Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A multidimensional approach to understanding lung tumour aggressiveness Enfield, Catherine S. 2017

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2017_september_enfield_catherine.pdf [ 2.68MB ]
Metadata
JSON: 24-1.0348376.json
JSON-LD: 24-1.0348376-ld.json
RDF/XML (Pretty): 24-1.0348376-rdf.xml
RDF/JSON: 24-1.0348376-rdf.json
Turtle: 24-1.0348376-turtle.txt
N-Triples: 24-1.0348376-rdf-ntriples.txt
Original Record: 24-1.0348376-source.json
Full Text
24-1.0348376-fulltext.txt
Citation
24-1.0348376.ris

Full Text

A MULTIDIMENSIONAL APPROACH TO UNDERSTANDING LUNG TUMOUR AGGRESSIVENESS  by Catherine S. Enfield BSc, University of Victoria, 2009 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Interdisciplinary Oncology) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) June 2017 © Catherine S. Enfield, 2017  ii  Abstract Lung cancer is a deadly disease with a 5-year survival of 18%. Patients with localized disease have improved outcomes, since they are eligible for surgery with curative intent. However, the majority of patients present with metastatic disease, where treatment options are limited. Genetic profiling of lung tumours has lead to the identification of driver mutations and the subsequent development of targeted therapies for some of these mutated genes, which has resulted in improved outcomes for late stage patients. Unfortunately, only subsets of patients harbour these mutations, and patients frequently acquire drug resistance. Novel therapeutics are desperately needed for patients with late stage disease in order to improve patient outcome. To achieve this goal, we require a deeper understanding of the biology that drives aggressive lung tumour biology. In this thesis, we employ two unique approaches to discover genes associated with lung tumour aggressiveness. Firstly, we move beyond the protein-coding landscape and characterize deregulation of small non-coding RNAs in metastatic lung cancer. We further assess the ability of small non-coding RNA expression patterns to classify patients into different outcome groupings. Secondly, we employ an integrative multi-omics approach in order to identify novel oncogenes that have been overlooked by conventional studies within a single dimension (e.g. mutation). Collectively, this work adds to our understanding of aggressive tumour biology. Further characterization of the aggressiveness-associated small non-coding RNAs identified here may inform on novel therapeutic avenues or gene signatures. Importantly, our discovery of a novel lung cancer oncogene may lead to a new therapy for lung cancer.   iii  Lay Abstract Lung cancer is one of the most deadly tumour types due to its ability to spread to distant organs. Treatment options for patients are limited. We discovered a new gene involved in aggressive lung cancer that might be a new drug target. Previous studies have focused on genes that code for proteins, but proteins make up less than 3% of the human genome. Genes that do not code for proteins, so-called non-coding RNAs, represent an opportunity for cancer gene discovery. We compared tumours that had spread to those that had not in order to find non-coding RNAs that control this deadly process. We also identified patterns of non-coding RNA expression that predict which tumours are more aggressive and that might need different treatment. This work adds to our understanding of aggressive lung tumour biology and our discoveries may help improve treatment and survival of lung cancer patients.  iv  Preface The research in this thesis was conducted with ethics approval from the UBC Research Ethics Board, Certificate Numbers: EDRN H09-00008, CCSRI H09-00934, W81XW-10-1-0634. Chapter 1 alludes to several manuscripts and reviews I have published during the course of my PhD, which are listed here. Co-first authorship is indicated by an underline. 1. Ng KW, Anderson C, Marshall EA, Minatel BC, [Enfield KSS], Saprunoff HL, Lam WL, Martinez VD (2016) Piwi-interacting RNAs in cancer: emerging functions and clinical utility. Molecular Cancer. Jan;15:5.  2. Firmino N, Martinez VD, Rowbotham DA, [Enfield KSS], Bennewith KL, Lam WL (2016) HPV status is associated with altered PIWI-interacing RNA expression in head and neck cancer. Oral Oncology. Apr;55, 43-8.  3. Martinez VD, [Enfield KSS], Rowbotham DA, Lam WL (2016) An atlas of gastric PIWI-Interacting RNA transcriptomes and their utility for identifying signatures of gastric cancer recurrence. Gastric Cancer. 19, 660-5. 4. Martinez VD, Vucic EA, Thu KL, Hubaux R, [Enfield KSS], Pikor LA, Becker-Santos DD, Brown CJ, Lam S, Lam WL (2015) Unique somatic and malignant expression patterns implicate PIWI-interacting RNAs in cancer-type specific biology. Scientific Reports. 5:10423, 1-17. 5. Rowbotham D, [Enfield KSS], Martinez VD, Thu KL, Vucic EA, Stewart GL, Bennewith K, Lam WL (2014) Multiple components of the VHL tumor suppressor complex are frequently affected by DNA copy number loss in pheochromocytoma. International Journal of Endocrinology. 2014:546347, 1-9. 6. Wilson IM, Vucic EA, [Enfield KSS], Thu KL, Zhang YA, Chari R, Lockwood WW, Radulovich N, Starczynowski DT, Banath J, Zhang M, Pusic A, Fuller M, Lonergan KM, Rowbotham D, Yee J, English J, Buys TPH, Selamat SA, Laird-Offringa I, Liu P, v  Anderson M, You M, Tsao MS, Brown C, Bennewith K, MacAulay CE, Karsan A, Gazdar AF, Lam S, Lam WL (2014) EYA4 is inactivated biallelically at a high frequency in sporadic lung cancer and is associated with familial lung cancer risk. Oncogene. 33, 4464-4473. 7. Hubaux R, Becker-Santos DD, [Enfield KSS], Rowbotham D, Lam S, Lam WL, Martinez VC (2013) Molecular features in arsenic-induced lung tumors. Molecular Cancer. 12:20, 1-11. 8. Hubaux R, Becker-Santos DD, [Enfield KSS], Lam S, Lam WL, Martinez VC (2012) Impact of Arsenic, Asbestos and Radon Exposure on the Lung Cancer Genome. Environmental Health. 11:89, 1-12. 9. Hubaux R, Becker-Santos DD, [Enfield KSS], Lam S, Lam WL, Martinez VD (2012) MicroRNAs as biomarkers for clinical features of lung cancer. Metabolomics. 2, 1-24. 10. Gibb EA, Vucic EA, [Enfield KSS], Stewart GL, Lonergan KM, Kennett JY, Becker-Santos DD, MacAulay CE, Lam S, Brown CJ, Lam WL (2011) Human cancer long non-coding RNA transcriptomes.  PLoS ONE. 6:e25915, 1-10 11. [Enfield KSS], Pikor LA, Martinez VD, Lam WL (2012) Mechanistic roles of non-coding RNAs in lung cancer biology and their clinical implications. Genetic Research International.  2012:737416, 1-16. Chapters 3, 4 and 5 were outlined as manuscripts for publication. The co-authors and author contributions are outlined below. Data from Chapter 3 has been presented at local, national and international conferences and is currently in preparation for publication. The authors involved are as follows: Enfield KSS, Kung SHY, Rowbotham DA, Pastrello C, Jurisica I, MacAualy CE, Lam S, Lam WL (2017). As first author I was responsible for the design and implementation of this project. I performed all data analyses as well as the majority of in vitro assays and molecular biology techniques, in addition to writing the manuscript. SHY Kung and DA Rowbotham assisted with in vitro assays. vi  C Pastrello assisted with in silico analysis and is supervised by I Jurisica. S Lam provided patient specimens. S Lam, CE MacAulay and WL Lam oversaw the project. Chapter 4 has been presented at international conferences and is published. [Enfield KS], Martinez VD, Marshall EA, Stewart GL, Kung SH, Enterina JR, Lam WL (2016). Deregulation of small non-coding RNAs at the DLK1-DIO3 imprinted locus predicts lung cancer patient outcome. Oncotarget. 2016; 7: 80957-80966. doi:10.18632/oncotarget.13133. I was responsible for study design and implementation. I performed the majority of the data analysis, and prepared the manuscript. VD Martinez generated the piRNA expression profiles, and provided input towards manuscript design and presentation. EA Marshall and GL Stewart assisted in data analysis and writing. SH Kung designed Figures 1 and 2. JR Enterina edited the document. WL Lam oversaw the project. Chapter 5 has been presented at local, national and international conferences and is in preparation for publication. [Enfield KSS], Rahmati S, Rowbotham DA, Fuller M, Anderson C, Marshall EA, Chari R, Becker-Santos DD, MacAulay CE, Lam S, Politi K, Lockwood WW, Karsan A, Jurisica I, Lam WL (2017). I was responsible for study design and implementation. I performed the majority of in vitro experiments and prepared the manuscript. S Rahmati assisted with in silico analysis and is supervised by I Jurisica. DA Rowbotham, C Anderson, EA Marshall, and DD Becker-Santos assisted with in vitro experiments. M Fuller performed the in vivo experiments and is supervised by A Karsan. K Politi and WW Lockwood provided the oncogene-inducible mouse tumour data. S Lam provided clinical specimens. CE MacAulay, S Lam, WW Lockwood, A Karsan, WL Lam all provided invaluable input to shape the project.   vii  Table of Contents Abstract ....................................................................................................................................................... ii Lay Abstract .............................................................................................................................................iii Preface ........................................................................................................................................................ iv Table of Contents ................................................................................................................................. vii List of Tables ...........................................................................................................................................xii List of Figures ....................................................................................................................................... xiii List of Abbreviations.......................................................................................................................... xvi Acknowledgements ............................................................................................................................. xix Dedication ................................................................................................................................................ xx 1 Introduction ..................................................................................................................................... 1 1.1 Lung cancer background .................................................................................................... 1 1.2 Lung cancer metastasis ....................................................................................................... 4 1.3 Molecular pathology of lung cancer and an era of targeted therapy ................. 4 1.4 Prognostic and predictive expression signatures ..................................................... 6 1.5 Non-coding RNAs in lung cancer biology ..................................................................... 8 1.5.1 microRNAs in lung cancer ........................................................................................................ 8 1.5.2 piRNAs in cancer ....................................................................................................................... 10 1.6 An integrative approach to understanding lung cancer biology ...................... 12 1.7 Outstanding clinical issues in lung cancer ................................................................ 14 1.8 Thesis objective and research approach ................................................................... 14 1.9 Overarching hypothesis ................................................................................................... 14 1.10 Specific aims ..................................................................................................................... 15 2 Methods .......................................................................................................................................... 17 viii  2.1 Patient tissue accrual ........................................................................................................ 17 2.2 Nucleic acid extraction ..................................................................................................... 17 2.3 Molecular profiling summary ........................................................................................ 17 2.3.1 Small RNA sequencing ............................................................................................................ 19 2.3.1.1 microRNA profiling in both cohorts ............................................................................. 19 2.3.1.2 piRNA profiling in both cohorts ..................................................................................... 19 2.3.2 DNA copy number .................................................................................................................... 20 2.3.3 DNA methylation data ............................................................................................................ 20 2.3.4 RNA expression data ............................................................................................................... 21 2.4 Cell lines ................................................................................................................................. 22 2.5 Basic laboratory techniques ........................................................................................... 23 2.5.1 Immunoblot ................................................................................................................................ 23 3 Identification of microRNAs deregulated in metastatic lung adenocarcinoma .. 25 3.1 Introduction.......................................................................................................................... 25 3.1.1 miRNA deregulation in cancer ............................................................................................ 25 3.1.2 Rationale for assessing miRNA deregulation in metastatic lung adenocarcinoma ........................................................................................................................................... 25 3.2 Methods .................................................................................................................................. 26 3.2.1 Patient cohort ............................................................................................................................. 26 3.2.2 miRNA expression analysis .................................................................................................. 28 3.2.3 Cell models .................................................................................................................................. 28 3.2.4 Migration and invasion assays ............................................................................................ 29 3.3 Results .................................................................................................................................... 29 3.3.1 Identification of miRNAs deregulated in metastatic lung adenocarcinoma ..... 29 3.3.2 Genomic positions of miRNAs deregulated in metastatic lung adenocarcinoma    .......................................................................................................................................................... 36 3.3.3 miRNA paralogs miR-106b and miR-106a are overexpressed in metastatic lung adenocarcinoma ................................................................................................................................. 38 3.3.4 In vitro assessment of metastatic phenotypes using cell models ......................... 41 ix  3.3.5 Associations of miR-106a and miR-106b expression with poor patient outcome in an external dataset .............................................................................................................. 47 3.4 Discussion ............................................................................................................................. 49 4 Deregulation of small non-coding RNAs at the DLK1-DIO3 imprinted locus predicts lung cancer patient outcome ......................................................................................... 51 4.1 Introduction.......................................................................................................................... 51 4.1.1 Metastasis-associated miRNAs are encoded at the imprinted DLK1-DIO3 locus    .......................................................................................................................................................... 51 4.1.2 miRNAs expressed from the DLK1-DIO3 locus are associated with poor lung adenocarcinoma patient outcome ........................................................................................................ 53 4.1.3 Expression of piRNAs from the DLK1-DIO3 locus remains unknown ................. 53 4.1.4 Rationale for examining associations of small ncRNAs expressed from the DLK1-DIO3 locus with lung cancer patient outcome ..................................................................... 53 4.2 Methods .................................................................................................................................. 54 4.2.1 Clinical Cohorts .......................................................................................................................... 54 4.2.2 Small ncRNA expression ........................................................................................................ 56 4.2.3 Small non-coding RNA differential expression analysis ........................................... 56 4.2.4 Survival Analysis ....................................................................................................................... 57 4.3 Results .................................................................................................................................... 57 4.3.1 The DLK1-DIO3 locus encodes somatically expressed piRNAs .............................. 57 4.3.2 A combined miRNA+piRNA signature better predicts overall survival of lung adenocarcinoma patients ......................................................................................................................... 64 4.3.3 The miRNA+piRNA signature is able to predict overall survival of lung squamous cell carcinoma patients ........................................................................................................ 69 4.3.4 The miRNA+piRNA signature identifies patients at risk of recurrence-free survival  .......................................................................................................................................................... 72 4.4 Discussion ............................................................................................................................. 77 5 ELF3 is a novel oncogene in lung adenocarcinoma ....................................................... 79 5.1 Introduction ............................................................................................................................... 79 5.1.1 Rationale for integrative multi-omics analysis ............................................................ 79 x  5.1.2 ELF3 is a putative lung cancer oncogene ........................................................................ 79 5.2 Methods .................................................................................................................................. 80 5.2.1 Patient cohorts ........................................................................................................................... 80 5.2.2 RNA profiling of LUSC samples in the BCCA cohort ................................................... 80 5.2.3 Survival analysis ....................................................................................................................... 83 5.2.4 GISTIC 2.0 copy number analysis ....................................................................................... 83 5.2.5 Oncogene mutation data ........................................................................................................ 83 5.2.6 Expression data from mouse models of tumourigenesis ......................................... 84 5.2.7 In vitro and in vivo experiments ......................................................................................... 84 5.2.8 Immunofluorescence microscopy ...................................................................................... 87 5.2.9 Soft agar colony formation ................................................................................................... 87 5.2.10 Cell proliferation ....................................................................................................................... 87 5.2.11 Cell apoptosis ............................................................................................................................. 88 5.3 Results .................................................................................................................................... 88 5.3.1 ELF3 is frequently overexpressed in lung adenocarcinoma and associated with patient outcome ................................................................................................................................. 88 5.3.2 The ELF3 locus at chromosome 1q32.1 is frequently altered by genetic and epigenetic mechanisms ............................................................................................................................. 95 5.3.3 DNA alterations are correlated with expression changes ....................................... 98 5.3.4 ELF3 overexpression is observed irrespective of driver oncogene.................. 105 5.3.5 ELF3 regulates cancer phenotypes of LUAD cell lines ........................................... 111 5.3.6 ELF3 expression is required for in vivo tumour growth ....................................... 116 5.4 Discussion ........................................................................................................................... 127 6 Conclusions .................................................................................................................................. 130 6.1 Summary of thesis findings .......................................................................................... 130 6.1.1 Chapter 3 summary with respect to Aim 1 ................................................................. 130 6.1.2 Chapter 4 summary with respect to Aim 1 ................................................................. 131 6.1.3 Chapter 5 summary with respect to Aim 2 ................................................................. 132 6.2 Strengths and limitations of thesis work ................................................................ 133 6.2.1 Chapter 3 ................................................................................................................................... 133 xi  6.2.2 Chapter 4 ................................................................................................................................... 135 6.2.3 Chapter 5 ................................................................................................................................... 135 6.3 Considerations and future directions ....................................................................... 136 6.3.1 Chapter 3 ................................................................................................................................... 136 6.3.2 Chapter 4 ................................................................................................................................... 137 6.3.3 Chapter 5 ................................................................................................................................... 138 References ............................................................................................................................................ 140 Appendix I: Publications ................................................................................................................. 160 Appendix II: Selected Abstracts.................................................................................................... 163  xii  List of Tables Table 2.1 Platforms used for multi-omic profiling of both clinical cohorts used in this thesis. ........................................................................................................................................................ 18 Table 2.2 Cell lines used for in vitro and in vivo assays in this thesis .............................. 23 Table 2.3 List of primary antibodies used in this thesis ....................................................... 24 Table 3.1 Clinical features of the BCCA lung adenocarcinoma cohort. ........................... 27 Table 3.2 List of 37 miRNAs significantly differentially expressed between localized and metastatic lung adenocarcinoma. ......................................................................................... 31 Table 4.1 Clinical features of the discovery and external cohorts. ................................... 55 Table 5.1 Summary of multi-omic profiling for the BCCA and TCGA datasets ............. 81 Table 5.2 Summary of clinical cohorts of lung adenocarcinoma and lung squamous cell carcinoma with ELF3 mRNA expression............................................................................. 82 Table 5.3 Cell lines used for in vitro and in vivo experiments............................................. 86  xiii  List of Figures Figure 1.1 Lung cancer survival rates decrease as stage increases .................................... 3 Figure 1.2 Frequencies of known driver oncogenes in lung adenocarcinoma ............... 5 Figure 1.3 Mechanism of action of miRNAs and piRNAs ...................................................... 11 Figure 1.4 Integrating the multi’omic dimensions of DNA level alterations in cancer ..    ............................................................................................................................................................ 13 Figure 3.1 Expression fold change of the most significantly differentially expressed microRNA, miR-106b .......................................................................................................................... 35 Figure 3.2 Paralog microRNA clusters harbouring metastasis-associated miRNAs.. 37 Figure 3.3 Expression of miR-106a and miR-106b in clinical specimens ...................... 39 Figure 3.4 Epithelial and mesenchymal expression profiles of LUAD cell lines .......... 42 Figure 3.5 Boyden chamber assay results .................................................................................. 43 Figure 3.6 miR-106a and miR-106b overexpression induces expression changes characteristic of epithelial-to-mesenchymal transition ........................................................ 45 Figure 3.7 Assessment of miR-106a and miR-106b expression and overall survival in The Cancer Genome Atlas cohort .................................................................................................. 48 Figure 4.1 UCSC genome browser screenshot of the imprinted DLK1-DIO3 locus .... 52 Figure 4.2 UCSC genome browser screenshot of the piRNAs encoded at the imprinted DLK1-DIO3 locus ............................................................................................................. 59 Figure 4.3 Histograms of expressed piRNAs in the discovery dataset ............................ 60 Figure 4.4 Histograms of expressed piRNAs in the validation cohort ............................ 62 Figure 4.5 Overall survival of risk groups as defined by the miRNA signature in lung adenocarcinoma ................................................................................................................................... 65 Figure 4.6 Risk classification of lung adenocarcinoma patients ........................................ 67 xiv  Figure 4.7 Overall survival of risk groups as defined by the miRNA+piRNA signature in lung adenocarcinoma .................................................................................................................... 68 Figure 4.8 The miRNA+piRNA signature predicts overall survival in lung squamous cell carcinoma patients (discovery dataset, n=27) ................................................................. 70 Figure 4.9 The miRNA+piRNA signature predicts overall survival in lung squamous cell carcinoma patients (external dataset, n=205) ................................................................. 71 Figure 4.10 Performance of small ncRNA-based signatures predicting recurrence-free survival in non-small cell lung cancer ................................................................................ 73 Figure 4.11 Log-rank p-value summary for overall survival predictions ...................... 74 Figure 4.12 Log-rank p-value summary for recurrence-free survival predictions in the external dataset ............................................................................................................................ 76 Figure 5.1 ELF3 expression in lung adenocarcinoma ............................................................ 90 Figure 5.2 ELF3 expression is correlated with tumour purity in TCGA.......................... 91 Figure 5.3 ELF3 expression in lung squamous cell carcinoma ........................................... 92 Figure 5.4 High ELF3 expression is associated with poor overall survival in lung adenocarcinoma ................................................................................................................................... 94 Figure 5.5 GISTIC 2.0 focal amplification plot .......................................................................... 96 Figure 5.6 Frequency of DNA-level alterations at the ELF3 locus across datasets .... 98 Figure 5.7 Multi’omic analysis of ELF3 in the BCCA dataset ............................................... 99 Figure 5.8 Multi’omic analysis of ELF3 in the TCGA dataset ............................................. 102 Figure 5.9 Multi’omic analysis of ELF3 in the TCGA dataset after filtering for >60% tumour purity ...................................................................................................................................... 104 Figure 5.10 ELF3 expression in lung adenocarcinomas with and without mutation in EGFR and KRAS ................................................................................................................................... 106 Figure 5.11 ELF3 expression in mouse models of tumourigenesis ................................ 108 xv  Figure 5.12 ELF3 expression is predominantly nuclear ..................................................... 109 Figure 5.13 Immunoblot of ELF3 expression across of panel of molecularly distinct LUAD cell lines .................................................................................................................................... 110 Figure 5.14 Experimental validation of ELF3 shRNA-mediated knock-down across biological replicates .......................................................................................................................... 112 Figure 5.15 In vitro results from shELF3 cell lines ............................................................... 113 Figure 5.16 In vitro results from ELF3 overexpression in HBECs .................................. 115 Figure 5.17 Immunoblot of ELF3 expression across A549 shELF3 clones ................. 117 Figure 5.18 A549 shEFL3 clone in vitro phenotypic results ............................................. 118 Figure 5.19 In vivo growth of A549 control cells and shELF3 clones (n=24) ............. 121 Figure 5.20 qPCR assessment of ELF3 expression in cell line input and tumour RNA ..    .......................................................................................................................................................... 122 Figure 5.21 In vivo growth of HCC827 control and shELF3 cells (n=12) ..................... 124 Figure 5.22 ELF3 knock-down is not maintained during the course of the HCC827 in vivo experiment .................................................................................................................................. 125  xvi  List of Abbreviations 3’ UTR   3 prime untranslated region ALK   Anaplastic Lymphoma Receptor Tyrosine Kinase ATCC   American Type Culture Collection asRNA   anti-sense RNA BCCA   British Columbia Cancer Agency BCGSC   British Columbia Genome Science Centre BPE   Bovine Pituitary Extract BRAF   B-Raf Proto-Oncogene, Serine/Threonine Kinase BrdU   bromodeoxyuridine BWA   Burrows-Wheeler Aligner CN   copy number DAPI   4’,6-diamidine-2’-phenylindole dihydrochloride DIO3   Iodothyronine Deiodinase 3 DLK1   Delta Like Non-Canonical Notch Ligand 1 DNA   deoxyribonucleic acid EGF   Epidermal Growth Factor EGFR   Epidermal Growth Factor Receptor ELF3   E74-Like Factor 3 EML4   Echinoderm Microtubule Associated Protein Like 4 EMT   epithelial to mesenchymal transition ERBB2  Erb-B2 Receptor Tyrosine Kinase ETS   E-twenty six FBS   fetal bovine serum xvii  FC   fold change FFPE   formalin fixed paraffin embedded FPKM   fragments per kilobase of transcript per million mapped reads FWER   family-wise error rate GEO   Gene Expression Omnibus GISTIC   Genomic Identification of Significant Targets in Cancer HBEC   human bronchial epithelial cells IHC   immunohistochemistry KD   knock-down KRAS   V-Ki-Ras2 Kirsten Rat Sarcoma 2 Viral Oncogene Homolog lncRNAs  long non-coding RNA LUAD   lung adenocarcinoma LUSC   lung squamous cell carcinoma MET   MET Proto-Oncogene, Receptor Tyrosine Kinase miRNA/miR  microRNA MOI   multiplicity of infection ncRNA   non-coding RNA NM-LUAD  non-malignant lung tissue from LUAD patients NM-LUSC  non-malignant lung tissue from LUSC patients NOTCH3  Neurogenic Locus Notch Homolog Protein 3 NSCLC   non-small cell lung cancer OE   overexpressed OS   overall survival PI   propidium iodide piRNA   PIWI-interacting RNA xviii  PIWI   P-element induced wimpy testis qPCR   polymerase chain reaction qRT-PCR  quantitative reverse transcription polymerase chain reaction RPKM   reads per kilobase of exon model per million mapped reads RPM   reads per million mapped reads RISC   RNA-induced silencing complex RNA   ribonucleic acid RFS   recurrence free survival SCLC   small cell lung cancer SDS-PAGE  sodium dodecyl sulfate polyacrylamide gel electrophoresis SEM   standard error of the mean shELF3  short hairpin RNA targeting ELF3 shRNA   short hairpin RNA siRNA   small interfering RNA snoRNA  small nucleolar RNA STAR   spliced transcripts alignment to a reference TCGA   The Cancer Genome Atlas TBS-T   tris-buffered saline with Tween® 20 TGFβ   Transforming Growth Factor Beta TKI   tyrosine kinase inhibitor tRNA   transfer RNA UE   underexpressed  xix  Acknowledgements I would like to acknowledge the immense support and guidance provided by past and present members of the Wan Lam Lab, especially those who contributed to the work presented in this thesis. I would also like to thank my collaborators, and the PIs and students at the BCCRC that fostered such an excellent research environment. I appreciate and acknowledge my scholarship support from the Interdisciplinary Oncology Program, University of British Columbia Graduate Entrance Scholarship, University of British Columbia Four Year Doctoral Fellowship, and the Canadian Institutes of Health Research Frederick Banting and Charles Best Canada Graduate Scholarship Doctoral Award. The research presented in this thesis was funded by: the Canadian Cancer Society Research Institute, Canadian Institutes of Health Research, Terry Fox Research Institute, the US Department of Defense Lung Cancer Research Program, and the US National Institutes of Health (NCI and NHLBI). I would especially like to thank my supervisor Dr. Wan Lam for providing me with the opportunity to join his lab, and my supervisory committee members Dr. Stephen Lam and Dr. Calum MacAulay for their guidance, insight, and support throughout my program. xx  Dedication I dedicate this thesis to my family who has always encouraged my aspirations, and to the amazing friends I have met along the way. A special dedication to my Grampa who always challenged me to achieve more and think critically, and my Granny who told me I could be anything, including a rocket scientist.1  1 Introduction 1.1 Lung cancer background Lung cancer is the leading cause of cancer-related death in Canada and worldwide, with a 5-year survival rate of 18% 1,2. Lung cancer largely remains an environmentally-induced malignancy, which is caused by tobacco smoking in up to 90% of cases. Lung cancer arising in never smokers accounts for 10-20% of cases and represents the seventh most prevalent cancer type when considered as a separate disease3. Lung tumours may also develop due to exposure to alternative environmental carcinogens including air pollution4, radon5-7, and arsenic4,8-10 (Appendix I, items 10-11), and have been associated with viral infection4 and inherited traits11-15. Lung cancer is divided into two major histological subtypes: small cell carcinoma (SCLC) and non-small cell carcinoma (NSCLC), which comprises over 80% of lung cancer cases16. The majority of NSCLC cases are of the adenocarcinoma histological subtype (LUAD) followed by squamous cell carcinoma (LUSC), with some cases of large cell carcinoma17. LUAD are thought to arise from Clara cells and/or type II pneumocytes in the distal airways, whereas LUSC and SCLC arise in the proximal airways from basal cells and pulmonary neuroendocrine cells, respectively18. Recent efforts by the international community (International Association for the Study of Lung Cancer, American Thoracic Society, and European Respiratory Society) have resulted in the further reclassification of LUAD based on predominant histological pattern into: minimally invasive adenocarcinoma, lepidic, acinar, papillary, micropapillary, solid, mucinous, colloid, fetal, and enteric adenocarcinoma19,20. LUAD comprises approximately 50% of all lung cancer cases and is the focus of this thesis. Poor survival of lung cancer patients is largely attributed to the frequent (>80%) discovery of late stage disease at time of diagnosis. Patients with late stage disease 2  are ineligible for surgical resection with curative intent17. Currently, surgical resection remains the most effective form of treatment and is available for patients with locoregional disease21,22.  Indeed, patient survival rates decrease as disease stage increases, and advanced stage patients typically receive a combination of chemotherapy or targeted therapy and/or radiotherapy23,24 (Figure 1.1). Treatment is complicated by the intra- and intertumoural heterogeneity observed at the histological and molecular level25.   3   Figure 1.1 Lung cancer survival rates decrease as stage increases (Left) Histogram illustrating the 5-year survival rates of Canadian lung cancer patients diagnosed with localized (blue), regional (lymph node involvement, green), and distant disease (Stage IV, red). The average 5-year survival is shown in black. (Right) Histogram illustrating the percent stage distribution at time of diagnosis for localized (blue), regional (lymph node involvement, green), and distant disease (Stage IV, red). Adapted from Jemal et al. 2.   4  1.2 Lung cancer metastasis Metastasis is the dissemination of cancer cells from the primary tumour to lymph nodes and distant organs, and is responsible for the majority of cancer-related deaths. The first site of metastasis is generally the tumour draining lymph node(s)26,27 with more aggressive or advanced stage tumours spreading to distal lymph nodes and distant organ sites. To achieve this spread, primary tumour cells must exit the tumour microenvironment, intravasate, travel through the bloodstream, extravasate into the new microenvironment, and colonize the new organ site28. The mechanisms driving tumour cell dissemination have remained unclear for decades, hindering therapeutic development. However, specific features of primary tumours are known to impact their metastatic potential such as tissue and cell of origin. For example, SCLC arises from pulmonary neuroendocrine cells and follows a more aggressive disease course than NSCLC18, while LUAD has an increased tendency to metastasize to the lymph nodes and brain as compared to LUSC29-31. In fact, LUAD is capable of metastasis to multiple organs, specifically contralateral lung, adrenal gland, brain, bone and liver, meaning certain LUAD are highly aggressive32,33. Beyond the metrics of clinical stage, histology and radiographic doubling time, there are currently no clinically implemented biomarkers to determine the aggressiveness of a given lung tumour. Furthermore, there is no tailored therapy specific to the treatment of metastasis. There is a clinical need to better understand the biology driving aggressive lung tumours to, in the long term, develop new biomarkers and therapeutic strategies to combat this devastating disease. 1.3 Molecular pathology of lung cancer and an era of targeted therapy Lung tumour genomes are characterized by one of the highest mutational burdens of all cancer types34,35, making it difficult to differentiate driver mutations, those that are required for tumourigenesis, from non-essential passenger mutations. Despite this, critical driver mutations have been identified for over half of LUAD and 5  include: KRAS, EGFR, PIK3CA, ERBB2, AKT, MAP2K1, BRAF, HRAS, NRAS, MET, and RIT136,37. Additional mechanisms of gene deregulation leading to oncogenic transformation include focal gene amplification of EGFR, MET, and ERBB238, and translocations resulting in ALK, ROS1, RET, and NRG1 fusion proteins39-45.  In addition to positive selection of oncogenes, tumour suppressor genes must be inactivated to achieve cell transformation. Tumour suppressors must often be inactivated on both alleles, as a single copy is often sufficient to repress cell transformation46. Examples of tumour suppressor genes in LUAD include TP53, CDKN2A, RB1, NF1, STK11, and KEAP1. The observed frequencies of known LUAD driver alterations are summarized in Figure 1.2.  Figure 1.2 Frequencies of known driver oncogenes in lung adenocarcinoma KRAS (32.2%)EGFR(11.3%) BRAF(11.3%)NF1(8.3%)NA(24.4%)6  By examining the lung cancer genome, one can define molecular subtypes of lung cancer. These molecular subtypes can appear histologically similar yet be associated with different clinical features19. For example, both KRAS and EGFR mutation are highly associated with LUAD but occur in smokers and non-smokers, respectively, with near mutual exclusivity. The discovery that molecular subtypes of lung cancer are “addicted” to the pathways activated by their respective driver genes presented novel therapeutic opportunities47. Successful targeted therapies in the form of tyrosine kinase inhibitors (TKIs) and antibodies were developed against EGFR48-51 and ALK (EML4-ALK fusion protein)39,52,53, among others. However, not all driver mutations, such as KRAS, are druggable. Furthermore, druggable targets such as EGFR and EML4-ALK are only present in 15-20%54 and 2-7%52 of LUAD, respectively. Nevertheless, these discoveries have shifted clinical practice to include routine molecular testing for these and other actionable mutations. The development of targeted lung cancer therapies resulted in important clinical revelations. (i) Patient pre-selection based on the molecular features of their tumour was found to be essential as clinical benefit was typically not observed in patients lacking the target alteration when compared to standard platinum doublet chemotherapy55-63. (ii) Most patients ultimately develop resistance to targeted therapies64-69. The strategy of targeting a single mutated protein has largely proved ineffective in achieving long term disease stability or cure, and is only applicable to a subset of lung cancer patients. Novel therapeutic targets are needed to improve lung cancer patient outcome. 1.4 Prognostic and predictive expression signatures The detection of EGFR mutation to guide treatment decisions is an example of a predictive marker, while specific tumour and patient features that are used to predict patient outcome independent of treatment are prognostic markers 70. Since tumour samples are routinely preserved in FPPE for pathological processing, an ideal marker is evaluable from such materials. Markers are typically protein-coding 7  genes, which are evaluable by immunohistochemistry (IHC). Some markers have both prognostic and predictive value including mutation of EGFR, which is also a good prognostic marker for both early and late stage NSCLC71-74. Besides specific targeted therapies, the identification and validation of robust, single gene prognostic and predictive markers has proved difficult. For example, KRAS mutation was initially demonstrated to be a marker of poor prognosis75; however, it failed to validate in additional studies76. The incorporation of multiple genes into a prognostic signature is thought to be a superior metric than the use of a single gene77,78. Several prognostic and predictive multigene classifiers are in clinical testing for NSCLC and non-squamous NSCLC, and involve between 12 and 160 protein-coding genes79-84. These were largely developed from publically available gene expression microarray data or quantitative PCR (qPCR) panels on FFPE-derived RNA and were generated and validated in large cohorts of hundreds of NSCLC samples. These signatures have been applied to not only predict patient prognosis, but also to predict which early stage (I-IIIA) patients would benefit from adjuvant chemotherapy following surgery. Approximately 30-40% of patients who undergo surgery with curative intent relapse, but there is currently no way to identify at-risk patients85,86. A biomarker panel would minimize risk of relapse for high risk patients while sparing low risk patients from unnecessary chemotherapy. Of note, there is not much gene overlap between signatures, which has raised concerns as to their robustness and reliability. Currently, there is no widely accepted or clinically implemented multigene signature for lung cancer. Newer multigene signatures have ventured into the landscape of non-coding genes, which encode RNA molecules that do not get translated into protein. Namely microRNAs (miRNAs), discussed in more detail below, have been studied for these purposes 87-92 (Appendix I, item 12). These small RNA species are attractive candidates due to their stability in FFPE tissues and biofluids, and their ability to 8  regulate multiple gene targets. By measuring expression of a small number of miRNAs one gains a more complete picture of the biology of a given tumour. Profiling studies have shown miRNAs are better able to stratify tissue types and cancer subtypes than protein-coding genes93,94, a promising finding for the improved stratification of NSCLC patients into risk or response groups. Despite a multitude of studies95-106, these multi-miRNA signatures have not yet undergone the rigorous testing required for integration into a clinical setting. The development of multi-gene signatures, either protein-coding or non-coding, to predict patient outcome remains an important clinical goal. 1.5 Non-coding RNAs in lung cancer biology Protein-coding genes represent less than 2% of the human genome, yet the majority of the genome is transcribed107,108. The past two decades of research have revealed a vast network of non protein-coding genes that are important to normal cell biology and that are deregulated in disease, including cancer109,110 (Appendix I, item 15). These non-coding RNAs (ncRNAs) can be divided into small and long categories simply based on length. Small ncRNAs are 17-200 nucleotides in length, and encompass microRNAs (miRNAs), PIWI-interacting RNAs (piRNAs), transfer RNAs (tRNAs), small interfering RNAs (siRNAs), small nucleolar RNAs (snoRNAs), and some ribosomal RNAs. Long ncRNAs (lncRNAs) are >200 nucleotides in length, and also include antisense RNAs (asRNAs) and pseudogenes111 (Appendix I, item 14). The role of ncRNAs in lung cancer biology is not yet fully understood; however, the multifaceted ability of these small and long ncRNAs to regulate cell biology and transcriptional programs renders them worthy of study. The deregulation of two classes of small ncRNAs, miRNAs and piRNAs, is examined in this thesis. There are an estimated 5,000 miRNAs encoded in the human genome112. These small ncRNAs exhibit highly tissue-specific expression patterns and show superior 1.5.1 microRNAs in lung cancer 9  clustering of tissues than protein-coding genes93,94. Furthermore, miRNA expression is able to distinguish between lung cancer subtypes with over 90% accuracy113-116. It is estimated that the majority of protein-coding genes are targeted by at least one miRNA117 . Similarly, a given miRNA is predicted to target dozens to hundreds of transcripts, making them powerful regulators of gene expression. miRNAs function to repress gene translation by binding to the 3’UTR of target transcripts through sequence homology. miRNA genes are transcribed by DNA polymerase II into long (100s-1000s nucleotides long) hairpin RNA structures called primary (pri)-miRNAs. These pri-miRNAs are cleaved by a microprocessor complex containing the endonuclease Drosha and DGCR8 into pre-miRNAs approximately 70 nucleotides in length that maintain a hairpin structure and are exported to the cytoplasm by Exportin-5 and RanGTP. Pre-miRNAs undergo further nuclease activity by Dicer to generate ~17-24 nucleotide long duplexes. The miRNA duplex is dissociated by a helicase allowing incorporation of the active mature strand into the RNA-induced silencing complex (RISC), which includes a member of the Argonaute protein family and facilitates binding of target sequences. Target sequences either undergo transcript degradation or inhibition of translation depending on the degree of sequence complementarity111,118 (Figure 1.3).  miRNAs can have tumour suppressive or oncogenic functions depending on their gene targets. If a miRNA targets an oncogene, it has tumour suppressive functions, and vice versa. Many oncogenic and tumour suppressive miRNAs have been identified in lung and other cancer types, such as the tumour suppressive let-7 and the oncogenic miR-21119,120. As discussed above, miRNA expression has been associated with patient prognosis and the use of miRNAs as biomarkers is appealing. There is potential to utilize miRNAs as therapeutics themselves, either through delivery of tumour suppressive miRNA mimics, or through delivery of inhibitors targeting oncogenic miRNAs. Both strategies are currently being pursued by pharmaceutical companies, including MIRNA Therapeutics (Austin, TX, USA) and Regulus (Carlsbad, CA, USA). The fact that miRNAs are naturally occurring 10  molecules that have the potential to broadly affect tumour cell viability by regulating expression of many target transcripts makes them an exciting new class of anti-cancer drug. piRNAs function in a manner analogous to miRNAs in that a small RNA interacts with a member of the Argonaute protein family to guide the RNA-protein complex to target sequences. While miRNAs interact with various Ago proteins and antagonize gene expression at the RNA level, piRNAs interact with PIWI and are thought to predominantly function in the epigenetic silencing of DNA121-123 (Figure 1.3, Appendix I, item 1). PIWI was first discovered in the germline of Drosophila melanogaster where the bulk of PIWI/piRNA research has been carried out. In germline cells, PIWI/piRNA silence selfish genetic elements such as transposable elements 124,125. Novel functions have been described in D. melanogaster and mice whereby piRNAs may regulate gene expression at the mRNA level 126-128. Recent sequencing studies have revealed piRNAs are expressed in somatic tissues in various species, including humans129-136 (Appendix I, items 3 & 6). Somatically expressed piRNAs have been found to display novel functions such as regulation of gene-specific methylation137-139, chromatin modification122,140, cell cycle regulation134, and targeting of mRNA transcripts127. In humans, piRNA expression can be deregulated in cancer and associated with patient outcome134,136,141-153 (Appendix I, items 4-6). Similar to other ncRNAs, piRNA expression patterns are highly tissue and disease-specific and correlate with clinical features136. Due to their similarity in size, the potential clinical applications of miRNAs also apply to piRNAs. 1.5.2 piRNAs in cancer 11   Figure 1.3 Mechanism of action of miRNAs and piRNAs (Top) miRNAs function at the RNA level by repressing gene translation. miRNA sequences are incorporated into the RISC which facilitates binding of target mRNA sequences at the 3’UTR via sequence complementarity. (Bottom) piRNAs function at the DNA level where they induce epigenetic silencing of transcripts. piRNAs interact with PIWI which facilitates binding of DNA target sequences in the 5’UTR region of genes. This can lead to methylation and silencing of DNA sequences123.   12  1.6 An integrative approach to understanding lung cancer biology Previous methodologies have succeeded in identifying essential lung cancer oncogenes disrupted by mutation, gene amplification, and genomic rearrangements for over half of all LUAD. These genes drive LUAD biology and therapeutic inhibition of these drivers has improved survival for a subset of patients. However, the majority of LUAD patients harbour driver alterations for which there is no targeted therapy. Furthermore, there is no targeted therapeutic strategy for the treatment of metastasis, which is the main cause of cancer-related death. Alternative approaches to cancer gene discovery are required to identify novel alterations that may be therapeutically actionable. The integration of multiple ‘omics levels represents an alternative approach to cancer gene discovery. This method allows for the identification of genes or protein complexes infrequently altered by a single mechanism, but frequently altered by multiple mechanisms that would have been overlooked by conventional analysis methods (Figure 1.4)154. This approach takes advantage of the fact that current technologies allow for whole-genome profiling on multiple high resolution platforms. The analysis of multiple data levels within a given tumour also allows for the identification of genes displaying bi-allelic disruption, which is a strong indicator of DNA-level selection and functional relevance. Furthermore, the impact of DNA-level disruption can be queried at the RNA level to filter out passenger alterations with no impact on gene expression. Our previous work utilizing this approach has lead to the discovery of novel oncogenes and tumour suppressor genes155-157 (Appendix I, item 9), and identified highly deregulated protein complexes158 (Appendix I, item 8). 13   Figure 1.4 Integrating the multi’omic dimensions of DNA level alterations in cancer  Genes are revealed to be more frequently disrupted when multiple mechanisms of DNA level alterations are considered. (E.g. mutation, copy number, methylation). Gene X is a hypothetical example.   14  1.7 Outstanding clinical issues in lung cancer As stated previously, the 5-year survival rate for lung cancer patients is 18%. The majority of cancer-related deaths are a result of metastasis for which there is no tailored therapy. There is a pressing need to better understand aggressive tumour biology in order to identify both (i) markers of aggressiveness, and (ii) biological targets to treat advanced disease. Furthermore, irrespective of stage, the majority of patients do not harbor actionable mutations, and those that do ultimately develop resistance to targeted therapies. Alternative methods of gene discovery must be pursued in order to identify novel therapeutic targets. 1.8 Thesis objective and research approach The main objective of this work is to identify novel genes important to aggressive lung tumour biology. In this thesis, “aggressiveness” refers to: metastasis, poor overall survival, poor recurrence free survival, or rapid tumour growth. “Genes” refers to protein-coding or non-protein-coding genes that are deregulated in or associated with aggressive lung tumours. To begin to address this objective, we take two alternative approaches to gene discovery moving beyond the traditional analysis of protein-coding gene deregulation in one dimension (e.g. mutation). Firstly, we investigate the landscape of small non-coding gene deregulation in metastatic lung cancer. Secondly, we employ a multidimensional integrative approach to search for novel lung cancer oncogenes that may be frequently disrupted by mechanisms other than mutation. 1.9 Overarching hypothesis To identify novel genes important to aggressive lung tumour biology, in both the non-coding and coding landscape, we test two hypotheses: 15  (1) Non-coding RNA genes are differentially expressed between aggressive and non-aggressive lung tumours, and functionally contribute to aggressive tumour biology. (2) Genes disrupted frequently by mechanisms other than mutation regulate aggressive lung tumour biology. 1.10 Specific aims Aim 1. Identify non-coding genes important to aggressive lung adenocarcinoma biology by characterizing non-coding RNAs differentially expressed in primary tumours with and without metastasis (Chapter 3 and 4) A strong indicator of aggressive tumour biology is its ability to metastasize. Chapter 3 describes the generation of small RNA sequencing profiles from metastatic and localized primary lung tumours and matched adjacent non-malignant lung tissues. In this analysis we sought to identify miRNAs specifically deregulated in metastatic tumours. At the time of analysis, there was a paucity of small RNA sequencing studies in general, with the majority of miRNA-related publications utilizing miRNA microarrays or qPCR plates. Small RNA sequencing allows for the identification of all annotated miRNAs, and allows for re-processing of data as new annotations become available. Importantly, the deregulation of miRNAs in metastatic lung cancer is not fully characterized. Another metric of aggressive tumour biology is patient outcome. In Chapter 4, results from Chapter 3 were utilized to identify miRNAs that predict patient overall survival and recurrence free survival. Furthermore, expression data for a novel class of small RNA, piRNA, was derived by re-mining the sequencing data. The ability of both miRNAs and piRNAs to predict lung cancer patient overall survival and recurrence free survival was assessed.  16  Aim 2. Search for novel lung adenocarcinoma oncogenes by querying components of known lung cancer pathways Driver mutations have been identified for over half of all LUAD; however, additional oncogenes remain undiscovered that may be deregulated by alternative mechanisms such as DNA copy number alterations and methylation changes. Chapter 5 investigates the relevance of a transcription factor, E74-like factor 3 (ELF3), in lung cancer biology. ELF3 is encoded on chromosome 1q, a region that undergoes frequent DNA gain in LUAD and is associated with aggressive clinical features159,160. Furthermore, ELF3 is a component of several oncogenic lung cancer pathways, namely ERBB2, TGFβ, and NOTCH161,162. ELF3 has not yet been identified as an oncogene by large-scale comparative gene expression or mutation studies. We hypothesize that ELF3 is a novel LUAD oncogene.   17  2 Methods 2.1 Patient tissue accrual Lung tumour specimens and non-malignant lung tissues were collected from treatment-naïve patients undergoing surgery with curative intent. Tissues were obtained from the Tumour Tissue Repository of the British Columbia Cancer Agency or Vancouver General Hospital under informed written patient consent and with approval from the University of British Columbia – BC Cancer Agency (BCCA) Research Ethics Board. Since Stage IV patients are not eligible for surgery, the majority of cases were Stage I-III. Distant metastases were discovered in three patients resulting in their reclassification as Stage VI. 2.2 Nucleic acid extraction Fresh frozen tissues were reviewed by a pathologist to assess histology and identify regions of >80% tumour cell content for tumour specimens.  Non-malignant tissues were confirmed to be histologically normal. DNA was extracted using standard phenol:chloroform procedures and RNA was extracted using Trizol, unless otherwise specified. 2.3 Molecular profiling summary DNA and RNA extracted from tumour and non-malignant tissues was subjected to whole genome profiling to carry out the Specific Aims. Data types include DNA copy number, DNA methylation, RNA expression, miRNA expression, and piRNA expression. In addition to data generated in house, publically available data was acquired from The Cancer Genome Atlas (TCGA) Data Portal (https://tcga-data.nci.nih.gov). A description of the profiling platforms utilized in each cohort is found in Table 2.1, and described in detail below. More detailed cohort summaries can be found in each data chapter. 18  Table 2.1 Platforms used for multi-omic profiling of both clinical cohorts used in this thesis. Cohort DNA copy number DNA methylation RNA expression miRNA & piRNA expression BCCA Affymetrix Genome-Wide Human SNP Array 6.0 Illumina Infinium Human Methylation 27 Array Illumina HT-12 Whole Genome 6, v3 BeadChip Array Illumina HiSeq 2000 small RNA sequencing TCGA Affymetrix Genome-Wide Human SNP Array 6.0 Illumina Infinium Human Methylation 450 Array Illumina HiSeq 2000 RNA sequencing Illumina HiSeq 2000 small RNA sequencing    19  2.3.1.1 microRNA profiling in both cohorts Total RNA from matched LUAD and non-malignant lung samples was extracted from fresh frozen tissue and subjected to miRNA sequencing. miRNA-sequencing libraries were constructed, bar-coded for sequencing, and sequenced using the Illumina HiSeq 2000 sequencing platform using a plate-based protocol developed at the British Columbia Genome Sciences Centre (BCGSC). miRNA alignment to the human genome (GRCh37/hg19) and quantification was performed using the BCGSC protocol163. Raw sequence reads per sample were identified by an assigned index, adapter sequences were removed, and reads were trimmed based on quality control metrics. In Chapter 3, high quality reads were aligned using the BWA algorithm. Reads were normalized per million (RPM), and RPM values for 1372 miRNAs (miRbase V20) were derived per sample. For Chapter 4, miRNA expression values were required to be comparable to piRNA expression values; therefore, reads were aligned using Spliced Transcripts Alignment to a Reference (STAR)164 and normalized as reads per kilobase of exon model per million mapped reads (RPKM). RPM or RPKM values for identical miRNAs mapping to more than one genomic location were summed, and for all miRNAs, an RPKM<1 was transformed to 0. 2.3.1.2 piRNA profiling in both cohorts Reads were first subject to quality control to exclude non-biological artifacts. Then, unaligned reads (in FASTQ format) were trimmed by size (retained reads ≥ 23 bp) and quality score (Phred quality scores ≥ 20) in order to enrich for high-quality reads mapping to piRNAs. Using the PartekFlow™ platform (Partek Inc., MO, USA), high-quality reads were mapped to the human genome (GRCh37/hg19) using the STAR aligner. Reads were quantified by an Expectation/Maximization (E/M) algorithm165 using a piRNA-specific annotation file generated from the piRNABank database (http://pirnabank.ibab.ac.in/)166. Partek Genome Suite was used to further 2.3.1 Small RNA sequencing 20  process and filter quantified files. Reads per kilobase of exon model per million mapped reads (RPKM) was used to scale and normalize read count167. BCCA DNA was hybridized to Affymetrix Genome-Wide Human SNP Array 6.0 arrays according to the manufacturer’s instructions. Raw CEL probe intensity files were processed and normalized using Partek Genomics Suite. Probe sequence, fragment length, GC content and background adjustments were applied to correct for biases in signal intensities. Tumour copy number profiles were processed using the corresponding non-malignant copy number profile as a baseline. Thresholds for DNA copy number alterations were applied as follows: copy number loss<1.7, copy number gain>2.3. TCGA Level 3 normalized Affymetrix Genome-Wide Human SNP Array 6.0 data, in the form of copy number per gene, per sample, was downloaded from the online data portal. Thresholds for DNA copy number alterations were applied as follows: loss<-0.3, gain>0.3, amplification>0.8. BCCA DNA was bisulfite converted and hybridized to the Illumina Infinium Human Methylation 27 array. This array assays over 27,000 CpG sites representing more than 14,000 unique genes. Raw methylation data was corrected for colour bias and normalized using SSN normalization with the Bioconductor package lumi in R statistical computing software. Probe hypermethylation was defined as a change in β-Value (∆βV) ≥ 0.15 (at least 15% more methylated in tumour), whereas probe 2.3.2 DNA copy number 2.3.3 DNA methylation data 21  hypomethylation was defined as a ∆βV ≤ -0.15 (at least 15% less methylated in tumour). TCGA Level 3 normalized Illumina Infinium Human Methylation 450 data, in the form of calculated β-values mapped to genome, per sample, was downloaded from the online data portal. This array interrogates more than 485,000 methylation sites at single nucleotide resolution and covers 99% of RefSeq genes and 96% of CpG islands. Probe hyper and hypomethylation was defined as described above. BCCA Expression profiles from total RNA were generated on the Illumina HT-12 Whole Genome 6, v3 BeadChip array according to the manufacturer’s instructions. This array contains over 48,000 probes measuring expression levels for over 25,000 genes. Bead-level data were pre-processed using the R package mbcb to perform background correction and probe summarization. Data were then quantile normalized and log2 transformed. Fold change was calculated by subtracting the normal log2(expression) value from the paired malignant log2(expression) value, and transforming the log2(fold change) value: Fold change=2^(log2(fold change)). TCGA Level 3 RNA sequencing data, defined as the calculated expression signal of a gene, per sample, was downloaded from the online data portal. Fold change was calculated as described above. 2.3.4 RNA expression data 22  2.4 Cell lines Cell lines were obtained from American Type Culture Collection (ATCC, Manassas, VA) and maintained according to ATCC guidelines. LUAD cell lines were cultured in RPMI 1640 (Gibco - ThermoFisher) supplemented with 10% FBS. Human Bronchial Epithelial Cells (HBECs) are untransformed cells that have been immortalized by expression of cyclin-dependent kinase (Cdk) 4 and humantelomerase reverse transcriptase (hTERT)168. HBECs were cultured in Opti-MEM media supplemented with 0.0002 ng/μl EGF and 30 μg/ml BPE.    23  Table 2.2 Cell lines used for in vitro and in vivo assays in this thesis  Cell Line Tissue In vitro In vivo Chapter H2347 Lung adenocarcinoma X  Chapter 3 HCC827 Lung adenocarcinoma X X Chapter 5 H1395 Lung adenocarcinoma X  Chapter 5 H1819 Lung adenocarcinoma X  Chapter 5 H1993 Lung adenocarcinoma X  Chapter 5 A549 Lung adenocarcinoma X X Chapter 5 HBEC Bronchial epithelial cells X  Chapter 5  2.5 Basic laboratory techniques Whole cell extracts were prepared in RIPA lysis buffer (20 mM Tris-HCl, pH 7.5; 150 mM NaCl; 0.5% DOC, 1% NP-40 and 0.1% SDS). For cytoplasmic and nuclear fractionation, cells were first pelleted and resuspended in Buffer A (10 mM HEPES, pH 7.9; 1.5 mM MgCl2; 10 mM KCl). After 10 minutes of incubation on ice, lysates were pelleted and supernatant was collected as the cytoplasmic extract. The pellet was washed two times with Buffer A and resuspended in Buffer C (20 mM HEPES, pH 7.9; 25% glycerol; 420 mM NaCl; 1.5 mM MgCl2 and 0.2 mM EDTA). Tubes were incubated for 20 minutes on ice and then centrifuged to clear the nuclear extract. All 2.5.1 Immunoblot 24  lysis buffers were supplemented with protease and phosphatase inhibitors (2 mM Na3VO4; 1 mM NaF; 2 mM β-glycerolphosphate; 0.2 mM PMSF; 0.5 mM DTT and Complete protease inhibitor cocktail (Roche Diagnostics, Laval, QC, Canada)). Protein concentrations were determined by the BCA Protein Assay (Pierce, Rockford, IL, USA) according to the manufacturer's recommendations. Immunoblot analysis was performed on cell-equivalent lysates subjected to SDS-PAGE and electrophoretic transfer to nitrocellulose membrane (Bio-Rad Laboratories, Mississauga, ON, Canada). Membranes were blocked in 5% milk + TBS-T for 1 hour, washed in TBS-T 3X for 10 minutes, and then incubated in primary antibody overnight at 4°C at an optimized concentration in 5% milk + TBS-T. Membranes were washed 3X for 10 minutes in TBS-T before incubation in secondary antibody for 1 hour at room temperature. Blots were developed using the ECL detection kit (Amersham, Mississauga, ON, Canada). Primary antibodies used in this thesis are summarized in Table 2.3. Table 2.3 List of primary antibodies used in this thesis Protein target Clone Company Catalog number ELF3 EPESER1 Abcam ab133621 Vimentin 5G3F10 Cell Signaling 3390S N-Cadherin H-4 Santa Cruz sc-271386 E-cadherin 24E10 Cell Signaling 3195S Histone H3 D1H2 Cell Signaling 4499S    25  3 Identification of microRNAs deregulated in metastatic lung adenocarcinoma 3.1 Introduction MicroRNAs (miRNAs) are small non-coding RNAs between 17-25 base pairs in length that function to inhibit the translation of target transcripts into protein. miRNA expression is frequently deregulated in cancer and involves the overexpression of oncogenic miRNAs and underexpression of tumour suppressive miRNAs. By analyzing predicted targets based on sequence complementarity at the 3’UTR of target mRNAs, certain miRNAs likely regulate pathways at multiple nodes and can even regulate multiple pathways169. In a sense, miRNAs are comparable to transcription factors, since they have the capacity to regulate hundreds of genes. The involvement of miRNAs in the process of lung cancer metastasis is not fully understood. One strategy to better understand the biology driving tumour metastasis is to compare expression profiles of non-metastatic tumours to metastatic tumours and identify differentially expressed genes. Given the contribution of miRNA deregulation to lung cancer biology, we hypothesized that miRNAs are differentially expressed between metastatic and localized primary LUAD, and contribute to metastatic phenotypes of LUAD cells. We applied a stringent set of criteria to the dataset in order to identify the most significantly deregulated miRNAs, and sought to validate the contribution of these miRNAs to the regulation of metastatic phenotypes in vitro. Finally, we sought to validate their clinical relevance in an external dataset (TCGA). 3.1.1 miRNA deregulation in cancer 3.1.2 Rationale for assessing miRNA deregulation in metastatic lung adenocarcinoma 26  3.2 Methods The BCCA has acquired numerous surgical specimens that were fresh frozen soon after collection, making them ideal for whole genome profiling (Chapter 2, Section 2.1). Lung tumour specimens as well as non-malignant lung tissues were collected at time of surgery, allowing for paired data analyses. For this study, primary LUAD tumours that had not metastasized (TNM stage N0 & M0) were compared to those that had metastasized (TNM stage N+ &/or M+). The final patient cohorts for this analysis included 61 cases with no locoregional or distant metastasis and 32 cases with metastasis. Patients were matched for age, gender, ethnicity, and smoking history. The clinical features of these patient groups are summarized in Table 3.1.   3.2.1 Patient cohort 27  Table 3.1 Clinical features of the BCCA lung adenocarcinoma cohort. Clinical Feature Localized (n=61) Metastatic (n=32) Test p-value Age range (median) 39-86 (71) 49-90 (65.5) t-test 0.1315 Gender         Male 21 8 Fisher's 0.4803 Female 40 24   Ethnicity         Caucasian 38 18 χ2 0.8043 Asian 16 9   NA 7 5   Smoking History         CS 29 14 χ2 0.9281 FS 15 8   NS 17 10   Stage         I 57 0 χ2 <0.0001 II 3 18   III (IV) 1 (0) 11 (3)   28  Generation of miRNA expression data was described in Section 2.4.1. The following expression criteria was applied to the dataset to identify expressed miRNAs: (i) miRNAs expressed in ≥10% of LUAD or non-malignant samples, and (ii) a median expression of ≥10 RPM in LUAD or non-malignant samples (n=222 miRNAs). This pre-processing was required to remove miRNAs that are not expressed in malignant or non-malignant tissues, and to exclude miRNAs with very low expression levels. A Wilcoxon sign-rank test was applied to compare expression of the 222 miRNAs between paired tumour and non-malignant tissues, and p-values were corrected using the Benjamini-Hochberg method. Expression fold change was calculated on a case-by-case basis, and fold change was compared between metastatic and localized cohorts using a two-tailed Mann Whitney U test. Differentially expressed miRNAs had a Mann Whitney U test p-value <0.05 and an average fold change of at least ±2-fold. Cells were cultured according to ATCC guidelines. SMARTchoice human lentivirus containing constitutive miRNA overexpression vector tagged with GFP, as well as a GFP-only control, was utilized to achieve overexpression of both miRNAs (Open Biosystems, Ottawa, ON, Canada). Following lentiviral delivery, cells were cultured in selective puromucin-containing media for 7 days to kill untransfected cells. As the fold change observed in non-metastatic and metastatic LUAD was between 2- to 4-fold, a low multiplicity of infection (Multiplicity of Infection (MOI) = 1.0) was used to achieve a similar increase in miRNA abundance in vitro. This MOI generated 2.3 and 2.9 fold increases for miR-106a and miR-106b, respectively.   3.2.2 miRNA expression analysis 3.2.3 Cell models 29  Boyden chamber assays were used to assess cell migration and invasion. Cells were cultured in serum-free media for 24 hours prior to seeding at optimized densities into the top of uncoated (migration) or Matrigel coated (invasion) Boyden chambers (8μm pore size, Fisher, Ottawa, ON, Canada). The top of the chamber contained serum-free media while the bottom of the chamber contained complete media to facilitate chemotaxis. After 24 hours, non-migrated cells were removed from the top of the chamber with a Q-tip, and the cells that had traversed to the bottom of the chamber were fixed in 100% methanol. Boyden chamber filters were removed and placed on glass microscope slides, mounted in DAPI-containing medium, and imaged using a fluorescent microscope. Images were then used for cell quantification. 3.3 Results A total of 222 miRNAs were expressed according to our criteria; 201 were significantly differentially expressed between LUAD and non-malignant lung. Next, miRNA expression was compared between metastatic and localized cohorts. To maintain valuable information available due to paired sampling, expression fold change was calculated for each of the 201 significantly differentially expressed miRNAs on a case-by-case basis. Fold change values for each miRNA were compared between the metastatic and localized groups using a two-tailed Mann Whitney U test. The fold change values of 37 miRNAs were significantly different between metastatic and localized cases (p<0.05; FC≥±2-fold) (Table 3.2). A total of 27 miRNAs were significantly overexpressed in metastatic cases, and 7 miRNAs were significantly underexpressed, relative to both non-malignant tissues and to localized cases. Three miRNAs were significantly overexpressed in LUAD relative to non-malignant lung, but with a higher average fold change in the localized group. When ranked by lowest p-value, the fold change of miR-106b was most significantly 3.2.4 Migration and invasion assays 3.3.1 Identification of miRNAs deregulated in metastatic lung adenocarcinoma 30  different between metastatic and non-metastatic samples (p=5.18E-05) with higher expression in the metastatic group (Figure 3.1).31  Table 3.2 List of 37 miRNAs significantly differentially expressed between localized and metastatic lung adenocarcinoma. miRNA Chr BP Start BP End Sign-rank B-H corrected p-value AVG Fold Change (Localized) AVG Fold Change (Metastatic) Mann Whitney U test p-value  Metastatic vs. Localized miR-1307 chr10 105154010 1.05E+08 0.00E+00 3.0 4.2 6.04E-03 OE miR-708 chr11 79113066 79113153 0.00E+00 7.2 9.8 1.65E-02 OE miR-34b chr11 111383663 1.11E+08 2.67E-04 -5.0 -9.0 2.66E-03 UE miR-34c chr11 111384164 1.11E+08 2.77E-09 -12.1 -84.9 1.87E-03 UE miR-18a chr13 92003005 92003075 0.00E+00 3.5 9.9 1.80E-02 OE miR-20a chr13 92003319 92003389 2.77E-09 1.5 2.1 1.96E-02 OE miR-337 chr14 101340830 1.01E+08 5.85E-03 1.1 3.6 3.73E-02 OE miR-127 chr14 101349316 1.01E+08 5.98E-07 2.1 5.8 4.48E-03 OE 32  miRNA Chr BP Start BP End Sign-rank B-H corrected p-value AVG Fold Change (Localized) AVG Fold Change (Metastatic) Mann Whitney U test p-value  Metastatic vs. Localized miR-136 chr14 101351039 1.01E+08 9.85E-07 4.1 13.4 3.51E-02 OE miR-758 chr14 101492357 1.01E+08 3.88E-05 2.7 6.4 3.88E-02 OE miR-381 chr14 101512257 1.02E+08 2.45E-02 26.1 1.5 1.58E-02 UE miR-382 chr14 101520643 1.02E+08 3.55E-08 2.9 6.7 2.53E-02 OE miR-134 chr14 101521024 1.02E+08 2.41E-08 2.7 6.9 1.51E-02 OE miR-409 chr14 101531637 1.02E+08 0.00E+00 11.1 18.8 2.93E-02 OE miR-1247 chr14 102026624 1.02E+08 0.00E+00 -5.4 -17.1 2.46E-03 UE miR-196a chr17 46709852 46709921 0.00E+00 105.9 187.0 9.74E-03 OE miR-217 chr2 56210102 56210211 2.25E-02 2.1 1.1 4.89E-02 UE miR-133a chr20 61162119 61162220 0.00E+00 -4.0 -10.0 4.99E-02 UE 33  miRNA Chr BP Start BP End Sign-rank B-H corrected p-value AVG Fold Change (Localized) AVG Fold Change (Metastatic) Mann Whitney U test p-value  Metastatic vs. Localized miR-130b chr22 22007593 22007674 0.00E+00 8.6 14.5 9.01E-04 OE miR-128 chr3 35785968 35786051 0.00E+00 3.4 2.3 9.77E-03 UE miR-425 chr3 49057581 49057667 0.00E+00 3.3 5.1 4.62E-02 OE miR-16 chr3 160122533 1.6E+08 0.00E+00 1.7 2.5 1.65E-02 OE miR-196b chr7 27209099 27209182 0.00E+00 10.7 56.8 7.79E-04 OE miR-93 chr7 99691391 99691470 0.00E+00 1.2 2.2 7.13E-04 OE miR-106b chr7 99691616 99691697 0.00E+00 1.4 2.4 5.18E-05 OE miR-182 chr7 129410223 1.29E+08 0.00E+00 5.1 6.7 1.96E-02 OE miR-96 chr7 129414532 1.29E+08 0.00E+00 23.6 27.4 4.12E-02 OE miR-183 chr7 129414745 1.29E+08 0.00E+00 7.4 10.1 3.74E-03 OE 34  miRNA Chr BP Start BP End Sign-rank B-H corrected p-value AVG Fold Change (Localized) AVG Fold Change (Metastatic) Mann Whitney U test p-value  Metastatic vs. Localized miR-598 chr8 10892716 10892812 0.00E+00 -3.9 -9.5 1.51E-02 UE let-7a chr9 96938239 96938318 0.00E+00 -2.0 -2.5 7.17E-03 UE miR-363 chrX 133303408 1.33E+08 2.65E-02 -2.3 -1.1 3.12E-03 OE miR-106a chrX 133304228 1.33E+08 0.00E+00 2.5 4.1 1.61E-02 OE miR-450b chrX 133674215 1.34E+08 2.77E-09 1.8 8.8 9.83E-04 OE miR-450a chrX 133674538 1.34E+08 0.00E+00 5.4 10.0 1.58E-02 OE miR-542 chrX 133675371 1.34E+08 3.26E-07 0.9 4.4 1.77E-03 OE miR-503 chrX 133680358 1.34E+08 0.00E+00 8.6 20.2 3.79E-04 OE miR-424 chrX 133680644 1.34E+08 0.00E+00 5.3 13.7 1.87E-03 OE 35   L o c a liz e d M e ta s ta t ic0123m iR -1 0 6 bFold Changep = 5 .1 8 E -0 5 Figure 3.1 Expression fold change of the most significantly differentially expressed microRNA, miR-106b Histogram (mean + SEM) of miR-106b expression fold change between paired non-malignant lung and lung adenocarcinoma samples in localized (green) and metastatic (red) samples. Mann Whitney U test p-value is shown.    36  Interestingly, miR-106b mapped to a cluster of three miRNAs (miR-106b, miR-93, and miR-25) on chromosome 7. miR-93 and miR-25 were also significantly deregulated between metastatic and non-metastatic groups; however, miR-25 did not meet the ≥2-fold cut-off and was underexpressed in LUAD. Two paralogs, genes resulting from duplication events within a genome, of the miR106b~25 cluster exist in the human genome: miR-106a~363 on chromosome X and miR-17~92 on chromosome 13 (Figure 3.2). miR106a and miR-363 of the cluster on chromosome X, and miR-18a and miR-20a of the cluster on chromosome 13 were significantly deregulated in metastatic LUAD. Similarly to the miR-106b~25 cluster, miR-106a was significantly overexpressed while miR-363 was underexpressed in metastatic LUAD.   3.3.2 Genomic positions of miRNAs deregulated in metastatic lung adenocarcinoma 37   Figure 3.2 Paralog microRNA clusters harbouring metastasis-associated miRNAs Schematic of three paralogous microRNA clusters: miR-17~92 on chromosome 13 is embedded within the protein coding gene C13ORF25; miR-106b~25 is embedded within the protein coding gene MCM7; miR-106a~363 is located on chromosome X and is not embedded in a protein coding gene. MicroRNA genes are colour-coded according to family, which is based on sequence homology. 3’ to 5’ directionality is indicated.   38  Nine of the miRNAs significantly deregulated in metastatic LUAD mapped to chromosome 14q32.1-32.31. This region is the imprinted DLK1-DIO3 locus which harbors the largest miRNA clusters in the genome. All of the miRNAs in the cluster were significantly overexpressed in LUAD, while miR-1247, which is located outside of the cluster at the 3’ end of the locus, was underexpressed. This observation will be further discussed in Chapter 4. The most significantly differentially expressed miRNA, miR-106b, and the related miR-106a, were chosen for further study since both miRNAs were statistically overexpressed in metastatic LUAD. The expression of miR-106a and miR-106b in non-malignant, localized, and metastatic lung samples is shown in Figure 3.3A-B, as well as the expression fold change of miR-106a in localized and metastatic samples (Figure 3.3C). Interestingly, the expression of miR-106a and miR-106b was significantly positively correlated (Spearman’s ɾ=0.395, p<1.0E-03) (Figure 3.3D).  3.3.3 miRNA paralogs miR-106b and miR-106a are overexpressed in metastatic lung adenocarcinoma 39   Figure 3.3 Expression of miR-106a and miR-106b in clinical specimens 40  Figure 3.3 Expression of miR-106a and miR-106b in clinical specimens. Box and whisker plots of log2(RPM) expression of (A) miR-106a and (B) miR-106b in localized LUAD (green) and corresponding non-malignant lung (light green), and metastatic LUAD (red) and corresponding non-malignant lung (pink). Whiskers represent the 5th-95th percentiles; the middle horizontal bar represents the mean. Expression between paired tissues was compared by sign rank test, expression between tumour groups was compared by Mann Whitney U test. (C) Histogram (mean + SEM) of miR-106a expression fold change between paired non-malignant lung and LUAD samples in localized (green) and metastatic (red) samples. Mann Whitney U test p-value is shown. (D) Spearman’s correlation of log2(miR-106a RPM) and log2(miR-106b RPM) in our cohort of 93 paired cases of LUAD and non-malignant lung.    41  To determine if miR-106a and miR-106b have the capacity to regulate metastatic phenotypes in vitro, we established constitutive overexpression models in a non-aggressive cell line to attempt to increase aggressive properties. We chose H2347 cells due to their inherently low miR-106a and miR-106b expression, epithelial morphology, and epithelial expression profile as defined by high E-cadherin expression, and low Vimentin and N-cadherin expression (Figure 3.4). Migration and invasion was assessed in miR-106a and miR-106b overexpressing cells and compared to empty vector controls by Boyden chamber assay. No significant change in migration was observed in miR-106a or miR-106b overexpressing cells compared to GFP only controls (Figure 3.5A). However, there was a significant increase in invasion of miR-106b overexpressing cells compared to GFP-only control cells (paired two-tailed Student’s t-test p=1.8E-02) (Figure 3.5B).   3.3.4 In vitro assessment of metastatic phenotypes using cell models  42    Figure 3.4 Epithelial and mesenchymal expression profiles of LUAD cell lines Immunoblot of epithelial marker E-cadherin and mesenchymal markers N-cadherin and Vimentin, with Histone H3 as a loading control. Lysates from the mesenchymal cell line H1792 were compared to lysates from H2347.   43   Figure 3.5 Boyden chamber assay results Histograms (mean + SEM) displaying (A) migrated and (B) invaded cell counts from three biological triplicate experiments for isogenic H2347 empty vector control cells (black), H2347 cells constitutively overexpressing miR-106a (blue), and H2347 cells constitutively overexpressing miR-106b (purple). Paired two-tailed Student’s t-test p-values are shown.     44  Another metric of aggressiveness is epithelial-to-mesenchymal transition (EMT). Cells with a more mesenchymal phenotype have an increased metastatic potential. The ability of miR-106a and miR-106b to induce EMT was assessed by immunoblot of standard epithelial and mesenchymal markers. Overexpression of miR-106a and to a greater extent miR-106b resulted in increased expression of mesenchymal markers N-cadherin and Vimentin, with some reduction in epithelial marker E-cadherin observed (Figure 3.6). Furthermore, cells overexpressing either miR-106a or miR-106b underwent a morphological change and appeared more mesenchymal (elongated) than the GFP-only control cells (Figure 3.6). Taken together, these results indicate miR-106a and miR-106b induce an EMT expression program, while miR-106b alone influences the invasiveness of H2347 cells.    45   Figure 3.6 miR-106a and miR-106b overexpression induces expression changes characteristic of epithelial-to-mesenchymal transition 46  Figure 3.6 miR-106a and miR-106b overexpression induces expression changes characteristic of epithelial-to-mesenchymal transition (A) Immunoblot of epithelial marker (E-Cadherin) and mesenchymal markers (N-Cadherin and Vimentin) in H2347 control, miR-106a overexpressing, and miR-106b overexpressing cells. Histone H3 was used as a loading control. Images of (B) control, (C) miR-106a overexpressing (left panel), and miR-106b overexpressing cells (right panel). Scale=400μm.   47  Associations of miR-106a and miR-106b with aggressive clinical features were queried in the publically available TCGA dataset. Only LUAD samples with miRNA expression, mRNA expression and clinical data were examined (n=268). Tumours with above median expression of both miR-106a and miR-106b were not associated with poor overall survival; however, when further stratified by above median expression of mesenchymal marker vimentin a significant association with survival was observed (log-rank p=2.9E-03). Vimentin alone did not demonstrate any significant association with survival. These cases were also significantly associated with higher tumour stage (χ2 p=2.72E-02) (Figure 3.7).   3.3.5 Associations of miR-106a and miR-106b expression with poor patient outcome in an external dataset 48   Figure 3.7 Assessment of miR-106a and miR-106b expression and overall survival in The Cancer Genome Atlas cohort Kaplan-Meier curves comparing overall survival of patients with high (red) vs. low (blue) expression of (A) miR-106a and miR-106b, (B) mesenchymal marker vimentin, and (C) miR-106a, miR-106b and vimentin. Log-rank p-values are shown. (D) Percent distribution of stage I (green), stage II (blue) and stage III (red) tumours with above or below median expression of all three markers (miR-106a, miR-106b and vimentin). Chi squared p-value is shown and was calculated using tumour sums per group.   49  3.4 Discussion Through generation of small RNA sequencing data from 186 clinical samples, we have identified a panel of 37 miRNAs significantly differentially expressed both between non-malignant and malignant lung tissues, and between metastatic and localized primary LUAD. To our knowledge, this is the largest comparative gene expression study to date that has profiled paired tumour and non-malignant tissues by small RNA sequencing. More recently, the TCGA has performed multi-omic profiling of approximately 500 LUAD, including small RNA sequencing; however, only 50 tumours have paired non-malignant profiles. Few studies have focused on small RNA sequencing of metastatic and localized lung tumours and involve small sample sizes. A previous study compared small RNA sequencing profiles of five LUAD with lymph node metastasis and five LUAD without lymph node metastasis, and identified 29 upregulated miRNAs and six downregulated miRNAs170. Another study compared small RNA sequencing profiles of not tumour but blood samples from 20 LUAD patients with bone metastasis and 20 without bone metastasis171. This blood-based study identified 21 upregulated and 7 downregulated miRNAs; findings were not validated in tumour specimens. While these studies and the study presented in Chapter 3 differed methodologically, two classes of miRNAs were consistently identified. All studies observed downregulation of members of the miR-34 family in metastatic cases, a tumour suppressor miRNA family with ties to NSCLC aggressiveness172. All studies also identified members of the miR-106a~363 cluster as overexpressed in the metastatic group, including miR-106a which was overexpressed in the blood of bone-metastatic patients. Interestingly, elevated blood-based detection of miR-106a has also been reported in breast cancer patients with metastatic disease, as well as in metastatic primary tumours and the metastases themselves173. In colorectal cancer, high expression of miR-106a was observed in metastatic cases and associated with shorter time to progression174. The discovery and validation of recurrently 50  deregulated miRNAs or miRNA families may lead to the development of prognostic biomarker panels or even therapeutic targets for lung and other cancers. In vitro validation of the most significantly differentially expressed miRNA, miR-106b, illustrates the utility of this approach in identifying miRNAs that functionally contribute to aggressive tumour biology. While miR-106b significantly increased the invasiveness of H2347 cells, miR-106a did not despite significant upregulation in metastatic cases. However, overexpression of either miR-106a or miR-106b resulted in increased mesenchymal marker expression that was accompanied by a change in cellular morphology characteristic of EMT. This suggests that these highly related miRNAs serve both similar and divergent functions. High expression of miR-106b has been shown to induce EMT in breast and hepatocellular cancer cell models 175,176. In colorectal cancer and lung cancer, inhibition of miR-106a resulted in reduced migration and invasion in vitro174,177. It is likely that in vitro inhibition of miR-106a or miR-106b in cell lines with high expression would have resulted in a stronger phenotype than was observed with overexpression. Similarly, it is possible that overexpression of the whole miR-106b~25 or miR-106a~363 cluster would have produced a stronger phenotype. A study overexpressing the miR-106b~25 cluster in H1299 NSCLC cells induced a more migratory and invasive cell phenotype178. A similar study was completed for all three clusters in Ewing Sarcoma, with whole cluster inhibition, rather than individual miRNA inhibition, showing the most significant decrease in clonogenic growth179. Overall, this study adds to our understanding of the biology behind metastatic LUAD by identifying a list of 37 metastasis associated miRNAs. We provide evidence that miR-106a and miR-106b contribute to the metastatic phenotype of yet another solid malignancy, by regulation of EMT and invasiveness. The utility of therapeutic inhibition or blood-based detection of miR-106a and miR-106b and their respective miRNA clusters warrants further investigation in lung and other cancer types.   51  4 Deregulation of small non-coding RNAs at the DLK1-DIO3 imprinted locus predicts lung cancer patient outcome 4.1 Introduction Approximately one quarter of the miRNAs significantly differentially expressed in metastatic LUAD mapped to the same region on chromosome 14q (Chapter 3). Interestingly, this is the genomic position of the imprinted DLK1-DIO3 locus, which encodes the largest miRNA clusters in the human genome in addition to other protein-coding and long and short non-coding genes (Figure 4.1). Genomic imprinting is the process by which the expression of an allele is silenced by methylation dependant on parental origin180.  Aberrant methylation patterns at imprinted loci resulting in expression changes of encoded transcripts are common in the pathogenesis of many diseases, including cancer181,182. In humans, anomalous imprinting at the DLK1-DIO3 locus at 14q32.1-14q.32.31 has been associated with respiratory insufficiency and reduced thorax development, amongst many other developmentally-related disorders183.    4.1.1 Metastasis-associated miRNAs are encoded at the imprinted DLK1-DIO3 locus 52   Figure 4.1 UCSC genome browser screenshot of the imprinted DLK1-DIO3 locus The DLK1-DIO3 locus maps to chromosome 14q32.1-32.31 (red box). Below, the red box is expanded to illustrate the protein coding (green), lncRNA (black), snoRNA (yellow), and miRNA (blue) genes encoded at this locus. In addition, the miRNAs identified in Chapter 3 as significantly associated with metastasis are shown53  Deregulation of small ncRNAs, mainly miRNAs, expressed from this locus has been associated with development and progression of different tumors, including lung, in both humans and mice184-186. While individual genes expressed from this locus have been associated with lung cancer patient outcome, a signature of three miRNAs (miR-370, miR-376a, miR-411) has been shown to better predict overall survival and recurrence-free survival187. This combined prediction signature suggests that the analysis of multiple genes encoded at DLK1-DIO3 may be more biologically informative than the analysis of any single gene. The role of other classes of small ncRNAs at this locus, such as piRNAs, which act primarily as transcriptional regulators, has not yet been investigated in lung cancer. piRNAs have highly-conserved functions across species, including epigenetic silencing of transposable elements and regulation of imprinting in mice188. Although originally discovered in germ cells, recent evidence of their somatic expression in non-malignant human tissues and tumours suggests alternative functions and clinical importance129,141,142,146,147,152. Importantly, emerging associations between piRNAs and patient outcome suggest piRNAs could be associated with outcome in lung cancer. Currently, the number and expression pattern of piRNAs encoded at DLK1-DIO3 remains undescribed. Although the nine metastasis-associated miRNAs identified in Chapter 3 are not part of the previously described miRNA prognostic signature (miR-370, miR-376a, miR-411), these two independent studies highlight a relationship between the DLK1-4.1.2 miRNAs expressed from the DLK1-DIO3 locus are associated with poor lung adenocarcinoma patient outcome 4.1.3 Expression of piRNAs from the DLK1-DIO3 locus remains unknown 4.1.4 Rationale for examining associations of small ncRNAs expressed from the DLK1-DIO3 locus with lung cancer patient outcome 54  DIO3 locus and lung cancer aggressiveness. In this chapter, we test the performance of the three miRNA signature in our discovery dataset (BCCA) and a validation dataset (TCGA). Next, we characterize piRNA expression at the DLK1-DIO3 locus, and determine whether their expression patterns enhance the prognostic value of miRNAs encoded at this clinically important locus. 4.2 Methods A description of discovery and external datasets and their clinical features can be found in Table 4.1.   4.2.1 Clinical Cohorts 55  Table 4.1 Clinical features of the discovery and external cohorts. Clinical Feature Discovery Cohort (BCCA)      n (%) External Cohort (TCGA)          n (%) Histological Subtype LUAD LUSC LUAD LUSC Tumour 84 34 163 220 Non-malignant 84 34 46 45 Smoking History     Current 35 (42) 11 (32) 44 (27) 57 (26) Never 25 (30) 1 (3) 17 (10) 7 (3) Former 20 (24) 22 (65) 98 (60) 147 (67) Gender     Male 24 (33) 10 (29) 86 (53) 165 (75) Female 56 (67) 24 (71) 77 (47) 47 (21) Age     Range 45–90 58–88 40–86 39–84 Median 71 70 64 68 Stage     IA 29 (35) 3 (9) 37 (23) 36 (16) IB 19 (23) 11 (32) 51 (31) 61 (28) 56  Clinical Feature Discovery Cohort (BCCA)      n (%) External Cohort (TCGA)          n (%) IIA 13 (15) 4 (12) 13 (8) 30 (14) IIB 5 (6) 7 (21) 23 (14) 40 (18) IIIA 10 (12) 2 (6) 21 (13) 31 (14) IIIB 0 (0) 1 (3) 6 (4) 10 (5) IV 2 (2) 1 (3) 12 (7) 3 (1)  See Chapter 2 Section 2.3.1. For the validation cohort, a purity cut-off of >70% was applied according to previously published studies189. This was to make expression profiles more comparable between datasets, and to reduce contaminating sequences derived from alternative cell types since piRNA and miRNA expression is highly tissue specific. Small ncRNAs were considered expressed if they had a scaled/normalized expression value ≥1 in at least 10% of both the discovery and external datasets. In the discovery cohort, small ncRNA expression from the paired tumour and non-malignant lung samples were compared by the sign-rank test, and between histological subtypes by the Mann Whitney U test. In the external cohort, all two-group comparisons were performed using the Mann Whitney U-test. Significance threshold was established at p-value<0.05.  4.2.2 Small ncRNA expression 4.2.3 Small non-coding RNA differential expression analysis 57  Univariate analysis: Cases were grouped based on piRNA expression tertiles, and survival analysis was conducted by log-rank test. For piRNAs with expression of 0 RPKM in >1/3 of samples, cases were dichotomized into those with RPKM=0 and those with RPKM>0. Cox proportional hazard model: Samples that had complete miRNA expression (RPKM), piRNA expression (RPKM), and survival data (overall survival or recurrence-free survival) were considered for Cox proportional hazard models. In addition to miRNAs previously associated with lung cancer patient outcome (miR-370, miR-376a, miR-411), Cox proportional hazard models including combinations of the seven expressed piRNAs, and combinations of the miRNAs and piRNAs were analyzed. The model with the lowest p-value was chosen for further analysis. Patient risk scores were generated per model by multiplying the expression value of a given gene by its hazard coefficient, and then summing the transformed gene expression values per sample190. Risk scores were ranked and divided into tertiles of high, intermediate, and low risk. Risk group Kaplan-Meier survival curves were then compared using the log-rank method. Significance threshold was established at p-value<0.05. Raw p-values were then adjusted using the Bonferroni method, resulting in an adjusted p-value cut-off of < 0.017. 4.3 Results Deregulation of the DLK1-DIO3 locus has been reported to be important to lung cancer biology, but the role of piRNAs derived from this locus has not yet been described. We analyzed expression data from two independent cohorts of LUAD, LUSC, and non-malignant lung samples to identify somatically-expressed piRNAs encoded at this locus. Of the 138 piRNAs encoded at the DLK1-DIO3 locus, seven 4.2.4 Survival Analysis 4.3.1 The DLK1-DIO3 locus encodes somatically expressed piRNAs 58  were expressed in LUAD, LUSC, and non-malignant lung samples in the discovery cohort (DQ596225, DQ596306, DQ596309, DQ596311, DQ596354, DQ596390, DQ596863) (Figure 4.2-4.3). Expression of all seven piRNAs was validated in the external cohort (Figure 4.4). Interestingly, these somatically expressed piRNAs are encoded exclusively in the imprinted locus (Figure 4.2). In the discovery cohort of paired tumour and non-malignant lung tissues, four of seven somatically expressed piRNAs (DQ596225, DQ596306, DQ596309, DQ596354) were significantly overexpressed in LUAD and one piRNA (DQ596309) was overexpressed in LUSC (Figure 4.3). In the external dataset, two piRNAs (DQ596225, DQ596390) were validated to be significantly differentially expressed. Furthermore, six of seven piRNAs were significantly differentially expressed between LUAD and LUSC, with higher expression observed in LUSC (Figure 4.4).  59    Figure 4.2 UCSC genome browser screenshot of the piRNAs encoded at the imprinted DLK1-DIO3 locus The DLK1-DIO3 locus maps to chromosome 14q32.1-32.31 (red box). Below, the red box is expanded to illustrate the 138 piRNAs encoded at this locus. In addition, single locus piRNAs, meaning piRNA sequences that are not duplicated throughout the genome, and the piRNAs determined to be expressed in this Chapter are shown. 60    Figure 4.3 Histograms of expressed piRNAs in the discovery dataset   61  Figure 4.3 Histograms of expressed piRNAs in the discovery dataset. Histograms display mean RPKM expression plus SEM in 84 paired non-malignant lung (NM-LUAD) and lung adenocarcinoma samples (LUAD), and 34 paired non-malignant lung (NM-LUSC) and lung squamous cell carcinoma (LUSC) samples. Significant p-values resulting from paired sign-rank analyses of gene expression are indicated as follows: *p<0.05 **p<0.01.   62   Figure 4.4 Histograms of expressed piRNAs in the validation cohort 63  Figure 4.4 Histograms of expressed piRNAs in the validation cohort. Histograms display mean RPKM expression plus SEM in 46 non-malignant lung (NM-LUAD) and 163 lung adenocarcinoma samples (LUAD), and 45 non-malignant lung (NM-LUSC) and 220 lung squamous cell carcinoma (LUSC) samples. Significant p-values resulting from Mann-Whitney U tests comparing gene expression are indicated as follows: *p<0.05 **p<0.01. ***p<0.001 ****p<0.0001.   64  Previous work has shown that a multi-miRNA classifier (miR-370, miR-376a, and miR-411) was able to predict LUAD patient outcome187. We applied this signature to our discovery dataset of LUAD and assessed the ability to predict patient overall survival (OS). Patient risk scores, indicating risk of death, were derived from a Cox proportional hazard model composed of these miRNAs. LUAD patients were divided into low, intermediate, and high risk groups and subjected to log-rank survival analysis. The miRNA signature is able to stratify LUAD patient risk groups and achieves marginal significance (low risk vs. high risk p=0.051) (Figure 4.5).However, in the external dataset the intermediate and high risk groups are not well segregated (Figure 4.5).  4.3.2 A combined miRNA+piRNA signature better predicts overall survival of lung adenocarcinoma patients 65    Figure 4.5 Overall survival of risk groups as defined by the miRNA signature in lung adenocarcinoma Kaplan-Meier curves of high (red), intermediate (green), and low (blue) risk groups as defined by the miRNA signature in (A) the discovery dataset (n=75) and (B) the external dataset (n=156). Log-rank p-values are shown.   66  Next, we investigated if piRNAs expressed from the DLK1-DIO3 locus could predict LUAD patient outcome. Just as the consideration of multiple miRNAs produced a signature, we hypothesize the consideration of multiple piRNAs expressed from this locus could result in a similar signature. Interestingly, while piRNA expression alone was unable to significantly predict OS in univariate or multivariate analysis, the incorporation of piRNA expression into the miRNA signature improved the stratification of patients into risk groups. The final survival model was selected by adding different combinations of the seven expressed piRNAs to the miRNA Cox proportional hazard model, and the model with the lowest p-value was used to calculate patient risk scores. The final survival model included the three-miRNA signature and four piRNAs encoded at this locus (DQ596306, DQ596309, DQ596390, and DQ596863), and will be referred to as the miRNA+piRNA signature.  Approximately one-third of patients from each risk group are reclassified by the miRNA+piRNA signature (Figure 4.6). In the discovery cohort, low risk LUAD patients had significantly improved outcome compared to both high (p=0.002) and intermediate (p=0.015) risk groups.  In the external cohort, high-risk LUAD patients had significantly worse outcome compared to both low (p=0.037) and intermediate (p=0.011) risk groups (Figure 4.7). In the external dataset the Kaplan-Meier curves of the low and medium risk groups were overlapping; suggesting the miRNA+piRNA signature is better at identifying high risk patients but cannot separate intermediate or low risk patients in this dataset. When the new low and intermediate risk groups are combined in this dataset, the OS prediction improves (p=0.004) (Figure 4.7). A family-wise error rate (FWER) adjustment was applied to the p-values using the stringent Bonferroni method in order to test the robustness of the signature. Even after adjustment, the majority of the miRNA+piRNA signature p-values passed the new significance threshold (Figure 4.11).  67   Figure 4.6 Risk classification of lung adenocarcinoma patients Patients are ordered by their miRNA signature-based risk classification (top) in order to illustrate the re-classification that occurs when the miRNA+piRNA signature (bottom) is applied to the (A) discovery (n=75) and (B) external (n=156) datasets.   68   Figure 4.7 Overall survival of risk groups as defined by the miRNA+piRNA signature in lung adenocarcinoma Kaplan-Meier curves of high (red), intermediate (green), low (blue), and low/intermediate (turquoise) risk groups as defined by miRNA signature in (A) the discovery dataset (n=75) and (B and C) the external dataset (n=156). Log-rank p-values are shown.   69  The previously-described miRNA signature has not been assessed in the other major subtype of NSCLC, LUSC.  In both our discovery and external datasets, LUSC patient risk groups as defined by the miRNA signature did not have significantly different OS outcomes (Figure 4.8A & 4.9A). Similarly, the LUSC patient risk groups stratified by piRNA expression did not have significantly different OS. However, as was shown in LUAD, the miRNA+piRNA signature was also able to classify LUSC patients into risk groups with distinct OS outcomes in both the discovery (Figure 4.8B) and external (Figure 4.9B-C) datasets. P-values remained significant after Bonferroni adjustment in the external dataset. All but one of the intermediate risk LUSC patients were reclassified into either high or low risk groups by the miRNA-piRNA signature in the discovery dataset (Figure 4.8C). Furthermore, the intermediate and high risk Kaplan-Meier curves overlap in the external dataset (Figure 4.9B-C), again suggesting that the miRNA+piRNA signature may identify two risk groups rather than three in some cases.   4.3.3 The miRNA+piRNA signature is able to predict overall survival of lung squamous cell carcinoma patients 70   Figure 4.8 The miRNA+piRNA signature predicts overall survival in lung squamous cell carcinoma patients (discovery dataset, n=27) Risk scores calculated based on (A) miRNA signatures and (B) miRNA+piRNA signatures. Patients were assigned to high (red), intermediate (green), and low (blue) risk groups and Kaplan-Meier survival curves were compared. (C) Risk group classifications were compared based on the miRNA-only (top) and miRNA+piRNA (bottom) signatures. Risk group colors are the same as in the above panels.  71   Figure 4.9 The miRNA+piRNA signature predicts overall survival in lung squamous cell carcinoma patients (external dataset, n=205) Kaplan-Meier curves of high (red), intermediate (green), and low (blue) risk groups as defined by (A) the miRNA signature and (B) the miRNA+piRNA signature are shown. Panel C combines the low and intermediate risk groups (turquoise) and compares survival to the high risk group. Log-rank p-values of select survival comparisons are shown. (D) Patients are ordered by their miRNA signature-based risk classification (top) in order to illustrate the reclassification that occurs when the miRNA+piRNA signature (bottom) is applied to the dataset.  72  Sufficient RFS data was not available for the discovery dataset. In the external dataset, we compared RFS data of risk groups defined by the miRNA signature, the piRNA signature, and the miRNA+piRNA signature. Only the miRNA+piRNA signature was able to stratify two patient risk groups with statistically different outcomes. Similarly to OS, RFS classifications by the miRNA+piRNA signature were statistically significant in both LUAD (p=0.018) and LUSC histological subtypes (p=0.037) (Figure 4.10), but did not pass Bonferroni adjustment (Figure 4.12).   4.3.4 The miRNA+piRNA signature identifies patients at risk of recurrence-free survival 73   Figure 4.10 Performance of small ncRNA-based signatures predicting recurrence-free survival in non-small cell lung cancer Risk groups as defined in the external dataset of (A) lung adenocarcinoma (LUAD, n=107) and (B) lung squamous cell carcinoma (LUSC, n=149). Kaplan-Meier curves of high risk groups (red) compared to the combined low and intermediate risk groups (turquoise) as defined by the miRNA+piRNA signature are shown. (C) LUAD patients and (D) LUSC patients are ordered by their miRNA signature-based risk classification (top) in order to illustrate the re-classification that occurs when the miRNA+piRNA signature (bottom) is applied to the dataset. Intermediate and low risk patients are represented by green and blue bars, respectively.   74   Figure 4.11 Log-rank p-value summary for overall survival predictions   75  Figure 4.11 Log-rank p-value summary for overall survival predictions Bar lengths represent the -log10(p-value) of each signature for LUAD (top) and LUSC (bottom) patients from (A) the discovery cohort and (B) the external cohort. Comparison across different risk groups are as follows: low vs. high risk (grey bars), intermediate vs. high risk (white bars), low + intermediate risk vs. high risk (black bars). Significance thresholds are established at p-value=0.05 (red dashed line indicates -log10 0.05), and at p-value=0.017 (Bonferroni-adjusted p-value) (green dashed line indicates –log10 0.017).   76   Figure 4.12 Log-rank p-value summary for recurrence-free survival predictions in the external dataset Bar lengths represent the -log10(p-value) of each signature for LUAD (top) and LUSC (bottom) patients from the discovery cohort. Comparison across different risk groups are as follows: low vs. high risk (grey bars), intermediate vs. high risk (white bars) low/intermediate risk vs. high risk (black bars). Significance thresholds are established at p-value=0.05 (red dashed line indicates -log10 0.05), and at p-value=0.017 (Bonferroni-adjusted p-value) (green dashed line indicates –log10 0.017).  77  4.4 Discussion This study further demonstrates the biological relevance of our analysis strategy outlined in Chapter 3. Furthermore, we establish that piRNAs are expressed at the DLK1-DIO3 locus, and suggest their relevance to lung cancer prognostics. We demonstrate the biological importance of multiple small ncRNA species through associations with NSCLC patient outcome. Incorporating both piRNA and miRNA expression in the classification of LUAD patients into risk groups improves classification compared to either small RNA species alone. In addition, stratification considering piRNA expression broadens the applicability of the signature to LUSC, which was not possible with miRNA expression. These findings highlight the complexity of the DLK1-DIO3 locus and underscore its clinical relevance to both major histological subtypes of NSCLC. The enhanced prediction of patient outcome may be linked to the additional level of regulation of gene expression provided by piRNAs, as well as the specific features of the seven piRNAs expressed from the DLK1-DIO3 locus. In order to regulate repetitive elements, single piRNAs are often encoded at multiple loci throughout the genome. However, piRNAs encoded at one locus are thought to function by regulating DNA methylation in target regions thereby acting as regulators of gene expression138,139. We identify seven somatically expressed piRNAs solely encoded at this locus, suggesting these piRNAs may function to regulate methylation of target genes. Malignancy-associated methylation changes at this locus were recently described in lung cancer 191; therefore, it is possible these piRNAs are involved in the deregulation of methylation patterns of this locus during lung tumourigenesis. Further studies will be required to determine whether deregulation of methylation at the DLK1-DIO3 locus is mediated by piRNAs or by alternative mechanisms. Although the function of somatically-expressed piRNAs has not yet been fully established, mounting evidence indicates they may serve as prognostic markers in a variety of tumor types, including  gastric (RFS), colon (RFS), breast (lymph node 78  positivity), kidney (OS) and head and neck (OS) cancer141,142,145,147,152,153,190. Moreover, piRNAs, as other small ncRNAs, are stable in biofluids and formalin-fixed paraffin-embedded material, highlighting the potential of piRNA-based prognostic markers across a variety of tumour types. To derive such a prognostic signature would require an alternative study design and much larger cohorts than the study presented here. However, this was not the goal of this study.  The goal of this study was to examine whether miRNAs and piRNAs expressed from the DLK1-DIO3 locus could hold prognostic value in lung cancer, which we have demonstrated in our discovery cohort and validated in the TCGA. In summary, our results provide further evidence of the involvement of imprinting-regulated small ncRNAs into lung cancer biology, and underscore the relevance of the DLK1-DIO3 locus to aggressive lung cancer biology.   79  5 ELF3 is a novel oncogene in lung adenocarcinoma 5.1 Introduction Large-scale sequencing efforts have identified driver oncogenes for approximately half of LUAD, while a substantial fraction of LUAD remains “driverless”. Even when the driver oncogene is known, targeted therapies are only available for a subset of mutations. Alternative approaches aimed at novel oncogene discovery must be explored in order to identify new therapeutic targets. By moving beyond the mutational landscape and integrating multiple ‘omics levels per sample (e.g. DNA, RNA), genes that are frequently disrupted by mechanisms other than mutation emerge. This approach allows for the identification of genes that may have been overlooked by previous sequencing studies. Furthermore, by profiling DNA and RNA ‘omics levels, this approach allows for the identification of DNA-level alterations that directly influence gene expression. LUAD is characterized by DNA level alterations other than mutation, such as DNA copy number gain and loss. Regions of DNA gain typically contain oncogenes, while regions of loss harbour tumour suppressor genes. Chromosome arm 1q is frequently gained in LUAD and is associated with aggressive clinical features159,160, although no single gene has been identified as the target of this DNA alteration.  One of the candidate oncogenes with potential roles in LUAD development that lies within the altered region on chromosome 1q is a transcription factor whose overexpression has been reported in several clinical samples of lung cancer and lung cancer cell lines192,193. This gene, E74-like factor 3 (ELF3), is characterized by epithelial-specific expression making it a unique member of the ETS family of transcription factors. 5.1.1 Rationale for integrative multi-omics analysis 5.1.2 ELF3 is a putative lung cancer oncogene 80  ELF3 falls within the signaling network of ERBB2, an established LUAD oncogene, and was recently tied to NOTCH and TGFβ signaling161,162.  Based on these collective observations and its genomic location on chromosome 1q, we hypothesize that ELF3 is a novel LUAD oncogene. To test this hypothesis, we interrogate clinical cohorts of LUAD and LUSC to determine whether ELF3 overexpression is specific to LUAD and associated with poor patient outcome. Through integrative multi-omic analysis, we examine the extent of ELF3 locus deregulation in LUAD and assess the impact of DNA level alterations on gene expression. Frequent DNA-level alterations at the ELF3 locus that correlate with expression would represent evidence that high ELF3 expression is selected-for in LUAD and is likely biologically relevant. To test biological relevance, we use isogenic cell models to determine whether manipulation of ELF3 expression affects oncogenic phenotypes and in vivo growth, which would prove its functional relevance to LUAD and indicate therapeutic potential. 5.2 Methods A profiling summary of the discovery and external datasets can be found in Table 5.1. For platform details, see Chapter 2 Table 2.1 and Section 2.3. In addition, public expression data from Duke University (GSE3141) and Samsung Medical Centre (GSE8894) were downloaded from GEO for comparative ELF3 mRNA expression analysis between LUAD and LUSC (Table 5.2). Paired tumour and non-malignant tissues were obtained as described in Section 2.1-2.2. Following library construction, total RNA was bar-coded and subjected to multiplexed RNA sequencing on the Illumina HiSeq 2000 sequencing platform using a plate-based protocol developed at the BCGSC. Reads were trimmed for quality and 5.2.1 Patient cohorts 5.2.2 RNA profiling of LUSC samples in the BCCA cohort 81  aligned to the human genome (GRCh38/hg19) using STAR. Quantification was performed using the BCGSC protocol163 and expression was normalized to Fragments per Kilobase of transcript per Million mapped reads (FPKM). Table 5.1 Summary of multi-omic profiling for the BCCA and TCGA datasets. Datatype LUAD NM-LUAD LUSC NM-LUSC BCCA Expression 83 83 NA NA DNA copy number 83 83 NA NA DNA methylation 77 77 NA NA All dimensions 77 77 NA NA TCGA Expression 513 58 504 51 DNA copy number 513 NA NA NA DNA methylation 452 21 NA NA Mutation 513 NA NA NA All dimensions 420 NA NA NA >60% purity 252 NA NA NA    82  Table 5.2 Summary of clinical cohorts of lung adenocarcinoma and lung squamous cell carcinoma with ELF3 mRNA expression Cohort # of samples TUMOUR NON-MALIGNANT Gene Expression Platform # LUAD # LUSC # LUAD # LUSC BC Cancer Agency (LUAD) 166 83 0 83 0 Affymetrix GeneChip Human Genome U133 Plus 2.0 Array BC Cancer Agency (LUSC) 58 0 29 0 29 Illumina HiSeq2000 RNA sequencing The Cancer Genome Atlas 1126 513 504 58 51 Illumina HiSeq2000 RNA sequencing GEO Duke University - GSE3141 111 58 53 0 0 Affymetrix GeneChip Human Genome U133 Plus 2.0 Array GEO Samsung Medical Center - GSE8894 138 63 75 0 0 Affymetrix GeneChip Human Genome U133 Plus 2.0 Array    83  Data for Kaplan Meier survival analysis were obtained from http://kmplot.com/analysis/index.php?p=service&cancer=lung194. Survival curves were compared by log-rank analysis. Level 3 whole genome SNP 6.0 copy number segmentation files (GRCg37/hg19) were downloaded from the TCGA Data Portal (https://tcga-data.nci.nih.gov). Specifically, 513 LUAD (tissue codes 01 and 02) *.nocnv_hg19.seg.txt files, from which a fixed set of germline-variable probes have been removed, were assembled into a master segmentation file for GISTIC 2.0 analysis195. Amplification and deletion thresholds were increased to 0.3; default settings were used for all other parameters. BCCA data: EGFR and KRAS mutations were identified by PCR amplification (Applied Biosystems GeneAmp PCR System 9700) and product sequencing (Applied Biosystems BigDye Terminator v3.1 cycle sequencing kit and capillary instrumentation) of tumour DNA. Exons 19 and 21 and exon 2 were screened in EGFR and KRAS, respectively. TCGA data: Level 2 Mutation Annotation Format (MAF) files for 513 LUAD were downloaded from the TCGA Data Portal. Only previously established driver mutations were considered196. Silent mutations were removed.   5.2.3 Survival analysis 5.2.4 GISTIC 2.0 copy number analysis 5.2.5 Oncogene mutation data 84  RNA was extracted from whole tumour bearing or non-tumour bearing lungs. RNA was profiled on the Affymetrix GeneChip Mouse Genome 430 2.0 expression microarray. Expression data was RMA normalized in bioconductor. Cell line mutational data was obtained from Broad’s Cancer Cell Line Encyclopedia (http://www.broadinstitute.org/ccle). Stable knockdown of ELF3 was achieved by lentiviral transfection of vectors encoding shRNA inserts directed against ELF3 mRNA as well as a puromycin resistance selectable marker (Sigma-Aldrich, St. Louis, MO, USA). Virus was prepared for ELF3-shRNAs and a control with no shRNA insert. LUAD cell lines were transfected, and after 24 hours media was replaced with puromycin-containing media. Following complete puromycin-induced death of control cells after 3-5 days, transfected cells were cultured in puromycin-media for an additional 7 days. Knockdown efficiencies were quantified by qRT-PCR (TaqMan - Applied Biosystems, Carlsbad, CA, USA) using 18S as an endogenous control: Hs00963881_m1 (ELF3) and Hs99999901_s1 (18S), and verified at the protein level by immunoblot (described below). Five shRNA clones were tested and the clone with the best knock-down of ELF3 transcript and protein levels was selected for phenotypic assays. Stable overexpression of ELF3 in Human Bronchial Epithelial Cells (HBECs) was achieved by lentiviral delivery of a vector containing the full ORF with a blasticidin resistance selectable marker or an empty vector control (Invitrogen, Carlsbad, CA, USA).  Cell lines used for in vivo experiments were subcutaneously injected (A549: 2.5 x 106cells per site; HCC827: 5 x 106 cells per site) into the left and right flanks of 6-8 week old NOD-SCID mice. Tumour volume was measured several times weekly until 5.2.6 Expression data from mouse models of tumourigenesis 5.2.7 In vitro and in vivo experiments 85  a total tumour burden of 1500mm3 was achieved or tumours became ulcerated, at which point mice were euthanized. The animal protocol  was  approved  by  the  Animal  Care  Committee  of  the  University  of  British Columbia (Vancouver, British Columbia, Canada). DNA from FFPE xenografts was extracted using the Biostic FFPE Tissue DNA Isolation Kit (MO BIO Laboratories, Inc., Carlsbad, CA, USA) and cleaned up using the MinElute Reaction Cleanup Kit (Qiagen, Hilden, Germany). DNA from cell line input was extracted using the DNeasy Blood & Tissue Kit (Qiagen, Hilden, Germany). A summary of cell line constructs and oncogene information can be found in Table 5.3.   86  Table 5.3 Cell lines used for in vitro and in vivo experiments.  The cell lines and their putative driver oncogene are listed. Also shown are the control vectors and vectors for the shRNA-mediated knock-down of ELF3 and stable overexpression of ELF3. Cell line Oncogene Control Vector ELF3 vector Optimal shRNA clone Company H1395 BRAF pLKO.1-puro pLKO.1-puro/ELF3-shRNA TRCN0000013863 Sigma-Aldrich HCC827 EGFR pLKO.1-puro pLKO.1-puro/ELF3-shRNA TRCN0000013863 Sigma-Aldrich H1993 MET pLKO.1-puro pLKO.1-puro/ELF3-shRNA TRCN0000013867 Sigma-Aldrich H1819 ERBB2 pLKO.1-puro pLKO.1-puro/ELF3-shRNA TRCN0000013867 Sigma-Aldrich A549 KRAS pLKO.1-puro pLKO.1-puro/ELF3-shRNA TRCN0000013863 Sigma-Aldrich  HBEC NA pLenti6.3/lacZ pLenti6.3/ELF3-ORF NA Invitrogen     87  For immunofluorescence analysis of 2-D monolayer cultures, cells were fixed in paraformaldehyde for 10 minutes. PBS containing 1% BSA was added to cells for 30 minutes to block non-specific interactions. Subsequently, anti-ELF3 (Sigma-Aldrich, Oakville, ON, Canada) was added for 2 hours at room temperature in goat serum/gelatin blocking buffer. Secondary antibodies were added for 1 hour also in blocking buffer at room temperature. Staining of F-actin and DNA were performed by adding, respectively, rhodamine-conjugated phalloidin and Hoechst (Invitrogen, Burlington, ON, Canada). Coverslips were mounted in Vectashield mounting medium (Vector Laboratories, Burlington, ON, Canada). Cell images were acquired with a Zeiss Colibri fluorescence microscope, AxioCam MRm camera and AxioVision Rel. 4.8 software (Carl Zeiss Canada Ltd., Toronto, ON, Canada).  Single cell suspensions were prepared in growth media supplemented with 20% fetal bovine serum (Invitrogen, Carlsbad, CA, USA), and 0.3% low-melting point agarose (Invitrogen, Carlsbad, CA, USA). One ml of cell suspension was plated onto an equal volume of supplemented media with a 0.5% low-melting point agarose concentration. Each cell line was seeded in triplicate in 12-well plates and cultured for 14-21 days at 37˚C. Experiments were repeated 3 times. The number of colonies formed by shELF3 cells was compared to controls using a paired Student’s t-test. Cell proliferation was quantified using the BD PharmingenTM Apoptosis, DNA Damage and Cell Proliferation Kit (BD Bioscience, Mississauga, ON, Canada). Cells were incubated with BrdU for 8-24h, and processed according to the manufacturer's instructions. Briefly, cells were fixed, permeabilized, and treated with DNase I before staining with PerCP-CyTM5.5 Mouse Anti-BrdU antibody and DAPI (1μg/ml). 5.2.8 Immunofluorescence microscopy 5.2.9 Soft agar colony formation 5.2.10 Cell proliferation 88  Cells were analyzed using the BD FACS CantoTM II cell analyzer (BD Bioscience, Mississauga, ON, Canada). The percent of BrdU+ singlet cells was compared between ELF3 altered cell lines and their respective isogenic controls using a paired two-tailed Student's t-test. Cell apoptosis was quantified using the BD PharmingenTM Annexin V Apoptosis Detection Kit I according to the manufacturer’s instructions (BD Bioscience, Mississauga, ON, Canada). Cells were grown in complete or serum free media for 72 hours prior to cell processing, staining, and analysis by flow cytometry on the BD FACS CantoTM II cell analyzer (BD Bioscience, Mississauga, ON, Canada). Live cells that were adherent at time of collection were used as a gating control. 5.3 Results The region harboring ELF3, chromosome 1q32.1, is gained in lung cancer with the most frequent occurrences in LUAD. However, DNA level events do not necessarily correlate with expression level alterations. To assess the prevalence of ELF3 overexpression across lung cancer subtypes, we assessed ELF3 expression in the two major NSCLC subtypes, LUAD and LUSC. In our paired cohort of 83 LUAD and matched non-malignant lung tissues, ELF3 was significantly overexpressed at least 2-fold in 73% of LUAD tumours (sign-rank p=1.64E-21). TCGA LUAD data validated these results; ELF3 was significantly overexpressed (n=513 LUAD, n=58 NM-LUAD; Mann-Whitney U test p=1.54E-07) and where paired analysis was available (n=57 pairs) ≥2-fold overexpression was observed in 40% of tumours (Figure 5.1). Consistent with ELF3 overexpression in tumour cells, TCGA samples with higher tumour cell purity (n=252)189 show higher ELF3 expression (Figure 5.2). 5.2.11 Cell apoptosis 5.3.1 ELF3 is frequently overexpressed in lung adenocarcinoma and associated with patient outcome 89  Interestingly, we find ELF3 expression levels were not significantly altered in LUSC (Mann-Whitney U test p=7.82E-01) (Figure 5.3A-B). When comparing tumour-specific ELF3 expression between LUAD and LUSC, we find ELF3 expression to be significantly higher in LUAD in the TCGA (Mann-Whitney U test p<1.0E-04) (Figure 5.3C). These findings were validated in two additional cohorts complete with LUAD and LUSC expression downloaded from GEO197,198 (Figure 5.3D-E).   90    Figure 5.1 ELF3 expression in lung adenocarcinoma Box and Whiskers plots comparing log2(ELF3 expression) between non-malignant lung (blue) and lung adenocarcinoma (red) samples in (A) the paired BCCA dataset (n=166), (B) unpaired TCGA dataset (non-malignant n=58, LUAD n=513), and paired samples from TCGA (n=114). Whiskers represent minimum and maximum values, with the mean represented by a horizontal bar in the center of the boxes. 91   Figure 5.2 ELF3 expression is correlated with tumour purity in TCGA (A) Scatter plot of log2(ELF3 expression) versus TCGA LUAD sample purity estimates189. Line of best fit and Spearman ɾ and p-values are shown. (B) Dot plot illustrating the mean log2(ELF3 expression) in non-malignant (NM) lung (blue, n=58) and LUAD samples grouped by tumour purity estimates (red).   92   Figure 5.3 ELF3 expression in lung squamous cell carcinoma   93  Figure 5.3 ELF3 expression in lung squamous cell carcinoma Box and whiskers plots comparing log2(ELF3 expression) between (A) 29 paired non-malignant lung (blue) and lung squamous cell carcinoma samples (green) in the BCCA dataset, and (B) unpaired non-malignant lung (blue, n=51) and lung squamous cell carcinoma samples (green, n=504) in the TCGA dataset. Log2(ELF3 expression) is compared between lung adenocarcinoma and lung squamous cell carcinoma in (C) the TCGA dataset, (D) the Duke University dataset, and (E) the Samsung Medical Centre dataset. Mann Whitney U test p-values are shown. Whiskers represent minimum and maximum values, with the mean represented by a horizontal bar in the center of the boxes.     94  The subtype-specific overexpression of ELF3 in LUAD translated to clinical outcomes. High ELF3 expression was only significantly correlated with poor overall survival in LUAD (log-rank p<1.0E-04), but not in LUSC (p>0.05). This observation also held true when the analysis was restricted to Stage I LUAD patients (log-rank p<1.0E-04) (Figure 5.4). These data demonstrate ELF3 is specifically overexpressed in LUAD at high frequency, and that high expression is associated with poor LUAD patient outcome prompting further study in this subtype.  Figure 5.4 High ELF3 expression is associated with poor overall survival in lung adenocarcinoma   95  Figure 5.4 High ELF3 expression is associated with poor overall survival in lung adenocarcinoma.  Kaplan-Meier plots comparing high (red) and low (blue) ELF3 expression tertiles derived from the meta survival data tool, K-M Plotter. Overall survival for (A) lung adenocarcinoma (all stages), (B) Stage I lung adenocarcinoma, and (C) lung squamous cell carcinoma (all stages) are shown.  We next searched for evidence of DNA-level alterations that could explain this high expression. We examined the frequency of DNA copy number gain at the ELF3 locus in our LUAD dataset and that of the TCGA. The ELF3 locus was gained in 44% and 49% of our dataset and the TCGA dataset, respectively, agreeing with previous reports of frequent gain at chromosome 1q. While DNA gain can result in increased gene expression, focal amplification of a smaller genomic region provides strong evidence of genetic selection. We analyzed over 500 LUAD copy number profiles using the GISTIC 2.0 algorithm195 and identify an amplicon at 1q32.1 that harbours ELF3. This region was amplified at significance comparable to both ERBB2 and MET focally amplified oncogenes196 (Figure 5.5).   5.3.2 The ELF3 locus at chromosome 1q32.1 is frequently altered by genetic and epigenetic mechanisms 96   Figure 5.5 GISTIC 2.0 focal amplification plot G-scores and q-values with respect to significantly focally amplified regions over the entire genome. G-scores represent the amplitude and frequency of each aberration across samples. FDR q-values are calculated per aberration and a q-value < 0.25 is considered significant (green line). Chromosomes are numbered and presented at the bottom of the figure as alternating white and black boxes. Centromeric positions are indicated by vertical dashed lines.    97  Gene deregulation by more than one mechanism provides further evidence to support its biological importance. We therefore examined the methylation status of the ELF3 promoter (represented by probes cg23901896 and cg12970084 on the Illumina platform) for evidence of selection at the epigenetic level. Changes in methylation levels were determined for individual patients in the BCCA cohort by subtracting the β-values observed in non-malignant lung from the paired tumour β-values. ELF3 probe hypomethylation (∆β≤-0.15) was observed at a high frequency, in 35% (cg12970084) and 38% (cg23901896) of cases. A similar analysis was conducted in the TCGA dataset whereby LUAD β-values were compared to a pooled average non-malignant lung β-value. In this analysis, hypomethylation was observed in 48% (cg12970084) and 21% (cg23901896) of LUAD, and remarkably in 71% (cg12970084) and 30% (cg23901896) of LUAD with higher tumour purity (>60%). Hypermethylation ((∆β≥0.15) was rarely observed (<1%). Interestingly, cases with DNA copy number gain often display simultaneous hypomethylation at the ELF3 locus (BCCA: 36%, TCGA: 36%, TCGA>60% purity: 54%). If either mechanism is considered, ELF3 is altered at the DNA-level in 79% of our BCCA cohort and 64% and 83% of the TCGA cohort, before and after filtering for tumour purity (Figure 5.6).  Mutation was infrequent (1.7%).    98   Figure 5.6 Frequency of DNA-level alterations at the ELF3 locus across datasets Histogram summarizing the frequencies of ELF3 copy number (CN) gain, hypomethylation, CN gain or hypomethylation, and CN gain and hypomethylation in the BCCA dataset (black bar), TCGA dataset (dark grey hatched bar), and TCGA dataset filtered for ≥60% tumour purity (white hatched bar). A tumour was considered hypomethylated if cg12970084 and/or cg23901896 was hypomethylated. The multi-‘omic molecular profiling of LUAD and non-malignant samples allowed us to determine whether DNA alterations corresponded to expression changes. Using our paired BCCA dataset, data from each tumour was compared to that of its paired non-malignant lung sample, resulting in DNA and RNA-level alterations normalized per case. These DNA and RNA-level alterations were then correlated within each tumour. For paired methylation analysis, the most significantly negatively correlated probe, cg12970084, was chosen (Spearman’s ɾ=-0.5548, p<1.0E-06) (Figure 5.7A-B). The comparison of cg12970084 methylation β-values and ELF3 expression separates LUAD from non-malignant samples in nearly dichotomous fashion (Figure 5.7B). ELF3 overexpression (>2-fold) was frequently accompanied 5.3.3 DNA alterations are correlated with expression changes 99  by DNA copy number gain and/or cg12970084 promoter hypomethylation within the same tumour (40 cases) (Figure 5.7C). Furthermore, we observed a trend towards higher ELF3 expression fold-change in LUAD with DNA-level alterations (Student’s t-test p=0.06) (Figure 5.7D).  Figure 5.7 Multi’omic analysis of ELF3 in the BCCA dataset    100  Figure 5.7 Multi’omic analysis of ELF3 in the BCCA dataset. Scatter plot of log2(ELF3 expression) and ELF3 promoter methylation β-values for (A) cg12970084 and (B) cg23901896 in the BCCA dataset of paired non-malignant (blue) and LUAD (red) samples (n=77 pairs). Spearman correlation ɾ and p-values are shown. (C) Venn diagram illustrating cases of ELF3 copy number gain, hypomethylation of cg12970084, and ELF3 overexpression (fold change ≥2) (n=77). The total number of cases with ELF3 copy number gain, hypomethylation, or overexpression are indicated as percentages, regardless of overlap. (D) Box and whiskers plot of ELF3 expression fold change in cases with no DNA-level alteration (no event, blue) and cases with DNA-level alteration (red).   101  In the TCGA dataset, tumour data was compared to a pooled average normal. Once again, expression was most significantly negatively correlated with methylation of cg12970084 (Spearman’s ɾ=-0.4153, p<1.0E-06), which was hypomethylated in approximately 80% of LUAD in both datasets (Figure 5.8A-B). The separation of LUAD from non-malignant samples using methylation and expression data was not as dramatic as in the BCCA dataset (Figure 5.8B). ELF3 expression was positively correlated with gene dosage (Spearman’s ɾ=0.2869; p<1.3E-04) (Figure 5.8C), and when grouped by copy number status, cases with gain and amplification had significantly higher expression than those with neutral copy number (Student’s t-testgain p=4.0E-04, t-testamplification p=3.7E-03) (Figure 5.8D). The majority of cases with gain or amplification of the ELF3 locus simultaneously harbored hypomethylated promoter regions. When either mechanism is considered, ELF3 expression was significantly higher in cases with DNA-level alteration at the ELF3 locus (t-test p=5.17E-07). Analysis of TCGA samples with higher tumour cell purity produced similar results and improved separation of methylation β-values between LUAD and non-malignant lung (Figure 5.9). Evidence from both datasets indicates high ELF3 expression is a genetically and epigenetically selected event in LUAD.   102   Figure 5.8 Multi’omic analysis of ELF3 in the TCGA dataset   103  Figure 5.8 Multi’omic analysis of ELF3 in the TCGA dataset. Scatter plot of log2(ELF3 expression) and ELF3 promoter methylation β-values for (A) cg12970084 (blue) and (B) cg23901896 in the TCGA dataset of non-malignant (blue, n=21) and LUAD (red, n=452) samples. Spearman correlation ɾ and p-values are shown. (C) Scatter plot of log2(ELF3 expression) and ELF3 copy number, with copy number thresholds for loss (-0.3), gain (0.3), and amplification (0.8) indicated by dashed vertical lines. (D) Box and whiskers plot of ELF3 expression by DNA category (loss, neutral, gain, amplification (amp)). Cases are colour coded by cg12970084 methylation status (hypermethylation=green, hypomethylation=red, no change=black). Significant Student’s t-test p-values are shown. (E) Box and whiskers plot of ELF3 expression in cases with no DNA-level alteration (no event, black) and cases with DNA-level alteration (red). Student’s t-test p-values are shown.   104   Figure 5.9 Multi’omic analysis of ELF3 in the TCGA dataset after filtering for >60% tumour purity  105  Figure 5.9 Multi’omic analysis of ELF3 in the TCGA dataset after filtering for >60% purity. (A) Scatter plot of log2(ELF3 expression) and cg12970084 β-values with non-malignant samples coloured in blue and LUAD samples coloured in red (n=273). Spearman correlation ɾ and p-values are shown. (B) Scatter plot of log2(ELF3 expression) and ELF3 copy number. Spearman correlation ɾ and p-values are shown. (C) Box and whiskers plot of ELF3 expression by DNA category (loss, neutral, gain, amplification (amp)). Cases are colour coded by cg12970084 methylation status (hypermethylation=green, hypomethylation=red, no change=black). Significant Student’s t-test p-value is shown. (D) Box and whiskers plot of ELF3 expression in cases with no DNA-level alteration (no event, black) and cases with DNA-level alteration (red). Student’s t-test p-value is shown. Our data indicate ELF3 upregulation is a frequent event in LUAD; however, its associations with molecular subtypes of LUAD are unknown. ELF3 expression was compared in LUAD with and without mutations in the two most frequently occurring oncogenes, KRAS and EGFR. We found no significant difference in ELF3 expression between tumours with or without these mutations in either dataset (Figure 5.10), implying ELF3 may be important to the biology of molecularly distinct LUAD.    5.3.4 ELF3 overexpression is observed irrespective of driver oncogene 106   Figure 5.10 ELF3 expression in lung adenocarcinomas with and without mutation in EGFR and KRAS Box and whiskers plots of log2(ELF3 expression) in LUAD with mutation in EGFR (green) or KRAS (red) and those without either mutation (grey) in (A) the BCCA dataset (n=83) and (B) the TCGA dataset (n=484). Kruskal-Wallis p-values are shown. Whiskers represent minimum and maximum values, with the mean represented by a horizontal bar in the center of the boxes.    107  To further test this theory, we assessed ELF3 expression in four doxycyline-inducible transgenic mouse models of LUAD tumourigenesis (EGFR∆L747-S752-CCSP, EGFRL858R-CCSP, KRASG12D-CCSP, and MYC-CCSP199-201. These models drive expression of the oncogenic transgenes through a Club cell specific promoter (CCSP). In all four models, we observe a dramatic increase in ELF3 expression compared to the lung tissue of doxycycline-treated mice lacking transgenes (EGFR∆L747-S752-CCSP p=9.56E-05, fold change (FC)=6.03; EGFRL858R-CCSP p=4.86E-05, FC=6.06; KRASG12D-CCSP p=5.66E-08, FC=8.03; MYC-CCSP p=1.08E-04, FC=5.83) (Figure 5.11). These data indicate high ELF3 expression may be beneficial to tumourigenesis in the context of these driver alterations.   108   Figure 5.11 ELF3 expression in mouse models of tumourigenesis Box and whiskers plots of log2(ELF3 expression) in lung tissue of doxycycline-treated mice lacking transgenes (white) compared to tumour tissue driven by LUAD oncogenes: EGFR∆L747-S75 (green), EGFRL858R (green), KRASG12D (red), MYC (blue). ****paired Student’s t-test p<1.0E-04.    109  Furthermore, we examined protein-level ELF3 expression in a panel of KRAS mutant, EGFR mutant, and KRAS/EGFR wild type LUAD cell lines. First, we confirm ELF3 expression to be predominantly nuclear, agreeing with its role as a transcription factor (Figure 5.12). Second, we observe variable expression in LUAD cell lines regardless of KRAS and EGFR status (Figure 5.13). These data demonstrate that high ELF3 expression is observed in LUAD with and without KRAS, EGFR, or MYC oncogenic drivers, suggesting functional relevance across molecular subtypes of LUAD.  Figure 5.12 ELF3 expression is predominantly nuclear  (A) Immunohistochemistry stain of ELF3 (green), cytoplasmic marker F-actin (red), nuclear stain Hoechst (blue), and merged channels for a LUAD cell line. (B) Immunoblot of cytoplasmic and nuclear LUAD cell line lysates with primary antibodies against ELF3, nuclear marker HDAC1, and cytoplasmic marker vinculin.  110   Figure 5.13 Immunoblot of ELF3 expression across of panel of molecularly distinct LUAD cell lines 111  Figure 5.13 Immunoblot of ELF3 expression across of panel of molecularly distinct LUAD cell lines.  Immunoblot of ELF3 expression in a panel of KRAS mutant, EGFR mutant and EML4-ALK positive LUAD cell lines, as well as in LUAD cell lines lacking these alterations (wild type). Histone H3 was used as a loading control. To explore the phenotypic impact of ELF3 in molecularly distinct LUAD, cell lines harboring different driver oncogenes with inherently high ELF3 expression were selected for in vitro assays (H1395 (BRAF), HCC827 (EGFR), H1993 (MET), H1819 (ERBB2), A549 (KRAS)) (Table 5.2). Stable knock-down of ELF3 was achieved by lentiviral delivery of a shRNA vector (shELF3) and compared to empty vector controls (Figure 5.14). In all evaluable lines, ELF3 knock-down resulted in reduced number of colonies formed in soft agar (HCC827 does not form colonies) (Figure 5.15). Proliferation of ELF3 knock-down lines was also reduced, as quantified by BrdU incorporation assay (Figure 5.15). Only ERBB2 amplified H1819 cells displayed a significant increase in proliferation following ELF3 knock-down.   5.3.5 ELF3 regulates cancer phenotypes of LUAD cell lines 112   Figure 5.14 Experimental validation of ELF3 shRNA-mediated knock-down across biological replicates (A) Representative immunoblot of ELF3 and loading control Histone H3 in our panel of LUAD cell lines. Driver mutation is indicated per line. C=control, KD=ELF3 knock-down. (B) qPCR results indicating the percent relative ELF3 expression of control and knock-down cell line. The results of at least three biologically independent knock-downs are shown.   113   Figure 5.15 In vitro results from shELF3 cell lines  (A) Representative image of soft agar colony formation experiment comparing H1395 control and shELF3 cells. (B) Histogram summarizing the results from three biological triplicate soft agar colony formation experiments. Number of colonies formed after 21 days by H1395, H1993, H1819, and A549 cell lines are shown for control (black) and shELF3 (red) cells. (C) Histogram summarizing the results from three biological triplicate BrdU proliferation experiments. Percent BrdU positive cells from H1395, HCC827, H1993, H1819, and A549 cell lines are shown for control (black) and shELF3 (red) cells.  Paired two-tailed t-test: *p<5.0E-02 ***p<1.0E-03 ****p<1.0E-04.  114  Overexpression of oncogenes in non-malignant cell lines can lead to increases in cancer phenotypes such as proliferation, and, if cell transformation occurs, anchorage-independent growth. To examine the phenotypic impact of ELF3 overexpression in a non-malignant setting, ELF3 was stably overexpressed in a human bronchial epithelial cell (HBEC) line immortalized by hTERT and CDK4 expression (Figure 5.16A-B). Baseline ELF3 expression in HBECs is almost undetectable by qPCR (Ct=cycle 34). Similarly to what was observed in the LUAD cell lines, high expression of ELF3 in HBECs significantly increased proliferation rates of this non-transformed cell line (Figure 5.16C). HBEC-ELF3 cells did not form colonies in soft agar; therefore, ELF3 overexpression alone is insufficient to cause cell transformation. These assays provide functional evidence that high ELF3 expression is important to the proliferation and anchorage-independent growth of various LUAD cell lines harbouring different oncogenes. Furthermore, elevated ELF3 expression can increase growth rates of non-transformed airway cells suggesting selection of ELF3 early in tumourigenesis could increase the fitness of precancerous cells.   115   Figure 5.16 In vitro results from ELF3 overexpression in HBECs (A) Immunoblot of ELF3 and loading control Histone H3 in control and ELF3 overexpressing HBEC cells. C=control, OE=ELF3 overexpression. (B) qPCR results indicating the fold change in ELF3 expression of overexpressing cells compared to isogenic control. (C) Histogram summarizing the results from triplicate BrdU proliferation experiments. Percent BrdU positive cells are shown for control (black) and ELF3-OE (white) HBECs.  Paired two-tailed t-test: **p<1.0E-02.    116  To examine the impact of ELF3 inhibition on tumour growth in vivo, we established a panel of shELF3 clones in KRAS mutant A549 cells. The clone with the lowest ELF3 expression was selected for further study. Importantly, knock-down was maintained after long-term culture; protein was undetectable by immunoblot after 4 weeks (Figure 5.17), and expression was at the limits of detection by qPCR (Ct ranged from cycle 38 to undetectable). The A549-shELF3 clone demonstrated reduced proliferation, as expected (Figure 5.18). In addition, the A549-shELF3 clone spontaneously died in complete culture medium. Cell death was quantified by flow cytometry (Annexin V/PI) following culture in complete or serum-free media. The A549-shELF3 clone underwent significantly more apoptosis in complete media (paired t-test p=3.71E-03), while only 35% of cells survived serum starvation compared to 75% of control cells (paired t-test p=1.08E-04) (Figure 5.18). Overall, elimination of ELF3 expression greatly reduced the viability and fitness of A549 cells when grown in 2D culture.   5.3.6 ELF3 expression is required for in vivo tumour growth 117   Figure 5.17 Immunoblot of ELF3 expression across A549 shELF3 clones (A) Immunoblot of ELF3 and loading control Histone H3 across 19 A549 shELF3 cloned cell line lysates and control (“C”) cell line lysate. Clone #3 (red font) had the lowest ELF3 expression and was chosen for further phenotypic assays. (B) Immunoblot of ELF3 and loading control Histone H3 in control and shELF3 clone #3 lysates after culturing for 28 days while phenotypic assays were conducted.   118   Figure 5.18 A549 shEFL3 clone in vitro phenotypic results  119  Figure 5.18 A549 shEFL3 clone in vitro phenotypic results. (A) Histogram summarizing the results from triplicate BrdU proliferation experiments. Percent BrdU positive cells are shown for control (black) and shELF3 (blue) A549 cells. (B) Histogram summarizing the results from triplicate cell death quantification assays. Percent Annexin V (AnnV) positive cells are shown for control (black) and shELF3 (blue) A549 cells grown in complete media (10% FBS) or serum depleted media (0% FBS) for 72 hours. (C) Representative flow cytometry scatter plot of cellular Annexin V staining (x-axis) and Propidium Iodide (PI) staining (y-axis) for control and shELF3 clones grown in complete media (10% FBS) or serum depleted media (0% FBS) for 72 hours. Gating was based on live unstained cells. Cells positive for Annexin V staining, which represents early (PI-) or late (PI+) apoptotic cells, are coloured red. Paired Student’s t-test ***p<1.0E-03 ****p<1.0E-04.   120  The impact of ELF3 loss on tumour growth was assessed by subcutaneous injection of the A549-shELF3 clone cells and controls into the flanks of NOD-SCID mice (n=24). Loss of ELF3 expression largely abolished the ability of A549 cells to form tumours in vivo, as A549-shELF3 cells showed no evidence of growth (Figure 5.19). Small lumps of tissue were discovered in 25% of shELF3 injection sites during surgical inspection at endpoint day 33 (Figure 5.19). To determine whether these tissue samples were small A549-derived tumours or mouse tissue, RNA was extracted from fresh frozen tissues and subjected to qPCR using human and mouse specific qPCR primers. These small tissue samples were confirmed to contain A549 cells that had restored human ELF3 expression to control levels (Figure 5.20).    121   Figure 5.19 In vivo growth of A549 control cells and shELF3 clones (n=24) (A) Growth curve indicating mean tumour volume (mm3) +SEM of control (black) and shELF3 clone (blue) tumours over time (days). Paired Student t-test p-values are shown for each time point. (B) Sample image of one mouse with a small shELF3-related mass detected at endpoint (black arrow, Mouse 13), and one mouse with no evidence of shELF3 tumour at endpoint (Mouse 13). (C) Sample image of the dissected control and shELF3 tumours from Mouse 15 (top panel) and control tumour from Mouse 13 (lower panel).  122   Figure 5.20 qPCR assessment of ELF3 expression in cell line input and tumour RNA qRT-PCR derived RQ values comparing ELF3 expression in (from left to right) shELF3 cell line expression to control cell line expression at input (black bars); shELF3 tumour expression to paired control tumour expression at endpoint (dark grey bars); control tumours at endpoint to control input cell line RNA (light grey bars); shELF3 tumours at endpoint to shELF3 input cell line RNA (white bars).    123  Tumour growth was also evaluated using the EGFR-driven HCC827 cell line (control vs. shELF3) by subcutaneous injection (n=12). While shELF3 tumour growth was initially slowed (Figure 5.21A), tumour growth eventually reached control levels (Figure 5.21B). Analysis of protein lysates from endpoint tumours showed that knock-down was not maintained (Figure 5.22A), and DNA analysis revealed the shRNA vector was negatively selected during the course of tumour growth as evidenced by reduced copies of the vector compared to input (Figure 5.22B). Conversely, the control vector was maintained indicating the reduction in shRNA vector copy number was not a result of the experimental conditions or technical issues. These experiments provide strong evidence that ELF3 is an oncogene that functionally contributes to the growth of molecularly distinct LUAD.   124   Figure 5.21 In vivo growth of HCC827 control and shELF3 cells (n=12) (A) Growth curve indicating mean tumour volume (mm3) +SEM of control (black) and shELF3 (red) tumours over the first 21 days of the experiment. (B) Growth curve indicating mean tumour volume (mm3) +SEM of control (black) and shELF3 (red) tumours over the entire experiment (42 days). Grey background indicates the time during which we postulate negative selection of knock-down (KD) vector occurs. Paired Student’s t-test +p<1.0E-01 **p<1.0E-2.   125   Figure 5.22 ELF3 knock-down is not maintained during the course of the HCC827 in vivo experiment   126  Figure 5.22 ELF3 knock-down is not maintained during the course of the HCC827 in vivo experiment. (A) Immunoblot of ELF3 and loading control Histone H3 across lysates extracted from all control and shELF3 HCC827 tumours. (B) Histogram of the DNA copy number of the puromycin resistance cassette found in the control and ELF3 knock-down vectors derived from custom TaqMan quantitative copy number assays. Copy number of each control tumour (black hatched bars) and shELF3 tumour (red hatched bars) is represented as a ratio of the copy number of HCC827 control cell line input DNA (solid black bar) and HCC827 shELF3 cell line input DNA (solid red bar).   127  5.4 Discussion The genomic and functional evidence presented in this study leads us to conclude that the transcription factor ELF3 is a novel LUAD oncogene. Overexpression was exclusively observed in 40-73% of LUAD and correlated significantly with patient survival. We observe frequent DNA level alterations at the ELF3 locus that contribute to its overexpression. While frequent copy number gain was to be expected at 1q, we identify a region of focal amplification at 1q32.1 that harbours ELF3. Furthermore, we detect concomitant deregulation of ELF3 by both increased gene dosage and promoter hypomethylation in 36-54% of LUAD. These DNA level alterations correlated significantly with expression levels excluding the possibility of passenger alterations. While ELF3 was the focus of this study, we do not wish to exclude the possibility that additional genes on chromosome 1q are biologically relevant as similar studies have produced alternative genes in nearby regions such as SHC1 (1q21.3)202. The phenotypic effects of ELF3 inhibition were demonstrated in vitro using shRNAs. Knock-down of ELF3 resulted in significantly reduced cancer phenotypes in a panel of LUAD cell lines, including contact-independent growth and proliferation. These experiments also support the biological relevance of ELF3 to molecularly diverse LUAD, since these cell lines harbored different oncogenes. Furthermore, high ELF3 expression was not associated with any of the LUAD driver mutations examined, and was observed across molecularly distinct LUAD cell lines, murine models, and clinical specimens. It is possible that therapeutic inhibition of ELF3 could benefit multiple molecular subtypes of LUAD. Evidence to support the potential utility of therapeutic inhibition came from our in vivo studies. Interestingly, the degree of ELF3 knock-down was highly associated with the growth potential of subcutaneously injected LUAD cells. In our experiment with nearly 100% ELF3 knock-down, cells were unable to establish tumours after 33 days. The fraction of ELF3 knock-down cells that were found to have formed very 128  small tumours at endpoint had restored ELF3 expression to control levels. These results indicate that ELF3 expression is required for tumour formation after subcutaneous injection. It is unclear whether or not these tumours would have proliferated if the experiment was carried out for a longer period of time. However, the overall tumour burden of the control tumour on the opposing flank was too great to allow the experiment to continue. These results are an intriguing supplement to the recent study that demonstrated ELF3 was essential for the maintenance of a NOTCH3 driven stem-like phenotype in KRAS mutant LUAD, including tumour initiating characteristics162. We show for the first time that ELF3 is required for tumour formation in KRAS mutant LUAD cells beyond the stem-like setting. In our experiment with 75% ELF3 knock-down, cells had no difficulty establishing tumours. Furthermore, the growth kinetics of ELF3 knock-down tumours ultimately reached control levels. This suggested there was a loss of knock-down which was confirmed at the DNA, RNA and protein levels. We hypothesize that the injected cell population was heterogeneous, with some cells having fewer copies of the shRNA vector and thus higher ELF3 expression. Due to its role in regulating cell proliferation shown in this study, it is likely that the cells with higher ELF3 expression (and fewer copies of the shRNA vector) proliferated more quickly than the cells with lower ELF3 expression (and more copies of the shRNA vector) and became the dominant population. This experiment demonstrates that clones with higher ELF3 expression out-compete clones with lower ELF3 expression, underscoring its importance to in vivo growth. High ELF3 expression may be pro-tumourigenic in the early stages of LUAD tumourigenesis. In vitro data demonstrated overexpression of ELF3 in non-transformed HBECs resulted in significantly increased proliferation, while clinical data indicated high ELF3 expression was associated with poor overall survival in patients with Stage I disease. This suggests that selection of ELF3 in a pre-malignant 129  setting may increase cell fitness allowing a cell to out-compete other clones. It also suggests that LUAD with high ELF3 expression are more aggressive, which could be attributed to regulation of cell growth and viability demonstrated in this study. Further experiments are needed to address the necessity of ELF3 expression in achieving cell transformation and tumour progression. The differential expression observed between LUAD and LUSC highlights the context-dependency of ELF3 function. In other tumour types, ELF3 exhibits both tumour suppressive and oncogenic properties depending on the tissue of origin.  In colorectal and breast cancer, ELF3 demonstrates oncogenic properties, with high expression attributed to gene amplification and associated with poor patient prognosis203-205. In billiary tract cancer, mucinous ovarian carcinoma, and cancers of the cervix, stomach and bladder, recurrent deleterious mutations in ELF3 have been identified, suggesting a tumour suppressive role in these cancer types206-210. Paradoxically, in prostate cancer ELF3 can act as an oncogene or a tumour suppressor, depending on the context211. The idea of therapeutic inhibition of ELF3 in LUAD is appealing. However, the context-dependent nature of ELF3 function must be taken into consideration. In summary, we find ELF3 is frequently deregulated in LUAD and conclude ELF3 is a novel LUAD oncogene on chromosome 1q. ELF3 promotes tumour aggressiveness in the form of rapid tumour growth and is associated with poor patient survival. ELF3 represents a novel therapeutic target that could be relevant to a significant proportion of LUAD patients. Future studies will elucidate the impact of ELF3 therapeutic inhibition on tumour growth and progression in molecularly distinct LUAD.   130  6 Conclusions 6.1 Summary of thesis findings The objective of this thesis was to identify novel genes important to aggressive lung tumour biology. Aggressiveness referred to metastasis, poor OS or RFS, and rapid tumour growth. Collectively, this work (i) identified miRNAs significantly deregulated in metastatic LUAD that not only regulated metastatic phenotypes of invasion and EMT in vitro, but were also associated with poor OS (Chapter 3), (ii) underscored the importance of small ncRNAs (miRNAs and piRNAs) expressed from the imprinted DLK1-DIO3 locus in predicting NSCLC patient OS and RFS (Chapter 4), and (iii) identified a novel LUAD oncogene, ELF3, that regulates tumour growth and is associated with poor patient outcome. Below, each of these findings is summarized in the context of Aims 1 and 2 as stated in Chapter 2, Section 2.8. Furthermore, we discuss how the findings support the two study hypotheses, restated here: (1) Non-coding RNA genes are differentially expressed between aggressive and non-aggressive lung tumours, and functionally contribute to aggressive tumour biology. (2) Genes disrupted frequently by mechanisms other than mutation regulate aggressive lung tumour biology. Aim 1: Identify non-coding genes important to aggressive lung adenocarcinoma biology by characterizing non-coding RNAs differentially expressed in primary tumours with and without metastasis. The role of miRNAs in regulating the process of metastasis is not fully understood. By comparing miRNA gene expression profiles of localized primary LUAD to those of metastatic LUAD, we wished to identify metastasis-associated miRNAs that are 6.1.1 Chapter 3 summary with respect to Aim 1  131  involved in regulating metastatic phenotypes. In Chapter 3, we identify a list of 37 metastasis-associated miRNAs including miR-106b, the most significantly deregulated miRNA. miR-106b was significantly overexpressed in metastatic LUAD compared to localized LUAD, and compared to non-malignant lung tissue. miR-106b shares sequence homology with miR-106a and miR-17, which are encoded on different chromosomes. Interestingly, miR-106a was significantly overexpressed in metastatic LUAD. In vitro manipulation of miR-106a and miR-106b levels demonstrated that while only miR-106b could significantly increase the invasiveness of H2347 cells, both miR-106a and miR-106b could influence the expression of EMT markers away from an epithelial pattern and towards a mesenchymal pattern. This observation had clinical implications in the external TCGA dataset. To observe a significant association between high miR-106a and miR-106b expression and poor OS, high vimentin expression had to be considered. Results from Chapter 3 support the first study hypothesis by demonstrating (i) miRNAs are significantly differentially expressed between metastatic and localized lung tumours, (ii) miR-106b, and perhaps miR-106a, regulates metastatic phenotypes in vitro, thereby functionally contributing to an aggressive phenotype, and (iii) miR-106a and miR-106b are associated with poor OS, another metric of aggressiveness. The list of metastasis-associated miRNAs identified in Chapter 3 largely mapped to the imprinted DLK1-DIO3 locus on chromosome 14q32.2-32.31. This was intriguing as miRNAs expressed from this locus have previously been associated with LUAD aggressiveness. While we did not investigate their influence on metastatic phenotypes in vitro, Chapter 4 validates the applicability of a published DLK1-DIO3 miRNA survival signature on both our discovery (BCCA) dataset and an external (TCGA) dataset. The relevance of an emerging class of small ncRNA, piRNA, to patient outcome signatures was also investigated. Several studies have shown that 6.1.2 Chapter 4 summary with respect to Aim 1  132  piRNA expression signatures are associated with patient outcome in other cancer types. Therefore, we re-mined our small RNA sequencing data and that of the TCGA dataset for piRNA expression, and discovered seven piRNAs were expressed from the DLK1-DIO3 locus. Expression of four of these piRNAs was able to improve upon patient risk classification for both LUAD and LUSC when combined with the miRNA signature. These results indicate that deregulation of both miRNA and piRNA expression at the DLK1-DIO3 locus is associated with the aggressiveness features poor OS and poor RFS, supporting our hypothesis that small ncRNAs are differentially expressed between aggressive and non-aggressive lung tumours. Aim 2: Search for novel lung adenocarcinoma oncogenes by querying components of known lung cancer pathways. Mutations in LUAD oncogenes such as EGFR can be treated by targeted therapeutics, but patients can display inherent or acquired resistance underscoring the need to identify novel therapeutic targets. Oncogenes deregulated by mechanisms other than mutation that are important to the growth and survival of LUAD may represent novel therapeutic targets. In Chapter 5, we identify the transcription factor ELF3 to be a LUAD oncogene. ELF3 was selected due to its (i) localization to chromosome 1q which is frequently gained in LUAD and associated with poor outcome, (ii) its association with several oncogenic lung cancer pathways including ERBB2, NOTCH3, and TGFβ, and (iii) due to the fact that ELF3 had not been identified in previous studies searching for frequently mutated genes. We discover that ELF3 is frequently altered at the DNA level by focal amplification or broader copy number gains, and by promoter hypomethylation. These alterations were significantly correlated with increases in expression levels. A greater than 2-fold increase in ELF3 expression was observed in 40-73% of LUAD and was associated with poor OS. We show that shRNA-mediated ELF3 inhibition reduced cell proliferation and anchorage-independent growth in a panel of LUAD cell lines, and that elimination of 6.1.3 Chapter 5 summary with respect to Aim 2  133  ELF3 expression dramatically reduces cell viability, cell proliferation, and the ability to form tumours in vivo. These data suggest that therapeutic inhibition of ELF3 could impede tumour growth. We provide evidence that ELF3 is a novel oncogene that regulates tumour growth and is associated with poor OS, supporting our hypothesis. 6.2 Strengths and limitations of thesis work The number of studies analyzing metastatic vs. non-metastatic lung cancer samples by small RNA sequencing are limited. Based on this, the strengths of the study presented in Chapter 3 is the sample size and profiling method. Small RNA sequencing profiles were generated for 186 patient specimens. These tissues were microdissected as directed by a pathologist in order to limit contamination from alternative cell types such as stroma. The paired nature of this dataset was valuable as it allowed for increased statistical power when comparing LUAD to non-malignant profiles. Comparing paired expression profiles limits the issue of inter-individual variability (i.e. comparing tumour data from one patient to normal data from unmatched patients). Tissues were fresh frozen, which allowed for better preservation of genomic materials as compared to FFPE derived samples. This translates to optimal sample quality for small RNA sequencing. Profiling by small RNA sequencing is superior to profiling by microarray in that expression of all annotated miRNAs can be derived, and the data can be re-analyzed as new annotations become available. This was particularly beneficial in Chapter 4, when piRNA expression reads were pulled from this data. This would not have been possible if we had utilized miRNA expression microarrays. We identified miRNAs that were not only deregulated in tumour compared to normal, but also deregulated between metastatic and localized disease. We identified 27 miRNAs that were upregulated and 7 miRNAs that were 6.2.1 Chapter 3 134  downregulated in metastatic LUAD, as well as 3 miRNAs overexpressed in cancer but with higher expression in the localized cases. To our knowledge, this study represents one of the largest analyses of metastatic and localized LUAD samples complete with paired non-malignant lung tissues profiled using small RNA sequencing to date. Future studies should search for miRNAs that are recurrently associated with lung cancer metastasis in multiple independent studies. Such miRNAs could represent novel diagnostic tools if their deregulated expression is detectable in the blood or sputum. One limitation of this analysis is the fact that while localized and metastatic samples were well matched in terms of gender, age, and smoking history, they were not matched for size. Size has long been associated with propensity to metastasize, but smaller tumours can also be metastatic. An ideal study would match localized and metastatic primary tumours for size, in order to eliminate the potential for biological variability induced by this feature. In addition, LUAD sub-classification by histology was not considered due to sample size. Certain LUAD histologies are clinically more aggressive. A study comparing aggressive LUAD histologies to less aggressive histologies would identify histology specific miRNAs, although these miRNAs may or may not be involved in regulating aggressive phenotypes. Alternatively, comparing matched histologies of LUAD that had or had not metastasized would be ideal in order to control for this variable. However, LUAD are more frequently made up of mixed histologies and certain histologies are quite rare, making sample collection for such an analysis difficult. With respect to the in vitro data, the associations of miR-106b with invasiveness, and of miR-106a and miR-106b with EMT, would be stronger if they were replicated in additional LUAD cell lines. Alternatively, the effect of miR-106a and miR-106b inhibition could be explored in a panel of LUAD cell lines, since this approach has proved effective in other studies. Such an experiment would provide important 135  information to either support or reject the idea that miR-106a or miR-106b could be therapeutic targets in LUAD. In Chapter 4 we derive one of the first combined miRNA and piRNA survival signatures, and further underscore the importance of imprinted DLK1-DIO3 locus deregulation to lung cancer aggressiveness. Importantly, the improvements made to the survival signatures by piRNAs were not only observed in our BCCA dataset, but also validated in a large external dataset (TCGA). These findings demonstrate the benefit of examining more than one class of gene at a time when deriving signatures. The goal of this work was to demonstrate the benefit of combining different classes of small ncRNAs into survival signatures, and to demonstrate the feasibility of utilizing piRNAs as prognostic tools. While the results are interesting, they have not undergone the rigorous testing required to derive a clinically applicable signature. To do this, large cohorts with hundreds to thousands of samples would be required. Furthermore, while piRNAs may be used in prognostic signatures in the future, their actual function remains largely unknown. Future work examining the mechanism of action of piRNAs in somatic tissues and cancer is required in order to fully appreciate their role in tumour biology. Lung tumour datasets complete with multi-omics data are valuable in order to extract as much information as possible about the tumour as a system, and in order to correlate DNA-level events with RNA-level expression changes. Chapter 5 utilizes data from both our paired multi-omic LUAD dataset, as well as the large publically available multi-omic TCGA dataset. We demonstrate for the first time the high frequency of ELF3 locus deregulation that occurs in LUAD. This data provided evidence that ELF3 is an oncogene, and was well supported by in vitro and in vivo 6.2.2 Chapter 4 6.2.3 Chapter 5 136  experiments in a panel of LUAD cell lines. However, high mRNA expression does not necessarily equate to high protein expression. While high protein expression was confirmed in LUAD cell lines, this study would be strengthened by examination of ELF3 protein-level expression in tumours by IHC. The loss of ELF3 knock-down observed in the HCC827 in vivo experiment indicated higher ELF3 expression is beneficial to tumour growth. However, this experiment should be repeated with a stable shELF3 clone with little to no ELF3 expression, as was done with A549 cells. This would show whether or not ELF3 is required for tumour growth in an additional cell line, and would provide stronger evidence that ELF3 is required for tumour growth in general. 6.3 Considerations and future directions The process of metastasis is a complex biological process which we do not yet fully understand. It is crucial we continue to study metastasis as it is the cause of the majority of cancer-related deaths. A better understanding of the underlying biology is needed to expose novel therapeutic vulnerabilities, as there is no targeted therapy for metastasis. This work adds to the growing body of literature that establishes miRNAs as key regulators of the hallmarks of cancer, including metastasis. The ability of miR-106b to regulate cell invasion, in lung and other cancer types, makes this miRNA a potential therapeutic target for advanced stage disease. However, the concept of miRNAs as therapeutic targets is still very new, and as such this branch of drug development is in its infancy. Future pre-clinical studies are required to examine the effect of miRNA inhibition on metastatic cancer. To date, drug delivery is the main hurdle for miRNA therapeutics in humans. These therapies typically accumulate in the liver and kidneys, which can lead to damaging off-target effects. For lung cancer, 6.3.1 Chapter 3 137  there is promise in the fact that therapies may be delivered by aerosol, thus limiting systemic distribution. Alternatively, by characterizing the networks of genes targeted by metastasis-associated miRNAs, pathways can emerge that may themselves be therapeutically targetable by more conventional means. Overall, miRNAs represent an exciting and novel branch of therapeutics. Lung cancer has a high mortality rate and a high recurrence rate, making clinical management difficult. At time of diagnosis, there is currently no way to predict which lung tumours will progress rapidly or recur. In other words, it is difficult to determine which patients should receive more aggressive treatment up front, or adjuvant chemotherapy after surgery with curative intent. A biomarker panel capable of addressing this issue would be invaluable to the field and is under investigation. The data presented in Chapter 4 highlights two key points relevant to this clinical issue: (i) genes encoded at the imprinted DLK1-DIO3 locus are significantly associated with lung cancer patient prognosis, and (ii) the combination of two classes of genes can improve gene signatures. Previously, piRNAs were not annotated within the DLK1-DIO3 locus. Chapter 4 shows that 138 piRNA genes are encoded and seven piRNAs are expressed at the locus, adding to our knowledge of its complexity. A growing number of studies have indicated piRNA expression patterns are correlated with patient outcome. As a proof-of-principle, we show that the addition of piRNAs to a miRNA-based signature improves classification of patients into risk groups with significantly different outcomes. We hypothesize that the combined examination of miRNAs (which act at the mRNA level) and piRNAs (which are believed to act predominantly at the DNA level) provides additional biological information that may improve risk stratification. In addition, miRNAs and piRNAs are expressed in a highly tissue and context specific manner; it is plausible that they display aggressiveness associated expression patterns as well. This study sets a precedent for combining different 6.3.2 Chapter 4 138  classes of small non-coding RNA molecules when designing prognostic gene expression signatures. The data presented in Chapter 4 adds to the growing literature that ties the imprinted DLK1-DIO3 locus to poor lung cancer patient outcome. Future large-scale studies would be needed to investigate whether or not this locus can be utilized as a prognostic biomarker in lung cancer. Due to their stability in biofluids, blood-based detection of locus specific miRNAs and piRNAs may be possible. Foreseeable challenges include the fact that changes in miRNA and piRNA expression levels can be subtle and may not be easily resolved within a population. Another outstanding question is whether or not the deregulated gene expression observed is due to loss of imprinting in tumours. If this is the case, and tumour-associated methylation changes are correlated with patient outcome, they may be detectable in the blood or sputum of patients. Such non-invasive means of assessing methylation biomarkers have been proposed in other studies212,213. Currently available targeted therapies are only available to the subpopulation of patients harboring an actionable mutation (e.g. EGFR in 15-20% of LUAD), meaning the majority of patients are treated with damaging chemotherapy and/or radiotherapy, and those on targeted therapy regimens typically display inherent or acquired resistance. To improve lung cancer patient prognosis, novel therapeutic targets are required. In Chapter 5, we show that ELF3 is a novel LUAD oncogene that is not frequently affected by mutation, but instead is altered by DNA copy number alterations and promoter methylation changes. We show that elimination of ELF3 is capable of preventing tumour formation in vivo, which suggests ELF3 could be an effective therapeutic target. Importantly ELF3 was overexpressed in upwards of 70% of LUAD cases; therefore, inhibition of ELF3 may be of benefit to a greater proportion of patients than any currently available targeted therapies. Future studies will focus on the identification and testing of compounds capable of ELF3 6.3.3 Chapter 5 139  inhibition, ideally through a combination of in silico and in vitro screens of FDA approved drugs. By testing FDA approved drugs, the time between pre-clinical testing and clinical development is reduced if a successful candidate is discovered.     140  References 1 Canadian-Cancer-Society-Advisory-Committee-on-Cancer-Statistics. Canadian Cancer Statistics 2015. 2015. 2 Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer statistics. CA: a cancer journal for clinicians. 2011; 61: 69-90. doi:10.3322/caac.20107. 3 Sun S, Schiller JH, Gazdar AF. Lung cancer in never smokers--a different disease. Nature reviews. Cancer. 2007; 7: 778-790. doi:10.1038/nrc2190. 4 Hajdu SI. Much overlooked causes of lung cancer. Annals of clinical and laboratory science. 2011; 41: 97-101. 5 Al-Zoughool M, Krewski D. Health effects of radon: a review of the literature. International journal of radiation biology. 2009; 85: 57-69. doi:10.1080/09553000802635054. 6 Krewski D, Lubin JH, Zielinski JM, Alavanja M, Catalan VS, Field RW, Klotz JB, Letourneau EG, Lynch CF, Lyon JL, Sandler DP, Schoenberg JB, Steck DJ et al. A combined analysis of North American case-control studies of residential radon and lung cancer. Journal of toxicology and environmental health. Part A. 2006; 69: 533-597. doi:10.1080/15287390500260945. 7 Lubin JH, Boice JD, Jr., Edling C, Hornung RW, Howe GR, Kunz E, Kusiak RA, Morrison HI, Radford EP, Samet JM, et al. Lung cancer in radon-exposed miners and estimation of risk from indoor exposure. Journal of the National Cancer Institute. 1995; 87: 817-827. 8 Hubaux R, Becker-Santos DD, Enfield KS, Rowbotham D, Lam S, Lam WL, Martinez VD. Molecular features in arsenic-induced lung tumors. Molecular cancer. 2013; 12: 20. doi:10.1186/1476-4598-12-20. 9 Hubaux R, Becker-Santos DD, Enfield KS, Lam S, Lam WL, Martinez VD. Arsenic, asbestos and radon: emerging players in lung tumorigenesis. Environmental health : a global access science source. 2012; 11: 89. doi:10.1186/1476-069X-11-89. 10 Smith AH, Lingas EO, Rahman M. Contamination of drinking-water by arsenic in Bangladesh: a public health emergency. Bulletin of the World Health Organization. 2000; 78: 1093-1103. 11 Bailey-Wilson JE, Amos CI, Pinney SM, Petersen GM, de Andrade M, Wiest JS, Fain P, Schwartz AG, You M, Franklin W, Klein C, Gazdar A, Rothschild H et al. A major lung cancer susceptibility locus maps to chromosome 6q23-25. American journal of human genetics. 2004; 75: 460-474. doi:10.1086/423857. 12 You M, Wang D, Liu P, Vikis H, James M, Lu Y, Wang Y, Wang M, Chen Q, Jia D, Liu Y, Wen W, Yang P et al. Fine mapping of chromosome 6q23-25 region in familial lung cancer families reveals RGS17 as a likely candidate gene. Clinical cancer research : an official journal of the American Association for Cancer Research. 2009; 15: 2666-2674. doi:10.1158/1078-0432.CCR-08-2335. 141  13 Sellers TA, Bailey-Wilson JE, Elston RC, Wilson AF, Elston GZ, Ooi WL, Rothschild H. Evidence for mendelian inheritance in the pathogenesis of lung cancer. Journal of the National Cancer Institute. 1990; 82: 1272-1279. 14 Shi J, Chatterjee N, Rotunno M, Wang Y, Pesatori AC, Consonni D, Li P, Wheeler W, Broderick P, Henrion M, Eisen T, Wang Z, Chen W et al. Inherited variation at chromosome 12p13.33, including RAD52, influences the risk of squamous cell lung carcinoma. Cancer discovery. 2012; 2: 131-139. doi:10.1158/2159-8290.CD-11-0246. 15 Amos CI, Pinney SM, Li Y, Kupert E, Lee J, de Andrade MA, Yang P, Schwartz AG, Fain PR, Gazdar A, Minna J, Wiest JS, Zeng D et al. A susceptibility locus on chromosome 6q greatly increases lung cancer risk among light and never smokers. Cancer research. 2010; 70: 2359-2367. doi:10.1158/0008-5472.CAN-09-3096. 16 Herbst RS, Heymach JV, Lippman SM. Lung cancer. The New England journal of medicine. 2008; 359: 1367-1380. doi:10.1056/NEJMra0802714. 17 Pao W, Girard N. New driver mutations in non-small-cell lung cancer. The Lancet. Oncology. 2011; 12: 175-180. doi:10.1016/S1470-2045(10)70087-5. 18 Sullivan JP, Minna JD, Shay JW. Evidence for self-renewing lung cancer stem cells and their implications in tumor initiation, progression, and targeted therapy. Cancer metastasis reviews. 2010; 29: 61-72. doi:10.1007/s10555-010-9216-5. 19 Travis WD, Brambilla E, Noguchi M, Nicholson AG, Geisinger K, Yatabe Y, Powell CA, Beer D, Riely G, Garg K, Austin JH, Rusch VW, Hirsch FR et al. International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society: international multidisciplinary classification of lung adenocarcinoma: executive summary. Proceedings of the American Thoracic Society. 2011; 8: 381-385. doi:10.1513/pats.201107-042ST. 20 Goldstraw P, Chansky K, Crowley J, Rami-Porta R, Asamura H, Eberhardt WE, Nicholson AG, Groome P, Mitchell A, Bolejack V, International Association for the Study of Lung Cancer S, Prognostic Factors Committee AB, Participating I et al. The IASLC Lung Cancer Staging Project: Proposals for Revision of the TNM Stage Groupings in the Forthcoming (Eighth) Edition of the TNM Classification for Lung Cancer. Journal of thoracic oncology : official publication of the International Association for the Study of Lung Cancer. 2016; 11: 39-51. doi:10.1016/j.jtho.2015.09.009. 21 Scott WJ, Howington J, Feigenberg S, Movsas B, Pisters K, American College of Chest P. Treatment of non-small cell lung cancer stage I and stage II: ACCP evidence-based clinical practice guidelines (2nd edition). Chest. 2007; 132: 234S-242S. doi:10.1378/chest.07-1378. 22 Lackey A, Donington JS. Surgical management of lung cancer. Seminars in interventional radiology. 2013; 30: 133-140. doi:10.1055/s-0033-1342954. 142  23 Scheff RJ, Schneider BJ. Non-small-cell lung cancer: treatment of late stage disease: chemotherapeutics and new frontiers. Seminars in interventional radiology. 2013; 30: 191-198. doi:10.1055/s-0033-1342961. 24 Hoffman PC, Mauer AM, Vokes EE. Lung cancer. Lancet. 2000; 355: 479-485. doi:10.1016/S0140-6736(00)82038-3. 25 Chen Z, Fillmore CM, Hammerman PS, Kim CF, Wong KK. Non-small-cell lung cancers: a heterogeneous set of diseases. Nature reviews. Cancer. 2014; 14: 535-546. doi:10.1038/nrc3775. 26 de Boer M, van Dijck JA, Bult P, Borm GF, Tjan-Heijnen VC. Breast cancer prognosis and occult lymph node metastases, isolated tumor cells, and micrometastases. Journal of the National Cancer Institute. 2010; 102: 410-425. doi:10.1093/jnci/djq008. 27 Morton DL, Thompson JF, Cochran AJ, Mozzillo N, Elashoff R, Essner R, Nieweg OE, Roses DF, Hoekstra HJ, Karakousis CP, Reintgen DS, Coventry BJ, Glass EC et al. Sentinel-node biopsy or nodal observation in melanoma. The New England journal of medicine. 2006; 355: 1307-1317. doi:10.1056/NEJMoa060992. 28 Thiery JP. Epithelial-mesenchymal transitions in tumour progression. Nature reviews. Cancer. 2002; 2: 442-454. doi:10.1038/nrc822. 29 Mujoomdar A, Austin JH, Malhotra R, Powell CA, Pearson GD, Shiau MC, Raftopoulos H. Clinical predictors of metastatic disease to the brain from non-small cell lung carcinoma: primary tumor size, cell type, and lymph node metastases. Radiology. 2007; 242: 882-888. doi:10.1148/radiol.2423051707. 30 Guo S, Reddy CA, Chao ST, Suh JH, Barnett GH, Vogelbaum MA, Videtic GM. Impact of non-small cell lung cancer histology on survival predicted from the graded prognostic assessment for patients with brain metastases. Lung cancer. 2012; 77: 389-393. doi:10.1016/j.lungcan.2012.03.028. 31 Holgersson G, Bergstrom S, Bergqvist M, Nyman J, Hoye E, Helsing M, Friesland S, Holgersson M, Birath E, Ekman S, Blystad T, Ewers SB, Morth C et al. Swedish lung cancer radiation study group: predictive value of histology for radiotherapy response in patients with non-small cell lung cancer. European journal of cancer. 2011; 47: 2415-2421. doi:10.1016/j.ejca.2011.06.011. 32 Hess KR, Varadhachary GR, Taylor SH, Wei W, Raber MN, Lenzi R, Abbruzzese JL. Metastatic patterns in adenocarcinoma. Cancer. 2006; 106: 1624-1633. doi:10.1002/cncr.21778. 33 Gaspar LE. Brain metastases in lung cancer. Expert review of anticancer therapy. 2004; 4: 259-270. doi:10.1586/14737140.4.2.259. 34 Alexandrov LB, Nik-Zainal S, Wedge DC, Aparicio SA, Behjati S, Biankin AV, Bignell GR, Bolli N, Borg A, Borresen-Dale AL, Boyault S, Burkhardt B, Butler AP et al. Signatures of mutational processes in human cancer. Nature. 2013; 500: 415-421. doi:10.1038/nature12477. 143  35 Vogelstein B, Papadopoulos N, Velculescu VE, Zhou S, Diaz LA, Jr., Kinzler KW. Cancer genome landscapes. Science. 2013; 339: 1546-1558. doi:10.1126/science.1235122. 36 Rodenhuis S, Slebos RJ, Boot AJ, Evers SG, Mooi WJ, Wagenaar SS, van Bodegom PC, Bos JL. Incidence and possible clinical significance of K-ras oncogene activation in adenocarcinoma of the human lung. Cancer research. 1988; 48: 5738-5741. 37 Sharma SV, Bell DW, Settleman J, Haber DA. Epidermal growth factor receptor mutations in lung cancer. Nature reviews. Cancer. 2007; 7: 169-181. doi:10.1038/nrc2088. 38 Greulich H, Kaplan B, Mertins P, Chen TH, Tanaka KE, Yun CH, Zhang X, Lee SH, Cho J, Ambrogio L, Liao R, Imielinski M, Banerji S et al. Functional analysis of receptor tyrosine kinase mutations in lung cancer identifies oncogenic extracellular domain mutations of ERBB2. Proceedings of the National Academy of Sciences of the United States of America. 2012; 109: 14476-14481. doi:10.1073/pnas.1203201109. 39 Soda M, Choi YL, Enomoto M, Takada S, Yamashita Y, Ishikawa S, Fujiwara S, Watanabe H, Kurashina K, Hatanaka H, Bando M, Ohno S, Ishikawa Y et al. Identification of the transforming EML4-ALK fusion gene in non-small-cell lung cancer. Nature. 2007; 448: 561-566. doi:10.1038/nature05945. 40 Uguen A, De Braekeleer M. ROS1 fusions in cancer: a review. Future oncology. 2016; 12: 1911-1928. doi:10.2217/fon-2016-0050. 41 Duruisseaux M, McLeer-Florin A, Antoine M, Alavizadeh S, Poulot V, Lacave R, Rabbe N, Cadranel J, Wislez M. NRG1 fusion in a French cohort of invasive mucinous lung adenocarcinoma. Cancer medicine. 2016; 5: 3579-3585. doi:10.1002/cam4.838. 42 Nakaoku T, Tsuta K, Ichikawa H, Shiraishi K, Sakamoto H, Enari M, Furuta K, Shimada Y, Ogiwara H, Watanabe S, Nokihara H, Yasuda K, Hiramoto M et al. Druggable oncogene fusions in invasive mucinous lung adenocarcinoma. Clinical cancer research : an official journal of the American Association for Cancer Research. 2014; 20: 3087-3093. doi:10.1158/1078-0432.CCR-14-0107. 43 Shim HS, Kenudson M, Zheng Z, Liebers M, Cha YJ, Hoang Ho Q, Onozato M, Phi Le L, Heist RS, Iafrate AJ. Unique Genetic and Survival Characteristics of Invasive Mucinous Adenocarcinoma of the Lung. Journal of thoracic oncology : official publication of the International Association for the Study of Lung Cancer. 2015; 10: 1156-1162. doi:10.1097/JTO.0000000000000579. 44 Dhanasekaran SM, Balbin OA, Chen G, Nadal E, Kalyana-Sundaram S, Pan J, Veeneman B, Cao X, Malik R, Vats P, Wang R, Huang S, Zhong J et al. Transcriptome meta-analysis of lung cancer reveals recurrent aberrations in NRG1 and Hippo pathway genes. Nature communications. 2014; 5: 5893. doi:10.1038/ncomms6893. 45 Takeuchi K, Soda M, Togashi Y, Suzuki R, Sakata S, Hatano S, Asaka R, Hamanaka W, Ninomiya H, Uehara H, Lim Choi Y, Satoh Y, Okumura S et al. 144  RET, ROS1 and ALK fusions in lung cancer. Nature medicine. 2012; 18: 378-381. doi:10.1038/nm.2658. 46 Knudson AG, Jr. Mutation and cancer: statistical study of retinoblastoma. Proceedings of the National Academy of Sciences of the United States of America. 1971; 68: 820-823. 47 Luo J, Solimini NL, Elledge SJ. Principles of cancer therapy: oncogene and non-oncogene addiction. Cell. 2009; 136: 823-837. doi:10.1016/j.cell.2009.02.024. 48 Ciardiello F, Tortora G. A novel approach in the treatment of cancer: targeting the epidermal growth factor receptor. Clinical cancer research : an official journal of the American Association for Cancer Research. 2001; 7: 2958-2970. 49 Grunwald V, Hidalgo M. Developing inhibitors of the epidermal growth factor receptor for cancer treatment. Journal of the National Cancer Institute. 2003; 95: 851-867. 50 Mendelsohn J, Baselga J. Status of epidermal growth factor receptor antagonists in the biology and treatment of cancer. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2003; 21: 2787-2799. doi:10.1200/JCO.2003.01.504. 51 Normanno N, Bianco C, De Luca A, Maiello MR, Salomon DS. Target-based agents against ErbB receptors and their ligands: a novel approach to cancer treatment. Endocrine-related cancer. 2003; 10: 1-21. 52 Kwak EL, Bang YJ, Camidge DR, Shaw AT, Solomon B, Maki RG, Ou SH, Dezube BJ, Janne PA, Costa DB, Varella-Garcia M, Kim WH, Lynch TJ et al. Anaplastic lymphoma kinase inhibition in non-small-cell lung cancer. The New England journal of medicine. 2010; 363: 1693-1703. doi:10.1056/NEJMoa1006448. 53 Gerber DE, Minna JD. ALK inhibition for non-small cell lung cancer: from discovery to therapy in record time. Cancer cell. 2010; 18: 548-551. doi:10.1016/j.ccr.2010.11.033. 54 Midha A, Dearden S, McCormack R. EGFR mutation incidence in non-small-cell lung cancer of adenocarcinoma histology: a systematic review and global map by ethnicity (mutMapII). American journal of cancer research. 2015; 5: 2892-2911. 55 Mok TS, Wu YL, Thongprasert S, Yang CH, Chu DT, Saijo N, Sunpaweravong P, Han B, Margono B, Ichinose Y, Nishiwaki Y, Ohe Y, Yang JJ et al. Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. The New England journal of medicine. 2009; 361: 947-957. doi:10.1056/NEJMoa0810699. 56 Lynch TJ, Bell DW, Sordella R, Gurubhagavatula S, Okimoto RA, Brannigan BW, Harris PL, Haserlat SM, Supko JG, Haluska FG, Louis DN, Christiani DC, Settleman J et al. Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to gefitinib. The New England journal of medicine. 2004; 350: 2129-2139. doi:10.1056/NEJMoa040938. 145  57 Paez JG, Janne PA, Lee JC, Tracy S, Greulich H, Gabriel S, Herman P, Kaye FJ, Lindeman N, Boggon TJ, Naoki K, Sasaki H, Fujii Y et al. EGFR mutations in lung cancer: correlation with clinical response to gefitinib therapy. Science. 2004; 304: 1497-1500. doi:10.1126/science.1099314. 58 Dancey JE, Freidlin B. Targeting epidermal growth factor receptor--are we missing the mark? Lancet. 2003; 362: 62-64. doi:10.1016/S0140-6736(03)13810-X. 59 Tsao MS, Sakurada A, Cutz JC, Zhu CQ, Kamel-Reid S, Squire J, Lorimer I, Zhang T, Liu N, Daneshmand M, Marrano P, da Cunha Santos G, Lagarde A et al. Erlotinib in lung cancer - molecular and clinical predictors of outcome. The New England journal of medicine. 2005; 353: 133-144. doi:10.1056/NEJMoa050736. 60 Ciardiello F, Tortora G. EGFR antagonists in cancer treatment. The New England journal of medicine. 2008; 358: 1160-1174. doi:10.1056/NEJMra0707704. 61 Shaw AT, Yeap BY, Mino-Kenudson M, Digumarthy SR, Costa DB, Heist RS, Solomon B, Stubbs H, Admane S, McDermott U, Settleman J, Kobayashi S, Mark EJ et al. Clinical features and outcome of patients with non-small-cell lung cancer who harbor EML4-ALK. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2009; 27: 4247-4253. doi:10.1200/JCO.2009.22.6993. 62 Shaw AT, Kim DW, Nakagawa K, Seto T, Crino L, Ahn MJ, De Pas T, Besse B, Solomon BJ, Blackhall F, Wu YL, Thomas M, O'Byrne KJ et al. Crizotinib versus chemotherapy in advanced ALK-positive lung cancer. The New England journal of medicine. 2013; 368: 2385-2394. doi:10.1056/NEJMoa1214886. 63 Solomon BJ, Mok T, Kim DW, Wu YL, Nakagawa K, Mekhail T, Felip E, Cappuzzo F, Paolini J, Usari T, Iyer S, Reisman A, Wilner KD et al. First-line crizotinib versus chemotherapy in ALK-positive lung cancer. The New England journal of medicine. 2014; 371: 2167-2177. doi:10.1056/NEJMoa1408440. 64 Kobayashi S, Boggon TJ, Dayaram T, Janne PA, Kocher O, Meyerson M, Johnson BE, Eck MJ, Tenen DG, Halmos B. EGFR mutation and resistance of non-small-cell lung cancer to gefitinib. The New England journal of medicine. 2005; 352: 786-792. doi:10.1056/NEJMoa044238. 65 Pao W, Miller VA, Politi KA, Riely GJ, Somwar R, Zakowski MF, Kris MG, Varmus H. Acquired resistance of lung adenocarcinomas to gefitinib or erlotinib is associated with a second mutation in the EGFR kinase domain. PLoS medicine. 2005; 2: e73. doi:10.1371/journal.pmed.0020073. 66 Kosaka T, Yatabe Y, Endoh H, Yoshida K, Hida T, Tsuboi M, Tada H, Kuwano H, Mitsudomi T. Analysis of epidermal growth factor receptor gene mutation in patients with non-small cell lung cancer and acquired resistance to gefitinib. Clinical cancer research : an official journal of the American Association for Cancer Research. 2006; 12: 5764-5769. doi:10.1158/1078-0432.CCR-06-0714. 146  67 Engelman JA, Zejnullahu K, Mitsudomi T, Song Y, Hyland C, Park JO, Lindeman N, Gale CM, Zhao X, Christensen J, Kosaka T, Holmes AJ, Rogers AM et al. MET amplification leads to gefitinib resistance in lung cancer by activating ERBB3 signaling. Science. 2007; 316: 1039-1043. doi:10.1126/science.1141478. 68 Choi YL, Soda M, Yamashita Y, Ueno T, Takashima J, Nakajima T, Yatabe Y, Takeuchi K, Hamada T, Haruta H, Ishikawa Y, Kimura H, Mitsudomi T et al. EML4-ALK mutations in lung cancer that confer resistance to ALK inhibitors. The New England journal of medicine. 2010; 363: 1734-1739. doi:10.1056/NEJMoa1007478. 69 Katayama R, Khan TM, Benes C, Lifshits E, Ebi H, Rivera VM, Shakespeare WC, Iafrate AJ, Engelman JA, Shaw AT. Therapeutic strategies to overcome crizotinib resistance in non-small cell lung cancers harboring the fusion oncogene EML4-ALK. Proceedings of the National Academy of Sciences of the United States of America. 2011; 108: 7535-7540. doi:10.1073/pnas.1019559108. 70 Shepherd FA, Tsao MS. Unraveling the mystery of prognostic and predictive factors in epidermal growth factor receptor therapy. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2006; 24: 1219-1220; author reply 1220-1211. doi:10.1200/JCO.2005.04.4420. 71 Eberhard DA, Johnson BE, Amler LC, Goddard AD, Heldens SL, Herbst RS, Ince WL, Janne PA, Januario T, Johnson DH, Klein P, Miller VA, Ostland MA et al. Mutations in the epidermal growth factor receptor and in KRAS are predictive and prognostic indicators in patients with non-small-cell lung cancer treated with chemotherapy alone and in combination with erlotinib. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2005; 23: 5900-5909. doi:10.1200/JCO.2005.02.857. 72 Riely GJ, Politi KA, Miller VA, Pao W. Update on epidermal growth factor receptor mutations in non-small cell lung cancer. Clinical cancer research : an official journal of the American Association for Cancer Research. 2006; 12: 7232-7241. doi:10.1158/1078-0432.CCR-06-0658. 73 Sasaki H, Shimizu S, Endo K, Takada M, Kawahara M, Tanaka H, Matsumura A, Iuchi K, Haneda H, Suzuki E, Kobayashi Y, Yano M, Fujii Y. EGFR and erbB2 mutation status in Japanese lung cancer patients. International journal of cancer. 2006; 118: 180-184. doi:10.1002/ijc.21301. 74 Kosaka T, Yatabe Y, Onozato R, Kuwano H, Mitsudomi T. Prognostic implication of EGFR, KRAS, and TP53 gene mutations in a large cohort of Japanese patients with surgically treated lung adenocarcinoma. Journal of thoracic oncology : official publication of the International Association for the Study of Lung Cancer. 2009; 4: 22-29. doi:10.1097/JTO.0b013e3181914111. 75 Mascaux C, Iannino N, Martin B, Paesmans M, Berghmans T, Dusart M, Haller A, Lothaire P, Meert AP, Noel S, Lafitte JJ, Sculier JP. The role of RAS oncogene in survival of patients with lung cancer: a systematic review of the literature with meta-analysis. British journal of cancer. 2005; 92: 131-139. doi:10.1038/sj.bjc.6602258. 147  76 Shepherd FA, Domerg C, Hainaut P, Janne PA, Pignon JP, Graziano S, Douillard JY, Brambilla E, Le Chevalier T, Seymour L, Bourredjem A, Le Teuff G, Pirker R et al. Pooled analysis of the prognostic and predictive effects of KRAS mutation status and KRAS mutation subtype in early-stage resected non-small-cell lung cancer in four trials of adjuvant chemotherapy. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2013; 31: 2173-2181. doi:10.1200/JCO.2012.48.1390. 77 Kwiatkowski DJ, Harpole DH, Jr., Godleski J, Herndon JE, 2nd, Shieh DB, Richards W, Blanco R, Xu HJ, Strauss GM, Sugarbaker DJ. Molecular pathologic substaging in 244 stage I non-small-cell lung cancer patients: clinical implications. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 1998; 16: 2468-2477. doi:10.1200/jco.1998.16.7.2468. 78 D'Amico TA, Massey M, Herndon JE, 2nd, Moore MB, Harpole DH, Jr. A biologic risk model for stage I lung cancer: immunohistochemical analysis of 408 patients with the use of ten molecular markers. The Journal of thoracic and cardiovascular surgery. 1999; 117: 736-743. 79 Zhu CQ, Ding K, Strumpf D, Weir BA, Meyerson M, Pennell N, Thomas RK, Naoki K, Ladd-Acosta C, Liu N, Pintilie M, Der S, Seymour L et al. Prognostic and predictive gene signature for adjuvant chemotherapy in resected non-small-cell lung cancer. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2010; 28: 4417-4424. doi:10.1200/JCO.2009.26.4325. 80 Kratz JR, He J, Van Den Eeden SK, Zhu ZH, Gao W, Pham PT, Mulvihill MS, Ziaei F, Zhang H, Su B, Zhi X, Quesenberry CP, Habel LA et al. A practical molecular assay to predict survival in resected non-squamous, non-small-cell lung cancer: development and international validation studies. Lancet. 2012; 379: 823-832. doi:10.1016/S0140-6736(11)61941-7. 81 Van Laar RK. Genomic signatures for predicting survival and adjuvant chemotherapy benefit in patients with non-small-cell lung cancer. BMC medical genomics. 2012; 5: 30. doi:10.1186/1755-8794-5-30. 82 Chen DT, Hsu YL, Fulp WJ, Coppola D, Haura EB, Yeatman TJ, Cress WD. Prognostic and predictive value of a malignancy-risk gene signature in early-stage non-small cell lung cancer. Journal of the National Cancer Institute. 2011; 103: 1859-1870. doi:10.1093/jnci/djr420. 83 Tang H, Xiao G, Behrens C, Schiller J, Allen J, Chow CW, Suraokar M, Corvalan A, Mao J, White MA, Wistuba, II, Minna JD, Xie Y. A 12-gene set predicts survival benefits from adjuvant chemotherapy in non-small cell lung cancer patients. Clinical cancer research : an official journal of the American Association for Cancer Research. 2013; 19: 1577-1586. doi:10.1158/1078-0432.CCR-12-2321. 84 Wistuba, II, Behrens C, Lombardi F, Wagner S, Fujimoto J, Raso MG, Spaggiari L, Galetta D, Riley R, Hughes E, Reid J, Sangale Z, Swisher SG et al. Validation of a proliferation-based expression signature as prognostic marker in early 148  stage lung adenocarcinoma. Clinical cancer research : an official journal of the American Association for Cancer Research. 2013; 19: 6261-6271. doi:10.1158/1078-0432.CCR-13-0596. 85 Feld R, Rubinstein LV, Weisenberger TH. Sites of recurrence in resected stage I non-small-cell lung cancer: a guide for future studies. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 1984; 2: 1352-1358. doi:10.1200/jco.1984.2.12.1352. 86 Martini N, Bains MS, Burt ME, Zakowski MF, McCormack P, Rusch VW, Ginsberg RJ. Incidence of local recurrence and second primary tumors in resected stage I lung cancer. The Journal of thoracic and cardiovascular surgery. 1995; 109: 120-129. doi:10.1016/S0022-5223(95)70427-2. 87 Hubaux R, Becker-Santos DD, Enfield KS, Lam S, Lam WL, Martinez VD. MicroRNAs As Biomarkers For Clinical Features Of Lung Cancer. Metabolomics : open access. 2012; 2: 1000108. doi:10.4172/2153-0769.1000108. 88 Bogner PN, Patnaik SK, Pitoniak R, Kannisto E, Repasky E, Hylander B, Yendamuri S, Ramnath N. Lung cancer xenografting alters microRNA profile but not immunophenotype. Biochemical and biophysical research communications. 2009; 386: 305-310. doi:10.1016/j.bbrc.2009.06.021. 89 Raponi M, Dossey L, Jatkoe T, Wu X, Chen G, Fan H, Beer DG. MicroRNA classifiers for predicting prognosis of squamous cell lung cancer. Cancer research. 2009; 69: 5776-5783. doi:10.1158/0008-5472.CAN-09-0587. 90 Hu Z, Chen X, Zhao Y, Tian T, Jin G, Shu Y, Chen Y, Xu L, Zen K, Zhang C, Shen H. Serum microRNA signatures identified in a genome-wide serum microRNA expression profiling predict survival of non-small-cell lung cancer. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2010; 28: 1721-1726. doi:10.1200/JCO.2009.24.9342. 91 Tan X, Qin W, Zhang L, Hang J, Li B, Zhang C, Wan J, Zhou F, Shao K, Sun Y, Wu J, Zhang X, Qiu B et al. A 5-microRNA signature for lung squamous cell carcinoma diagnosis and hsa-miR-31 for prognosis. Clinical cancer research : an official journal of the American Association for Cancer Research. 2011; 17: 6802-6811. doi:10.1158/1078-0432.CCR-11-0419. 92 Duncavage E, Goodgame B, Sezhiyan A, Govindan R, Pfeifer J. Use of microRNA expression levels to predict outcomes in resected stage I non-small cell lung cancer. Journal of thoracic oncology : official publication of the International Association for the Study of Lung Cancer. 2010; 5: 1755-1763. doi:10.1097/JTO.0b013e3181f3909d. 93 Lu J, Getz G, Miska EA, Alvarez-Saavedra E, Lamb J, Peck D, Sweet-Cordero A, Ebert BL, Mak RH, Ferrando AA, Downing JR, Jacks T, Horvitz HR et al. MicroRNA expression profiles classify human cancers. Nature. 2005; 435: 834-838. doi:10.1038/nature03702. 94 Rosenfeld N, Aharonov R, Meiri E, Rosenwald S, Spector Y, Zepeniuk M, Benjamin H, Shabes N, Tabak S, Levy A, Lebanony D, Goren Y, Silberschein E 149  et al. MicroRNAs accurately identify cancer tissue origin. Nature biotechnology. 2008; 26: 462-469. doi:10.1038/nbt1392. 95 Shen J, Stass SA, Jiang F. MicroRNAs as potential biomarkers in human solid tumors. Cancer letters. 2013; 329: 125-136. doi:10.1016/j.canlet.2012.11.001. 96 Russo F, Di Bella S, Nigita G, Macca V, Lagana A, Giugno R, Pulvirenti A, Ferro A. miRandola: extracellular circulating microRNAs database. PloS one. 2012; 7: e47786. doi:10.1371/journal.pone.0047786. 97 Shah MY, Calin GA. The mix of two worlds: non-coding RNAs and hormones. Nucleic acid therapeutics. 2013; 23: 2-8. doi:10.1089/nat.2012.0375. 98 Cortez MA, Bueso-Ramos C, Ferdin J, Lopez-Berestein G, Sood AK, Calin GA. MicroRNAs in body fluids--the mix of hormones and biomarkers. Nature reviews. Clinical oncology. 2011; 8: 467-477. doi:10.1038/nrclinonc.2011.76. 99 Etheridge A, Lee I, Hood L, Galas D, Wang K. Extracellular microRNA: a new source of biomarkers. Mutation research. 2011; 717: 85-90. doi:10.1016/j.mrfmmm.2011.03.004. 100 Cortez MA, Calin GA. MicroRNA identification in plasma and serum: a new tool to diagnose and monitor diseases. Expert opinion on biological therapy. 2009; 9: 703-711. doi:10.1517/14712590902932889. 101 Keller A, Backes C, Leidinger P, Kefer N, Boisguerin V, Barbacioru C, Vogel B, Matzas M, Huwer H, Katus HA, Stahler C, Meder B, Meese E. Next-generation sequencing identifies novel microRNAs in peripheral blood of lung cancer patients. Molecular bioSystems. 2011; 7: 3187-3199. doi:10.1039/c1mb05353a. 102 Keller A, Leidinger P, Bauer A, Elsharawy A, Haas J, Backes C, Wendschlag A, Giese N, Tjaden C, Ott K, Werner J, Hackert T, Ruprecht K et al. Toward the blood-borne miRNome of human diseases. Nature methods. 2011; 8: 841-843. doi:10.1038/nmeth.1682. 103 Keller A, Leidinger P, Borries A, Wendschlag A, Wucherpfennig F, Scheffler M, Huwer H, Lenhof HP, Meese E. miRNAs in lung cancer - studying complex fingerprints in patient's blood cells by microarray experiments. BMC cancer. 2009; 9: 353. doi:10.1186/1471-2407-9-353. 104 Keller A, Leidinger P, Gislefoss R, Haugen A, Langseth H, Staehler P, Lenhof HP, Meese E. Stable serum miRNA profiles as potential tool for non-invasive lung cancer diagnosis. RNA biology. 2011; 8: 506-516. 105 Leidinger P, Keller A, Borries A, Huwer H, Rohling M, Huebers J, Lenhof HP, Meese E. Specific peripheral miRNA profiles for distinguishing lung cancer from COPD. Lung cancer. 2011; 74: 41-47. doi:10.1016/j.lungcan.2011.02.003. 106 Leidinger P, Keller A, Backes C, Huwer H, Meese E. MicroRNA expression changes after lung cancer resection: a follow-up study. RNA biology. 2012; 9: 900-910. doi:10.4161/rna.20107. 107 Consortium EP, Birney E, Stamatoyannopoulos JA, Dutta A, Guigo R, Gingeras TR, Margulies EH, Weng Z, Snyder M, Dermitzakis ET, Thurman RE, Kuehn 150  MS, Taylor CM et al. Identification and analysis of functional elements in 1% of the human genome by the ENCODE pilot project. Nature. 2007; 447: 799-816. doi:10.1038/nature05874. 108 Djebali S, Davis CA, Merkel A, Dobin A, Lassmann T, Mortazavi A, Tanzer A, Lagarde J, Lin W, Schlesinger F, Xue C, Marinov GK, Khatun J et al. Landscape of transcription in human cells. Nature. 2012; 489: 101-108. doi:10.1038/nature11233. 109 Gibb EA, Vucic EA, Enfield KS, Stewart GL, Lonergan KM, Kennett JY, Becker-Santos DD, MacAulay CE, Lam S, Brown CJ, Lam WL. Human cancer long non-coding RNA transcriptomes. PloS one. 2011; 6: e25915. doi:10.1371/journal.pone.0025915. 110 Dhamija S, Diederichs S. From junk to master regulators of invasion: lncRNA functions in migration, EMT and metastasis. International journal of cancer. 2016; 139: 269-280. doi:10.1002/ijc.30039. 111 Enfield KS, Pikor LA, Martinez VD, Lam WL. Mechanistic Roles of Noncoding RNAs in Lung Cancer Biology and Their Clinical Implications. Genetics research international. 2012; 2012: 737416. doi:10.1155/2012/737416. 112 Londin E, Loher P, Telonis AG, Quann K, Clark P, Jing Y, Hatzimichael E, Kirino Y, Honda S, Lally M, Ramratnam B, Comstock CE, Knudsen KE et al. Analysis of 13 cell types reveals evidence for the expression of numerous novel primate- and tissue-specific microRNAs. Proceedings of the National Academy of Sciences of the United States of America. 2015; 112: E1106-1115. doi:10.1073/pnas.1420955112. 113 Fassina A, Cappellesso R, Fassan M. Classification of non-small cell lung carcinoma in transthoracic needle specimens using microRNA expression profiling. Chest. 2011; 140: 1305-1311. doi:10.1378/chest.11-0708. 114 Huang W, Hu J, Yang DW, Fan XT, Jin Y, Hou YY, Wang JP, Yuan YF, Tan YS, Zhu XZ, Bai CX, Wu Y, Zhu HG et al. Two microRNA panels to discriminate three subtypes of lung carcinoma in bronchial brushing specimens. American journal of respiratory and critical care medicine. 2012; 186: 1160-1167. doi:10.1164/rccm.201203-0534OC. 115 Gilad S, Lithwick-Yanai G, Barshack I, Benjamin S, Krivitsky I, Edmonston TB, Bibbo M, Thurm C, Horowitz L, Huang Y, Feinmesser M, Hou JS, St Cyr B et al. Classification of the four main types of lung cancer using a microRNA-based diagnostic assay. The Journal of molecular diagnostics : JMD. 2012; 14: 510-517. doi:10.1016/j.jmoldx.2012.03.004. 116 Du L, Schageman JJ, Irnov, Girard L, Hammond SM, Minna JD, Gazdar AF, Pertsemlidis A. MicroRNA expression distinguishes SCLC from NSCLC lung tumor cells and suggests a possible pathological relationship between SCLCs and NSCLCs. Journal of experimental & clinical cancer research : CR. 2010; 29: 75. doi:10.1186/1756-9966-29-75. 117 Friedman RC, Farh KK, Burge CB, Bartel DP. Most mammalian mRNAs are conserved targets of microRNAs. Genome research. 2009; 19: 92-105. doi:10.1101/gr.082701.108. 151  118 Hutvagner G, Simard MJ. Argonaute proteins: key players in RNA silencing. Nature reviews. Molecular cell biology. 2008; 9: 22-32. doi:10.1038/nrm2321. 119 Feng YH, Tsao CJ. Emerging role of microRNA-21 in cancer. Biomedical reports. 2016; 5: 395-402. doi:10.3892/br.2016.747. 120 Wang X, Cao L, Wang Y, Wang X, Liu N, You Y. Regulation of let-7 and its target oncogenes (Review). Oncology letters. 2012; 3: 955-960. doi:10.3892/ol.2012.609. 121 Huang XA, Yin H, Sweeney S, Raha D, Snyder M, Lin H. A major epigenetic programming mechanism guided by piRNAs. Developmental cell. 2013; 24: 502-516. doi:10.1016/j.devcel.2013.01.023. 122 Yin H, Lin H. An epigenetic activation role of Piwi and a Piwi-associated piRNA in Drosophila melanogaster. Nature. 2007; 450: 304-308. doi:10.1038/nature06263. 123 Enfield KS, Martinez VD, Marshall EA, Stewart GL, Kung SH, Enterina JR, Lam WL. Deregulation of small non-coding RNAs at the DLK1-DIO3 imprinted locus predicts lung cancer patient outcome. Oncotarget. 2016; 7: 80957-80966. doi:10.18632/oncotarget.13133. 124 Lin H, Spradling AC. A novel group of pumilio mutations affects the asymmetric division of germline stem cells in the Drosophila ovary. Development. 1997; 124: 2463-2476. 125 Vagin VV, Sigova A, Li C, Seitz H, Gvozdev V, Zamore PD. A distinct small RNA pathway silences selfish genetic elements in the germline. Science. 2006; 313: 320-324. doi:10.1126/science.1129333. 126 Post C, Clark JP, Sytnikova YA, Chirn GW, Lau NC. The capacity of target silencing by Drosophila PIWI and piRNAs. Rna. 2014; 20: 1977-1986. doi:10.1261/rna.046300.114. 127 Watanabe T, Lin H. Posttranscriptional regulation of gene expression by Piwi proteins and piRNAs. Molecular cell. 2014; 56: 18-27. doi:10.1016/j.molcel.2014.09.012. 128 Gou LT, Dai P, Yang JH, Xue Y, Hu YP, Zhou Y, Kang JY, Wang X, Li H, Hua MM, Zhao S, Hu SD, Wu LG et al. Pachytene piRNAs instruct massive mRNA elimination during late spermiogenesis. Cell research. 2014; 24: 680-700. doi:10.1038/cr.2014.41. 129 Ng KW, Anderson C, Marshall EA, Minatel BC, Enfield KS, Saprunoff HL, Lam WL, Martinez VD. Piwi-interacting RNAs in cancer: emerging functions and clinical utility. Molecular cancer. 2016; 15: 5. doi:10.1186/s12943-016-0491-9. 130 Gebert D, Ketting RF, Zischler H, Rosenkranz D. piRNAs from Pig Testis Provide Evidence for a Conserved Role of the Piwi Pathway in Post-Transcriptional Gene Regulation in Mammals. PloS one. 2015; 10: e0124860. doi:10.1371/journal.pone.0124860. 131 Kwon C, Tak H, Rho M, Chang HR, Kim YH, Kim KT, Balch C, Lee EK, Nam S. Detection of PIWI and piRNAs in the mitochondria of mammalian cancer 152  cells. Biochemical and biophysical research communications. 2014; 446: 218-223. doi:10.1016/j.bbrc.2014.02.112. 132 Rouget C, Papin C, Boureux A, Meunier AC, Franco B, Robine N, Lai EC, Pelisson A, Simonelig M. Maternal mRNA deadenylation and decay by the piRNA pathway in the early Drosophila embryo. Nature. 2010; 467: 1128-1132. doi:10.1038/nature09465. 133 Barckmann B, Pierson S, Dufourt J, Papin C, Armenise C, Port F, Grentzinger T, Chambeyron S, Baronian G, Desvignes JP, Curk T, Simonelig M. Aubergine iCLIP Reveals piRNA-Dependent Decay of mRNAs Involved in Germ Cell Development in the Early Embryo. Cell reports. 2015; 12: 1205-1216. doi:10.1016/j.celrep.2015.07.030. 134 Mei Y, Wang Y, Kumari P, Shetty AC, Clark D, Gable T, MacKerell AD, Ma MZ, Weber DJ, Yang AJ, Edelman MJ, Mao L. A piRNA-like small RNA interacts with and modulates p-ERM proteins in human somatic cells. Nature communications. 2015; 6: 7316. doi:10.1038/ncomms8316. 135 Ha H, Song J, Wang S, Kapusta A, Feschotte C, Chen KC, Xing J. A comprehensive analysis of piRNAs from adult human testis and their relationship with genes and mobile elements. BMC genomics. 2014; 15: 545. doi:10.1186/1471-2164-15-545. 136 Martinez VD, Vucic EA, Thu KL, Hubaux R, Enfield KS, Pikor LA, Becker-Santos DD, Brown CJ, Lam S, Lam WL. Unique somatic and malignant expression patterns implicate PIWI-interacting RNAs in cancer-type specific biology. Scientific reports. 2015; 5: 10423. doi:10.1038/srep10423. 137 Siddiqi S, Matushansky I. Piwis and piwi-interacting RNAs in the epigenetics of cancer. Journal of cellular biochemistry. 2012; 113: 373-380. doi:10.1002/jcb.23363. 138 Fu A, Jacobs DI, Zhu Y. Epigenome-wide analysis of piRNAs in gene-specific DNA methylation. RNA biology. 2014; 11: 1301-1312. doi:10.1080/15476286.2014.996091. 139 Fu A, Jacobs DI, Hoffman AE, Zheng T, Zhu Y. PIWI-interacting RNA 021285 is involved in breast tumorigenesis possibly by remodeling the cancer epigenome. Carcinogenesis. 2015; 36: 1094-1102. doi:10.1093/carcin/bgv105. 140 He X, Chen X, Zhang X, Duan X, Pan T, Hu Q, Zhang Y, Zhong F, Liu J, Zhang H, Luo J, Wu K, Peng G et al. An Lnc RNA (GAS5)/SnoRNA-derived piRNA induces activation of TRAIL gene by site-specifically recruiting MLL/COMPASS-like complexes. Nucleic acids research. 2015; 43: 3712-3725. doi:10.1093/nar/gkv214. 141 Martinez VD, Enfield KS, Rowbotham DA, Lam WL. An atlas of gastric PIWI-interacting RNA transcriptomes and their utility for identifying signatures of gastric cancer recurrence. Gastric cancer : official journal of the International Gastric Cancer Association and the Japanese Gastric Cancer Association. 2016; 19: 660-665. doi:10.1007/s10120-015-0487-y. 153  142 Huang G, Hu H, Xue X, Shen S, Gao E, Guo G, Shen X, Zhang X. Altered expression of piRNAs and their relation with clinicopathologic features of breast cancer. Clinical & translational oncology : official publication of the Federation of Spanish Oncology Societies and of the National Cancer Institute of Mexico. 2013; 15: 563-568. doi:10.1007/s12094-012-0966-0. 143 Hashim A, Rizzo F, Marchese G, Ravo M, Tarallo R, Nassa G, Giurato G, Santamaria G, Cordella A, Cantarella C, Weisz A. RNA sequencing identifies specific PIWI-interacting small non-coding RNA expression patterns in breast cancer. Oncotarget. 2014; 5: 9901-9910. doi:10.18632/oncotarget.2476. 144 Zhang H, Ren Y, Xu H, Pang D, Duan C, Liu C. The expression of stem cell protein Piwil2 and piR-932 in breast cancer. Surgical oncology. 2013; 22: 217-223. doi:10.1016/j.suronc.2013.07.001. 145 Cheng J, Deng H, Xiao B, Zhou H, Zhou F, Shen Z, Guo J. piR-823, a novel non-coding small RNA, demonstrates in vitro and in vivo tumor suppressive activity in human gastric cancer cells. Cancer letters. 2012; 315: 12-17. doi:10.1016/j.canlet.2011.10.004. 146 Cheng J, Guo JM, Xiao BX, Miao Y, Jiang Z, Zhou H, Li QN. piRNA, the new non-coding RNA, is aberrantly expressed in human cancer cells. Clinica chimica acta; international journal of clinical chemistry. 2011; 412: 1621-1625. doi:10.1016/j.cca.2011.05.015. 147 Busch J, Ralla B, Jung M, Wotschofsky Z, Trujillo-Arribas E, Schwabe P, Kilic E, Fendler A, Jung K. Piwi-interacting RNAs as novel prognostic markers in clear cell renal cell carcinomas. Journal of experimental & clinical cancer research : CR. 2015; 34: 61. doi:10.1186/s13046-015-0180-3. 148 Li Y, Wu X, Gao H, Jin JM, Li AX, Kim YS, Pal SK, Nelson RA, Lau CM, Guo C, Mu B, Wang J, Wang F et al. Piwi-Interacting RNAs (piRNAs) Are Dysregulated in Renal Cell Carcinoma and Associated with Tumor Metastasis and Cancer-Specific Survival. Molecular medicine. 2015; 21: 381-388. doi:10.2119/molmed.2014.00203. 149 Law PT, Qin H, Ching AK, Lai KP, Co NN, He M, Lung RW, Chan AW, Chan TF, Wong N. Deep sequencing of small RNA transcriptome reveals novel non-coding RNAs in hepatocellular carcinoma. Journal of hepatology. 2013; 58: 1165-1173. doi:10.1016/j.jhep.2013.01.032. 150 Muller S, Raulefs S, Bruns P, Afonso-Grunz F, Plotner A, Thermann R, Jager C, Schlitter AM, Kong B, Regel I, Roth WK, Rotter B, Hoffmeier K et al. Next-generation sequencing reveals novel differentially regulated mRNAs, lncRNAs, miRNAs, sdRNAs and a piRNA in pancreatic cancer. Molecular cancer. 2015; 14: 94. doi:10.1186/s12943-015-0358-5. 151 Yan H, Wu QL, Sun CY, Ai LS, Deng J, Zhang L, Chen L, Chu ZB, Tang B, Wang K, Wu XF, Xu J, Hu Y. piRNA-823 contributes to tumorigenesis by regulating de novo DNA methylation and angiogenesis in multiple myeloma. Leukemia. 2015; 29: 196-206. doi:10.1038/leu.2014.135. 152 Cui L, Lou Y, Zhang X, Zhou H, Deng H, Song H, Yu X, Xiao B, Wang W, Guo J. Detection of circulating tumor cells in peripheral blood from patients with 154  gastric cancer using piRNAs as markers. Clinical biochemistry. 2011; 44: 1050-1057. doi:10.1016/j.clinbiochem.2011.06.004. 153 Firmino N, Martinez VD, Rowbotham DA, Enfield KS, Bennewith KL, Lam WL. HPV status is associated with altered PIWI-interacting RNA expression pattern in head and neck cancer. Oral oncology. 2016; 55: 43-48. doi:10.1016/j.oraloncology.2016.01.012. 154 Chari R, Thu KL, Wilson IM, Lockwood WW, Lonergan KM, Coe BP, Malloff CA, Gazdar AF, Lam S, Garnis C, MacAulay CE, Alvarez CE, Lam WL. Integrating the multiple dimensions of genomic and epigenomic landscapes of cancer. Cancer metastasis reviews. 2010; 29: 73-93. doi:10.1007/s10555-010-9199-2. 155 Wilson IM, Vucic EA, Enfield KS, Thu KL, Zhang YA, Chari R, Lockwood WW, Radulovich N, Starczynowski DT, Banath JP, Zhang M, Pusic A, Fuller M et al. EYA4 is inactivated biallelically at a high frequency in sporadic lung cancer and is associated with familial lung cancer risk. Oncogene. 2014; 33: 4464-4473. doi:10.1038/onc.2013.396. 156 Pikor LA, Lockwood WW, Thu KL, Vucic EA, Chari R, Gazdar AF, Lam S, Lam WL. YEATS4 is a novel oncogene amplified in non-small cell lung cancer that regulates the p53 pathway. Cancer research. 2013; 73: 7301-7312. doi:10.1158/0008-5472.CAN-13-1897. 157 Thu KL, Pikor LA, Chari R, Wilson IM, Macaulay CE, English JC, Tsao MS, Gazdar AF, Lam S, Lam WL, Lockwood WW. Genetic disruption of KEAP1/CUL3 E3 ubiquitin ligase complex components is a key mechanism of NF-kappaB pathway activation in lung cancer. Journal of thoracic oncology : official publication of the International Association for the Study of Lung Cancer. 2011; 6: 1521-1529. doi:10.1097/JTO.0b013e3182289479. 158 Rowbotham DA, Enfield KS, Martinez VD, Thu KL, Vucic EA, Stewart GL, Bennewith KL, Lam WL. Multiple Components of the VHL Tumor Suppressor Complex Are Frequently Affected by DNA Copy Number Loss in Pheochromocytoma. International journal of endocrinology. 2014; 2014: 546347. doi:10.1155/2014/546347. 159 Petersen I, Bujard M, Petersen S, Wolf G, Goeze A, Schwendel A, Langreck H, Gellert K, Reichel M, Just K, du Manoir S, Cremer T, Dietel M et al. Patterns of chromosomal imbalances in adenocarcinoma and squamous cell carcinoma of the lung. Cancer research. 1997; 57: 2331-2335. 160 Zhu H, Wong MP, Tin V. High-resolution detection of recurrent aberrations in lung adenocarcinomas by array comparative genomic hybridization and expression analysis of selective genes by quantitative PCR. International journal of oncology. 2014; 44: 2068-2076. doi:10.3892/ijo.2014.2384. 161 Liu J, Cho SN, Akkanti B, Jin N, Mao J, Long W, Chen T, Zhang Y, Tang X, Wistub, II, Creighton CJ, Kheradmand F, DeMayo FJ. ErbB2 Pathway Activation upon Smad4 Loss Promotes Lung Tumor Growth and Metastasis. Cell reports. 2015. doi:10.1016/j.celrep.2015.02.014. 155  162 Ali SA, Justilien V, Jamieson L, Murray NR, Fields AP. Protein Kinase Ciota Drives a NOTCH3-dependent Stem-like Phenotype in Mutant KRAS Lung Adenocarcinoma. Cancer cell. 2016; 29: 367-378. doi:10.1016/j.ccell.2016.02.012. 163 Cancer Genome Atlas N. Comprehensive molecular portraits of human breast tumours. Nature. 2012; 490: 61-70. doi:10.1038/nature11412. 164 Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M, Gingeras TR. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013; 29: 15-21. doi:10.1093/bioinformatics/bts635. 165 Xing Y, Yu T, Wu YN, Roy M, Kim J, Lee C. An expectation-maximization algorithm for probabilistic reconstructions of full-length isoforms from splice graphs. Nucleic acids research. 2006; 34: 3150-3160. doi:10.1093/nar/gkl396. 166 Sai Lakshmi S, Agrawal S. piRNABank: a web resource on classified and clustered Piwi-interacting RNAs. Nucleic acids research. 2008; 36: D173-177. doi:10.1093/nar/gkm696. 167 Mortazavi A, Williams BA, McCue K, Schaeffer L, Wold B. Mapping and quantifying mammalian transcriptomes by RNA-Seq. Nature methods. 2008; 5: 621-628. doi:10.1038/nmeth.1226. 168 Ramirez RD, Sheridan S, Girard L, Sato M, Kim Y, Pollack J, Peyton M, Zou Y, Kurie JM, Dimaio JM, Milchgrub S, Smith AL, Souza RF et al. Immortalization of human bronchial epithelial cells in the absence of viral oncoproteins. Cancer research. 2004; 64: 9027-9034. doi:10.1158/0008-5472.CAN-04-3703. 169 Shirdel EA, Xie W, Mak TW, Jurisica I. NAViGaTing the micronome--using multiple microRNA prediction databases to identify signalling pathway-associated microRNAs. PloS one. 2011; 6: e17429. doi:10.1371/journal.pone.0017429. 170 Meng W, Ye Z, Cui R, Perry J, Dedousi-Huebner V, Huebner A, Wang Y, Li B, Volinia S, Nakanishi H, Kim T, Suh SS, Ayers LW et al. MicroRNA-31 predicts the presence of lymph node metastases and survival in patients with lung adenocarcinoma. Clinical cancer research : an official journal of the American Association for Cancer Research. 2013; 19: 5423-5433. doi:10.1158/1078-0432.CCR-13-0320. 171 Xie L, Yang Z, Li G, Shen L, Xiang X, Liu X, Xu D, Xu L, Chen Y, Tian Z, Chen X. Genome-wide identification of bone metastasis-related microRNAs in lung adenocarcinoma by high-throughput sequencing. PloS one. 2013; 8: e61212. doi:10.1371/journal.pone.0061212. 172 Kim YH, Lee WK, Lee EB, Son JW, Kim DS, Park JY. Combined Effect of Metastasis-Related MicroRNA, miR-34 and miR-124 Family, Methylation on Prognosis of Non-Small-Cell Lung Cancer. Clinical lung cancer. 2017; 18: e13-e20. doi:10.1016/j.cllc.2016.06.005. 156  173 Gong C, Qu S, Liu B, Pan S, Jiao Y, Nie Y, Su F, Liu Q, Song E. MiR-106b expression determines the proliferation paradox of TGF-beta in breast cancer cells. Oncogene. 2015; 34: 84-93. doi:10.1038/onc.2013.525. 174 Feng B, Dong TT, Wang LL, Zhou HM, Zhao HC, Dong F, Zheng MH. Colorectal cancer migration and invasion initiated by microRNA-106a. PloS one. 2012; 7: e43452. doi:10.1371/journal.pone.0043452. 175 Smith AL, Iwanaga R, Drasin DJ, Micalizzi DS, Vartuli RL, Tan AC, Ford HL. The miR-106b-25 cluster targets Smad7, activates TGF-beta signaling, and induces EMT and tumor initiating cell characteristics downstream of Six1 in human breast cancer. Oncogene. 2012; 31: 5162-5171. doi:10.1038/onc.2012.11. 176 Yau WL, Lam CS, Ng L, Chow AK, Chan ST, Chan JY, Wo JY, Ng KT, Man K, Poon RT, Pang RW. Over-expression of miR-106b promotes cell migration and metastasis in hepatocellular carcinoma by activating epithelial-mesenchymal transition process. PloS one. 2013; 8: e57882. doi:10.1371/journal.pone.0057882. 177 Xie X, Liu HT, Mei J, Ding FB, Xiao HB, Hu FQ, Hu R, Wang MS. miR-106a promotes growth and metastasis of non-small cell lung cancer by targeting PTEN. International journal of clinical and experimental pathology. 2015; 8: 3827-3834. 178 Savita U, Karunagaran D. MicroRNA-106b-25 cluster targets beta-TRCP2, increases the expression of Snail and enhances cell migration and invasion in H1299 (non small cell lung cancer) cells. Biochemical and biophysical research communications. 2013; 434: 841-847. doi:10.1016/j.bbrc.2013.04.025. 179 Dylla L, Jedlicka P. Growth-promoting role of the miR-106a~363 cluster in Ewing sarcoma. PloS one. 2013; 8: e63032. doi:10.1371/journal.pone.0063032. 180 Brannan CI, Bartolomei MS. Mechanisms of genomic imprinting. Current opinion in genetics & development. 1999; 9: 164-170. doi:10.1016/S0959-437X(99)80025-2. 181 Jelinic P, Shaw P. Loss of imprinting and cancer. The Journal of pathology. 2007; 211: 261-268. doi:10.1002/path.2116. 182 Hirasawa R, Feil R. Genomic imprinting and human disease. Essays in biochemistry. 2010; 48: 187-200. doi:10.1042/bse0480187. 183 Kagami M, Sekita Y, Nishimura G, Irie M, Kato F, Okada M, Yamamori S, Kishimoto H, Nakayama M, Tanaka Y, Matsuoka K, Takahashi T, Noguchi M et al. Deletions and epimutations affecting the human 14q32.2 imprinted region in individuals with paternal and maternal upd(14)-like phenotypes. Nature genetics. 2008; 40: 237-242. doi:10.1038/ng.2007.56. 184 Lempiainen H, Couttet P, Bolognani F, Muller A, Dubost V, Luisier R, Del Rio Espinola A, Vitry V, Unterberger EB, Thomson JP, Treindl F, Metzger U, Wrzodek C et al. Identification of Dlk1-Dio3 imprinted gene cluster noncoding RNAs as novel candidate biomarkers for liver tumor promotion. 157  Toxicological sciences : an official journal of the Society of Toxicology. 2013; 131: 375-386. doi:10.1093/toxsci/kfs303. 185 Benetatos L, Hatzimichael E, Londin E, Vartholomatos G, Loher P, Rigoutsos I, Briasoulis E. The microRNAs within the DLK1-DIO3 genomic region: involvement in disease pathogenesis. Cellular and molecular life sciences : CMLS. 2013; 70: 795-814. doi:10.1007/s00018-012-1080-8. 186 Valdmanis PN, Roy-Chaudhuri B, Kim HK, Sayles LC, Zheng Y, Chuang CH, Caswell DR, Chu K, Zhang Y, Winslow MM, Sweet-Cordero EA, Kay MA. Upregulation of the microRNA cluster at the Dlk1-Dio3 locus in lung adenocarcinoma. Oncogene. 2015; 34: 94-103. doi:10.1038/onc.2013.523. 187 Nadal E, Zhong J, Lin J, Reddy RM, Ramnath N, Orringer MB, Chang AC, Beer DG, Chen G. A MicroRNA cluster at 14q32 drives aggressive lung adenocarcinoma. Clinical cancer research : an official journal of the American Association for Cancer Research. 2014; 20: 3107-3117. doi:10.1158/1078-0432.CCR-13-3348. 188 Watanabe T, Tomizawa S, Mitsuya K, Totoki Y, Yamamoto Y, Kuramochi-Miyagawa S, Iida N, Hoki Y, Murphy PJ, Toyoda A, Gotoh K, Hiura H, Arima T et al. Role for piRNAs and noncoding RNA in de novo DNA methylation of the imprinted mouse Rasgrf1 locus. Science. 2011; 332: 848-852. doi:10.1126/science.1203919. 189 Aran D, Sirota M, Butte AJ. Systematic pan-cancer analysis of tumour purity. Nature communications. 2015; 6: 8971. doi:10.1038/ncomms9971. 190 Becker-Santos DD, Thu KL, English JC, Pikor LA, Martinez VD, Zhang M, Vucic EA, Luk MT, Carraro A, Korbelik J, Piga D, Lhomme NM, Tsay MJ et al. Developmental transcription factor NFIB is a putative target of oncofetal miRNAs and is associated with tumour aggressiveness in lung adenocarcinoma. The Journal of pathology. 2016. doi:10.1002/path.4765. 191 Molina-Pinelo S, Salinas A, Moreno-Mata N, Ferrer I, Suarez R, Andres-Leon E, Rodriguez-Paredes M, Gutekunst J, Jantus-Lewintre E, Camps C, Carnero A, Paz-Ares L. Impact of DLK1-DIO3 imprinted cluster hypomethylation in smoker patients with lung cancer. Oncotarget. 2016. doi:10.18632/oncotarget.10611. 192 Tymms MJ, Ng AY, Thomas RS, Schutte BC, Zhou J, Eyre HJ, Sutherland GR, Seth A, Rosenberg M, Papas T, Debouck C, Kola I. A novel epithelial-expressed ETS gene, ELF3: human and murine cDNA sequences, murine genomic organization, human mapping to 1q32.2 and expression in tissues and cancer. Oncogene. 1997; 15: 2449-2462. doi:10.1038/sj.onc.1201427. 193 Oettgen P, Alani RM, Barcinski MA, Brown L, Akbarali Y, Boltax J, Kunsch C, Munger K, Libermann TA. Isolation and characterization of a novel epithelium-specific transcription factor, ESE-1, a member of the ets family. Molecular and cellular biology. 1997; 17: 4419-4433. 194 Gyorffy B, Surowiak P, Budczies J, Lanczky A. Online survival analysis software to assess the prognostic value of biomarkers using transcriptomic 158  data in non-small-cell lung cancer. PloS one. 2013; 8: e82241. doi:10.1371/journal.pone.0082241. 195 Mermel CH, Schumacher SE, Hill B, Meyerson ML, Beroukhim R, Getz G. GISTIC2.0 facilitates sensitive and confident localization of the targets of focal somatic copy-number alteration in human cancers. Genome biology. 2011; 12: R41. doi:10.1186/gb-2011-12-4-r41. 196 Cancer Genome Atlas Research N. Comprehensive molecular profiling of lung adenocarcinoma. Nature. 2014; 511: 543-550. doi:10.1038/nature13385. 197 Bild AH, Yao G, Chang JT, Wang Q, Potti A, Chasse D, Joshi MB, Harpole D, Lancaster JM, Berchuck A, Olson JA, Jr., Marks JR, Dressman HK et al. Oncogenic pathway signatures in human cancers as a guide to targeted therapies. Nature. 2006; 439: 353-357. doi:10.1038/nature04296. 198 Lee ES, Son DS, Kim SH, Lee J, Jo J, Han J, Kim H, Lee HJ, Choi HY, Jung Y, Park M, Lim YS, Kim K et al. Prediction of recurrence-free survival in postoperative non-small cell lung cancer patients by using an integrated model of clinical information and gene expression. Clinical cancer research : an official journal of the American Association for Cancer Research. 2008; 14: 7397-7404. doi:10.1158/1078-0432.CCR-07-4937. 199 Politi K, Zakowski MF, Fan PD, Schonfeld EA, Pao W, Varmus HE. Lung adenocarcinomas induced in mice by mutant EGF receptors found in human lung cancers respond to a tyrosine kinase inhibitor or to down-regulation of the receptors. Genes & development. 2006; 20: 1496-1510. doi:10.1101/gad.1417406. 200 Podsypanina K, Politi K, Beverly LJ, Varmus HE. Oncogene cooperation in tumor maintenance and tumor recurrence in mouse mammary tumors induced by Myc and mutant Kras. Proceedings of the National Academy of Sciences of the United States of America. 2008; 105: 5242-5247. doi:10.1073/pnas.0801197105. 201 Fisher GH, Wellen SL, Klimstra D, Lenczowski JM, Tichelaar JW, Lizak MJ, Whitsett JA, Koretsky A, Varmus HE. Induction and apoptotic regression of lung adenocarcinomas by regulation of a K-Ras transgene in the presence and absence of tumor suppressor genes. Genes & development. 2001; 15: 3249-3262. doi:10.1101/gad.947701. 202 Lockwood WW, Chari R, Coe BP, Girard L, Macaulay C, Lam S, Gazdar AF, Minna JD, Lam WL. DNA amplification is a ubiquitous mechanism of oncogene activation in lung and other cancers. Oncogene. 2008; 27: 4615-4624. doi:10.1038/onc.2008.98. 203 Wang JL, Chen ZF, Chen HM, Wang MY, Kong X, Wang YC, Sun TT, Hong J, Zou W, Xu J, Fang JY. Elf3 drives beta-catenin transactivation and associates with poor prognosis in colorectal cancer. Cell death & disease. 2014; 5: e1263. doi:10.1038/cddis.2014.206. 204 Mesquita B, Lopes P, Rodrigues A, Pereira D, Afonso M, Leal C, Henrique R, Lind GE, Jeronimo C, Lothe RA, Teixeira MR. Frequent copy number gains at 1q21 and 1q32 are associated with overexpression of the ETS transcription 159  factors ETV3 and ELF3 in breast cancer irrespective of molecular subtypes. Breast cancer research and treatment. 2013; 138: 37-45. doi:10.1007/s10549-013-2408-2. 205 Prescott JD, Poczobutt JM, Tentler JJ, Walker DM, Gutierrez-Hartmann A. Mapping of ESE-1 subdomains required to initiate mammary epithelial cell transformation via a cytoplasmic mechanism. Molecular cancer. 2011; 10: 103. doi:10.1186/1476-4598-10-103. 206 Ryland GL, Hunter SM, Doyle MA, Caramia F, Li J, Rowley SM, Christie M, Allan PE, Stephens AN, Bowtell DD, Australian Ovarian Cancer Study G, Campbell IG, Gorringe KL. Mutational landscape of mucinous ovarian carcinoma and its neoplastic precursors. Genome medicine. 2015; 7: 87. doi:10.1186/s13073-015-0210-y. 207 Nakamura H, Arai Y, Totoki Y, Shirota T, Elzawahry A, Kato M, Hama N, Hosoda F, Urushidate T, Ohashi S, Hiraoka N, Ojima H, Shimada K et al. Genomic spectra of biliary tract cancer. Nature genetics. 2015; 47: 1003-1010. doi:10.1038/ng.3375. 208 Ojesina AI, Lichtenstein L, Freeman SS, Pedamallu CS, Imaz-Rosshandler I, Pugh TJ, Cherniack AD, Ambrogio L, Cibulskis K, Bertelsen B, Romero-Cordoba S, Trevino V, Vazquez-Santillan K et al. Landscape of genomic alterations in cervical carcinomas. Nature. 2014; 506: 371-375. doi:10.1038/nature12881. 209 Wang K, Yuen ST, Xu J, Lee SP, Yan HH, Shi ST, Siu HC, Deng S, Chu KM, Law S, Chan KH, Chan AS, Tsui WY et al. Whole-genome sequencing and comprehensive molecular profiling identify new driver mutations in gastric cancer. Nature genetics. 2014; 46: 573-582. doi:10.1038/ng.2983. 210 Nordentoft I, Lamy P, Birkenkamp-Demtroder K, Shumansky K, Vang S, Hornshoj H, Juul M, Villesen P, Hedegaard J, Roth A, Thorsen K, Hoyer S, Borre M et al. Mutational context and diverse clonal development in early and late bladder cancer. Cell reports. 2014; 7: 1649-1663. doi:10.1016/j.celrep.2014.04.038. 211 Shatnawi A, Norris JD, Chaveroux C, Jasper JS, Sherk AB, McDonnell DP, Giguere V. ELF3 is a repressor of androgen receptor action in prostate cancer cells. Oncogene. 2014; 33: 862-871. doi:10.1038/onc.2013.15. 212 Anglim PP, Alonzo TA, Laird-Offringa IA. DNA methylation-based biomarkers for early detection of non-small cell lung cancer: an update. Molecular cancer. 2008; 7: 81. doi:10.1186/1476-4598-7-81. 213 Guo S, Yan F, Xu J, Bao Y, Zhu J, Wang X, Wu J, Li Y, Pu W, Liu Y, Jiang Z, Ma Y, Chen X et al. Identification and validation of the methylation biomarkers of non-small cell lung cancer (NSCLC). Clinical epigenetics. 2015; 7: 3. doi:10.1186/s13148-014-0035-3.    160  Appendix I: Publications This appendix lists all published manuscripts I was involved in during my degree. Co-first authorship is indicated by an underline. Relevance to specific Chapters is indicated, if applicable. 1. Enfield KSS, Martinez VD, Marshall E, Stewart GL, Kung SHY, Enterina JR, Lam WL (2016) Deregulation of small non-coding RNAs at the DLK1-DIO3 imprinted locus predicts lung adenocarcinoma patient outcome. Oncotarget. Dec;49, 80957-80966. This publication is the basis of Chapter 4. 2. Marshall EA, Ng KW, Kung SHY, Conway EM, Martinez VD, Halvorsen EC, Rowbotham DA, Vucic EA, Plumb AW, Becker-Santos DD, Enfield KSS, Bennewith KL, Lockwood WW, Lam S, English JC, Abraham N, Lam WL (2016) Emerging roles of T Helper 17 and Regulatory T Cells in lung cancer progression and metastasis. Molecular Cancer. 1:67. 3. Ng KW, Anderson C, Marshall EA, Minatel BC, Enfield KSS, Saprunoff HL, Lam WL, Martinez VD (2016) Piwi-interacting RNAs in cancer: emerging functions and clinical utility. Molecular Cancer. Jan;15:5. This publication is cited in Chapter 1. 4. Firmino N, Martinez VD, Rowbotham DA, Enfield KSS, Bennewith KL, Lam WL (2016) HPV status is associated with altered PIWI-interacing RNA expression in head and neck cancer. Oral Oncology. Apr;55, 43-8. This publication is cited in Chapter 1. 5. Martinez VD, Enfield KSS, Rowbotham DA, Lam WL (2016) An atlas of gastric PIWI-Interacting RNA transcriptomes and their utility for identifying signatures of gastric cancer recurrence. Gastric Cancer. 19, 660-5. This publication is cited in Chapter 1. 6. Martinez VD, Vucic EA, Thu KL, Hubaux R, Enfield KSS, Pikor LA, Becker-Santos DD, Brown CJ, Lam S, Lam WL (2015) Unique somatic and malignant expression 161  patterns implicate PIWI-interacting RNAs in cancer-type specific biology. Scientific Reports. 5:10423, 1-17. This publication is cited in Chapter 1. 7. Vucic EA, Thu KL, Pikor LA, Enfield KSS, Yee J, English JC, MacAulay CE, Lam S, Jurisica I, Lam WL (2014) Smoking status influences miRNA deregulation in lung adenocarcinoma affecting patient survival. BMC Cancer. 14:778, 1-14. 8. Rowbotham D, Enfield KSS, Martinez VD, Thu KL, Vucic EA, Stewart GL, Bennewith K, Lam WL (2014) Multiple components of the VHL tumor suppressor complex are frequently affected by DNA copy number loss in pheochromocytoma. International Journal of Endocrinology. 2014:546347, 1-9. This publication is cited in Chapter 1. 9. Wilson IM, Vucic EA, Enfield KSS, Thu KL, Zhang YA, Chari R, Lockwood WW, Radulovich N, Starczynowski DT, Banath J, Zhang M, Pusic A, Fuller M, Lonergan KM, Rowbotham D, Yee J, English J, Buys TPH, Selamat SA, Laird-Offringa I, Liu P, Anderson M, You M, Tsao MS, Brown C, Bennewith K, MacAulay CE, Karsan A, Gazdar AF, Lam S, Lam WL (2014) EYA4 is inactivated biallelically at a high frequency in sporadic lung cancer and is associated with familial lung cancer risk. Oncogene. 33, 4464-4473. This publication is cited in Chapter 1. 10. Hubaux R, Becker-Santos DD, Enfield KSS, Rowbotham D, Lam S, Lam WL, Martinez VC (2013) Molecular features in arsenic-induced lung tumors. Molecular Cancer. 12:20, 1-11. This publication is cited in Chapter 1. 11. Hubaux R, Becker-Santos DD, Enfield KSS, Lam S, Lam WL, Martinez VC (2012) Impact of Arsenic, Asbestos and Radon Exposure on the Lung Cancer Genome. Environmental Health. 11:89, 1-12. This publication is cited in Chapter 1. 12. Hubaux R, Becker-Santos DD, Enfield KSS, Lam S, Lam WL, Martinez VD (2012) MicroRNAs as biomarkers for clinical features of lung cancer. Metabolomics. 2, 1-24. This publication is cited in Chapter 1. 13. Gibb EA, Becker-Santos DD, Enfield KSS, Guillaud M, van Niekerk D, Matisic JP, MacAulay CE, Lam WL (2012) Aberrant expression of long non-coding RNAs in 162  cervical intaepithelial neoplasia. International Journal of Gynecological Cancer. 22:9, 1557-63. This publication is cited in Chapter 1. 14. Enfield KSS, Pikor LA, Martinez VD, Lam WL (2012) Mechanistic roles of non-coding RNAs in lung cancer biology and their clinical implications. Genetic Research International.  2012:737416, 1-16. This publication is cited in Chapter 1. 15. Gibb EA, Vucic EA, Enfield KSS, Stewart GL, Lonergan KM, Kennett JY, Becker-Santos DD, MacAulay CE, Lam S, Brown CJ, Lam WL (2011) Human cancer long non-coding RNA transcriptomes.  PLoS ONE. 6:e25915, 1-10. 16. Gibb EA, Enfield KSS, Stewart GL, Lonergan KM, Chari R, Ng RT, Zhang L, MacAulay CE, Rosin MP, and Lam WL (2011)  Long non-coding RNAs are expressed in oral mucosa and altered in oral premalignant lesions.  Oral Oncology. 47:11, 1055-61. 17. Enfield KSS, Stewart GL, Pikor LA, Alvarez CE, Lam S, Lam WL, Chari R (2011) MicroRNA gene dosage alterations and drug response in lung cancer. Journal of Biomedicine and Biotechnology. 2011: 474632, 1-15. 18. Enfield KSS, Pikor LA, Heryet C, Lam WL (2010) DNA Extraction from Paraffin Embedded Materials. Journal of Visualized Experiments. 49: 2763. 19. Gibb E, Enfield KSS, Tsui I, Chari R, Lam S, Alvarez CE, Lam WL (2010) Deciphering Squamous Cell Carcinoma Using Multi-Dimensional Genomic Approaches. Journal of Skin Cancer. 2011:541405, 1-16.    163  Appendix II: Selected Abstracts This appendix lists a selection of abstracts I presented at local, national, or international conferences. They were selected based on their relevance to Chapters 3-5 of this thesis. Presenting author is designated by an asterisk. 1. Kung SHY*, Enfield KSS, Rowbotham DA, Marshall EA, Holly A, Pastrello C, Minatel BC, Dellaire G, Jurisica I, Lam WL (2016) Combined overexpression of miR-106 paralogs regulates metastasis and is associated with poor outcome. Cell Symposia: Hallmarks of Cancer. Ghent, Belgium. 2. Enterina JR*, Enfield KSS, Martinez VD, Marshall E, Stewart GL, Lam WL (2016) Piwi-interacting RNAs are expressed from the DLK1-DIO3 locus and predict lung cancer patient outcome. Cell Symposia: Hallmarks of Cancer. Ghent, Belgium. 3. Minatel BC*, Martinez VM, Marshall EA, Ng K, Enfield KSS, Lam S, Lam WL (2016) PIWI-interacting RNAs expression networks associated with clinically-relevant tumor features and patient prognosis. Cell Symposia: Hallmarks of Cancer. Ghent, Belgium. 4. Enfield KSS*, Rahmati S, Rowbotham DA, Fuller M, Anderson C, Kennett JY, Marshall EA, Chari R, Becker-Santos DD, Ng K, MacAulay CE, Lam S, Politi K, Lockwood WW, Karsan A, Jurisica I, Lam WL (2016) ELF3 overexpression leads to oncogenic reprogramming of protein interactions exposing therapeutically actionable targets. 17th World Conference of Lung Cancer. Vienna, Austria. 5. Kung SHY*, Enfield KSS, Rowbotham DA, Marshall EA, Suzuki M, Holly A, Pastrello C, Minatel BC, Dellaire G, Jurisica I, MacAulay CE, Lam S, Lam WL (2016) Expression of miR-106 paralogs improves prognostic value of mesenchymal signatures but only miR-106b promotes invasiveness. 17th World Conference of Lung Cancer. Vienna, Austria. 6. Enterina JR*, Enfield KSS, Martinez VD, Marshall E, Stewart GL, Lam WL (2016) Deregulation of small non-coding RNAs at the DLK1-DIO3 imprinted locus  164  predicts lung adenocarcinoma patient outcome. 17th World Conference of Lung Cancer. Vienna, Austria. 7. Minatel BC*, Martinez VM, Marshall EA, Ng K, Enfield KSS, Lam S, Lam WL (2016) A PIWI-interacting RNAs co-expression networks as a prognostic factor in lung cancer. 17th World Conference of Lung Cancer. Vienna, Austria. 8. Enfield KSS*, Rowbotham DA, Rahmati S, Anderson C, Kennett JY, Marshall EA, Chari R, Becker-Santos DD, Ng K, MacAulay CE, Lam S, Lockwood WW, Jurisica I, Lam WL (2016) Alterations to the ELF3 oncogene circumvent upstream regulation in lung cancer. ASHG 2016 Annual Meeting. Vancouver, BC. 9. Kung SHY*, Enfield KSS, Rowbotham DA, Pastrello C, Minatel BC, Jurisica I, MacAulay CE, Lam S, Lam WL (2016) Characterization of microRNAs upregulated in metastatic lung adenocarcinoma. ASHG 2016 Annual Meeting. Vancouver, BC. 10. Enterina JR*, Enfield KSS, Martinez VD, Lam WL (2016) A combined miRNA-piRNA signature derived from the DLK1-DIO3 imprinted locus predicts lung adenocarcinoma patient outcome. ASHG 2016 Annual Meeting. Vancouver, BC.  11. Enfield KSS*, Rowbotham DA, Anderson C, Marshall E, Ng K, Minatel BC, Chari R, Fuller M, Becker-Santos DD, MacAulay CE, Lam S, Lockwood WW, Karsan A, Lam WL (2016) ELF3 amplification circumvents dependency on upstream driver mutations in lung adenocarcinoma. Terry Fox Research Institute 7th Annual Meeting. Vancouver, BC. 12. Enfield KSS*, Anderson C, Marshall E, Ng K, Minatel BC, Rowbotham DA, Chari R, Fuller M, Milne K, Becker-Santos DD, MacAulay CE, Karsan A, Lam S, Lam WL (2016) ELF3 amplification at 1q32.1 promotes SMAD4-independent tumourigenesis. Fourth AACR-IASLC International Joint Conference: Lung Cancer Translational Science from the Bench to the Clinic. San Diego, CA. Abstract B10. 13. Martinez VD*, Enfield KSS, Vucic EA, Firminio N, Bennewith KL, Lam S, Lam WL (2016) Piwi-interacting RNA transcriptome analyses identify cancer type-specific expression and signatures predicting lung tumor behavior. Fourth  165  AACR-IASLC International Joint Conference: Lung Cancer Translational Science from the Bench to the Clinic. San Diego, CA. Abstract A22; PR03 - invited talk. 14. Enfield KSS*, Rowbotham DA, Holly A, Anderson C, Ng KW, Minatel BdC, Dellaire G, Pastrello C, Jurisica I, MacAulay C, Lam S, Lam WL (2015) miR-106a and miR-106b affect growth and metastasis of lung adenocarcinoma. AACR Non-coding RNAs and Cancer. Boston, MA. 15. Enfield KSS*, Rowbotham DA, Holly A, Anderson C, Ng KW, Minatel BdC, Dellaire G, Pastrello C, Jurisica I, MacAulay C, Lam S, Lam WL (2015) miR-106a and miR-106b affect growth and metastasis of lung adenocarcinoma. TFRI BC Node Research Day. Vancouver, BC. 16. Enfield KSS*, Anderson C, Ng KW, Minatel BdC, Rowbotham DA, Becker-Santos DD, Chari R, Fuller M, MacAulay C, Karsan A, Lam S, Lam WL (2015) ELF3 is a novel lung adenocarcinoma oncogene. BC Cancer Centre Research Day. Vancouver, BC. 17. Enfield KSS*, Ng KW, Anderson C, Minatel BdC, Rowbotham DA, Holly A, Dellaire G, Pastrello C, Jurisica I, MacAulay C, Lam S, Lam WL (2015) miR-106a and miR-106b are overexpressed in metastatic lung adenocarcinoma. BC Cancer Centre Research Day. Vancouver, BC. 18. Enfield KSS*, Rowbotham DA, Holly A, Dellaire G, Pastrello C, Jurisica I, MacAulay C, Lam S, Lam WL (2015) miR-106a and miR-106b affect metastasis of lung adenocarcinoma via EMT-dependent and EMT-independent pathways. Terry Fox Research Institute 6th Annual Meeting. St. John's, NL. 19. Enfield KSS*, Rowbotham DA, Holly A, Dellaire G, Pastrello C, Jurisica I, MacAulay C, Lam S, Lam WL (2015) miR-106a and miR-106b affect metastasis of lung adenocarcinoma via EMT-dependent and EMT-independent pathways. GSAT/BTB/IOP Retreat. Vancouver, BC. 20. Rowbotham D*, Enfield KSS, Chari R, Becker-Santos D, Lam S, Lam WL (2014) ELF3 is a novel oncogene frequently activated by genetic and epigenetic  166  mechanisms in lung adenocarcinoma. 4th Annual TFRI-BC Node Research Day. Vancouver, BC. 21. Enfield KSS*, Hubaux R, Vucic EA, Jurisica I, Lam S, Lam WL (2014) MicroRNA deregulation associated with node positive lung adenocarcinoma. GSAT/BTB/IOP Retreat. Vancouver, BC. 22. Enfield KSS*, Ramnarine VR, Rowbotham DA, Lam S, Lam WL (2013) MicroRNA deregulation associated with node positive lung adenocarcinoma. 15th World Conference of Lung Cancer. Sydney, Australia. Presentation number P2.02-020. 23. Enfield KSS*, Rowbotham DA, Becker-Santos DD, Chari R, Fuller M, Zhang M, Suzuki M, MacAulay CE, Karsan A, Lam S, Lam WL (2013) ELF3 is a novel oncogene frequently activated by genetic and epigenetic mechanisms in lung adenocarcinoma. 15th World Conference of Lung Cancer. Sydney, Australia. Presentation number MO15.10 – invited talk. 24. Enfield KSS*, Ramnarine VR, Rowbotham DA, Lam S, Lam WL (2013) Deregulation of microRNAs associated with node positive lung adenocarcinoma. BC Cancer Agency Research Day. Vancouver, BC. 25. Enfield KSS*, Ramnarine VR, Rowbotham DA, Lam S, Lam WL (2013) Deregulation of microRNAs associated with node positive lung adenocarcinoma. Bioinformatics and Oncology Research Day. Vancouver, BC. 26. Enfield KSS*, Ramnarine VR, Rowbotham DA, Lam S, Lam WL (2013) MicroRNA deregulation associated metastasis and recurrence of non-small cell lung cancer. AACR-JCA Breakthroughs in Translational Research. Maui, HI. 27. Rowbotham D*, Enfield KSS, Chari R, Becker-Santos D, Lam S, Lam WL (2012) ELF3, a fetal lung transcription factor, is frequently re-activated in lung adenocarcinoma. BCCA Annual Conference. Vancouver, BC. 28. Enfield KSS*, Ramnarine VR, Lam S, Lam WL (2012) MicroRNA deregulation associated with non-small cell lung cancer metastasis. BCCA Annual Conference. Vancouver, BC.  167  29. Enfield KSS*, Lam S, Lam WL (2012) MicroRNA deregulation associated with non-small cell lung cancer metastasis. Cold Spring Harbor: Mechanisms and Models of Cancer. Cold Spring Harbor Labs, NY. 30. Enfield KSS*, Lam S, Lam WL (2012) The paralogous microRNA clusters, miR-17-92 and miR-106-25, are specifically overexpressed in metastatic non-small cell lung carcinomas. AACR Special Conference: Noncoding RNAs and Cancer, Miami, FL. 31. Enfield KSS*, Chari R, Becker-Santos DD, Lam S, Lam WL (2011) ELF3, a fetal lung transcription factor, is frequently re-activated in lung adenocarcinoma. BC Annual Cancer Conference, Vancouver, BC. 32. Enfield KSS*, Chari R, Becker-Santos D, Lam S, Lam WL (2011) ELF3 is a putative oncogene frequently gained and hypomethylated in the adenocarcinoma subtype of non-small cell lung cancer. American Society of Human Genetics Annual Meeting, Montreal, QC. 33. Wilson I*, Vucic EA, Enfield KSS, Chari R, Zhang YA, Radulovich N, Starczynowski D, Banath J, Zhang M, Pusic A, Fuller M, Lonergan K, Buys T, Yee J, Laird-Offringa IA, Liu P, Tsao MS, Anderson M, You M, Bennewith K, Karsan A, MacAulay CE, Lam S, Gazdar AF, Lam WL (2011) EYA4 is a frequently inactivated tumour suppressor gene within the 6q lung cancer susceptibility locus. 14th World Conference on Lung Cancer, Amsterdam, The Netherlands. 34. Enfield KSS*, Wilson IM, Vucic EA, Chari R, Zhang YA, You M, MacAulay CA, Lam SL, Gazdar AF, Lam WL (2010) EYA4 is a putative lung tumor suppressor on 6q23-25 with a role in DNA repair and apoptosis. BC Annual Cancer Conference, Vancouver, BC. 35. Enfield KSS*, Wilson IM, Vucic EA, Chari R, Zhang YA, You M, MacAulay CA, Lam SL, Gazdar AF, Lam WL (2010) Deletion and hypermethylation of a tumor suppressor gene EYA4 in 6q23-25 is associated with lung cancer risk and poor survival. AACR Frontiers in Cancer Prevention, Philadelphia, PA.  168  36. Wilson IM, Enfield KSS*, Vucic EA, Chari R, Zhang YA, You M, MacAulay CA, Lam SL, Gazdar AF, Lam WL (2010) A novel lung tumor suppressor implicated in somatic and familial cancers. AACR 101st Annual Meeting, Washington, D.C. 37. Wilson IM, Enfield KSS*, Vucic EA, Chari R, Zhang YA, You M, MacAulay CA, Lam SL, Gazdar AF, Lam WL (2010) A novel lung tumor suppressor implicated in somatic and familial cancers. Pathology Research Day, Vancouver, BC.  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0348376/manifest

Comment

Related Items