. 36. Iyer NG, Ozdag H, Caldas C. p300/CBP and cancer. Oncogene 2004;23:4225-31. 37. Pasqualucci L, Dominguez-Sola D, Chiarenza A, Fabbri G, Grunn A, Trifonov V, et al. Inactivating mutations of acetyltransferase genes in B-lymphoma. Nature 2011;471:189-95. 38. Johnstone RW, Licht JD. Histone deacetylase inhibitors in cancer therapy: is transcription the primary target? Cancer Cell 2003;4:13-8 39. Bereshchenko OR, Gu W, Dalla-Favera R. Acetylation inactivates the transcriptional repressor BCL6. Nat Genet 2002;32:606-13. 40. Ley TJ, Ding L, Walter MJ, McLellan MD, Lamprecht T, Larson DE, et al. DNMT3A mutations in acute myeloid leukemia. N Engl J Med. 2010;363:2424-2433. 41. Thol F, Damm F, Lüdeking A, Winschel C, Wagner K, Morgan M, et al. Incidence and prognostic influence of DNMT3A mutations in acute myeloid leukemia. J Clin Oncol. 2011;29:2889-2896. 36 42. Metzeler KH, Maharry K, Radmacher MD, Mrózek K, Margeson D, Becker H, et al. TET2 mutations improve the new European LeukemiaNet risk classification of acute myeloid leukemia: a Cancer and Leukemia Group B study. J Clin Oncol. 2011;29:1373- 1381. 43. Bejar R, Stevenson K, Abdel-Wahab O, Galili N, Nilsson B, Garcia-Manero G, et al. Clinical effect of point mutations in myelodysplastic syndromes. N Engl J Med. 2011;364:2496-2506. 44. Abdel-Wahab O, Pardanani A, Patel J, Wadleigh M, Lasho T, Heguy A, et al. Concomitant analysis of EZH2 and ASXL1 mutations in myelofibrosis, chronic myelomonocytic leukemia and blast-phase myeloproliferative neoplasms. Leukemia. 2011;25:1200-1202. 45. Metzeler KH, Becker H, Maharry K, Radmacher MD, Kohlschmidt J, Mrózek K, et al. ASXL1 mutations identify a high-risk subgroup of older patients with primary cytogenetically normal AML within the ELN favorable genetic category. Blood. 2011;118:6920-6929. 46. Guglielmelli P, Biamonte F, Score J, Hidalgo-Curtis C, Cervantes F, Maffioli M, et al. EZH2 mutational status predicts poor survival in myelofibrosis. Blood. 2011;118:5227- 5234. 47. Tefferi A, Pardanani A, Lim KH, Abdel-Wahab O, Lasho TL, Patel J, et al. TET2 mutations and their clinical correlates in polycythemia vera, essential thrombocythemia and myelofibrosis. Leukemia. 2009;23:905-911. 48. Tefferi A, Jimma T, Sulai NH, Lasho TL, Finke CM, Knudson RA, et al. IDH mutations in primary myelofibrosis predict leukemic transformation and shortened survival: clinical evidence for leukemogenic collaboration with JAK2V617F. Leukemia. 2012;26:475-480. 49. Tefferi A, Lasho TL, Abdel-Wahab O, Guglielmelli P, Patel J, Caramazza D, et al. IDH1 and IDH2 mutation studies in 1473 patients with chronic-, fibrotic- or blast-phase essential thrombocythemia, polycythemia vera or myelofibrosis. Leukemia. 2010;24:1302-1309. 50. De Carvalho DD, Sharma S, You JS, Su SF, Taberlay PC, Kelly TK, et al. DNA methylation screening identifies driver epigenetic events of cancer cell survival. Cancer Cell. 2012;21:655-667. 51. Baylin SB, Jones PA. A decade of exploring the cancer epigenome - biological and translational implications. Nat Rev Cancer. 2011;11:726-734. 52. Sorm F, Vesely J. Effect of 5-aza-2′-deoxycytidine against leukemic and hemopoietic tissues in AKR mice. Neoplasma. 1968;15:339-343. 37 53. Notari RE, DeYoung JL. Kinetics and mechanisms of degradation of the antileukemic agent 5-azacytidine in aqueous solutions. J Pharm Sci. 1975;64:1148-1157. 54. Jones PA, Taylor SM. Cellular differentiation, cytidine analogs and DNA methylation. Cell. 1980;20:85-93. 55. Kaminskas E, Farrell A, Abraham S, Baird A, Hsieh LS, Lee SL, et al. Approval summary: azacitidine for treatment of myelodysplastic syndrome subtypes. Clin Cancer Res. 2005;11:3604-3608. 56. Silverman LR, Demakos EP, Peterson BL, Kornblith AB, Holland JC, OdchimarReissig R, et al. Randomized controlled trial of azacitidine in patients with the myelodysplastic syndrome: a study of the cancer and leukemia group B. J Clin Oncol. 2002;20:2429-2440. 57. Kaminskas E, Farrell AT, Wang YC, Sridhara R, Pazdur R. FDA drug approval summary: azacitidine (5-azacytidine, Vidaza) for injectable suspension. Oncologist. 2005;10:176-182. 58. Kantarjian H, Issa JP, Rosenfeld CS, Bennett JM, Albitar M, DiPersio J, et al. Decitabine improves patient outcomes in myelodysplastic syndromes: results of a phase III randomized study. Cancer. 2006;106:1794-1803. 59. Azad N, Zahnow CA, Rudin CM, Baylin SB. The future of epigenetic therapy in solid tumours--lessons from the past. Nat Rev Clin Oncol 2013;10:256-66. 60. Gabbara S, Bhagwat AS. The mechanism of inhibition of DNA (cytosine-5-)methyltransferases by 5-azacytosine is likely to involve methyl transfer to the inhibitor. Biochem J. 1995;307:87-92. 61. Santi DV, Norment A, Garrett CE. Covalent bond formation between a DNA-cytosine methyltransferase and DNA containing 5-azacytosine. Proc Natl Acad Sci USA. 1984;81:6993-6997. 62. Ferguson AT, Vertino PM, Spitzner JR, Baylin SB, Muller MT, Davidson NE. Role of estrogen receptor gene demethylation and DNA methyltransferase. DNA adduct formation in 5-aza-2'deoxycytidine-induced cytotoxicity in human breast cancer cells. J Biol Chem. 1997;272:32260-32266. 63. Ghoshal K, Datta J, Majumder S, Bai S, Kutay H, Motiwala T, et al. 5-Azadeoxycytidine induces selective degradation of DNA methyltransferase 1 by a proteasomal pathway that requires the KEN box, bromo-adjacent homology domain, and nuclear localization signal. Mol Cell Biol. 2005;25:4727-4741. 38 64. Tsai HC, Li H, Van Neste L, Cai Y, Robert C, Rassool FV, et al. Transient low doses of DNAdemethylating agents exert durable antitumor effects on hematological and epithelial tumor cells. Cancer Cell. 2012;21:430-446. 65. Silverman LR, Fenaux P, Mufti GJ, Santini V, Hellström-Lindberg E, Gattermann N, et al. Continued azacitidine therapy beyond time of first response improves quality of response in patients with higher-risk myelodysplastic syndromes. Cancer 2011;117:2697702. 66. Piekarz RL, Frye R, Turner M, Wright JJ, Allen SL, Kirschbaum MH, et al. Phase II multi-institutional trial of the histone deacetylase inhibitor romidepsin as monotherapy for patients with cutaneous T-cell lymphoma. J Clin Oncol. 2009;27:5410-5417. 67. Whittaker SJ, Demierre MF, Kim EJ, Rook AH, Lerner A, Duvic M, et al. Final results from a multicenter, international, pivotal study of romidepsin in refractory cutaneous T-cell lymphoma. J Clin Oncol. 2010;28:4485-4491. 68. O'Connor OA, Heaney ML, Schwartz L, Richardson S, Willim R, MacGregorCortelli B, et al. Clinical experience with intravenous and oral formulations of the novel histone deacetylase inhibitor suberoylanilide hydroxamic acid in patients with advanced hematologic malignancies. J Clin Oncol. 2006;24:166-173. 69. Duvic M, Talpur R, Ni X, Zhang C, Hazarika P, Kelly C, et al. Phase 2 trial of oral vorinostat (suberoylanilide hydroxamic acid, SAHA) for refractory cutaneous T-cell lymphoma (CTCL). Blood. 2007;109:31-39. 70. Lane AA, Chabner BA. Histone deacetylase inhibitors in cancer therapy. J Clin Oncol 2009;27:5459-68. 71. Hochedlinger K, Plath K. Epigenetic reprogramming and induced pluripotency. Development 2009;136:509-23. 72. Lister R, Pelizzola M, Kida YS, Hawkins RD, Nery JR, Hon G, et al. Hotspots of aberrant epigenomic reprogramming in human induced pluripotent stem cells. Nature 2011;471:68-73. 73. Feng B, Ng JH, Heng JC, Ng HH. Molecules that promote or enhance reprogramming of somatic cells to induced pluripotent stem cells. Cell Stem Cell 2009;4:301-12. 74. Gray SG, Ekström TJ. The human histone deacetylase family. Exp Cell Res. 2001;262:75-83. 75. Solomon JM, Pasupuleti R, Xu L, McDonagh T, Curtis R, DiStefano PS, et al. Inhibition of SIRT1 catalytic activity increases p53 acetylation but does not alter cell survival following DNA damage. Mol Cell Biol. 2006;26:28-38. 39 76. Brahmer JR, Tykodi SS, Chow LQ, Hwu WJ, Topalian SL, Hwu P, et al. Safety and activity of anti-PD-L1 antibody in patients with advanced cancer. N Engl J Med. 2012;366:2455-65. 77. Juergens RA, Wrangle J, Vendetti FP, Murphy SC, Zhao M, Coleman B, et al. Combination epigenetic therapy has efficacy in patients with refractory advanced nonsmall cell lung cancer. Cancer Discov. 2011;1:598-607. 78. Sharma SV, Lee DY, Li B, Quinlan MP, Takahashi F, Maheswaran S, et al. A chromatin-mediated reversible drug-tolerant state in cancer cell subpopulations. Cell 2010;141:69-80. 79. Delmore JE, Issa GC, Lemieux ME, Rahl PB, Shi J, Jacobs HM, et al. BET bromodomain inhibition as a therapeutic strategy to target c-Myc. Cell 2011;146:904-17. multiple myeloma 80. Mertz JA, Conery AR, Bryant BM, Sandy P, Balasubramanian S, Mele DA, et al. Targeting MYC dependence in cancer by inhibiting BET bromodomains. Proc Natl Acad Sci U S A 2011;108:16669-74. 81. Dawson MA, Prinjha RK, Dittmann A, Giotopoulos G, Bantscheff M, Chan WI, et al. Inhibition of BET recruitment to chromatin as an effective treatment for MLL-fusion leukaemia. Nature 2011;478:529-33. 82. Ott CJ, Kopp N, Bird L, Paranal RM, Qi J, Bowman T, et al. BET bromodomain inhibition targets both c-Myc and IL7R in high-risk acute lymphoblastic leukemia. Blood 2012;120:2843-52. 83. Puissant A, Frumm SM, Alexe G, Bassil CF, Qi J, Chanthery YH, et al. Targeting MYCN in neuroblastoma by BET bromodomain inhibition. Cancer Discov. 2013;3:30823. 84. Wyce A, Ganji G, Smitheman KN, Chung CW, Korenchuk S, Bai Y, et al. BET inhibition silences expression of MYCN and BCL2 and induces cytotoxicity in neuroblastoma tumor models. PLoS One 2013;8:e72967. 85. Shimamura T, Chen Z, Soucheray M, Carretero J, Kikuchi E, Tchaicha JH, et al. Efficacy of BET bromodomain inhibition in Kras-mutant non-small cell lung cancer. Clin Cancer Res 2013;19:6183-92. 86. Bandopadhayay P, Bergthold G, Nguyen B, Schubert S, Gholamin S, Tang Y, et al. BET bromodomain inhibition of MYC-amplified medulloblastoma. Clin Cancer Res 2014;20:912-25. 40 87. Chapuy B, McKeown MR, Lin CY, Monti S, Roemer MG, Qi J, et al. Discovery and characterization of super-enhancer-associated dependencies in diffuse large B cell lymphoma. Cancer Cell 2013;24:777-90. 88. Asangani IA, Dommeti VL, Wang X, Malik R, Cieslik M, Yang R, et al. Therapeutic targeting of BET bromodomain proteins in castration-resistant prostate cancer. Nature 2014;doi: 10.1038/nature13229. 89. ClinicalTrials.gov [Internet]. Available from: http://www.cancer.gov/clinicaltrials/ search/results?protocolsearchid=10430216 90. Tensha Therapeutics Announces First Clinical Trial of BET Bromodomain Inhibitor TEN-010 for Treatment of Cancer [Internet]. Cambridge: BusinessWire; c2013 [updated 2013 Nov 20; cited 2014 May 3]. Available from: http://www.businesswire.com/ news/home/20131120005823/en/Tensha-Therapeutics-Announces-Clinical-Trial-BETBromodomain 91. Constellation Pharmaceuticals Initiates Clinical Development of CPI-0610, a Novel BET Protein Bromodomain Inhibitor, in Patients with Lymphoma [Internet]. Cambridge: Constellation Pharmacueitcal, Pure Communications, Inc.; c2014 [updated 2013 Sept 10; cited 2014 May 3]. Available from: http://www.constellationpharma.com/2013/09/ constellation-pharmaceuticals-initiates-clinical-development-of-cpi-0610-a-novel-betprotein-bromodomain-inhibitor-in-patients-with-lymphoma/. 92. Byvoet P, Shepherd GR, Hardin JM, Noland BJ. The distribution and turnover of labelled methyl groups in histone fractions of cultured mammalian cells. Arch Biochem Biophys 1972;148:558-567. 93. Shi Y, Lan F, Matson C, Mulligan P, Whetstine JR, Cole PA, et al. Histone demethylation mediated by the nuclear amine oxidase homolog LSD1. Cell 2004;119:941-53. 94. Murray K. The occurrence of ɛ-N-methyl lysine in histones. Biochemistry 1964;3:10- 15. 95. Paik WK, Kim S. ɛ-N-dimethyllysine in histones. Biochem Biophys Res Commun 1967;27:479-483. 96. Haempel K, Lange HW, Birkofer L. ɛ-N-trimethyllysine, a new amino acid in histones. Die Naturwissenschaften 1968:55:37. 97. Zee BM, Levin RS, Xu B, LeRoy G, Wingreen NS, Garcia BA. In vivo residuespecific histone methylation dynamics. J Biol Chem 2010;285:3341-50. 41 98. Rea S, Eisenhaber F, O'Carroll D, Strahl BD, Sun ZW, Schmid M, et al. Regulation of chromatin structure by site-specific histone H3 methyltransferases. Nature 2000;406:5939. 99. Feng Q, Wang H, Ng HH, Erdjument-Bromage H, Tempst P, Struhl K, et al. Methylation of H3 lysine 79 is mediated by a new family of HMTases without a SET domain. Curr Biol 2002;15:1052-8. 100. Tsukada Y, Fang J, Erdjument-Bromage H, Warren ME, Borchers CH, Tempst P, et al. Histone demethylation by a family of JmjC domain-containing proteins. Nature 2006;439:811-6. 101. Whetstine JR, Nottke A, Lan F, Huarte M, Smolikov S, Chen Z, et al. Reversal of histone lysine trimethylation by the JMJD2 family of histone demethylases. Cell 2006;125:467-81. 102. Albert M, Helin K. Histone methyltransferases in cancer. Semin Cell Dev Biol 2010;21:209-20. 103. Chi P, Allis CS, Wang GG. Covalent histone modifications-miswritten, misinterpreted and mis-erased in human cancers. Nat Rev Cancer 2010;10:457-69. 104. Greer EL, Shi Y. Histone methylation: a dynamic mark in health, disease, and inheritance. Nat Rev Genet 2012;13:343-57. 105. Daigle SR, Olhava EJ, Therkelsen CA, Basavapathruni A, Jin L, Boriack-Sjodin A, et al. Potent inhibition of DOT1L as a treatment of MLL-fusion leukemia. Blood 2013;122:1017-25. 106. DOT1L inhibitor [Internet]. Cambridge: Epizyme, Inc.; c2013 [updated 2013 Dec; cited 2014 May 3]. Available from: http://www.epizyme.com/programs/dot1l-inhibitor/. 107. Kleer CG, Cao Q, Varambally S, Shen R, Ota I, Tomlins SA, et al. EZH2 is a marker of aggressive breast cancer and promotes neoplastic transformation of breast epithelial cells. Proc Natl Acad Sci USA 2003;100:11606-11. 108.Varambally S, Dhanasekaran SM, Zhou M, Barrette TR, Kumar-Sinha C, Sanda MG, et al. The polycomb group protein EZH2 is involved in progression of prostate cancer. Nature 2002;419:624-9. 109. Bracken AP, Helin K. Polycomb group proteins: navigators of lineage pathways led astray in cancer. Nat Rev Cancer 2009;9:773-84. 110. Visser HP, Gunster MJ, Kluin-Nelemans HC, Manders EM, Raaphorst FM, Meijer CJ, et al. The Polycomb group protein EZH2 is upregulated in proliferating, cultured human mantle cell lymphoma. Br J Haematol 2001;112:950-958. 42 111. van Haaften G, Dalgliesh GL, Davies H, Chen L, Bignell G, Greenman C, et al. Somatic mutations of the histone H3K27 demethylase gene UTX in human cancer. Nature Genet 2009;41:521-3. 112. Ernst T, Chase AJ, Score J, Hidalgo-Curtis CE, Bryant C, Jones AV, et al. Inactivating mutations of the histone methyltransferase gene EZH2 in myeloid disorders. Nature Genet 2010;42:722-6. 113. EZH2 inhibitor [Internet]. Cambridge: Epizyme, Inc.; c2013 [updated 2013 Dec; cited 2014 May 3]. Available from: http://www.epizyme.com/programs/ezh2-inhibitor/. 114. GSK Clinical Study Register [Internet]. Brentford, Middlesex, UK: GlaxoSmith Kline plc.; c2001-2014 [updated 2014 Mar 13; cited 2014 May 3]. Available from: http://www.gsk-clinicalstudyregister.com/study/117208#ps 115. Suzuki T, Terashima M, Tange S, Ishimura A. Roles of histone methyl-modifyind enzymes in development and progression of cancer. Cancer Sci 2013;104:795-800. 116. ClinicalTrials.gov [Internet]. Available from: http://clinicaltrials.gov/ct2/show/ NCT02034123. 117. Lim S, Janzer A, Becker A, Zimmer A, Schüle R, Buettner R, et al. Lysine-specific demethylase 1 (LSD1) is highly expressed in ER-negative breast cancers and a biomarker predicting aggressive biology. J Carcinogenesis 2010;31:512-20. 118. Metzger E, Wissmann M, Yin N, Müller JM, Schneider R, Peters AHFM, et al. LSD1 demethylates repressive histone marks to promote androgen-receptor-dependent transcription. Nature 2005;437:436-9. 119. Wissmann M, Yin N, Müller JM, Greschik H, Fodor BD, Jenuwein T, et al. Cooperative demethylation by JMJD2C and LSD1 promotes androgen receptordependent gene expression. Nat Cell Biol 2007;9:347-53. 120. Cai C, He HH, Chen S, Coleman I, Wang H, Fang Z, et al. Androgen receptor gene expression in prostate cancer is directly suppressed by the androgen receptor through recruitment of lysine-specific demethylase 1. Cancer Cell 2011;20:457-71. 121. Kahl P, Gullotti L, Heukamp LC, Wolf S, Friedrichs N, Vorreuther R, et al. Androgen receptor coactivators lysine-specific histone demethylase 1 and four and a half LIM domain protein 2 predict risk of prostate cancer recurrence. Cancer Res 2006;66:11341-7. 122. Schulte JH, Lim S, Schramm A, Friedrichs N, Koster J, Versteeg R, et al. Lysinespecific demethylase 1 is strongly expressed in poorly differentiated neuroblastoma: implications for therapy. Cancer Research 2009;69:2065-71. 43 123. Hu X, Li X, Valverde K, Fu X, Noguchi C, Qiu Y, et al. LSD1-mediated epigenetic modification is required for TAL1 function and hematopoiesis. Proc Natl Acad Sci USA 2009;106:10141-6. 124. Li Y, Deng C, Hu X, Patel B, Fu X, Qiu Y, et al. Dynamic interaction between TAL1 oncoprotein and LSD1 regulates TAL1 function in hematopoiesis and leukemogenesis. Oncogene 2012;31:5007-18. 125. Schenk T, Chen WC, Göllner S, Howell L, Jin L, Hebestreit K, et al. Inhibition of the LSD1 (KDM1A) demethylase reactivates the all-trans-retinoic acid differentiation pathway in acute myeloid leukemia. Nat Med 2012;18:605-11. 126. Harris WJ, Huang X, Lynch JT, Spencer GJ, Hitchin JR, Li Y, et al. The histone demethylase KDM1A sustains the oncogenic potential of MLL-AF9 leukemia stem cells. Cancer Cell 2012;21:473-87. 127. Hayami S, Kelly JD, Cho HS, Yoshimatsu M, Unoki M, Tsunoda T, et al. Overexpression of LSD1 contributes to human carcinogenesis through chromatin regulation in various cancers. Int J Cancer 2011;128:574-86. 128. Kauffman EC, Robinson BD, Downes MJ, Powell LG, Lee MM, Scherr DS, et al. Role of androgen receptor and associated lysine-demethylase coregulators, LSD1 and JMJD2A, in localized and advanced human bladder cancer. Mol Carcinog 2011;50:93144. 129. Bennani-Baiti IM, Machado I, Llombart-Bosch A, Kovar H. Lysine-specific demethylase 1 (LSD1/KDM1A/AOF2/BHC110) is expressed and is an epigenetic drug target in chondrosarcoma, Ewing's sarcoma, osteosarcoma, and rhabdomyosarcoma. Hum Pathol 2012;43:1300-7. 130. Schildhaus HU, Riegel R, Hartmann W, Steiner S, Wardelmann E, MerkelbachBruse S, et al. Lysine-specific demethylase 1 is highly expressed in solitary fibrous tumors, synovial sarcomas, rhabdomyosarcomas, desmoplastic small round cell tumors, and malignant peripheral nerve sheath tumors. Hum Pathol 2011;42:1667-75. 131. Magerl C, Ellinger J, Braunschweig T, Kremmer E, Koch LK, Höller T, et al. H3K4 dimethylation in hepatocellular carcinoma is rare compared with other hepatobiliary and gastrointestinal carcinomas and correlates with expression of the methylase Ash2 and the demethylase LSD1. Hum Pathol 2010;41:181-9. 132. Zhao ZK, Yu HF, Wang DR, Dong P, Chen L, Wu WG, et al. Overexpression of lysine specific demethylase 1 predicts worse prognosis in primary hepatocellular carcinoma patients. World J Gastroenterol 2012;18:6651-6. 44 133. Yang M, Gocke CB, Luo X, Borek D, Tomchick DR, Machius M, et al. Structural basis for CoREST-dependent demethylation of nucleosomes by the human LSD1 histone demethylase. Molecular Cell 2006, 23:377-387. 134. Stavropoulos P, Blobel G, Hoelz A. Crystal structure and mechanism of human lysine-demethylase 1. Nat Struct Mol Biol 2006;13:626-32. 135. Forneris F, Binda C, Vanoni MA, Mattevi A, Battaglioli E. Histone demethylation catalysed by LSD1 is a flavin-dependent oxidative process. FEBS Lett 2005;579:(2203-7. 136. Baron R, Binda C, Tortorici M, McCammon JA, Mattevi A. Molecular mimicry and ligand recongnition in binding and catalysis by the histone demethylase LSD1-CoREST complex. Structure 2011;19:212-220. 137. Yang M, Culhane JC, Szewczuk LM, Gocke CB, Brautigam CA, Tomchick DR, et al. Structural basis of histone demethylation by LSD1 revealed by suicide inactivation. Nat Struct Mol Biol 2007;14:535-9. 138. Forneris F, Binda C, Adamo A, Battaglioli E, Mattevi A. Structural basis of LSD1CoREST selectivity in histone H3 recognition. J Biol Chem 2007;282:20070-4. 139. Wang Y, Zhang H, Chen Y, Sun Y, Yang F, Yu W, et al. LSD1 is a subunit of the NuRD complex and targets the metastasis programs in breast cancer. Cell 2009;138:660672. 140. Tsai MC, Manor O, Wan Y, Mosammaparast N, Wang JK, Lan F, et al. Long noncoding RNA as modular scaffold of histone modification complexes. Science 2010;329:689-693. 141. Forneris F, Binda C, Vanoni MA, Battaglioli E, Mattevi A. Human histone demethylase LSD1 reads the histone code. J Biol Chem 2005;208:41360-5. 142. Metzger E, Imhof A, Patel D, Kahl P, Hoffmeyer K, Friedrichs N, et al. Phosphorylation of histone H3T6 by PKCβI controls demethylation at histone H3K4. Nature 2010;464:792-7. 143. Zibetti C, Adamo A, Binda C, Forneris F, Toffolo E, Verpelli C, et al. Alternative splicing of the histone demethylase LSD1/KDM1 contributes to the modulation of neurite morphogenesis in the mammalian nervous system. J Neuroscience 2010;30:2521-32. 144. Toffolo E, Rusconi F, Paganini L, Tortorici M, Pilotto S, Heise C, et al. Phosphorylation of neuronal lysine-specific demethylase 1 LSD1/KDM1A impairs transcriptional repression by regulating interaction with CoREST and histone deacetylases HDAC1/2. J Neurochem 2014;128:603-16. 45 145. Nam HJ, Boo K, Kim D, Han DH, Choe HK, Kim CR, et al. Phosphorylation of LSD1 by PKCα is crucial for circadian rhythmicity and phase resetting. Mol Cell 2014;53:791-805. 146. Costa R, Arrigoni G, Cozza G, Lolli G, Battistutta R, Izpisua Belmonte JC, et al. The lysine-specific demethylase 1 is a novel substrate of protein kinase CK2. Biochim Biophys Acta 2014;1844:722-9. 147. Lan F, Collins RE, De Cegil R, Alpatov R, Horton JR, Shi X, et al. Recognition of unmethylated histone H3 lysine 4 links BHC80 to LSD1-mediated gene repression. Nature 2007;448:718-722. 148. Garcia-Bassets I, Kwon Y-S, Telese F, Prefontaine GG, Hutt KR, Cheng CS, et al. Histone methylation-dependent mechanisms impose ligand dependency for gene activation by nuclear receptors. Cell 2007;128:505-518. 149. Lin T, Ponn A, Hu X, Law BK, Lu J. Requirement of the histone demethylase LSD1 in Snai1-mediate transcriptional repression during epithelial-mesenchymal transition. Oncogene 2010;29:4896-4904. 150. Lin Y, Wu Y, Li J, Dong C, Ye X, Chi YI, et al. The SNAG domain of Snail1 functions as a molecular hook for recruiting lysine-specific demethylase 1. EMBO J 2010;29:1803-16. 151. 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. 152. Rudolph T, Yonezawa M, Lein S, Heidrich K, Kubicek S, Schäfer C, et al. Heterochromatin formation in Drosophila is initiated through active removal of H3K4 methylation by the LSD1 homolog SU(VAR)3-3. Mol Cell 2007;26:103-15. 153. Lan F, Zaratiegui M, Villén J, Vaughn MW, Verdel A, Huarte M, et al. S. pombe LSD1 homologs regulate heterochromatin propagation and euchromatic gene transcription. Mol Cell 2007;26:89-101. 154. Stefano LD, Walker JA, Burgio G, Corona DFV, Mulligan P, Naar AM, et al. Functional antagonism between histon H3K4 demethylases in vivo. Genes Dev 2011;25:17-28. 155. Wang J, Scully K, Zhu X, Cai L, Zhang J, Prefontaine GG, et al. Opposing LSD1 complexes function in developmental gene activation and repression programmes. Nature 2007;446:882-7. 156. Huang J, Sengupta R, Espejo AB, Lee MG, Dorsey JA, Richter M, et al. p53 is regulated by the lysine demethylase LSD1. Nature 2007;449:105-8. 46 157. Wang J, Hevi S, Kurash JK, Lei H, Gay F, Bajko J, et al. The lysine demethylase LSD1 (KDM1) is required for maintenance of global DNA methylation. Nat Genet 2009;41:125-9. 158. Nair SS, Li DQ, Kumar R. A core chromatin remodeling factor instructs global chromatin signaling through multivalent reading of nucleosome codes. Mol Cell 2013;49:704-18. 159. Huang Y, Greene E, Stewart TM, Goodwin AC, Baylin SB, Woster PM, et al. Inhibition of lysine-specific demethylase 1 by polyamine analogues results in reexpression of aberrantly silenced genes. Proc Natl Acad Sci 2007;104:8023-8. 160. Willmann D, Lim S, Wetzel S, Metzger E, Jandausch A, Wilk W, et al. Impairment of prostate cancer cell growth by a selective and reversible LSD1 inhibitor. Int J Cancer 2012;131:2704-9. 161. Ram O, Goren A, Amit I, Shoresh N, Yosef N, Ernst J, et al. Combinatorial patterning of chromatin regulators uncovered by genome-wide location analysis in human cells. Cell 2011;147:1628-39. 162. Robertson JC, Hurley NC, Tortorici M, Ciossani G, Borrello MT, Vellore NA, et al. Expanding the druggable space of the LSD1/CoREST epigenetic target: new potential binding regions for drug-like molecules, peptides, protein partners, and chromatin. PLoS Comput Biol 2013;9:e1003158. 163. Foster CT, Dovey OM, Lezina L, Luo JL, Gant TW, Barlev N, et al. Lysine-specific demethylase 1 regulates the embryonic transcriptome and CoREST stability. Mol Cell Biol 2010;30:4851-63. 164. Musri MM, Carmona MC, Hanzu FA, Kaliman P, Gomis R, Párrizas M. Histone demethylase LSD1 regulates adipogenesis. J Biol Chem 2010;285:30034-41. 165. Fuentes P, Cánovas J, Berndt FA, Noctor SC, Kukuljan M. CoREST/LSD1 control the development of pyramidal cortical neurons. Cereb Cortex 2012;22:1431-41. 166. Sun G, Alzayady K, Stewart R, Ye P, Yang S, Li W, et al. Histone demethylase LSD1 regulates neural stem cell proliferation. Mol Cell Biol 2010;30:1997-2005. 167. Adamo A, Sesé B, Boue S, Castaño J, Paramonov I, Barrero MJ, et al. LSD1 regulates the balance between self-renewal and differentiation in human embryonic stem cells. Nat Cell Biol 2011;13:652-9. 168. Yin F, Lan R, Zhang X, Zhu L, Chen F, Xu Z, et al. LSD1 regulates pluripotency of embryonic stem/carcinoma cells through histone deacetylase 1-mediated deacetylation of histone H4 at lysine 16. Mol Cell Biol 2014;34:158-79. 47 169. Nair VD, Ge Y, Balasubramaniyan N, Kim J, Okawa Y, Chikina M, et al. Involvement of histone demethylase LSD1 in short-time-scale gene expression changes during cell cycle progression in embryonic stem cells. Mol Cell Biol 2012;32:4861-76. 170. Whyte WA, Bilodeau S, Orlando DA, Hoke HA, Frampton GM, Foster CT, et al. Enhancer decommissioning by LSD1 during embryonic stem cell differentiation. Nature 2012;482:221-5. 171. Sprüssel A, Schulte JH, Weber S, Necke M, Händschke K, Thor T, et al. Lysinespecific demethylase 1 restricts hematopoietic progenitor proliferation and is essential for terminal differentiation. Leukemia 2012;26:2039-51. 172. Takagi T, Ramachandran C, Bermejo M, Yamashita S, Yu LX, Amidon GL. A provisional biopharmaceutical classification of the top 200 oral drug products in the United States, Great Britain, Spain, and Japan. Mol Pharmaceutics 2006;3:631-643. 173. Li P, Zhao L. Developing early formulations: practice and perspective. Int J Pharm 2007;341:1-19. 174. Ledford H. Translational research: 4 ways to fix the clinical trial. Nature 2011;477:526-8. 48 Table 1.1 FDA-approved epigenetic therapies. Agent Name (Trade Name) Target Indication FDA-Approval Date Azacitidine (Vidaza) DNMT Myelodysplastic syndrome 2004 May 19 Decitabine (Dacogen) DNMT Myelodysplastic syndrome 2006 May 2 Vorinostat (Zolinza) HDAC Cutaneous T-Cell Lymphoma 2006 Oct 6 Romidepsin (Istodax) HDAC Cutaneous T-Cell Lymphoma 2009 Nov 5 Table 1.2 Histone methylation is globally misregulated in cancer. (Adapted from (104)) Cancer Type Methyl Mark Consequence Prostate Cancer ↓H3K4me2 High recurrence Prostate Cancer ↓H4K2me2 High recurrence Lung Cancer ↓H3K4me2 Poorer survival Kidney Cancer ↓H3K4me2 Poorer survival Breast Cancer ↓H3K4me2 Poorer survival Breast Cancer ↓H3K27me3 Poorer survival Breast Cancer ↓H4K20me3 Worse clinical outcomes Pancreatic Cancer ↓H3K4me2 Poorer survival Pancreatic Cancer ↓H3K9me2 Poorer survival Pancreatic Cancer ↓H3K27me3 Poorer survival Gastric Adenocarcinoma ↑H3K9me3 Poorer survival Ovarian Cancer ↓H3K27me3 Poorer survival Lymphomas ↓H4K20me3 Associated with Colon Adenocarcinoma ↓H4K20me3 Associated with 49 Table 1.3 Altered expression of KDMs in cancer. (Adapted from (115)) Name Synonym Alteration Associated Cancer KDM1A LSD1, AOF2 Overexpression Prostate, neuroblastoma, lung, colorectal, bladder KDM1A LSD1, AOF2 Downregulation Breast KDM2B FBXL10, JHDM1B Overexpression Pancreatic, leukemia KDM2B FBXL10, JHDM1B Downregulation Glioblastoma KDM3A JMJD1A, JHDM2A Overexpression Colorectal, renal cell carcinoma KDM4A JMJD2A Overexpression Lung, breast KDM4B JMJD2B Overexpression ER + breast, bladder, lung KDM4C JMJD2C, GASC1 Amplification Esophageal, breast, medulloblastoma, primary mediastinal KDM4C JMJD2C, GASC1 Overexpression Lymphoma KDM5A JARID1A, RBP2 Overexpression Gastric, breast KDM5A JARID1A, RBP2 Translocation Acute myeloid leukemia KDM5B JARID1B, PLU1 Overexpression Breast, prostate, bladder, lung KDM5B JARID1B, PLU1 Downregulation Melanoma KDM6A UTX Mutation Multiple myeloma, esophageal squamous cell, renal cell, chronic myelomonocytic leukemia KDM6B JMJD3 Downregulation Lung, liver PHF8 JHDM1F Overexpression Prostate 50 Figure 1.1 Histone methylation machinery. The histone methylation "writers" (methyltransferases; light gray) and "erasers" (demethylases; dark gray) for H3K4, H3K9, and H3K27. 51 Figure 1.2 The crystal structure of lysine-specific demethylase 1. LSD1 (blue) in complex with both Co-REST (yellow) and an N-terminal H3 peptide (magenta). FAD is shown bound by the amine oxidse domain (red). PDB ID: 2V1D 52 Figure 1.3 Catalytic oxidative demethylation of histone H3 lysine 4 by LSD1. Demethylation results in the generation of both H2O2 and formaldehyde. Adapted from Forneris, et al. (135). 53 Figure 1.4 Other proteins bind LSD1 through molecular mimicry. Comparison of SNAG domain (green) and histone H3 (magenta) binding to the LSD1 (blue) and CoREST (yellow) complex. CoREST and LSD1 are visualized with a van der Waals surface. 54 CHAPTER 2 HIGH-THROUGHPUT VIRTUAL SCREENING IDENTIFIES NOVEL N'-(1-PHENYLETHYLIDENE)-BENZOHYDRAZIDES AS POTENT, SPECIFIC, AND REVERSIBLEINHIBITORS OF LSD1 Venkataswamy Sorna and Emily R. Theisen are co-first authors of this work. VS was responsible for the chemical synthesis and purification of the compound series. ERT wrote the manuscript and was responsible for biochemical assessment of HCI2509 as well as cell-based assays. Bret Stephens performed the initial compound screen of 121 compounds. ERT and BS screened compounds synthesized in-house. Reproduced with permission from Venkataswamy Sorna, Emily R. Theisen, Bret Stephens, Steven L. Warner, David J. Bearss, Hariprasad Vankayalapati, and Sunil Sharma. Journal of Medicinal Chemistry 2013 56 (23), 9496-9508. Copyright 2013 American Chemical Society. 55 56 57 58 59 60 61 62 63 64 65 66 67 68 2.8 Supplementary Materials 2.8.1 Supplementary Methods 2.8.1.1 Detailed Virtual Screening Methods 2.8.1.1.1 Preparation of the Binding Site Model There were several X-ray crystal complex structures of LSD1 (PDB ID: 2Y48, 3BAT, 3BAU, 2XAF, 2XAG, 2XAH, 2XAJ, 2XAQ and 2XAS) at the beginning of our work, we used the LSD1 complex model with an X-ray crystal structure (PDB code 2Z5U) and protein coordinates used for fragment-based and structure-based virtual screening. Water molecules were then removed and the missing bond order and geometries were edited. Hydrogen atoms were added and the combined complex structure was submitted for protein preparation and energy minimization calculation using ICM and Schrodinger. The fully refined structure with bound ligand molecule was further submitted for grids calculation to define the active site as the collection of amino acids enclosed within a 12 Å radius sphere centered on the bound ligand (Figure 2.1). The target LSD1was optimized using Monte Carlo simulation and energy optimizations. 2.8.1.1.2 Preprocessing of three-dimensional ligand databases The external source database in the form of sdf format was processed using the ligand preparation tools. The final coordinates were stored in multi-sdf files. Custom filters included Lipinski's rule-of-five (Ro5) and manual filtering to remove very large molecules, dimers, polymers, molecules containing unusual heteroatoms, and highly reactive functional groups. Each of the databases were combined together with a final library of ~2 million molecules commercially available from 26 vendors were considered 69 for virtual screening using Glide SP/XP, ICM, and GOLD. 2.8.1.1.3 Virtual Screening Method A flow scheme indicating the steps of the virtual screening (VS) process is shown in Figure 2.2. The database of 13 million library compounds was curated using Ligprep, the filters from Section 2.8.1.1.2, and Glide HTVS methods to attain the set of ~ 2 million compounds screened against the prepared target. Grid potentials were rapidly generated which accounted for shape of the binding pocket, hydrophobicity, electrostatic potentials, and hydrogen-bonding profile. The compounds were screened using our own workflow (Figure 2.2) for LSD1 binding properties using a rigid target and flexible ligands in the internal coordinate's space. The compounds experimentally confirmed as LSD1 inhibitors were used for regular docking into LSD1. Docking calculations of the LSD1 inhibitors were performed using the ICM and Glide docking module with default setup and rescoring with GOLD. The structures of the active compounds were energy minimized in the same environment and saved in PDB format. These energy-minimized inhibitors were then reposed into ICM and converted into ICM object, and MMFF charges were assigned for each of the ligand. Docking took an average of 3-4 min/molecule on a four AMD 64-bit processors RedHat linux server with 4 GB of RAM. The speed for each compound was dependent on the number of torsional degrees of freedom. 70 2.8.1.1.4 Postprocessing and Compound Selection Criteria Compounds having desired scores, hydrogen bond formation and hydrophobic interactions were estimated by interatomic distances for further analysis. The conformational stability of each candidate was also estimated by force field energy difference between the complexes conformation and freely minimized conformation, and the top-scoring candidates from this category were selected for further analysis. Compounds in each of the three categories were visually inspected to eliminate candidates without ideal hydrogen bond geometry, hydrophobic molecular surfaces, or torsion angles. The resulting 121 focused screening structures were further analyzed using molecular property filters in QikProp, with calculated log S, permeability (Caco2 and MDCK) and Lipinski like criteria. 2.8.2 Analytical Data for Purchased Hits The commercially available hit compounds given in Table 1 (1-10) were purchased from ChemBridge, http://www.hit2lead.com, and Enamine, http://www.enamine.net. Their characterizations were confirmed using 1H NMR and Mass Spec and their purity was determined by HPLC.  (E)-4-hydroxy-N'-(2-hydroxybenzylidene)benzohydrazide (1): 1 H NMR (400 MHz, DMSO-d6): δ 11.84 (s, 1H), 11.39 (s, 1H), 10.11 (s, 1H), 8.54 (s, 1H), 7.87 (m, 2H), 7.49 (d, 1H, J = 8.4 Hz), 7.28 (t, 1H, J = 8.4 Hz), 6.91 (m, 4H). ESI-MS: 256.2 [M+H]+.  (E)-N'-(5-chloro-2-hydroxybenzylidene)-4-hydroxybenzohydrazide (2): 1H NMR (400 MHz, DMSO-d6): δ 11.89 (s, 1H), 11.32 (bs, 1H), 10.04 (s, 1H), 8.57 (s, 71 1H),7.82 (m, 2H), 7.61 (d, 1H, J = 2.4 Hz), 7.29 (dd, 1H, J = 2.4 & 8.8 Hz), 6.95 (d, 1H, J = 8.4 Hz), 6.88 (m, 2H). ESI-MS: 290.7 [M+H]+.  (E)-4-hydroxy-N'-(1-(2-hydroxyphenyl)ethylidene)benzohydrazide (3): 1H NMR (400 MHz, DMSO-d6): δ 10.94 (s, 1H), 10.03 (bs, 1H), 7.84 (m, 2H), 7.62 (dd, 1H, J = 1.6 & 8.0 Hz), 7.29 (t, 1H, J = 8.4 Hz), 6.90 (m, 4H), 2.47 (s, 3H). ESIMS: 270.28 [M+H]+.  (E)-4-bromo-N'-(2-hydroxybenzylidene)benzohydrazide (4): 1H NMR (400 MHz, DMSO-d6): δ 12.01 (s, 1H), 11.15 (s, 1H), 8.62 (s, 1H), 7.88 (m, 2H), 7.75 (m, 2H), 7.53 (d, 1H, J = 8.8 Hz), 7.31 (t, 1H, J = 8.8 Hz), 6.93 (m, 2H). ESI-MS: 319.16 [M+H]+.  (E)-3-chloro-N'-(2-hydroxybenzylidene)benzohydrazide (5): 1H NMR (400 MHz, DMSO-d6): δ 12.09 (s, 1H), 11.12 (s, 1H), 8.64 (s, 1H), 7.98 (s, 1H), 7.90 (m, 1H), 7.67 (m, 1H), 7.57 (m, 2H), 7.31 (t, 1H, J = 7.6 Hz), 6.93 (m, 2H). ESI-MS: 274.70 [M+H]+.  (E)-N'-(1-(2-hydroxyphenyl)ethylidene)-3-(morpholinosulfonyl)benzohydrazide (6): ESI-MS: 403.4 [M+H]+, purity by HPLC 97.25%.  (E)-3-(morpholinosulfonyl)-N'-(1-(naphthalen-1-yl)ethylidene)benzohydrazide (7): ESI-MS: 437.5 [M+H]+.  5-chloro-N'-(2-fluoro-5-(morpholinosulfonyl)benzoyl)-2-methoxybenzohydrazide (8): 1H NMR (400 MHz, DMSO-d6): δ 7.90 (m, 3H), 7.51 (m, 2H), 7.22 (m, 1H), 4.02 (s, 3H), 3.71 (m, 4H), 2.97 (m, 4H). ESI-MS: 471.8 [M+H]+.  N'-(3-chlorobenzoyl)-2-fluoro-5-(morpholinosulfonyl)benzohydrazide (9): 1 H NMR (400 MHz, DMSO-d6): δ 8.01 (m, 2H), 7.92 (m, 2H), 7.53 (m, 3H), 3.69 72 (m, 4H), 2.96 (m, 4H). ESI-MS: 441.8 [M+H]+.  (E)-N,N-diethyl-3-(2-(1-(p-tolyl)ethylidene)hydrazinecarbonyl)benzenesulfonamide (10): ESI-MS: 387.5 [M+H]+, purity by HPLC 94%. 73 Supplementary Table S2.1. Docking scores of compounds 1-10 S. No ICM score Glide score Gold fitness score 1 -42.25 -8.14 56.26 2 -42.25 -7.92 58.21 3 -21.91 -7.87 51.29 4 -37.77 -8.64 57.69 5 -36.3 -8.84 47.98 6 -24 -6.26 43.26 7 -20.97 -6.14 46.64 8 -18.39 -6.63 49.93 9 -8.16 -7.21 41.86 10 -8.5 -6.81 52.19 Structure 74 Supplementary Table S2.2. Commercially available LSD1 hits (111) from the list of 121 compounds selected. S. No Structure 11 IC50 (μM) LSD1 >100 ICM Score GLIDE Score GOLD Fitness (kcal/mol) (kcal/mol) Score (kcal/mol) -16.89 -4.87 27.21 12 >100 -16.34 -4.89 29.21 13 >100 -16.21 -4.76 24.21 14 >100 -21.21 -5.27 28.23 15 12.2 -17.22 -5.12 18.21 16 18.6 -26.81 -6.96 28.21 17 67.3 -27.28 -5.23 29.81 18 >100 -17.79 -7.43 22.74 19 >100 -14.34 -5.99 31.04 20 >100 -17.24 -4.76 20.17 21 >100 -28.21 -6.29 30.61 75 Table S2.2 Continued S. Structure No 22 IC50 (μM) LSD1 >100 23 >100 ICM Score GLIDE Score GOLD Fitness (kcal/mol) (kcal/mol) Score (kcal/mol) -26.24 -7.14 34.82 -23.23 -5.86 32.76 24 >100 -21.28 -6.13 32.52 25 >100 -14.93 -4.21 20.12 26 >100 -13.34 -7.24 32.61 27 >100 -11.21 -6.21 25.21 28 >100 -11.29 -5.34 23.78 29 >100 -16.25 -4.88 30.22 30 0.196 -42.25 -8.14 56.26 31 22 -27.29 -6.77 31.55 32 >100 -21.89 -6.69 32.31 33 >100 -13.29 -6.86 39.03 76 Table S2.2 Continued S. Structure No 34 IC50 (μM) LSD1 >100 ICM Score GLIDE Score GOLD Fitness (kcal/mol) (kcal/mol) Score (kcal/mol) -22.95 -6.29 37.19 35 37 -21.38 -7.22 36 17 -19.29 -8.66 25.94 29.93 37 >100 -13.12 -5.13 22.14 38 >100 -17.37 -4.77 25.65 39 >100 -18.58 -4.86 25.92 40 >100 -16.43 -5.16 22.74 41 >100 -19.99 -6.16 21.04 42 >100 -16.19 -4.42 20.17 43 >100 -17.23 -5.66 20.61 44 >100 -13.87 -3.33 24.82 45 >100 -11.81 -5.77 22.76 46 >100 -17.99 -5.99 22.52 77 Table S2.2 Continued S. Structure No 47 IC50 (μM) LSD1 >100 ICM Score GLIDE Score GOLD Fitness (kcal/mol) (kcal/mol) Score (kcal/mol) -14.39 -4.96 20.12 48 >100 -53.29 -4.77 22.61 49 >100 -17.64 -4.52 25.21 50 >100 -17.31 -4.86 23.78 51 >100 -21.28 -7.33 30.22 52 >100 -19.73 -4.97 31.55 53 >100 -17.34 -3.59 22.31 54 >100 -20.21 -6.76 39.03 55 >100 -26.29 -5.23 37.19 56 >100 -26.25 -6.22 35.94 57 67 -21.28 -6.66 39.93 58 >100 -19.33 -6.33 32.14 59 >100 -29.84 -5.67 25.65 78 Table S2.2 Continued S. Structure No 60 IC50 (μM) LSD1 >100 ICM Score GLIDE Score GOLD Fitness (kcal/mol) (kcal/mol) Score (kcal/mol) -16.23 -4.19 25.92 61 >100 -11.97 -3.16 19.22 62 >100 -14.27 -3.17 20.71 63 18 -21.88 -4.42 22.23 64 >100 -17.13 -4.22 22.07 65 >100 -16.55 -5.13 26.62 66 >100 -17.11 -4.37 30.49 67 >100 -19.39 -2.79 33.71 68 >100 -16.87 -4.69 31.58 69 32 -21.88 -3.17 30.98 70 >1 uM -24.43 -6.52 30.62 71 >1 uM -23.94 -6.33 30.97 72 >1 uM -21.41 -7.23 31.28 79 Table S2.2 Continued S. Structure No 73 IC50 ICM Score GLIDE Score GOLD Fitness (μM) (kcal/mol) (kcal/mol) Score LSD1 (kcal/mol) >1 uM -21.99 -9.47 32.23 74 >1 uM -26.25 -8.99 35.26 75 >1 uM -29.18 -7.79 36.75 76 >1 uM -24.23 -7.17 30.42 77 >1 uM -23.37 -7.43 38.68 78 >1 uM -21.81 -7.46 30.59 79 >1 uM -26.54 -7.13 30.29 80 >1 uM -26.45 -8.17 35.82 81 >1 uM -27.31 -8.21 38.72 82 >1 uM -26.99 -7.06 31.62 83 >1 uM -26.35 -6.20 30.01 84 >10 uM -28.18 -6.42 31.67 84 >10 uM -22.33 -8.93 31.27 86 >10 uM -26.39 -8.13 34.82 87 >10 uM -31.96 -6.17 30.89 88 >10 uM -29.64 -6.86 31.04 80 Table S2.2 Continued S. Structure No 89 IC50 ICM Score GLIDE Score GOLD Fitness (μM) (kcal/mol) (kcal/mol) Score LSD1 (kcal/mol) >10 uM -21.75 -7.36 33.14 90 >10 uM -29.81 -7.77 32.02 91 >10 uM -26.79 -7.44 33.02 92 >10 uM -32.55 -8.16 33.69 93 >10 uM -19.28 -6.29 34.21 94 >10 uM -25.66 -7.67 31.48 95 >10 uM -21.77 -8.16 37.94 96 >10 uM -23.61 -7.16 33.75 97 >10 uM -29.59 -7.97 30.41 98 >10 uM -29.435 -8.89 32.92 99 >10 uM -31.41 -6.16 33.19 100 >10 uM -31.89 -.923 32.41 101 >10 uM -36.29 -7.87 34.16 102 >10 uM -24.19 -7.22 33.67 81 Table S2.2 Continued S. Structure No 103 IC50 ICM Score GLIDE Score GOLD Fitness (μM) (kcal/mol) (kcal/mol) Score LSD1 (kcal/mol) >10 uM -27.11 -7.66 34.92 104 >10 uM -22.17 -7.17 31.97 105 >10 uM -32.77 -6.95 31.62 106 >10 uM -36.74 -6.16 30.52 107 >10 uM -36.75 -6.14 30.65 108 >10 uM -31.56 -7.27 41.62 109 >10 uM -21.87 -8.19 31.56 110 >10 uM -34.21 -8.79 32.61 111 >10 uM -39.88 -8.29 33.41 112 >10 uM -34.13 -7.42 37.79 113 >10 uM -33.39 -6.76 32.73 114 >10 uM -31.21 -8.29 32.51 115 >100 uM -16.44 -4.19 21.99 116 >10 uM -24.21 -7.39 32.08 82 Table S2.2 Continued S. Structure No 117 IC50 ICM Score GLIDE Score GOLD Fitness (μM) (kcal/mol) (kcal/mol) Score LSD1 (kcal/mol) >10 uM -29.78 -6.49 35.05 118 >10 uM -24.43 -6.41 30.12 119 >10 uM -23.89 -7.99 32.45 120 >10 uM -21.29 -6.16 22.08 121 >10 uM -16.74 -5.19 25.05 83 Supplementary Table S2.3. Tanimoto similarity coefficients comparing compound 12 and known LSD1 inhibitors from Chart 1 Compound A B C D E F G H I J K L M N O P Tanimoto Similarity score 0.26 0.21 0.31 0.26 0.22 0.36 0.24 0.28 0.38 0.35 0.39 0.29 0.11 0.11 0.32 0.11 84 Supplementary Table S2.4. Off-target inhibition assay results. Activity % ± SD [12] (nM) CYP1A2 CYP2C19 CYP2C9 CYP2D6 CYP3A4 30000 10000 3333 1111 370 123 41.2 13.7 4.57 1.52 0.51 60±3 86±1 87±2 96±1 97±8 93±3 105±7 106±4 111±3 110±2 53±1 91±2 105±0.4 111±3n/a 110±0.1 107±12 114±1 106±11 117±3 89±3 46±2 84±1 98±3 108±0.2 111±1 107±5 109±1 107±8 112±1 107±6 94±2 101±11 102±10 105±12 110±8 105±12 108±6 104±12 109±6 106±2 21±2 46±2 71±4 87±10 100±3 98±9 98±1 99±9 105±1 96±5 100000 30000 10000 3000 1000 300 100 D-LDH 103±7 103±6 106±5 112±5 121±6 118±5 119±5 GO 105±3 106±3 108±6 92±3 97±0.3 109±3 102±1 hERG 98±4 104±2 100±6 102±7 98±3 95±4 98±2 107±0.1 93±2 100±2 Supplementary Table S2.5. Different model fits for enzyme kinetics results. Model vmax (F/s) ± SE Km (μM) ± SE ki (nM) ± SE DMSO R2 1 nM R2 3 nM R2 10 nMR2 30 nM R2 100 nM R2 Global R2 Competitive 635.8±12.83 0.919±0.1128 4.136±0.7027 0.9033 0.8788 0.9022 0.8492 0.2257 -0.6184 0.8599 Noncompetitive 688.9±10.80 1.310±0.0928 39.04±3.046 0.9269 0.8774 0.9077 0.9154 0.9289 0.6953 0.9239 Uncompetitive 695.8±11.67 1.411±0.1042 32.76±2.67 0.9275 0.8758 0.9076 0.9060 0.9146 0.5652 0.9198 85 Supplementary Figure S2.1. Binding Site Model and definition of active site of LSD1 structure generated from PDB ID 2Z5U. Supplementary Figure S2.2. Flow diagram for the Virtual Ligand Screening (VLS) using ICM-VLS, Schrodinger workflow GOLD programs. 86 Supplementary Figure S2.3. Complete reaction schemes for compounds 11-22 87 Supplementary Figure S2.4. LC-MS Data for Compound 12 (96% purity). 88 CHAPTER 3 REVERSIBLE LSD1 INHIBITION INTERFERES WITH GLOBAL EWS/ETS TRANSCRIPTIONAL ACTIVITY AND IMPEDES EWING SARCOMA TUMOR GROWTH Savita Sankar and Emily R. Theisen are co-first authors of this work. SS and ERT executed RNA-seq studies with HCI2509. SS profiled EWS/ETS knockdown by RNAseq. SS and ERT jointly analyzed sequencing data. SS and ERT generated EWS/ETS knockdown constructs to assay the cell viability shift. ERT performed viability experiments with EWS/FLI expressed in NIH3T3cells. SS performed qPCR validation of target genes. Laura M. Hoffman stained, imaged, and analyzed cells by immunofluorescence microscopy. SS and ERT performed soft agars. ERT performed apoptosis assays. Timothy Mulvihill performed global methylation analysis. SS, ERT, and TM peformed HMOX1 analysis. ERT and Jared Bearss wer responsible for xenograft experiments. ERT and SS evaluated HMOX1 levels in tumors. ERT wrote the manuscript. Reproduced with permission from Sankar S and Theisen ER, et al. Clinical Cancer Research 2014 20 (17), 4584-4597. Copyright 2014 American Association for Cancer Research. 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 Supplementary Table S3.1. Primer Sequences for qRT-PCR analysis from RNA. Gene NKX22 Forward Reverse 5' CTACGACAGCAGCGACAACC 3' 5' GCCTTGGAGAAAAGCACTCG 3' 5' CGAAGTAAATGCCCCAGATGA 3' 5' GGTGGGGAAAGGCTGATGAAC 3' CAV1 5' ATCGACCTGGTCAACCGCGAC 3' E2F1 5' GCCACTGACTCTGCCACCATA 3' IGF1 5' GAAGATGCACACCATGTCCTC 3' 5' CTCCAGCCTCCTTAGATCACA 3' GSTM4 HMOX 1 IGFBP3 5' GCTGCCCTACTTGATTGATGG 3' 5'TGATTGGAGACGTCCATAGCC 3' 5' AACTTTCAGAAGGGCCAGGT 3' 5' GTAGACAGGGGCGAAGACTG 3' 5' CATCAAGAAAGGGCATGCTAA 3' 5' CTACGGCAGGGACCATATTCT 3' CDH1 5' TGCCCAGAAAATGAAAAAGG 3' 5' GTGTATGTGGCAATGCGTTC 3' RUNX2 5' CCTCGGAGAGGTACCAGATG 3' 5' AAACTCTTGCCTCGTCCACT 3' 104 Supplementary Figure S3.1. Transcriptional Profiling of HCI2509 in A673 and TTC466. (A,B) Cell viability assay showing the difference in HCI2509 sensitivity between (A) TTC-466 cells with control and EWS/ERG knockdown or (B) NIH 3T3 cells with control and EWS/FLI expression. The dose-response curves were determined after 96 hours of treatment and normalized to the vehicle controls. n=3 for each point. Error bars denote standard deviation. EC50 and 95% CI were determined using GraphPad Prism 6. Note the line for ERG-RNAi data in (A) is a connecting line, not a curve fit. (C) Venn diagram representations of the overlap between the EWS/FLI and EWS/ERG transcription profiles, both generated by RNA-seq. Chi-square determined p-values are indicated with the observed contingency tables shown. (D) Gene set enrichment analysis (GSEA) using genes regulated by EWS/FLI in A673 cells (RNA-seq) as the rank-ordered dataset and the EWS/ERG-upregulated or the EWS/ERG-downregulated genesets (RNAseq). Normalized enrichment scores (NES) and p-values are shown. (E,F) Venn diagram representations generated from respective RNA-seq data sets using default cutoffs (2-fold change, FDR=10%). (E) represents the overlap between the HCI2509 and the EWS/FLIknockdown transcription profiles, both generated in A673 cells; (F) the overlap between the HCI2509 and the EWS/ERG-knockdown transcription profiles, both generated in TTC-466 cells. Chi-square determined p-values are indicated with the observed contingency tables shown. (G,H) GSEA using genes directly regulated by EWS/FLI in A673 cells (ChIP-chip and RNA-seq overlap) as the geneset and HCI2509 regulated genes in A673 cells (RNA-seq) as the rank-ordered dataset in (G) or the vorinostat regulated genes in A673 cells (microarray) as the rank-ordered dataset in (H). Normalized enrichment scores (NES) and p-values are shown. (I,J) Top ten categories from DAVID functional analysis of the (I) EWS/FLI up-/HCI2509 down- and EWS/FLI down/HCI2509 upregulated genesets and (J) EWS/ERG up-/HCI2509 down- and EWS/ERG down-/HCI2509 upregulated genesets. The log transformed enrichment scores for each category are indicated on the x-axis. (K) Validation of NKX2.2, CAV1, GSTM4, E2F1, IGF-1, RUNX2, IGFBP3, HMOX1 and CDH1 as HCI2509 targets by qRT-PCR analysis using EWS-502, SK-ES-1, SK-N-MC, and TC-71 cells treated for 48 hours with vehicle or HCI2509 at 2xEC50. The p-value for each fold change is < 0.05 (n=3). Individual pvalues are reported in Supplementary Table S2. 105 106 107 108 109 Supplementary Figure S3.2. Morphological changes with HCI2509 treatment (A) Whole-field immunofluorescence images of A673 cells treated with increasing doses of HCI2509 for 72 hours. Staining was performed for F-actin stress fibers (red - phalloidin) and for focal adhesions (green - paxillin), and nuclei (blue). HCI2509 induced a dosedependent increase in the cell spreading and morphology. (B,C) Immunofluorescence images of A673 cells treated with either control siRNA or LSD1 siRNA 50 nM for 48 hours. Staining was performed for (B) LSD1 and (C) F-actin stress fibers. Measurements of LSD1 nuclear signal and cell area were performed on at least 6 fields for each transfection. Decrease in LSD1 nuclear signal correlated with more organized actin fibers and cell spreading. (D) Whole-field and (E) close up immunofluorescence images of TTC-466 cells treated with increasing doses of HCI2509 for 3 days. Staining was performed for F-actin stress fibers (red - phalloidin) and for focal adhesions (green - paxillin), and nuclei (blue). HCI2509 induced a dose-dependent increase in the cell spreading and morphology. (F) Measurement of cell area in pixels in phalloidin images shows a dose-dependent increase in cell spreading with of HCI2509. TTC-466 cells were fixed and stained with phalloidin. Cell area was quantified as previously described (39). Data is shown as scatter plot with mean plus standard deviation, and unpaired parametric t-test was used to determine p-values (* p <0.05, *** p < 0.0001). 110 111 Supplementary Figure S3.3. Effects of HCI2509 on Transformation, Methylation, and Apoptosis. (A,B,C) Quantification of colonies formed by (A) EWS-502, (B) TC71, and (C) SK-ES-1 cells treated with either vehicle (0.3% DMSO) or varying doses of HCI2509. Error bars indicate SD of duplicate assays. EC50 values were determined using GraphPad Prism 6. (D,E,F) Cell viability and caspase activation at 0, 24, and 48 hours in (D) SK-N-MC, (E) TC71, and (F) SK-ES-1 cells treated with 2xEC50 HCI2509. Measurements were normalized to their respective vehicle (0.3% DMSO) sample at the appropriate time point. Error bars indicate SD (n=3). 112 Supplementary Figure S3.4. Regulation of HMOX1 in Ewing sarcoma. (A) qRT-PCR for HMOX1 induction following treatment with candidate LSD1 inhibitors with respect to inhibitor biochemical potency against LSD1 in a biochemical assay (Cayman Chemical). (B) Western blot analysis to demonstrate expression of the RNAi-resistant 3xFLAG tagged EWS/FLI, Δ22, or R2L2 cDNA constructs using an anti-FLAG antibody in A673 cells expressing a control shRNA (Luc) or an EWS/FLI shRNA. Tubulin was used as the loading control. (C) qRT-PCR analysis to assess level of knockdown of various corepressors or HMOX1 induction in A673 cells treated with either Luc-RNAi or RNAi for REST (REST p=3.53E-6, HMOX1 p=1.92E-2), RCoR1 (RCoR1 p=1.18E-4, HMOX1 p=3.67E-2), NCoR/SMRT (NCoR/SMRT p=2.60E-7, HMOX1 p=5.85E-1), or Sin3A (Sin3A p=1.27E-6, HMOX1 p=1.57E-1). Error bars indicate SD and p-values were determined using students t-test (n=3). (D) Western blot analysis for HMOX1 expression in A673 cells infected either with empty vector or an HA-tagged HMOX1 cDNA using an anti-HMOX1 antibody. Tubulin was used as a loading control.(E) Growth assays (3T5) for A673 cells described in (D). Student's t-test showed no significant difference in growth curves. (F) Quantification of colonies formed by A673 cells described in (D). Error bars indicate SD of duplicate assays. 113 114 Supplementary Figure S3.5. Tumor volume, body weight and blood counts. (A) In vivo subcutaneous hind-flank xenograft studies measuring tumor volume for animals bearing tumors grown from (A) SK-ES-1 cells. The p-value was determined by 2-way ANOVA comparing the treatment curve to the vehicle curve. Individual tumor growth curves are shown for the vehicle-treated (blue) and HCI2509-treated (red) groups. (B,C,D) Body weight measurements for animals bearing tumors grown from (B) A673 cells, (C) SK-NMC cells, and (D) SK-ES-1 cells. N=10 for all groups, with the exception of SK-N-MC HCI2509 treated group as noted. For body weights, the change in body weight normalized to day 0 was considered and a student's t-test was used to determine the pvalue. (E) Blood counts for white blood cells (WBC), hematocrit (HCT), and platelets (PLT) from immunodeficient mice treated intraperitoneally either with vehicle or 40 mg/kg HCI2509 MWF for 24 days ± SD. Blood was drawn using a cheek draw and assayed at both day 0 and day 24. The normal range is reported. 115 CHAPTER 4 REVERSIBLE INHIBITION OF LYSINE SPECIFIC DEMETHYLASE 1 IS A NOVEL ANTITUMOR STRATEGY FOR POORLY DIFFERNTIATED ENDOMETRIAL CARCINOMA Emily Rose Theisen and Snehal Gajiwala performed the experiments and wrote the manuscript. Jared Bearss provided insight in experimental design. Sunil Sharma and Margit Janat-Amsbury designed and supervised the experiments and wrote the manuscript. Venkataswamy Sorna and Margit Janat-Amsbury contributed reagents, facilities, and personnel. All authors read and approved the final manuscript. Reproduced with permission from Theisen ER, et al. BMC Cancer 2014, 14:752. Copyright 2014 BioMed Central. 116 117 118 119 120 121 122 123 124 125 126 127 128 Supplementary Figure S4.1. Time course evaluation of cell cycle perturbations caused by HCI2509 treatment. (A,B) Cell cycle populations of (A) AN3CA and (B) KLE cell lines after exposure to vehicle (0 and 48 hours) or 3 μM HCI2509 (6, 12, 24, and 48 hours). For AN3CA and KLE cells, 2 x 104 counts and 1 x 104 counts were used, respectively. Data are representative of two biological replicates. 129 130 131 Supplementary Figure S4.2. TUNEL assay replicates and controls. (A,B) Fluorescence microscopy images of (A) AN3CA and (B) KLE cell lines after exposure to either vehicle or 3 X IC50 HCI2509 and then stained with TUNEL for apoptotic nuclei (green), DAPI for nuclei (blue), and phalloidin for actin (red). HCI2509 treatment induced apoptosis with apoptotic cells marked with (*). (C) Fluorescence microscopy images of TUNEL negative and positive controls with untreated AN3CA and KLE cells. Negative controls were generated by adding labeled nucleotide with no enzyme and positive controls were generated by pretreating DNase before TUNEL labeling. Cells are stained with TUNEL (green), DAPI (blue), and phalloidin for actin (red). 132 133 134 Supplementary Figure S4.3. In depth xenograft model analysis. (A) Individual mouse images from study day 7 (day 0 of treatment). All images are on the same luminescence scale from 1.54 x 104 p/s to 8.66 x 106 p/s. (B) Quantified bioluminscence measurements of both the vehicle and HCI2509 treatment groups pooled. Total flux (photons/second) was rank ordered and plotted on a semilog plot. The linearity of the log-transformed data supports a log-normal distribution. (C) Fisher's exact test shows significant association of HCI2509 treatment with tumor regression. Both the observed and expected contingency tables are shown with the reported p-value. (D) Tumor volume and body weight measurements including both the untreated and unimplanted control. Tumor volumes are plotted as the geometric mean of the observed luminescent signal and body weight is plotted as the average and SD. 135 136 CHAPTER 5 CONCLUSIONS AND OUTLOOK 5.1 Conclusions This work described the discovery of a novel N'-(1-phenylethylidene)benzohydrazides compound series as potent, specific and reversible LSD1 inhibitors. Hitto-lead optimization identified HCI2509 as a lead compound with nanomolar binding affinity for LSD1. HCI2509 is a noncompetitive LSD1 inhibitor with no detectable activity against either of the monoamine oxidases or other tested flavoenzymes. Treatment with HCI2509 decreased viability across numerous cancer cell lines. The overarching purpose of this work was to address the hypothesis that by targeting the key histone demethylase, potent and specific LSD1 inhibitors will exhibit single-agent efficacy in solid tumor models. We chose both Ewing sarcoma and endometrial carcinoma as model systems to test this hypothesis with HCI2509. 5.1.1 High-throughput Virtual Screening Leads to HCI2509 This work started with a high-throughput virtual screen of approximately 2 million small molecules against the crystal structure of LSD1 (PDB ID: 2Z5U). The compound library used was built with commercially available small molecules prefiltered for favorable physicochemical properties. An initial hit, compound 1 in Chapter 137 2, was identified and showed LSD1 inhibition with a biochemical IC50 of ~100-300 nM. Compound 1 was then optimized through iterative medicinal chemistry and biochemical testing to arrive at HCI2509 ("compound 12" in Chapter 2) which has a Ki of ~30 nM. The hit to lead optimization process improved the specificity profile of the N'-(1phenylethylidene)-benzohydrazides series considerably. Compound 1 showed activity against MAO B comparable to that of the MAO inhibitor tranylcypromine, while HCI2509 showed no detectable activity against either MAO A or B. HCI2509 was shown to be a noncompetitive inhibitor that perturbed LSD1 conformation in a manner distinct from tranylcypromine. The protein purification optimized to support these studies is reported in Appendix A. While the binding site of HCI2509 is predicted to fall in or near the FAD binding pocket of LSD1 based on the docking setup used, no evidence has been gathered to date which shows FAD displacement. This work showed little off-target activity against the cytochrome P450 enzymes as well as human Ether-à-go-go (hERG). Importantly and finally, HCI2509 showed activity in cell-based assays, with two additional and unpublished cell-line panels reported in Appendix B. These results informed the decision to test HCI2509 in Ewing sarcoma and endometrial carcinoma. 5.1.2 HCI2509 is Uniquely Active in Ewing Sarcoma HCI2509 showed particularly potent activity in multiple Ewing sarcoma cell lines. LSD1 was recently discovered as critical for EWS/FLI-mediated target gene repression through recruitment as a member of the NuRD complex (1). We hypothesized that through reactivation of EWS/FLI-repressed tumor suppressors, we were inducing apoptosis and impairing transformation in Ewing sarcoma cell lines. In order to test this 138 we looked at the transcriptional changes induced by HCI2509 treatment by RNAsequencing and compared them to the transcriptional changes caused by EWS/FLI knockdown in the same cell line. We found that not only was HCI2509 disrupting EWS/FLI-mediated repression globally, but that it also impaired EWS/FLI-mediated transcriptional activation of critical oncogenes. This effect was observed at both direct and indirect target genes, and contrasted with HDAC inhibition with vorinostat, which only affected the EWS/FLI-repressed targets. The effect was so striking that we next asked whether or not it was specific to EWS/FLI, or whether HCI2509 would show comparable effects in a cell line containing and alternative EWS/ETS fusion, specifically EWS/ERG. We found the same phenomena where HCI2509 treatment disrupted the global EWS/ERG transcriptional program both at activated and repressed target genes. This was also the first time that the EWS/ERG transcriptional profile was published and, while unsurprising, it was observed that both EWS/FLI and EWS/ERG regulated similar transcriptional programs. We further hypothesized that these transcriptional changes would be consistent across cell lines. Most Ewing sarcoma cell lines tolerate EWS/FLI knockdown poorly as compared to A673 cells and RNA-sequencing experiments are costly in both time and money. We thus performed qPCR after treatment with HCI2509 in several cell lines to assess the transcriptional changes in a 9-gene panel representing characteristic EWS/FLI target genes and saw consistent downregulation of EWS/FLI activated targets and vice versa across Ewing sarcoma cell lines. Ewing sarcoma kills patients after relapse and metastasis. The cellular morphologies associated with Ewing sarcoma metastasis have been characterized by 139 Chaturvedi, et al. (2) and show that EWS/FLI most likely promotes metastatic phenotypes through decreased cellular adhesion, though not through migration and invasiveness. This mechanism is different from epithelial cancers, and so while it does not involve the epithelial-to-mesenchymal transition that LSD1 helps to regulate (3-5), we wanted to evaluate the effects in the same in vitro morphology assay used in (2). HCI2509 and siRNA-mediated knockdown of LSD1 both recapitulated the EWS/FLI knockdown phenotype here and HCI2509 prevented transformation in colony forming assays, confirming that LSD1 inhibition reverses some of the cellular phenotypes driven by EWS/FLI as would be predicted by the transcriptional profiling. This was also the first study to demonstrate target engagement in cells, though indirectly, through HMOX1 induction. LSD1 was found associated with the HMOX1 promoter and siRNA-mediated knockdown of LSD1 showed dose-dependent increases in HMOX1 transcript. HMOX1 was of interest, because not only is it repressed by EWS/FLI and activated by HCI2509, but it is also downregulated in primary patient samples (6). This suggests that LSD1 inhibition may prove relevant to the transcriptional mechanism at work in human patients. Interestingly, even with target engagement demonstrated, the changes in global methylation at LSD1 substrates H3K4 and H3K9 were more subtle than anticipated in both the A673 and TTC466 cell lines. No significant changes were observed at H3K4, though H3K9me2 and H3K9me3 both significantly increased. A mechanistic understanding is yet to be determined. Most importantly, LSD1 inhibition with HCI2509 showed single-agent antitumor efficacy in multiple Ewing sarcoma xenograft models. These models are notoriously difficult to carry out, as tumors typically show a wide range of growth rates, and are 140 likely log-normal distributed in this manner. However, HCI2509 showed a clear effect in all three models tested, with two complete regressions observed, and no observed toxicity. These data are extremely promising, as few treatment options work as single agents in mouse models of Ewing sarcoma, and they suggest a therapeutic window may be wide enough in this disease. 5.1.3 Antitumor Activity of HCI2509 in Endometrial Carcinoma Having shown both target engagement and ruled out a nonspecific cytotoxic mechanism of action, we wanted to test whether HCI2509 would also show efficacy in a second malignancy also associated with epigenetic misregulation, though perhaps through more diverse mechanisms than a single translocation. Type II endometrial carcinoma was recently shown to commonly have driver mutations in chromatin regulatory enzymes, and was also sensitive to LSD1 inhibition in our cell-line screens with HCI2509. In this model system, LSD1 inhibition showed decreased proliferation and transformation in two cell lines that were refractory to hormone treatment. LSD1 inhibition caused perturbation of the cell cycle, though the mechanisms through which this occurs remain undetermined. However, unlike Ewing sarcoma cell lines, H3K4me3 was significantly upregulated with HCI2509 treatment, and H3K9me2 varied between cell lines. HCI2509 induced HMOX1 and the adhesion gene CDH1, further supporting that the proliferation, transformation, and cell cycle effects were the result of LSD1 target engagement. Significantly, LSD1 inhibition induced apoptotic cell death. We chose to pursue the question of in in vivo efficacy with the KLE cell line in an orthotopic model. Type II 141 endometrial cancer is commonly disseminated in the peritoneal cavity and subcutaneous models simply cannot recapitulate this tumor environment. Epigenetics represents the intersection of the genes with the environment and in a disease driven by genes, environment, and epigenetics, evaluation of an epigenetic therapy requires testing in the most representative system feasible. In this model system of endometrial cancer, HCI2509 treatment was significantly associated with tumor regression, demonstrating single agent efficacy. 5.1.4 Noteworthy Observations Taken together, this body of work both supports epigenetic therapy as a powerful tool in the treatment of cancer and underscores the amount of work left to be done in order to fully understand the mechanistic basis for therapeutic efficacy. For example, the EWS/ETS-based activity seen in Ewing sarcoma suggests LSD1 may act in a manner that is Ewing sarcoma specific. While the precise mechanisms are the topic for future studies, this particular instance is proof-of-concept that in at least some malignancies, epigenomic misregulation is so central to the disease etiology that it may prove feasible to hijack the whole oncogenic program. Ewing sarcoma is unique, however, in being relatively mutationally silent at onset (7), such that malignant reprogramming happens exclusively through epigenomic mechanisms. In other malignancies, we need better tools to identify which patients might benefit from particular epigenetic therapies. While LSD1 is broadly observed to be upregulated across dedifferentiated cancers, which perhaps hints at a common role in malignancy, it should be noted that the changes to global histone methylation with HCI2509 were markedly different in Ewing sarcoma and endometrial 142 carcinoma. Each disease will likely have its own etiology requiring deep sequencing experiments to fully grasp. However, some progress is being made, with SOX2 overexpression suggested as a biomarker denoting sensitivity to LSD1 inhibition in lung, breast, and ovarian carcinomas (8). Another interesting observation not fully discussed in Chapters 2-4 is the fact that while both Ewing sarcoma and endometrial carcinoma showed in vitro sensitivity which translated in vivo, this is not true for all cell lines. In fact, the PC3 cell line showed very little sensitivity in the cell line screens up to 3 μM (Appendix B), but was exquisitely sensitive in xenograft models when the plasma concentrations were consistently 200-400 nM (9). Further, this is consistent with drug development efforts in GlaxoSmithKline's LSD1 inhibitor program. Similarly insensitive cell lines in vitro show delayed tumor growth in xenograft models (10). The reasons for this remain unclear, but are likely related to poor correlation between in vitro culture conditions and the in vivo tumor microenvironment and how these signals are integrated at the epigenomic level. It also underscores the importance of well-designed in vivo studies which recapitulate the disease state as can best be achieved before human trials as the differences between mice and men are certainly as large as that between plastic and an immunodeficient mouse. 5.2 Future Studies While on the whole the studies reported here show significant promise for LSD1 inhibitors as they progress to and through the clinic, several lines of inquiry became apparent as a result of this work. There are important questions to be asked from both a basic and translational perspective to build an understanding of the causal relationships 143 by which LSD1 drives cancer and how different modes of inhibition may impair those. This will enable more precise clinical science to get the right LSD1 inhibitor to the right patient as LSD1 inhibition reaches the clinic. 5.2.1 Biophysical and Biochemical Inquiries Several biochemical questions remain unanswered, the most obvious of which is, is there a better way to show LSD1 binding within cells, preferably direct binding? A recently reported technique utilizing an in-cell version of the thermal shift assay (CETSA) (11) for binding may provide the answer. In essence, this assay calls for drug exposure either in cell lysates or whole cells. After equilibration, the lysates or cells are incubated at temperatures varying from 40-64°C, at which point cells are lysed if necessary. Protein which has unfolded and crashed out is then removed by centrifugation and the protein of interest is detected in the supernatant by western blot. The antibody used needs to be fairly sensitive and specific for the protein of interest. Ultimately, similar to the fluorescent thermal shift assay used in Chapter 2, the readout for binding is increased amounts of protein at higher temperatures than the vehicle control, to demonstrate the stabilizing effect of ligand binding. The most immediate follow on question is whether or not the compound binds any other proteins in the cell. This would require synthesis of a biotinylated derivative and confirmation that biotinylation does not completely abolish binding affinity. Any proteins that pull down with the compound could be further confirmed with CETSA. The real elephant in the room is the binding mode of HCI2509. Given the complexities of LSD1 regulation discussed in Chapter 1, knowing the binding pocket 144 would enable the generation and testing of point-mutants, which should rescue enzymatic activity in the context of drug exposure. It would also identify potential mechanisms for the development of resistance. LSD1 has been crystallized in a number of hands (12-16) and in collaboration with the Chris Hill lab; crystallography is possible, though not required for clinical translation. 5.2.2 Further Routes of Inquiry in Ewing Sarcoma LSD1 inhibition in and the epigenomics of Ewing sarcoma are major areas of research that remain incompletely explored, but show great promise. At a basic level, the histone methylation data reported in Chapter 3 comprise a crude and preliminary evaluation of the epigenetic impacts of LSD1 inhibition in Ewing sarcoma cells. Really, the "epigenetics" of HCI2509 in Ewing sarcoma are not yet worked out in mechanistic detail and the observed changes in global histone methylation marks are quite subtle. In order to understand what changes, and whether these changes are associated with EWS/FLI or LSD1 requires a minimum of directed ChIP studies at candidate loci, but more likely ChIP-seq for the histone marks of interest, LSD1 and EWS/FLI in the absence or presence or drug or various knockdowns. The very first questions to address are the genomic co-localization patterns of EWS/FLI and LSD1, specifically at active targets, as well as the sites of histone methylation mark changes in the context of drug treatment. Other pieces of data show different transcriptional effects with different classes of LSD1 inhibitors. The irreversible inhibitors and HCI2509 appear to both modulate HMOX1 in a dose-dependent fashion around the cell viability IC50, which supports 145 HMOX1 as a bonfide LSD1 target. They both leave NR0B1 unaffected. Interestingly, the EWS/FLI activated gene NKX2.2 is downregulated by HCI2509 and the EWS/FLIrepressed CTGF is upregulated by HCI2509, but not the irreversible inhibitors (Figure 5.1). This is where identification of the binding mode and relevant protein-protein interactions partners would be potentially illuminating. Given that the binding mode is unknown and that LSD1 is bound, two possibilities exist, either there is something unknown about LSD1 biology or there is an unknown off-target effect, or both, and that these unknowns are particularly impactful in Ewing sarcoma. The next step is to test other potent and reversible inhibitors for the same transcriptional activity. While the discussion in Chapter 1 of positive controls for novel modes of LSD1 inhibition still applies, it is worth investigating whether knockdown of LSD1 by siRNA or shRNA recapitulate the transcriptional and other effects of HCI2509 in Ewing sarcoma cells by RNA-seq, morphology assessments, colony forming assays, and tumorigenic studies. Optimization of siRNA in Chapter 3 showed a 50% reduction in LSD1 protein levels as the maximum knockdown attainable. If LSD1 is as critical as the inhibition data would suggest, it may be difficult to optimize a system with 80-90% knockdown of LSD1. Moreover, if an shRNA can be optimized, to what extent do different LSD1 mutants, for example, enzymatically dead or truncated, rescue LSD1 knockdown? Answers to these questions would help map the relevant domains on LSD1 for more mechanistic biochemical studies to clarify the role of LSD1 in Ewing sarcoma. As mentioned, metastasis and recurrent disease are the killers in Ewing sarcoma, and the xenograft studies reported in Chapter 3 address only primary subcutaneous tumors. Intratibial models were used by Chaturvedi, et al. (2) to investigate the metastatic 146 behavior of Ewing sarcoma cells. Investigation of HCI2509 treatment in intratibial models of both mice and rats may prove useful in understaning the impact of LSD1 inhibition on the development of metastatic disease. Preliminary studies in nude rats show drastic differences in the development of metastases between vehicle and HCI2509 treated animals (Figure 5.2). The lung metastases in Figure 5.2 were undetectable using bioluminsecent imaging, such that optimization of a different imaging modality is required. Both CTGF and HMOX1 were observed to be downregulated in primary patient samples of Ewing sarcoma and upregulated with HCI2509 treatment. As was reported in Chapter 3, secreted proteins were significantly upregulate by HCI2509. IL8 is another secreted protein similarly downregulated by EWS/FLI and observed to be decreased in patient samples, and upregulated by HCI2509. TGFβR2 was validated as a tumor suppressor gene repressed by EWS/FLI, both in cell lines and the clinic, and was also induced by HCI2509 treatment (1). The continued optimization of animal models provides an opportunity to explore the pharmacodynamics of secreted proteins, such as IL8 or CTGF, which may correlate with pharmacological modulation of HMOX1 or TGFβR2 levels in tumor samples. In addition to response biomarkers, in order to translate LSD1 inhibition to the clinic for Ewing sarcoma, the use of HCI2509 needs to be assessed in combination with the other standards of care. This is required both in vitro and in vivo to evaluate the potential for drug synergy, antagonism, or unforeseen toxicity. The most likely candidates for these studies are irinotecan and temozolomide. Preliminary data from an SK-ES-1 xenograft model suggest that HCI2509 may show synergistic effects in 147 combination with temozolomide and larger studies are required to validate these results in other models of the disease (Figure 5.3). All in all, the results reported in Chapter 3 represent a major advance in our ability to target EWS/ETS-mediated oncogenic transcriptional programs. It has pushed forward in a way that raises several fundamental basic science questions to address the mechanisms by which one small molecule can flip a switch recently thought unflippable with such strategies. Moving toward the clinic will likewise require efforts to develop appropriate preclinical models to truly test in animals what we hope to test in the clinic in order maximize the predictive value of preclinical work. Ewing sarcoma is a rare disease that affects young adults, and any human trial in this population needs to be as tightly designed as possible to determine whether this could provide better therapeutic options for this aggressive malignancy. 5.2.3 Further Routes of Inquiry for Endometrial Carcinoma Given the promising results both in vitro and in vivo in endometrial carcinoma, several follow up studies are warranted. From the perspective of basic LSD1 biology, it is interesting, though perhaps not unexpected, that the observed changes in global histone methylation are different in endometrial cancer than those shown for Ewing sarcoma. Moreover, the two cell lines tested showed slightly different global histone methylation changes in response to HCI2509 exposure, notably H3K9me2 increased in AN3CA cells while remaining unchanged in KLE cells. The biological mechanisms driving this difference may provide insight about the varying role of LSD1 in different cell lines and differentiate which phenomena are most associated with the antitumor effects of 148 HCI2509. Testing this pharmacodynamic marker for LSD1 inhibition across more endometrial carcinoma cell lines, both Type I and Type II, could shed further light on the impact of HCI2509 on global histone methylation in endometrial cancer, and allow analysis of whether changes in any one mark predict sensitivity to HCI2509. It would also be helpful to know how global epigenetic changes translate to transcriptional changes in endometrial carcinoma, such that RNA-seq and ChIP-seq similar to those proposed for Ewing sarcoma should be pursued. Further, validation that LSD1 inhibition phenocopies siRNA-mediated LSD1 knockdown would strengthen this work, though this approach needs to be undertaken with careful consideration due to the complexities of LSD1 biology, described in Chapter 1. Translating this work to the clinic will required addressing whether the antitumor efficacy observed across cell lines also holds true in multiple xenograft models, including those derived from primary tumor tissue. Additionally, HCI2509 should be screened for synergy both in vitro and in vivo with the current standards of care. Pretreatment with HCI2509 for 24 hours showed no sensitization to progesterone in vitro for AN3CA or KLE (Figure 5.4), though this may not be true for other Type II endometrial carcinoma cell lines or models of Type I disease. Overall, the data reported for Type II endometrial cancer warrant continued preclinical evaluation of LSD1 inhibition in this aggressive gynecologic malignancy. 5.2.4 Formulation One of the major remaining hurdles for translation of this compound series to the clinic is the optimization of a more clinically acceptable formulation. The relative 149 hydrophobicity (logP=3.96) and high melting temperature (>220°C) of HCI2509 classify the compound as "brick dust." Moreover, the hydrazone core is amenable to a relatively planar conformation and contains several hydrogen bond donors and acceptors for interand intramolecular hydrogen bonding. Together, these factors contribute to drive HCI2509 to form needle-like crystals in aqueous formulations. Ideally, translation to the clinic would involve the development of an oral formulation or tablet. This would likely require the development of an amorphous form or salt, as well as inclusion of surfactants or wetting agents to promote dissolution. The hydrazone moiety also necessitates the use of enteric coating to prevent acid-catalyzed hydrolysis of the compound. Taken together, these factors have limited the available preclinical formulation strategies. Salt formation has been attempted multiple times by Venkataswamy Sorna to no avail. Ultimately, we utilized several different cosolvent strategies for the preclinical studies reported here, with both stable solutions and suspension tested in pharmacokinetic studies. Efficacy studies utilized a stable suspension of HCI2509 crystals dosed 30 mg/kg directly into the peritoneal cavity. There are different features of Ewing sarcoma and endometrial carcinoma to consider in the design of formulations to enable future studies in these diseases. Ewing sarcoma requires systemic treatment to target metastatic and micrometastatic disease as adjuvant therapy in concert with surgical resection. Oral tablets or intravenous routes of delivery would be appropriate. However, while metastatic endometrial cancer may benefit from a similar strategy, in cases where the disease remains localized to the peritoneal cavity or uterus, delivery via intravaginal gel may offer an attractive alternative route. This type of localized delivery minimizes the risk of off-target systemic toxicities or undesirable epigenetic reprogramming. Further, a 150 vaginal gel might be formulated to contain varying combinations of agents which show synergistic effect, for example, progesterone therapy. Varying the formulation strategy for the same epigenetic agent could tailor the desired epigenetic reprogramming effects to the specific needs of the disease of interest. 5.3 Outlook The overarching goal of this lab is to provide cancer patients with innovative targeted therapeutic options. This project addressed that larger goal through the identification and validation of a novel series of potent, specific, and reversible LSD1 inhibitors, both biochemically and in preclinical models of cancer. Ultimately the N'-(1phenylethylidene)-benzohydrazides series of inhibitors have cast new light on the biology of Ewing sarcoma that may lead to improved clinical care for this rare and aggressive disease through both translational and basic research. Additionally, the compound series identified herein showed single agent efficacy in Type II endometrial cancer, suggesting epigenetic inhibition may provide therapeutic benefit in this aggressive gynecologic malignancy. Detailed mechanistic studies are still required in both disease areas to fully elucidate the biological role for LSD1 and the mechanism by which HCI2509 acts. HCI2509 faces many hurdles on the road to the clinic. While HCI2509 is highly permeable, solubility remains a major challenge. Either analogues more amenable to salt formation or possessing more favorable solubility characteristics would be preferable for additional preclinical studies and clinical development. Candidate derivatives are currently under investigation at the Center for Investigational Therapeutics and remain a promising topic for future study. Until then, HCI2509 is a useful tool for proof-of- 151 concept and mechanistic studies in both in vitro and in vivo model systems and will inform the clinical development and use of novel reversible LSD1 inhibitors for Ewing sarcoma, endometrial carcinoma, and other malignancies. 5.4 References 1. Sankar S, Bell R, Stephens B, Zhuo R, Sharma S, Bearss DJ, et al. Mechanism and relevance of EWS/FLI-mediated transcriptional repression in Ewing sarcoma. Oncogene 2012;32:5089-100. 2. Chaturvedi A, Hoffman LM, Welm AL, Lessnick SL, Beckerle MC. The EWS/FLI oncogene drives changes in cellular morphology, adhesion, and migration in Ewing sarcoma. Genes Cancer 2012;3:102-16. 3. Lin T, Ponn A, Hu X, Law BK, Lu J. Requirement of the histone demethylase LSD1 in Snai1-mediate transcriptional repression during epithelial-mesenchymal transition. Oncogene 2010;29:4896-4904. 4. Lin Y, Wu Y, Li J, Dong C, Ye X, Chi YI, et al. The SNAG domain of Snail1 functions as a molecular hook for recruiting lysine-specific demethylase 1. EMBO J 2010;29:180316. 5. 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. 6. Kauer M, Ban J, Kofler R, Walker B, Davis S, Meltzer P, et al. A molecular function map of Ewing's sarcoma. PLoS ONE 2009;4:e5415. 7. Crompton B, Stewart C, Taylor-Weiner A, Alexa G, Kurek K, Calicchio M, et al. The genomic landscape of pediatric Ewing sarcoma. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Abstract nr 999. 8. Zhang X, Lu F, Wang J, Yin F, Xu Z, Qi D, et al. Pluripotent stem cell protein Sox2 confers sensitivity to LSD1 inhibition in cancer cells. Cell Rep 2013;5:445-57. 9. Theisen ER, Bearss J, Sorna V, Bearss DJ, Sharma S. Targeted inhibition of LSD1 in castration-resistant prostate cancer. In: Proceedings of the American Association for Cancer Research 104th Annual Meeting 2013: 6-10 April 2013; Washington, DC. 152 10. Kruger R. Novel anti-tumor activity of targeted LSD1 inhibition. [Presentation]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Abstract nr SY35-03. 11. Martinez Molina D, Jafari R, Ignatushchenko M, Seki T, Larsson EA, Dan C, et al. Monitoring drug target engagement in cells and tissues using the cellular thermal shift assay. Science. 2013;341:84-7. 12. Yang M, Gocke CB, Luo X, Borek D, Tomchick DR, Machius M, et al. Structural basis for CoREST-dependent demethylation of nucleosomes by the human LSD1 histone demethylase. Molecular Cell 2006, 23:377-387. 13. Stavropoulos P, Blobel G, Hoelz A. Crystal structure and mechanism of human lysinedemethylase 1. Nat Struct Mol Biol 2006;13:626-32. 14. Baron R, Binda C, Tortorici M, McCammon JA, Mattevi A. Molecular mimicry and ligand recongnition in binding and catalysis by the histone demethylase LSD1-CoREST complex. Structure 2011;19:212-220. 15. Yang M, Culhane JC, Szewczuk LM, Gocke CB, Brautigam CA, Tomchick DR, et al. Structural basis of histone demethylation by LSD1 revealed by suicide inactivation. Nat Struct Mol Biol 2007;14:535-9. 16. Forneris F, Binda C, Adamo A, Battaglioli E, Mattevi A, Structural basis of LSD1CoREST selectivity in histone H3 recognition. J Biol Chem 2007;282:20070-4. 153 Figure 5.1 The effects of different classes of LSD1 inhibitors on EWS/FLI targets. (A,B,C) The change in gene expression of EWS/FLI-activated targets NKX2-2 and NR0B1 and EWS/FLI-repressed targets HMOX1 and CTGF induced by (A) HCI2509, (B) the irreversible inhibitor OG-L002 (biochemical IC50 ~ 20 nM), and (C) tranylcypromine (biochemical IC50 ~ 20 uM). A673 cells were treated with varying concentrations of inhibitor for 48 hours before RNA was harvested. Doses were chosen based on the IC50 of the inhibitor in a 96-hour cell viability assay such that the dose range is centered around the IC50. HCI2509 results in decreased expression in NKX2-2, no effect on NR0B1, and dose-dependent increasesd in CTGF and HMOX1, consistent with the result in Chapter 3. Interestingly, both other LSD1 inhibitors are much less potent at decreasing cell viability and only recapitulate HMOX1 induction. 154 155 Figure 5.2 HCI2509 decreases metastasis in a nude rat model of Ewing sarcoma. A dose 5 x 106 SK-N-MC cells were implanted in the tibia of nude rats (n=6) and allowed to engraft for 7 days. At that time the primary tumor was imaged by bioluminescence and animals were randomized into vehicle or treatment groups. Treatment animals received daily intraperitoneal injections of 60 mg/kg of HCI2509. After 4 weeks of treatment, animals were taken off of the study and monitored. Thirty days later, rats were sacrificed and lung metastases were observed in vehicle-treated animals, while very little metastatic disease was observed in HCI2509-treated animals. 156 Figure 5.3 Potential synergy between HCI2509 and temozolomide in vivo. In a subcutaneous hindflank SK-ES-1 xenograft study of HCI2509 dosed with vehicle (n=10), 50 mg/kg HCI2509 (n=10), 25 mg/kg temozolomide (n=10), or a combination (n=10) orally showed potential synergistic activity between HCI2509 and temozolomide. The tumor model displays a fair amount of variability, such that error bars were removed for clarity. The mean for each group at their respective time point is plotted. 157 Figure 5.4 HCI2509 does not sensitize cells to treatment with medroxyprogesterone 17acetate (MPA). (A,B) AN3CA (A) and KLE (B) cells were pretreated with varying concentrations of HCI2509 for 24 hours before being treated with either mock or 10 µM MPA for an additional 72 hours. Controls included vehicle (0.5% DMSO) and vehicle treatment for 24 hours followed by 10 µM MPA for an additional 72 hours (MPA alone). Overall, KLE cells were slightly affected by MPA, but no additive effects are observed in the context of HCI2509. 158 APPENDIX A PURIFICATION OF LYSINE-SPECIFIC DEMETHYLASE 1 159 A.1 Protein Purification Protocol: Full Length LSD1 Transform BL21*(DE3) cells with pET15b-hLSD1 (Amp resistance) and allow colonies to grow overnight at 37°C. Innoculate 5 mL of LB+Amp with one colony and allow to grow at 37°C+shaking until OD600~0.6-1. Innoculate 50 mL of LB+Amp with desired amount of previous culture to grow at 37°C+shaking overnight. In the morning, take 6 mL for every liter of induction media to innoculate and centrifuge at 3000 rcf for 20 minutes. Innoculate 1L of LB+Amp with desired amount of bacteria and grow until culture reaches OD600~0.8-0.9. Induce expression with 0.5 mM IPTG and reduce temperature to 22°C and rpm to 180. Shake for 20 hours and then collect pellets. Thaw pellet halfway in cold running water and the remaining halfway on ice. Once thawed, add 1 mg/mL lysozyme and Dnase (optional). Sonicate to lyse 7 cycles of [45 seconds on, 1 minute off]. Clean sonicator tip between tubes/beakers to improve lysis of later samples. Ultracentrifuge lysate at 40K rcf, 4°C, for 45 minutes and discard supernatant. Resuspend pellet in His Extraction Buffer. Ultracentrifuge lysate at 40K rcf, 4°C, for 45 minutes and collect supernatant. Equilibrate column by rinsing off with His-B Buffer followed by His-A Buffer. Run supernatant over desired His column to load protein. Instead, wash with His-A Buffer until back to baseline. Elute with a 15-20 column gradient. Collect appropriate fractions. Dialyze eluent into TGEK-50 depending on your final goal (you may need to concentrate your sample after dialysis; I would just do two size exclusion runs and combine everything). Ultracentrifuge dialyzed protein at 40K rcf, 4°C, for 45 minutes 160 and collect supernatant. Equilibrate S-column (S) with TGEK-1000 and -50 (1000 and 50 mM KCl, respectively). Run protein over ion exchange column and elute with a salt gradient over 15-20 columns. Dialyze ion exchange eluent into TGEK-300. Equilibrate size exclusion column in the TGEK-300. Ultracentrifuge dialyzed protein at 40K rcf, 4°C, for 45 minutes before running size exchange. Collect appropriate fractions for use, concentrate in stirred cell if desired, aliquot, and flash freeze. Typical yield ~1 mg or less/L. A.2 Buffers A.2.1 His Extraction Buffer  25 mM Tris  1 M NaCl  0.1 mM EDTA  10 mM Imidazole  +Fresh BME (7 uL per 100 mL buffer) and PMSF (0.5 mM) A.2.2 His A Buffer  25 mM Tris  1 M NaCl  0.1 mM EDTA  10 mM Imidazole  +Fresh BME (7 uL per 100 mL buffer) and PMSF (0.5 mM) 161 A.2.3 His B Buffer  25 mM Tris  1 M NaCl  0.1 mM EDTA  500 mM Imidazole  +Fresh BME (7 uL per 100 mL buffer) and PMSF (0.5 mM) A.2.4 T/CGEK (For T/CEK do not add glycerol)  T=tris (pH~7) C=citrate (pH~5.2)  25 mM T or C  10% glycerol (G)  1 mM EDTA (E)  desired concentration of KCl  +Fresh BME (7 uL per 100 mL buffer) and PMSF (0.5 mM) 162 A.3 Results Figure A.1 Chromatography tracking LSD1 purifcation. The columns on the first gel are as follows: 1=ladder, 2=lysate loaded onto column, 3=flow through, 4-10=discarded elution fractions during imidazole gradient, 11-12=fractions containing LSD1 band at 96 kDa, 13=ladder. The columns on the second gel are as follows: 14=ladder, 15=loaded sample, 16=collected fractions, 17=concentrated final product, 18=positive control from Hontao Yu Lab, 19=Cayman Chemical hLSD1, 20=ladder. Figure A.2 Purified protein is active. Confirmation of active enzyme using the MichaelisMenten conditions described in Chapter 1. Substrate is the H3K4me2 H3 peptide (residues 1-21). 163 APPENDIX B IN-HOUSE CELL LINE SCREEN AND XCELLIGENCE PROFILING 164 B.1 96-Cell Line Panel Table B.1 A 96-cell line panel. Ninety-six cell lines were assayed for decreased cell viability by ATP-Lite after 96-hours of HCI2509 treatment. Cell Line SK-ES-1 NCCIT Raji IC50 (µM HCI2509) 0.47 0.47 0.49 S-16 Ramos U-937 0.52 0.55 H647 0.57 Skov-3 MCF-7 0.65 0.66 BT-20 RL-95-2 0.77 0.78 LNCap AN3-CA 0.78 0.73 Her-218 C-6 TC-32 A673 Hs-B2 Jurkat SaOS-2 LOX BT-549 0.62 0.74 0.85 0.93 0.99 0.93 0.89 0.90 1.05 Hep-G2 H1666 F98 EWS-502 Hs700-T MV4-11 0.94 0.97 0.99 0.92 0.93 0.97 AGS C-33A Ovcar-8 0.92 1.11 0.85 Malignancy Ewing's sarcoma embryonal teratoma B-Lymphocyte; Burkitt's Lymphoma Schwann cells Lymphoblastoid Macrophage; histiocytic lymphoma metastatic adenosquamous lung carcinoma ovarian adenocarcinoma metastatic breast adenocarcinoma mammary gland; carcinoma uterine endometrial carcinoma prostate carcinoma uterine endometrial adenocarcinoma breast glioma Ewing's sarcoma Ewing's sarcoma Leukemic t-cell T-cell osteosarcoma malignant melanoma mammary gland ductal carcinoma hepatocellular carcinoma colorectal carcinoma glioma Ewing's sarcoma Pancreatic biphenotypic B myelomonocytic leukemia gastric adenocarcinoma cervical carcinoma ovary 165 Table B.1 Continued Cell Line Hs578-T Kato-III Capan-1 U251 K562 IC50 (µM HCI2509) 1.15 1.06 1.07 1.19 1.22 HL-60 1.05 MDA-MB-468 Molt-4 1.17 1.10 RD-ES HeLa Hec-1-A 1.25 1.22 1.25 T98-G Colo-205 Hs-B2 Su-DHL6 HPAF-2 Kasumi -1 KG-1 A2780 HCT-116 HT-29 BxPc-3 SK-MEL-5 Hel U87-MG Yugen-8 MG-63 TC-71 MDA-MB-231 1.19 1.29 1.33 1.32 1.40 1.27 1.27 1.10 1.13 1.20 1.20 1.27 1.46 1.73 1.15 1.59 1.42 1.50 AsPc-1 1.55 RKO J82 Malme-3M H1781 IOMM-1 PSN-1 HCT-15 1.74 1.68 1.74 1.89 1.68 1.74 1.77 Malignancy Breast gastric carcinoma liver met of panc primary glioblastoma chronic myelogenous leukemia acute promyelocytic leukemia breast carcinoma acute lymphoblastic leukemia Ewing's sarcoma cervical adenocarcinoma uterine endometrial adenocarcinoma glioblastoma multiforme colorectal adenocarcinoma T-lymphoblastic leukemia B-cell Pancreas AML AML Ovarian carcinoma Colorectal carcinoma colorectal adenocarcinoma Pancreas adenocarcinoma melanoma Bone-erythro leukemia Glioblastoma-astrocytoma metastatic melanoma osteosarcoma Ewing's sarcoma Metastatic breast adenocarcinoma Metastatic pancreatic adenocarcinoma Colorectal carcinoma bladder carcinoma malignant melanoma NSCLC malignant meningioma pancreatic adenocarcinoma colorectal carcinoma 166 Table B.1 Continued Cell Line SW-480 Hs822-T MiaPaCa-2 A498 OPM-2 Caki-1 HPAC Wi-58 786-0 Du-145 SK-UT-1 SK-MEL-2 PC-12 IC50 (µM HCI2509) 1.60 1.96 2.10 2.20 2.51 1.83 2.38 2.56 2.28 2.45 2.34 2.62 2.88 H460 Panc-1 Mut-J H1975 H522 SNU-16 A549 Hek-293 Hs-766-T 2.39 2.19 2.39 2.89 2.94 1.78 2.65 2.27 2.41 MDA-MB-435 H441 2.81 2.80 PC-3 CFPAC-1 Hup-T4 Panc-02-03 2.82 3.16 2.67 3.00 Malignancy colorectal adenocarcinoma Ewing's sarcoma pancreatic carcinoma kidney carcinoma myeloma kidney carcinoma pancreatic adenocarcinoma Normal Lung renal adenocarcinoma brain met of prostate uterine sarcoma melanoma adrenal phaeochromocytoma Large cell lung carcinoma pancreatic carcinoma pancreatic NSCLC NSCLC gastric cancer lung carcinoma Embryonic kidney Metastatic pancreatic adenocarcinoma Melanoma lung papillary adenocarcinoma Bone met of prostate Pancreas Pancreatic adenocarcinoma pancreatic adenocarcinoma 167 B.2 xCELLigence Profiling Figure B.1 xCelligence screen of Ewing sarcoma cell lines. Real-time measurement of cellular index in four Ewing sarcoma cell lines after treatment with 0.3% DMSO (blue), 300 nM HCI2509 (purple), 1 µM HCI2509 (green), or 3 µM HCI2509 (red) following 24 hours of cell seeding. Measurements were taken every 2 hours. Data are presented as mean and standard deviation (n=3).These data were used to pick the appropriate dose and timing for many experiments reported in Chapter 3, specifically RNA-seq. Cellualr index is a measure of electrical impedance of current passed through the media caused by the adherence of cells to gold electrode on the base of the tissue culture plate. 168 169 APPENDIX C PHARMACOKINETIC MEASUREMENTS IN MICE Emily R. Theisen, Jared Bearss, Adam Hollerbach Emily R Theisen and Jared Bearss performed the animal dosing, plasma collection, extraction, and sample preparation. Adam Hollerbach of the Department of Chemistry Mass Spectrometry Core designed and optimized the mass spec detection protocol. 170 C.1 Mass Spectrometry Methodology C.1.1 Quantitation of HCI-2509 in Rat and Mouse Plasma C.1.1.1 Preparation of HCI2509 Stock Solution Prepare stock solution of 1.25 mg/mL HCI2509 in DMSO. Add 40µL of stock solution to 960µL of plasma, to make a 50.0 μg/mL spiking solution. C.1.1.2 Preparation of Standard Curve Standard (Final Conc, ng/mL) 25,000 18,750 12,500 5,000 2500 1250 500 250 125 50.0 25.0 12.5 Spiking Solution (μg/mL) 50.0 50.0 50.0 50.0 50.0 50.0 5.00 5.00 5.00 0.500 0.500 0.500 Aliquot Volume (μL) 200 150 100 50 20 10 50 20 10 40 20 10 Blank Plasma Volume (μL) 200 250 300 450 380 390 450 380 390 360 380 390 Final Volume (μL) 400 400 400 500 400 400 500 400 400 400 400 400 C.1.1.3 Preparation of HCI2528 (Compound 14 Chapter 2) Stock Make a stock of 1.25 mg/mL HCI-2528 (internal standard) in DMSO. Add 10µL of stock solution to 1990µL of DMSO, to make a 6.25 μg/mL spiking solution. C.1.1.4 Sample Preparation All samples should be kept on ice until processing. Pipette 50µL of blank plasma (double blank and blank), standard, or subject sample into a 1.5mL microcentrifuge tube. 171 Add 5µL of internal standard to each tube except double blank. Prepare a tube with just serum and the internal standard as a control. Add 150µL acetonitrile and vortex vigorously for 60 seconds. Centrifuge at top speed for 5 minutes in refrigerated centrifuge set to 4ºC. Pipette 150µL water into a separate 1.5mL microcentrifuge tube. Add 150µL supernatant (top layer) from the extract and vortex vigorously for ~10 sec. Store samples and standard curve at -20ºC until analysis. C.1.1.5 Liquid Chromatography Parameters Waters ACQUITY H-CLASS  Column: Waters Xbridge C18 3.5 um, 4.6x50mm  Solvent C Name: Formic Acid  Solvent D Name: Acetonitrile  Low Pressure Limit: 0 psi  High Pressure Limit: 15000 psi  Seal Wash Period 5.00 min  Gradient: Time (min) 1. Initial 2. 1.00 3. 2.50 4. 4.00 5. 5.00 Flow Rate (mL/min) 1.000 1.000 1.000 1.000 1.000 C.1.1.6 Mass Spec Parameters Waters ACQUITY TQD  Capillary (kV) 3.50 %C 60.0 60.0 00.0 60.0 60.0 %D 40.0 40.0 100.0 40.0 40.0 Curve Initial 6 6 6 6 172  Source Temperature (°C) 100  Desolvation Temperature (°C) 215  Cone Gas Flow (L/Hr) 5  Desolvation Gas Flow (L/Hr) 550  Collision Gas Flow (mL/Min) 0.22 C.1.1.7 MRM Parameters Compound HCI-2528 HCI-2509 Parent (m/z) Daughter (m/z) Dwell (s) 396.1 168 0.100 468.1 168 0.100 Cone (V) 30 35 Col (V) 23 25 C.2 Pharmacokinetic Measurements in Mice C.2.1 Dosing Mice were dosed by intravenous tail injection (IV) or by oral gavage (PO) with HCI2509. At the desired time point mice were sacrificed and blood was collected by cardiac puncture. Plasma was then stored at -80 ºC until analysis. Formulations for 5mg/kg IV and 20 mg/kg PO dosing were stable solutions. Formulations for 50 mg/kg PO and 40 mg/kg IP were stable suspensions of HCI2509 crystals which were administered through a 22 ½ gauge needle. Generally speaking, in mice, HCI2509 appears to be rapidly cleared (Cl=24.33 ml/min/kg) with a half-life of 0.87 hour. The bioavailability of HCI2509 from oral formulations varied depending on whether the form was a solution and a suspension. The solution form (F=27%) showed much greater bioavailability than the suspension (F=4.5%), consistent with high predicted permeability for HCI2509 and related 173 compounds. In the case of the suspension, the low bioavailability is likely due to the stability of HCI2509 crystals as they pass through the GI tract. The acid lability of the hydrazone moiety likely decreases the observed F for both oral forms. Efficacy studies utilized a suspension dosed intraperitoneally, which provided a depot for release of drug out to 4 hours. C.2.1.1 Formulations C.2.1.1.1 5 mg/kg IV and 20 mg/kg PO - Clear, Yellow Solution  15% N,N-dimethylacetamide  20% Propylene glycol  25% Water for injection  40% Polyethylene glycol 400 MW C.2.1.1.2 50 mg/kg PO - Stable, Crystalline Suspension  10% Ethanol  40% Propylene glycol  50% Phosphate-buffered saline (pH 7.4) C.2.1.1.3 40 mg/kg IP - Stable, Crystalline Suspension  50% PEG400  50% PBS  Drug was completely dissolved in PEG400 using sonication and PBS added. 174 Table C.1 Pharmacokinetic parameters for 5 mg/kg HCI2509 dosed as solution IV. Parameter Time (hr) 0.08333 Cp HCI2509 (ng/ml) Mice 1-8 Mice 9-16 Mice 17-24 7391 8228 6338 Mean 7319 STDEV %CV 947 12.9 0.25 0.5 3251 1374 2578 1768 3797 1589 3209 1577 610 197 19.0 12.5 1 2 655 119 414 88 891 235 653 147 239 77.5 36.5 52.6 4 8 24 0 0 0 24 17 0 35 12 0 19.7 9.67 0 17.9 8.74 0 91.0 90.4 11143.85 14698.90 8188.44 11343.73 3259.83 28.7 3131.48 3389.52 3670.09 3397.03 269.39 7.9 3203.97 3416.66 3688.64 3436.43 242.94 7.1 0.6348 0.8337 1.001 0.8231 0.1833 22.3 26.01 24.39 22.59 24.33 1.710 7.0 0.4067 0.5697 0.7385 0.5716 0.1659 29.0 0.4222 1.107 1.072 0.8669 0.3855 44.5 C0 (ng/mL) AUC0-t (ng*hr/mL) AUC0-∞ (ng*hr/mL) Vss (L/kg) Cl (mL/min/kg) MRT (hr) T1/2 (hr) 175 Table C.2 Pharmacokinetic parameters for 20 mg/kg HCI2509 dosed as solution PO. Parameter Time (hr) 0.25 Cp HCI2509 (ng/ml) Mice 1-7 Mice 8-14 Mice 15-21 638 1162 852 Mean 884 STDEV 263 %CV 29.8 0.5 1 851 346 1163 440 792 999 935 595 199 353 21.3 59.3 2 4 246 69 219 302 231 155 232 175 13.5 117 5.8 67.2 8 24 56 3 422 3 64 10 181 5.33 209 4.04 115.7 75.8 851 1163 999 1004.33 156.07 15.5 0.5 0.5 1 0.6667 0.28868 43.3 1898.13 6535.13 2790.75 3741.33 2460.32 65.8 1916.39 6546.98 2862.96 3775.44 2446.44 64.8 7630.51 39569.1 12798.41 19999.3 17143.75 85.7 3.982 6.044 4.470 4.832 1.078 22.3 4.220 0.1397 2.739 0.4809 5.005 0.2054 3.988 0.2753 1.151 0.1811 28.8 65.7 Cmax (ng/mL) Tmax (hr) AUC0-t (ng*hr/mL) AUC0-∞ (ng*hr/mL) AUMC0-∞ (ng*hr^2/mL) MRT (hr) T1/2 (hr) F 0-t 176 Table C.3 Pharmacokinetic parameters for 50 mg/kg HCI2509 dosed as suspension PO. Parameter Time (hr) 0.25 0.5 1 2 4 8 12 24 Cmax (ng/mL) Tmax (hr) AUC0-t (ng*hr/mL) AUC0-∞ (ng*hr/mL) AUMC0-∞ (ng*hr^2/mL) MRT (hr) T1/2 (hr) F 0-t Cp HCI2509 (ng/ml) Mice 1-8 Mice 9-16 Mice 17-24 129 99 69 151 56 82 79 154 66 122 192 94 251 202 126 90 52 207 25 27 40 10 13 11 Mean 99 96.33 99.67 136 193 116.33 30.67 11.33 STDEV 30 49.10 47.50 50.48 62.98 80.79 8.14 1.53 %CV 30.3 51.0 47.7 37.1 32.6 69.4 26.6 13.5 251 202 207 220 26.96 12.3 4 4 8 5.33 2.309 43.3 1704.13 1557.25 1830.5 1697.29 136.75 8.1 1768.92 1718.33 1896.80 1794.7 91.98 5.1 12168.5 15122.1 15901.8 14397.5 1969.4 13.7 6.879 8.800 8.384 8.0210 1.010 12.6 4.491 0.0502 8.589 0.0458 4.172 0.0538 5.752 0.0450 2.461 0.0040 42.8 8.1 177 Figure C.1 Typical standard curves to quantitate HCI2509 by LC-MS/MS. Known concentrations of HCI2509 are extracted from mouse plasma, plotted as a standard curve and used to determine the concentrations of HCI2509 in mouse plasma. The limit of detection was determine using LOD=3.3*(SD/S), where SD is the standard deviation in the y-intercept and S is the slope. LOD was determined to be 43 ng/mL. Linearity was observed for all tested ranges. 178 Figure C.2 Plasma concentration-time curves for HCI2509 in mice as determined by LCMS/MS. The data in Tables C1-C3 plotted on a linear scale. Data are visualized as mean and standard deviation (n=3). Figure C.3 Plasma concentration-time curves for HCI2509 in mice as determined by LCMS/MS - semilog. The data in Tables C1-C3 plotted on a linear scale. Data are visualized on a semilog plot as mean and standard deviation (n=3). 179 Figure C.4 Plasma concentration-time curves for 40 mg/kg HCI2509 in mice as determined by LC-MS/MS. Linear plot of 50/50 PEG400/PBS formulation used for efficacy studies. Data are visualized as mean and standard deviation (n=3). Figure C.5 Plasma concentration-time curves for 40 mg/kg HCI2509 in mice as determined by LC-MS/MS - semilog. Semilog plot of 50/50 PEG400/PBS formulation used for efficacy studies. Data are visualized as mean and standard deviation (n=3).